The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman...

103
The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astron autics

Transcript of The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman...

Page 1: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

The Coming Transition inAutomobile Cockpits

Insights from Aerospace

Prof R John HansmanDepartment of Aeronautics amp Astronautics

Evolution of Cockpit Displays

bull Software size doubles every 18 months

bull Compensating for ldquoFBW offsetrdquo reduces doubling to 33 months

Software Growth in Aircraft

Hypothesis

bull We are entering a period of significant change in automobile Human-Machine Interaction driven by Information Technologies

bull Automobiles will undergo a change more substantial than the change in aircraft from ldquosteam gaugerdquo to ldquoglass cockpitsrdquo

Car Aircraft Comparison

bull Market

bull Number of vehicles (US)

bull Safety (US)

bull Threat response time constant

bull Hazard density

bull System complexity

Capital investment (ROI)Consumer product

300000200000000

663 fatalities (1998)41000 fatalities (1997)

Order 5-60 secOrder 1 sec

Low 3-D collision (vehicle terrain animal) WXHigh 2-D collision (vehicle object person animal hellip)

HighMedlow

Car Aircraft Comparison (cont)

bull Operator selectivitytrainingmedical

bull Tracking precision (Heading)

bull Recurrent training

bull Operations procedure

bull Impaired operators (Alcohol Drugs)

HighLow

Order 5degOrder 1deg

YesNo

YesNo

Order 1107 - 109

Order 1104 - 105

Aerospace Systems Applicable to Cars

bull Control systemsndash ABSndash Stability augmentation

bull Fly by WireLight (FBWFBL)ndash Integrity Concerns (eg

777)bull Critical software systemsbull Fault tolerant systemsbull Head up displays (HUD)bull Helmet mounted displays

(HMD)bull Synthetic Vision Systemsbull Sensor Fusionbull Hands on throttle and stic

k (HOTAS)bull Dark cockpit

bull Navigation systemsndash GPS DGPSndash IRSGPS

bull Situation awareness displaysndash Moving mapndash Database

bull Caution and Warning Systemsbull Collision Alerting Systemsbull Tactile alerting

ndash Stick shakerbull Master caution

ndash Information accessibilityndash Maintenance Diagnostics

Track Hardware Layout

bull Two 2 GPS antennas were mounted on the car to form a single baseline

bull Data-Linc Group (SRM6000) Modem antenna also attached to roll barndash Real-time communication with ground station

MIT Run Results

Typical Performance

bull Precise state determination ndash 2 - 5 cm position error ndash 1 - 2 cms velocity error ndash 1 - 2 degrees heading ndash 5 - 10 Hz

Track Results -Slip Measurements

bull Car heading and

velocity vectors

not aligned

Track Results -Slip Measurements

Comparative Lap Results

Acceleration vs Position

ldquoHuman Centeredrdquo InformationRequirements Analysis

bull Integrated Human Centered Systems Approachbull ldquoSemi-Structuredrdquo Decision Theorybull Driver Distraction Analysis

Driving Task Definition

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Vehicle Operation Tasks

bull Vehicle control tasks [skill based]ndash Lateral control (steering)ndash Longitudinal control (accel braking)

bull Tactical decisions [rule based]ndash Maneuveringndash Systems management

bull Strategic decisions [knowledge based]ndash Route selectionndash Goal management

bull Monitoring [skill rule knowledge]ndash Situation awareness

[Rasmussen Skill Rule Knowledge Hierarchy]

Lateral Tracking Loop

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

Driver InputOutput Modes

Visual- External Scene - Forward - Roadway - Traffic - Lights - Signage - Environment - Peripheral - Rear- Internal - Instrument Cluster

TactileProprioceptive- Lateral Accel- Longitudinal Accel- Road Surface- Control Force Feedback- Vibration- Switch Feedback

Audio- Roadway- Alerts - Horns (ext) - ChimesClicks- Velocity (Wind)- Engine- Radio- Passenger

Olfactory- Gasoline- SmokeFire- Passenger

Other

Manual (Hand)- Wheel- Steering Column- Gear Shift- Switch (Panel)- Switch (Other)

Manual (Foot)- Throttle- Brake- Other

Voice

Other- Eye Tracking- Blink- Gesture- Thought

Driver

ldquoSituation Awarenessrdquo

bull Term originally defined for air combatbull Working Definition Sufficiently detailed mental picture of the

vehicle and environment (ie world model) to allow the operator to make well-informed (ie conditionally correct) decisions

Driver Situation AwarenessComponents

PersonalFactors

EquipmentStatus

VehiclePerformance

ControlSettings

Passengers

Non-drivingElements

Internal

SituationAwareness

Weather

AdjacentEnvironment

Signage

RoadwaySurface

LocationRoute

AdjacentTraffic

External

Endsley Situation Awareness Model

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Interaction Metaphors

bull Car as homebull Car as kitchenbull Car as bathroombull Car as bedroombull Car as music roombull Car as playroombull Car as entertainment ctrbull Car as toolboxbull Car as closet

bull Car as officebull Car as comm center

bull Car as image statementbull Car as clothingbull Car as jewelrybull Car as sports equipmentbull Car as safe spacebull Car as cocoon

Trends in Driver Attentional Demand

Driv

er A

tten

tion

al D

emDr

iver

Att

enti

onal

Dem

and

and

VehicleVehicleOperation OperationOperation Operation

SecondarySecondaryTasks Tasks

TimeTime

Time Spent in Vehicles (US)Time Spent in Vehicles (US)- Average 350 hrsyrperson Average 350 hrsyrperson - 500 Million hrsweek- 500 Million hrsweek

76 of Drivers Report Activities HaveldquoCausedNearly Causedrdquo an Accident

TALKING ON CELLPHONE

TURNING TO SPEAK

USING COMPUTER

CIGARETTE ASHES

BREAKING UP FIGHTBETWEEN KIDS

SPILLING COFFEE

Source Opinion Research Corp Interviews Time5800 (N=1016)

Concerns Regarding High SecondaryTask Loads

bull Growing Evidence and Public Perception of Safety Problem

bull Cell Phone use in USndash 115-120 Million Active Cell Phonesndash 50-70 Use in Vehiclesndash 39 of Drivers Using (Daylight Hours)

bull NE Journal of Medicine Estimatendash 4 fold increase in collision risk using cell phone

bull NHTSA Estimatesndash 12 Million Accidents (25-30) caused by Distracted

Driverbull Not limited to cell phones

bull Note These effects may be latent in normal operations and may only manifest in non-normal or emergency situations

Typical Performance vs TaskLoad Curve

Per

form

ance

Task Load

Distraction Components

bull Manualndash Inability or delay in operation of vehicle controlndash Hands occupied or out of position (cell phone)

bull Visualndash Head Down Problemsndash Visual Accommodationndash Visual Clutterndash Visual Compulsion

bull Cognitivendash Lack of cognitive engagement with primary Taskndash Latency in mental context shiftsndash Multi-tasking capabilitiesndash Prioritization

bull Emotionalbull High Individual Variability in Multi-Tasking Capability

Distraction Data Sources

bull Controlled Experimentsndash Vary Secondary Task Load (Independent Variable)

bull Visualbull Cognitive

ndash Measurebull Performancebull Response to Disturbancebull Situation Awareness

ndash Need to Increase Task Loading to Saturationndash Need to Include Unanticipated Eventsndash Need Lowest Common Denominator Population

bull Field Datandash Event Recordersndash Cell Phone Triggered

bull Subjective Survey Data

Controlled Experiment Issues (1)

bull Simulator Testingndash Controlled Scenarios +ndash Safe to go to Task Saturationndash Face and Cue Validity Issues ndashndash Simulator Sickness ndashndash Cost -

bull Dual Control Vehicle Testing (Test Track)ndash Good Validity +ndash Safety Issues at Task Saturation ndashndash Cost +

Controlled Experiment Issues (2)

bull Variability in Primary and Secondary Taskingndash How do you measure Secondary Task Loadndash How do you control Secondary Task Load

bull Performance Measuresndash Trackingndash Reaction Timendash Side Task Performance

bull Subjective Workload Measuresbull Situation Awareness Measures

ndash Testable Response Method

Simulator Studies of Driver CognitiveDistraction Caused by Cell Phone Use

bull Independent Variablesndash Cognitive Loadingndash Hands freeHands Fixed

bull Dependant Variablesndash Situation Awarenessndash Reaction Timendash Tracking MIT Age Lab Simulator

Cognitive Loading Levels

bull Lowndash Small talk

bull bullMediumndash Discuss movieoperaballet plot-line

bull bullHighndash Edit document by phone

Results Situational Awareness

bull Significant reduction in SA with Cell Phone Use

bull No significant difference between HFamp HH

Ave

rag

e F

ract

ion

Co

rrec

t

Normal HF HHPhone Setup

Stop Sign Response Time Resultsby Age Group

bull Significant effect of Age

bull No significant effect of Cell Phone Use

Re

sp

on

se

Tim

e [

se

c]

Normal1

Normal2

HFSmall

HFMovie

HF Edit HHSmall

HHMovie

HH Edit

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 2: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Evolution of Cockpit Displays

bull Software size doubles every 18 months

bull Compensating for ldquoFBW offsetrdquo reduces doubling to 33 months

Software Growth in Aircraft

Hypothesis

bull We are entering a period of significant change in automobile Human-Machine Interaction driven by Information Technologies

bull Automobiles will undergo a change more substantial than the change in aircraft from ldquosteam gaugerdquo to ldquoglass cockpitsrdquo

Car Aircraft Comparison

bull Market

bull Number of vehicles (US)

bull Safety (US)

bull Threat response time constant

bull Hazard density

bull System complexity

Capital investment (ROI)Consumer product

300000200000000

663 fatalities (1998)41000 fatalities (1997)

Order 5-60 secOrder 1 sec

Low 3-D collision (vehicle terrain animal) WXHigh 2-D collision (vehicle object person animal hellip)

HighMedlow

Car Aircraft Comparison (cont)

bull Operator selectivitytrainingmedical

bull Tracking precision (Heading)

bull Recurrent training

bull Operations procedure

bull Impaired operators (Alcohol Drugs)

HighLow

Order 5degOrder 1deg

YesNo

YesNo

Order 1107 - 109

Order 1104 - 105

Aerospace Systems Applicable to Cars

bull Control systemsndash ABSndash Stability augmentation

bull Fly by WireLight (FBWFBL)ndash Integrity Concerns (eg

777)bull Critical software systemsbull Fault tolerant systemsbull Head up displays (HUD)bull Helmet mounted displays

(HMD)bull Synthetic Vision Systemsbull Sensor Fusionbull Hands on throttle and stic

k (HOTAS)bull Dark cockpit

bull Navigation systemsndash GPS DGPSndash IRSGPS

bull Situation awareness displaysndash Moving mapndash Database

bull Caution and Warning Systemsbull Collision Alerting Systemsbull Tactile alerting

ndash Stick shakerbull Master caution

ndash Information accessibilityndash Maintenance Diagnostics

Track Hardware Layout

bull Two 2 GPS antennas were mounted on the car to form a single baseline

bull Data-Linc Group (SRM6000) Modem antenna also attached to roll barndash Real-time communication with ground station

MIT Run Results

Typical Performance

bull Precise state determination ndash 2 - 5 cm position error ndash 1 - 2 cms velocity error ndash 1 - 2 degrees heading ndash 5 - 10 Hz

Track Results -Slip Measurements

bull Car heading and

velocity vectors

not aligned

Track Results -Slip Measurements

Comparative Lap Results

Acceleration vs Position

ldquoHuman Centeredrdquo InformationRequirements Analysis

bull Integrated Human Centered Systems Approachbull ldquoSemi-Structuredrdquo Decision Theorybull Driver Distraction Analysis

Driving Task Definition

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Vehicle Operation Tasks

bull Vehicle control tasks [skill based]ndash Lateral control (steering)ndash Longitudinal control (accel braking)

bull Tactical decisions [rule based]ndash Maneuveringndash Systems management

bull Strategic decisions [knowledge based]ndash Route selectionndash Goal management

bull Monitoring [skill rule knowledge]ndash Situation awareness

[Rasmussen Skill Rule Knowledge Hierarchy]

Lateral Tracking Loop

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

Driver InputOutput Modes

Visual- External Scene - Forward - Roadway - Traffic - Lights - Signage - Environment - Peripheral - Rear- Internal - Instrument Cluster

TactileProprioceptive- Lateral Accel- Longitudinal Accel- Road Surface- Control Force Feedback- Vibration- Switch Feedback

Audio- Roadway- Alerts - Horns (ext) - ChimesClicks- Velocity (Wind)- Engine- Radio- Passenger

Olfactory- Gasoline- SmokeFire- Passenger

Other

Manual (Hand)- Wheel- Steering Column- Gear Shift- Switch (Panel)- Switch (Other)

Manual (Foot)- Throttle- Brake- Other

Voice

Other- Eye Tracking- Blink- Gesture- Thought

Driver

ldquoSituation Awarenessrdquo

bull Term originally defined for air combatbull Working Definition Sufficiently detailed mental picture of the

vehicle and environment (ie world model) to allow the operator to make well-informed (ie conditionally correct) decisions

Driver Situation AwarenessComponents

PersonalFactors

EquipmentStatus

VehiclePerformance

ControlSettings

Passengers

Non-drivingElements

Internal

SituationAwareness

Weather

AdjacentEnvironment

Signage

RoadwaySurface

LocationRoute

AdjacentTraffic

External

Endsley Situation Awareness Model

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Interaction Metaphors

bull Car as homebull Car as kitchenbull Car as bathroombull Car as bedroombull Car as music roombull Car as playroombull Car as entertainment ctrbull Car as toolboxbull Car as closet

bull Car as officebull Car as comm center

bull Car as image statementbull Car as clothingbull Car as jewelrybull Car as sports equipmentbull Car as safe spacebull Car as cocoon

Trends in Driver Attentional Demand

Driv

er A

tten

tion

al D

emDr

iver

Att

enti

onal

Dem

and

and

VehicleVehicleOperation OperationOperation Operation

SecondarySecondaryTasks Tasks

TimeTime

Time Spent in Vehicles (US)Time Spent in Vehicles (US)- Average 350 hrsyrperson Average 350 hrsyrperson - 500 Million hrsweek- 500 Million hrsweek

76 of Drivers Report Activities HaveldquoCausedNearly Causedrdquo an Accident

TALKING ON CELLPHONE

TURNING TO SPEAK

USING COMPUTER

CIGARETTE ASHES

BREAKING UP FIGHTBETWEEN KIDS

SPILLING COFFEE

Source Opinion Research Corp Interviews Time5800 (N=1016)

Concerns Regarding High SecondaryTask Loads

bull Growing Evidence and Public Perception of Safety Problem

bull Cell Phone use in USndash 115-120 Million Active Cell Phonesndash 50-70 Use in Vehiclesndash 39 of Drivers Using (Daylight Hours)

bull NE Journal of Medicine Estimatendash 4 fold increase in collision risk using cell phone

bull NHTSA Estimatesndash 12 Million Accidents (25-30) caused by Distracted

Driverbull Not limited to cell phones

bull Note These effects may be latent in normal operations and may only manifest in non-normal or emergency situations

Typical Performance vs TaskLoad Curve

Per

form

ance

Task Load

Distraction Components

bull Manualndash Inability or delay in operation of vehicle controlndash Hands occupied or out of position (cell phone)

bull Visualndash Head Down Problemsndash Visual Accommodationndash Visual Clutterndash Visual Compulsion

bull Cognitivendash Lack of cognitive engagement with primary Taskndash Latency in mental context shiftsndash Multi-tasking capabilitiesndash Prioritization

bull Emotionalbull High Individual Variability in Multi-Tasking Capability

Distraction Data Sources

bull Controlled Experimentsndash Vary Secondary Task Load (Independent Variable)

bull Visualbull Cognitive

ndash Measurebull Performancebull Response to Disturbancebull Situation Awareness

ndash Need to Increase Task Loading to Saturationndash Need to Include Unanticipated Eventsndash Need Lowest Common Denominator Population

bull Field Datandash Event Recordersndash Cell Phone Triggered

bull Subjective Survey Data

Controlled Experiment Issues (1)

bull Simulator Testingndash Controlled Scenarios +ndash Safe to go to Task Saturationndash Face and Cue Validity Issues ndashndash Simulator Sickness ndashndash Cost -

bull Dual Control Vehicle Testing (Test Track)ndash Good Validity +ndash Safety Issues at Task Saturation ndashndash Cost +

Controlled Experiment Issues (2)

bull Variability in Primary and Secondary Taskingndash How do you measure Secondary Task Loadndash How do you control Secondary Task Load

bull Performance Measuresndash Trackingndash Reaction Timendash Side Task Performance

bull Subjective Workload Measuresbull Situation Awareness Measures

ndash Testable Response Method

Simulator Studies of Driver CognitiveDistraction Caused by Cell Phone Use

bull Independent Variablesndash Cognitive Loadingndash Hands freeHands Fixed

bull Dependant Variablesndash Situation Awarenessndash Reaction Timendash Tracking MIT Age Lab Simulator

Cognitive Loading Levels

bull Lowndash Small talk

bull bullMediumndash Discuss movieoperaballet plot-line

bull bullHighndash Edit document by phone

Results Situational Awareness

bull Significant reduction in SA with Cell Phone Use

bull No significant difference between HFamp HH

Ave

rag

e F

ract

ion

Co

rrec

t

Normal HF HHPhone Setup

Stop Sign Response Time Resultsby Age Group

bull Significant effect of Age

bull No significant effect of Cell Phone Use

Re

sp

on

se

Tim

e [

se

c]

Normal1

Normal2

HFSmall

HFMovie

HF Edit HHSmall

HHMovie

HH Edit

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 3: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

bull Software size doubles every 18 months

bull Compensating for ldquoFBW offsetrdquo reduces doubling to 33 months

Software Growth in Aircraft

Hypothesis

bull We are entering a period of significant change in automobile Human-Machine Interaction driven by Information Technologies

bull Automobiles will undergo a change more substantial than the change in aircraft from ldquosteam gaugerdquo to ldquoglass cockpitsrdquo

Car Aircraft Comparison

bull Market

bull Number of vehicles (US)

bull Safety (US)

bull Threat response time constant

bull Hazard density

bull System complexity

Capital investment (ROI)Consumer product

300000200000000

663 fatalities (1998)41000 fatalities (1997)

Order 5-60 secOrder 1 sec

Low 3-D collision (vehicle terrain animal) WXHigh 2-D collision (vehicle object person animal hellip)

HighMedlow

Car Aircraft Comparison (cont)

bull Operator selectivitytrainingmedical

bull Tracking precision (Heading)

bull Recurrent training

bull Operations procedure

bull Impaired operators (Alcohol Drugs)

HighLow

Order 5degOrder 1deg

YesNo

YesNo

Order 1107 - 109

Order 1104 - 105

Aerospace Systems Applicable to Cars

bull Control systemsndash ABSndash Stability augmentation

bull Fly by WireLight (FBWFBL)ndash Integrity Concerns (eg

777)bull Critical software systemsbull Fault tolerant systemsbull Head up displays (HUD)bull Helmet mounted displays

(HMD)bull Synthetic Vision Systemsbull Sensor Fusionbull Hands on throttle and stic

k (HOTAS)bull Dark cockpit

bull Navigation systemsndash GPS DGPSndash IRSGPS

bull Situation awareness displaysndash Moving mapndash Database

bull Caution and Warning Systemsbull Collision Alerting Systemsbull Tactile alerting

ndash Stick shakerbull Master caution

ndash Information accessibilityndash Maintenance Diagnostics

Track Hardware Layout

bull Two 2 GPS antennas were mounted on the car to form a single baseline

bull Data-Linc Group (SRM6000) Modem antenna also attached to roll barndash Real-time communication with ground station

MIT Run Results

Typical Performance

bull Precise state determination ndash 2 - 5 cm position error ndash 1 - 2 cms velocity error ndash 1 - 2 degrees heading ndash 5 - 10 Hz

Track Results -Slip Measurements

bull Car heading and

velocity vectors

not aligned

Track Results -Slip Measurements

Comparative Lap Results

Acceleration vs Position

ldquoHuman Centeredrdquo InformationRequirements Analysis

bull Integrated Human Centered Systems Approachbull ldquoSemi-Structuredrdquo Decision Theorybull Driver Distraction Analysis

Driving Task Definition

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Vehicle Operation Tasks

bull Vehicle control tasks [skill based]ndash Lateral control (steering)ndash Longitudinal control (accel braking)

bull Tactical decisions [rule based]ndash Maneuveringndash Systems management

bull Strategic decisions [knowledge based]ndash Route selectionndash Goal management

bull Monitoring [skill rule knowledge]ndash Situation awareness

[Rasmussen Skill Rule Knowledge Hierarchy]

Lateral Tracking Loop

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

Driver InputOutput Modes

Visual- External Scene - Forward - Roadway - Traffic - Lights - Signage - Environment - Peripheral - Rear- Internal - Instrument Cluster

TactileProprioceptive- Lateral Accel- Longitudinal Accel- Road Surface- Control Force Feedback- Vibration- Switch Feedback

Audio- Roadway- Alerts - Horns (ext) - ChimesClicks- Velocity (Wind)- Engine- Radio- Passenger

Olfactory- Gasoline- SmokeFire- Passenger

Other

Manual (Hand)- Wheel- Steering Column- Gear Shift- Switch (Panel)- Switch (Other)

Manual (Foot)- Throttle- Brake- Other

Voice

Other- Eye Tracking- Blink- Gesture- Thought

Driver

ldquoSituation Awarenessrdquo

bull Term originally defined for air combatbull Working Definition Sufficiently detailed mental picture of the

vehicle and environment (ie world model) to allow the operator to make well-informed (ie conditionally correct) decisions

Driver Situation AwarenessComponents

PersonalFactors

EquipmentStatus

VehiclePerformance

ControlSettings

Passengers

Non-drivingElements

Internal

SituationAwareness

Weather

AdjacentEnvironment

Signage

RoadwaySurface

LocationRoute

AdjacentTraffic

External

Endsley Situation Awareness Model

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Interaction Metaphors

bull Car as homebull Car as kitchenbull Car as bathroombull Car as bedroombull Car as music roombull Car as playroombull Car as entertainment ctrbull Car as toolboxbull Car as closet

bull Car as officebull Car as comm center

bull Car as image statementbull Car as clothingbull Car as jewelrybull Car as sports equipmentbull Car as safe spacebull Car as cocoon

Trends in Driver Attentional Demand

Driv

er A

tten

tion

al D

emDr

iver

Att

enti

onal

Dem

and

and

VehicleVehicleOperation OperationOperation Operation

SecondarySecondaryTasks Tasks

TimeTime

Time Spent in Vehicles (US)Time Spent in Vehicles (US)- Average 350 hrsyrperson Average 350 hrsyrperson - 500 Million hrsweek- 500 Million hrsweek

76 of Drivers Report Activities HaveldquoCausedNearly Causedrdquo an Accident

TALKING ON CELLPHONE

TURNING TO SPEAK

USING COMPUTER

CIGARETTE ASHES

BREAKING UP FIGHTBETWEEN KIDS

SPILLING COFFEE

Source Opinion Research Corp Interviews Time5800 (N=1016)

Concerns Regarding High SecondaryTask Loads

bull Growing Evidence and Public Perception of Safety Problem

bull Cell Phone use in USndash 115-120 Million Active Cell Phonesndash 50-70 Use in Vehiclesndash 39 of Drivers Using (Daylight Hours)

bull NE Journal of Medicine Estimatendash 4 fold increase in collision risk using cell phone

bull NHTSA Estimatesndash 12 Million Accidents (25-30) caused by Distracted

Driverbull Not limited to cell phones

bull Note These effects may be latent in normal operations and may only manifest in non-normal or emergency situations

Typical Performance vs TaskLoad Curve

Per

form

ance

Task Load

Distraction Components

bull Manualndash Inability or delay in operation of vehicle controlndash Hands occupied or out of position (cell phone)

bull Visualndash Head Down Problemsndash Visual Accommodationndash Visual Clutterndash Visual Compulsion

bull Cognitivendash Lack of cognitive engagement with primary Taskndash Latency in mental context shiftsndash Multi-tasking capabilitiesndash Prioritization

bull Emotionalbull High Individual Variability in Multi-Tasking Capability

Distraction Data Sources

bull Controlled Experimentsndash Vary Secondary Task Load (Independent Variable)

bull Visualbull Cognitive

ndash Measurebull Performancebull Response to Disturbancebull Situation Awareness

ndash Need to Increase Task Loading to Saturationndash Need to Include Unanticipated Eventsndash Need Lowest Common Denominator Population

bull Field Datandash Event Recordersndash Cell Phone Triggered

bull Subjective Survey Data

Controlled Experiment Issues (1)

bull Simulator Testingndash Controlled Scenarios +ndash Safe to go to Task Saturationndash Face and Cue Validity Issues ndashndash Simulator Sickness ndashndash Cost -

bull Dual Control Vehicle Testing (Test Track)ndash Good Validity +ndash Safety Issues at Task Saturation ndashndash Cost +

Controlled Experiment Issues (2)

bull Variability in Primary and Secondary Taskingndash How do you measure Secondary Task Loadndash How do you control Secondary Task Load

bull Performance Measuresndash Trackingndash Reaction Timendash Side Task Performance

bull Subjective Workload Measuresbull Situation Awareness Measures

ndash Testable Response Method

Simulator Studies of Driver CognitiveDistraction Caused by Cell Phone Use

bull Independent Variablesndash Cognitive Loadingndash Hands freeHands Fixed

bull Dependant Variablesndash Situation Awarenessndash Reaction Timendash Tracking MIT Age Lab Simulator

Cognitive Loading Levels

bull Lowndash Small talk

bull bullMediumndash Discuss movieoperaballet plot-line

bull bullHighndash Edit document by phone

Results Situational Awareness

bull Significant reduction in SA with Cell Phone Use

bull No significant difference between HFamp HH

Ave

rag

e F

ract

ion

Co

rrec

t

Normal HF HHPhone Setup

Stop Sign Response Time Resultsby Age Group

bull Significant effect of Age

bull No significant effect of Cell Phone Use

Re

sp

on

se

Tim

e [

se

c]

Normal1

Normal2

HFSmall

HFMovie

HF Edit HHSmall

HHMovie

HH Edit

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 4: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Hypothesis

bull We are entering a period of significant change in automobile Human-Machine Interaction driven by Information Technologies

bull Automobiles will undergo a change more substantial than the change in aircraft from ldquosteam gaugerdquo to ldquoglass cockpitsrdquo

Car Aircraft Comparison

bull Market

bull Number of vehicles (US)

bull Safety (US)

bull Threat response time constant

bull Hazard density

bull System complexity

Capital investment (ROI)Consumer product

300000200000000

663 fatalities (1998)41000 fatalities (1997)

Order 5-60 secOrder 1 sec

Low 3-D collision (vehicle terrain animal) WXHigh 2-D collision (vehicle object person animal hellip)

HighMedlow

Car Aircraft Comparison (cont)

bull Operator selectivitytrainingmedical

bull Tracking precision (Heading)

bull Recurrent training

bull Operations procedure

bull Impaired operators (Alcohol Drugs)

HighLow

Order 5degOrder 1deg

YesNo

YesNo

Order 1107 - 109

Order 1104 - 105

Aerospace Systems Applicable to Cars

bull Control systemsndash ABSndash Stability augmentation

bull Fly by WireLight (FBWFBL)ndash Integrity Concerns (eg

777)bull Critical software systemsbull Fault tolerant systemsbull Head up displays (HUD)bull Helmet mounted displays

(HMD)bull Synthetic Vision Systemsbull Sensor Fusionbull Hands on throttle and stic

k (HOTAS)bull Dark cockpit

bull Navigation systemsndash GPS DGPSndash IRSGPS

bull Situation awareness displaysndash Moving mapndash Database

bull Caution and Warning Systemsbull Collision Alerting Systemsbull Tactile alerting

ndash Stick shakerbull Master caution

ndash Information accessibilityndash Maintenance Diagnostics

Track Hardware Layout

bull Two 2 GPS antennas were mounted on the car to form a single baseline

bull Data-Linc Group (SRM6000) Modem antenna also attached to roll barndash Real-time communication with ground station

MIT Run Results

Typical Performance

bull Precise state determination ndash 2 - 5 cm position error ndash 1 - 2 cms velocity error ndash 1 - 2 degrees heading ndash 5 - 10 Hz

Track Results -Slip Measurements

bull Car heading and

velocity vectors

not aligned

Track Results -Slip Measurements

Comparative Lap Results

Acceleration vs Position

ldquoHuman Centeredrdquo InformationRequirements Analysis

bull Integrated Human Centered Systems Approachbull ldquoSemi-Structuredrdquo Decision Theorybull Driver Distraction Analysis

Driving Task Definition

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Vehicle Operation Tasks

bull Vehicle control tasks [skill based]ndash Lateral control (steering)ndash Longitudinal control (accel braking)

bull Tactical decisions [rule based]ndash Maneuveringndash Systems management

bull Strategic decisions [knowledge based]ndash Route selectionndash Goal management

bull Monitoring [skill rule knowledge]ndash Situation awareness

[Rasmussen Skill Rule Knowledge Hierarchy]

Lateral Tracking Loop

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

Driver InputOutput Modes

Visual- External Scene - Forward - Roadway - Traffic - Lights - Signage - Environment - Peripheral - Rear- Internal - Instrument Cluster

TactileProprioceptive- Lateral Accel- Longitudinal Accel- Road Surface- Control Force Feedback- Vibration- Switch Feedback

Audio- Roadway- Alerts - Horns (ext) - ChimesClicks- Velocity (Wind)- Engine- Radio- Passenger

Olfactory- Gasoline- SmokeFire- Passenger

Other

Manual (Hand)- Wheel- Steering Column- Gear Shift- Switch (Panel)- Switch (Other)

Manual (Foot)- Throttle- Brake- Other

Voice

Other- Eye Tracking- Blink- Gesture- Thought

Driver

ldquoSituation Awarenessrdquo

bull Term originally defined for air combatbull Working Definition Sufficiently detailed mental picture of the

vehicle and environment (ie world model) to allow the operator to make well-informed (ie conditionally correct) decisions

Driver Situation AwarenessComponents

PersonalFactors

EquipmentStatus

VehiclePerformance

ControlSettings

Passengers

Non-drivingElements

Internal

SituationAwareness

Weather

AdjacentEnvironment

Signage

RoadwaySurface

LocationRoute

AdjacentTraffic

External

Endsley Situation Awareness Model

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Interaction Metaphors

bull Car as homebull Car as kitchenbull Car as bathroombull Car as bedroombull Car as music roombull Car as playroombull Car as entertainment ctrbull Car as toolboxbull Car as closet

bull Car as officebull Car as comm center

bull Car as image statementbull Car as clothingbull Car as jewelrybull Car as sports equipmentbull Car as safe spacebull Car as cocoon

Trends in Driver Attentional Demand

Driv

er A

tten

tion

al D

emDr

iver

Att

enti

onal

Dem

and

and

VehicleVehicleOperation OperationOperation Operation

SecondarySecondaryTasks Tasks

TimeTime

Time Spent in Vehicles (US)Time Spent in Vehicles (US)- Average 350 hrsyrperson Average 350 hrsyrperson - 500 Million hrsweek- 500 Million hrsweek

76 of Drivers Report Activities HaveldquoCausedNearly Causedrdquo an Accident

TALKING ON CELLPHONE

TURNING TO SPEAK

USING COMPUTER

CIGARETTE ASHES

BREAKING UP FIGHTBETWEEN KIDS

SPILLING COFFEE

Source Opinion Research Corp Interviews Time5800 (N=1016)

Concerns Regarding High SecondaryTask Loads

bull Growing Evidence and Public Perception of Safety Problem

bull Cell Phone use in USndash 115-120 Million Active Cell Phonesndash 50-70 Use in Vehiclesndash 39 of Drivers Using (Daylight Hours)

bull NE Journal of Medicine Estimatendash 4 fold increase in collision risk using cell phone

bull NHTSA Estimatesndash 12 Million Accidents (25-30) caused by Distracted

Driverbull Not limited to cell phones

bull Note These effects may be latent in normal operations and may only manifest in non-normal or emergency situations

Typical Performance vs TaskLoad Curve

Per

form

ance

Task Load

Distraction Components

bull Manualndash Inability or delay in operation of vehicle controlndash Hands occupied or out of position (cell phone)

bull Visualndash Head Down Problemsndash Visual Accommodationndash Visual Clutterndash Visual Compulsion

bull Cognitivendash Lack of cognitive engagement with primary Taskndash Latency in mental context shiftsndash Multi-tasking capabilitiesndash Prioritization

bull Emotionalbull High Individual Variability in Multi-Tasking Capability

Distraction Data Sources

bull Controlled Experimentsndash Vary Secondary Task Load (Independent Variable)

bull Visualbull Cognitive

ndash Measurebull Performancebull Response to Disturbancebull Situation Awareness

ndash Need to Increase Task Loading to Saturationndash Need to Include Unanticipated Eventsndash Need Lowest Common Denominator Population

bull Field Datandash Event Recordersndash Cell Phone Triggered

bull Subjective Survey Data

Controlled Experiment Issues (1)

bull Simulator Testingndash Controlled Scenarios +ndash Safe to go to Task Saturationndash Face and Cue Validity Issues ndashndash Simulator Sickness ndashndash Cost -

bull Dual Control Vehicle Testing (Test Track)ndash Good Validity +ndash Safety Issues at Task Saturation ndashndash Cost +

Controlled Experiment Issues (2)

bull Variability in Primary and Secondary Taskingndash How do you measure Secondary Task Loadndash How do you control Secondary Task Load

bull Performance Measuresndash Trackingndash Reaction Timendash Side Task Performance

bull Subjective Workload Measuresbull Situation Awareness Measures

ndash Testable Response Method

Simulator Studies of Driver CognitiveDistraction Caused by Cell Phone Use

bull Independent Variablesndash Cognitive Loadingndash Hands freeHands Fixed

bull Dependant Variablesndash Situation Awarenessndash Reaction Timendash Tracking MIT Age Lab Simulator

Cognitive Loading Levels

bull Lowndash Small talk

bull bullMediumndash Discuss movieoperaballet plot-line

bull bullHighndash Edit document by phone

Results Situational Awareness

bull Significant reduction in SA with Cell Phone Use

bull No significant difference between HFamp HH

Ave

rag

e F

ract

ion

Co

rrec

t

Normal HF HHPhone Setup

Stop Sign Response Time Resultsby Age Group

bull Significant effect of Age

bull No significant effect of Cell Phone Use

Re

sp

on

se

Tim

e [

se

c]

Normal1

Normal2

HFSmall

HFMovie

HF Edit HHSmall

HHMovie

HH Edit

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 5: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Car Aircraft Comparison

bull Market

bull Number of vehicles (US)

bull Safety (US)

bull Threat response time constant

bull Hazard density

bull System complexity

Capital investment (ROI)Consumer product

300000200000000

663 fatalities (1998)41000 fatalities (1997)

Order 5-60 secOrder 1 sec

Low 3-D collision (vehicle terrain animal) WXHigh 2-D collision (vehicle object person animal hellip)

HighMedlow

Car Aircraft Comparison (cont)

bull Operator selectivitytrainingmedical

bull Tracking precision (Heading)

bull Recurrent training

bull Operations procedure

bull Impaired operators (Alcohol Drugs)

HighLow

Order 5degOrder 1deg

YesNo

YesNo

Order 1107 - 109

Order 1104 - 105

Aerospace Systems Applicable to Cars

bull Control systemsndash ABSndash Stability augmentation

bull Fly by WireLight (FBWFBL)ndash Integrity Concerns (eg

777)bull Critical software systemsbull Fault tolerant systemsbull Head up displays (HUD)bull Helmet mounted displays

(HMD)bull Synthetic Vision Systemsbull Sensor Fusionbull Hands on throttle and stic

k (HOTAS)bull Dark cockpit

bull Navigation systemsndash GPS DGPSndash IRSGPS

bull Situation awareness displaysndash Moving mapndash Database

bull Caution and Warning Systemsbull Collision Alerting Systemsbull Tactile alerting

ndash Stick shakerbull Master caution

ndash Information accessibilityndash Maintenance Diagnostics

Track Hardware Layout

bull Two 2 GPS antennas were mounted on the car to form a single baseline

bull Data-Linc Group (SRM6000) Modem antenna also attached to roll barndash Real-time communication with ground station

MIT Run Results

Typical Performance

bull Precise state determination ndash 2 - 5 cm position error ndash 1 - 2 cms velocity error ndash 1 - 2 degrees heading ndash 5 - 10 Hz

Track Results -Slip Measurements

bull Car heading and

velocity vectors

not aligned

Track Results -Slip Measurements

Comparative Lap Results

Acceleration vs Position

ldquoHuman Centeredrdquo InformationRequirements Analysis

bull Integrated Human Centered Systems Approachbull ldquoSemi-Structuredrdquo Decision Theorybull Driver Distraction Analysis

Driving Task Definition

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Vehicle Operation Tasks

bull Vehicle control tasks [skill based]ndash Lateral control (steering)ndash Longitudinal control (accel braking)

bull Tactical decisions [rule based]ndash Maneuveringndash Systems management

bull Strategic decisions [knowledge based]ndash Route selectionndash Goal management

bull Monitoring [skill rule knowledge]ndash Situation awareness

[Rasmussen Skill Rule Knowledge Hierarchy]

Lateral Tracking Loop

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

Driver InputOutput Modes

Visual- External Scene - Forward - Roadway - Traffic - Lights - Signage - Environment - Peripheral - Rear- Internal - Instrument Cluster

TactileProprioceptive- Lateral Accel- Longitudinal Accel- Road Surface- Control Force Feedback- Vibration- Switch Feedback

Audio- Roadway- Alerts - Horns (ext) - ChimesClicks- Velocity (Wind)- Engine- Radio- Passenger

Olfactory- Gasoline- SmokeFire- Passenger

Other

Manual (Hand)- Wheel- Steering Column- Gear Shift- Switch (Panel)- Switch (Other)

Manual (Foot)- Throttle- Brake- Other

Voice

Other- Eye Tracking- Blink- Gesture- Thought

Driver

ldquoSituation Awarenessrdquo

bull Term originally defined for air combatbull Working Definition Sufficiently detailed mental picture of the

vehicle and environment (ie world model) to allow the operator to make well-informed (ie conditionally correct) decisions

Driver Situation AwarenessComponents

PersonalFactors

EquipmentStatus

VehiclePerformance

ControlSettings

Passengers

Non-drivingElements

Internal

SituationAwareness

Weather

AdjacentEnvironment

Signage

RoadwaySurface

LocationRoute

AdjacentTraffic

External

Endsley Situation Awareness Model

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Interaction Metaphors

bull Car as homebull Car as kitchenbull Car as bathroombull Car as bedroombull Car as music roombull Car as playroombull Car as entertainment ctrbull Car as toolboxbull Car as closet

bull Car as officebull Car as comm center

bull Car as image statementbull Car as clothingbull Car as jewelrybull Car as sports equipmentbull Car as safe spacebull Car as cocoon

Trends in Driver Attentional Demand

Driv

er A

tten

tion

al D

emDr

iver

Att

enti

onal

Dem

and

and

VehicleVehicleOperation OperationOperation Operation

SecondarySecondaryTasks Tasks

TimeTime

Time Spent in Vehicles (US)Time Spent in Vehicles (US)- Average 350 hrsyrperson Average 350 hrsyrperson - 500 Million hrsweek- 500 Million hrsweek

76 of Drivers Report Activities HaveldquoCausedNearly Causedrdquo an Accident

TALKING ON CELLPHONE

TURNING TO SPEAK

USING COMPUTER

CIGARETTE ASHES

BREAKING UP FIGHTBETWEEN KIDS

SPILLING COFFEE

Source Opinion Research Corp Interviews Time5800 (N=1016)

Concerns Regarding High SecondaryTask Loads

bull Growing Evidence and Public Perception of Safety Problem

bull Cell Phone use in USndash 115-120 Million Active Cell Phonesndash 50-70 Use in Vehiclesndash 39 of Drivers Using (Daylight Hours)

bull NE Journal of Medicine Estimatendash 4 fold increase in collision risk using cell phone

bull NHTSA Estimatesndash 12 Million Accidents (25-30) caused by Distracted

Driverbull Not limited to cell phones

bull Note These effects may be latent in normal operations and may only manifest in non-normal or emergency situations

Typical Performance vs TaskLoad Curve

Per

form

ance

Task Load

Distraction Components

bull Manualndash Inability or delay in operation of vehicle controlndash Hands occupied or out of position (cell phone)

bull Visualndash Head Down Problemsndash Visual Accommodationndash Visual Clutterndash Visual Compulsion

bull Cognitivendash Lack of cognitive engagement with primary Taskndash Latency in mental context shiftsndash Multi-tasking capabilitiesndash Prioritization

bull Emotionalbull High Individual Variability in Multi-Tasking Capability

Distraction Data Sources

bull Controlled Experimentsndash Vary Secondary Task Load (Independent Variable)

bull Visualbull Cognitive

ndash Measurebull Performancebull Response to Disturbancebull Situation Awareness

ndash Need to Increase Task Loading to Saturationndash Need to Include Unanticipated Eventsndash Need Lowest Common Denominator Population

bull Field Datandash Event Recordersndash Cell Phone Triggered

bull Subjective Survey Data

Controlled Experiment Issues (1)

bull Simulator Testingndash Controlled Scenarios +ndash Safe to go to Task Saturationndash Face and Cue Validity Issues ndashndash Simulator Sickness ndashndash Cost -

bull Dual Control Vehicle Testing (Test Track)ndash Good Validity +ndash Safety Issues at Task Saturation ndashndash Cost +

Controlled Experiment Issues (2)

bull Variability in Primary and Secondary Taskingndash How do you measure Secondary Task Loadndash How do you control Secondary Task Load

bull Performance Measuresndash Trackingndash Reaction Timendash Side Task Performance

bull Subjective Workload Measuresbull Situation Awareness Measures

ndash Testable Response Method

Simulator Studies of Driver CognitiveDistraction Caused by Cell Phone Use

bull Independent Variablesndash Cognitive Loadingndash Hands freeHands Fixed

bull Dependant Variablesndash Situation Awarenessndash Reaction Timendash Tracking MIT Age Lab Simulator

Cognitive Loading Levels

bull Lowndash Small talk

bull bullMediumndash Discuss movieoperaballet plot-line

bull bullHighndash Edit document by phone

Results Situational Awareness

bull Significant reduction in SA with Cell Phone Use

bull No significant difference between HFamp HH

Ave

rag

e F

ract

ion

Co

rrec

t

Normal HF HHPhone Setup

Stop Sign Response Time Resultsby Age Group

bull Significant effect of Age

bull No significant effect of Cell Phone Use

Re

sp

on

se

Tim

e [

se

c]

Normal1

Normal2

HFSmall

HFMovie

HF Edit HHSmall

HHMovie

HH Edit

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 6: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Car Aircraft Comparison (cont)

bull Operator selectivitytrainingmedical

bull Tracking precision (Heading)

bull Recurrent training

bull Operations procedure

bull Impaired operators (Alcohol Drugs)

HighLow

Order 5degOrder 1deg

YesNo

YesNo

Order 1107 - 109

Order 1104 - 105

Aerospace Systems Applicable to Cars

bull Control systemsndash ABSndash Stability augmentation

bull Fly by WireLight (FBWFBL)ndash Integrity Concerns (eg

777)bull Critical software systemsbull Fault tolerant systemsbull Head up displays (HUD)bull Helmet mounted displays

(HMD)bull Synthetic Vision Systemsbull Sensor Fusionbull Hands on throttle and stic

k (HOTAS)bull Dark cockpit

bull Navigation systemsndash GPS DGPSndash IRSGPS

bull Situation awareness displaysndash Moving mapndash Database

bull Caution and Warning Systemsbull Collision Alerting Systemsbull Tactile alerting

ndash Stick shakerbull Master caution

ndash Information accessibilityndash Maintenance Diagnostics

Track Hardware Layout

bull Two 2 GPS antennas were mounted on the car to form a single baseline

bull Data-Linc Group (SRM6000) Modem antenna also attached to roll barndash Real-time communication with ground station

MIT Run Results

Typical Performance

bull Precise state determination ndash 2 - 5 cm position error ndash 1 - 2 cms velocity error ndash 1 - 2 degrees heading ndash 5 - 10 Hz

Track Results -Slip Measurements

bull Car heading and

velocity vectors

not aligned

Track Results -Slip Measurements

Comparative Lap Results

Acceleration vs Position

ldquoHuman Centeredrdquo InformationRequirements Analysis

bull Integrated Human Centered Systems Approachbull ldquoSemi-Structuredrdquo Decision Theorybull Driver Distraction Analysis

Driving Task Definition

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Vehicle Operation Tasks

bull Vehicle control tasks [skill based]ndash Lateral control (steering)ndash Longitudinal control (accel braking)

bull Tactical decisions [rule based]ndash Maneuveringndash Systems management

bull Strategic decisions [knowledge based]ndash Route selectionndash Goal management

bull Monitoring [skill rule knowledge]ndash Situation awareness

[Rasmussen Skill Rule Knowledge Hierarchy]

Lateral Tracking Loop

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

Driver InputOutput Modes

Visual- External Scene - Forward - Roadway - Traffic - Lights - Signage - Environment - Peripheral - Rear- Internal - Instrument Cluster

TactileProprioceptive- Lateral Accel- Longitudinal Accel- Road Surface- Control Force Feedback- Vibration- Switch Feedback

Audio- Roadway- Alerts - Horns (ext) - ChimesClicks- Velocity (Wind)- Engine- Radio- Passenger

Olfactory- Gasoline- SmokeFire- Passenger

Other

Manual (Hand)- Wheel- Steering Column- Gear Shift- Switch (Panel)- Switch (Other)

Manual (Foot)- Throttle- Brake- Other

Voice

Other- Eye Tracking- Blink- Gesture- Thought

Driver

ldquoSituation Awarenessrdquo

bull Term originally defined for air combatbull Working Definition Sufficiently detailed mental picture of the

vehicle and environment (ie world model) to allow the operator to make well-informed (ie conditionally correct) decisions

Driver Situation AwarenessComponents

PersonalFactors

EquipmentStatus

VehiclePerformance

ControlSettings

Passengers

Non-drivingElements

Internal

SituationAwareness

Weather

AdjacentEnvironment

Signage

RoadwaySurface

LocationRoute

AdjacentTraffic

External

Endsley Situation Awareness Model

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Interaction Metaphors

bull Car as homebull Car as kitchenbull Car as bathroombull Car as bedroombull Car as music roombull Car as playroombull Car as entertainment ctrbull Car as toolboxbull Car as closet

bull Car as officebull Car as comm center

bull Car as image statementbull Car as clothingbull Car as jewelrybull Car as sports equipmentbull Car as safe spacebull Car as cocoon

Trends in Driver Attentional Demand

Driv

er A

tten

tion

al D

emDr

iver

Att

enti

onal

Dem

and

and

VehicleVehicleOperation OperationOperation Operation

SecondarySecondaryTasks Tasks

TimeTime

Time Spent in Vehicles (US)Time Spent in Vehicles (US)- Average 350 hrsyrperson Average 350 hrsyrperson - 500 Million hrsweek- 500 Million hrsweek

76 of Drivers Report Activities HaveldquoCausedNearly Causedrdquo an Accident

TALKING ON CELLPHONE

TURNING TO SPEAK

USING COMPUTER

CIGARETTE ASHES

BREAKING UP FIGHTBETWEEN KIDS

SPILLING COFFEE

Source Opinion Research Corp Interviews Time5800 (N=1016)

Concerns Regarding High SecondaryTask Loads

bull Growing Evidence and Public Perception of Safety Problem

bull Cell Phone use in USndash 115-120 Million Active Cell Phonesndash 50-70 Use in Vehiclesndash 39 of Drivers Using (Daylight Hours)

bull NE Journal of Medicine Estimatendash 4 fold increase in collision risk using cell phone

bull NHTSA Estimatesndash 12 Million Accidents (25-30) caused by Distracted

Driverbull Not limited to cell phones

bull Note These effects may be latent in normal operations and may only manifest in non-normal or emergency situations

Typical Performance vs TaskLoad Curve

Per

form

ance

Task Load

Distraction Components

bull Manualndash Inability or delay in operation of vehicle controlndash Hands occupied or out of position (cell phone)

bull Visualndash Head Down Problemsndash Visual Accommodationndash Visual Clutterndash Visual Compulsion

bull Cognitivendash Lack of cognitive engagement with primary Taskndash Latency in mental context shiftsndash Multi-tasking capabilitiesndash Prioritization

bull Emotionalbull High Individual Variability in Multi-Tasking Capability

Distraction Data Sources

bull Controlled Experimentsndash Vary Secondary Task Load (Independent Variable)

bull Visualbull Cognitive

ndash Measurebull Performancebull Response to Disturbancebull Situation Awareness

ndash Need to Increase Task Loading to Saturationndash Need to Include Unanticipated Eventsndash Need Lowest Common Denominator Population

bull Field Datandash Event Recordersndash Cell Phone Triggered

bull Subjective Survey Data

Controlled Experiment Issues (1)

bull Simulator Testingndash Controlled Scenarios +ndash Safe to go to Task Saturationndash Face and Cue Validity Issues ndashndash Simulator Sickness ndashndash Cost -

bull Dual Control Vehicle Testing (Test Track)ndash Good Validity +ndash Safety Issues at Task Saturation ndashndash Cost +

Controlled Experiment Issues (2)

bull Variability in Primary and Secondary Taskingndash How do you measure Secondary Task Loadndash How do you control Secondary Task Load

bull Performance Measuresndash Trackingndash Reaction Timendash Side Task Performance

bull Subjective Workload Measuresbull Situation Awareness Measures

ndash Testable Response Method

Simulator Studies of Driver CognitiveDistraction Caused by Cell Phone Use

bull Independent Variablesndash Cognitive Loadingndash Hands freeHands Fixed

bull Dependant Variablesndash Situation Awarenessndash Reaction Timendash Tracking MIT Age Lab Simulator

Cognitive Loading Levels

bull Lowndash Small talk

bull bullMediumndash Discuss movieoperaballet plot-line

bull bullHighndash Edit document by phone

Results Situational Awareness

bull Significant reduction in SA with Cell Phone Use

bull No significant difference between HFamp HH

Ave

rag

e F

ract

ion

Co

rrec

t

Normal HF HHPhone Setup

Stop Sign Response Time Resultsby Age Group

bull Significant effect of Age

bull No significant effect of Cell Phone Use

Re

sp

on

se

Tim

e [

se

c]

Normal1

Normal2

HFSmall

HFMovie

HF Edit HHSmall

HHMovie

HH Edit

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 7: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Aerospace Systems Applicable to Cars

bull Control systemsndash ABSndash Stability augmentation

bull Fly by WireLight (FBWFBL)ndash Integrity Concerns (eg

777)bull Critical software systemsbull Fault tolerant systemsbull Head up displays (HUD)bull Helmet mounted displays

(HMD)bull Synthetic Vision Systemsbull Sensor Fusionbull Hands on throttle and stic

k (HOTAS)bull Dark cockpit

bull Navigation systemsndash GPS DGPSndash IRSGPS

bull Situation awareness displaysndash Moving mapndash Database

bull Caution and Warning Systemsbull Collision Alerting Systemsbull Tactile alerting

ndash Stick shakerbull Master caution

ndash Information accessibilityndash Maintenance Diagnostics

Track Hardware Layout

bull Two 2 GPS antennas were mounted on the car to form a single baseline

bull Data-Linc Group (SRM6000) Modem antenna also attached to roll barndash Real-time communication with ground station

MIT Run Results

Typical Performance

bull Precise state determination ndash 2 - 5 cm position error ndash 1 - 2 cms velocity error ndash 1 - 2 degrees heading ndash 5 - 10 Hz

Track Results -Slip Measurements

bull Car heading and

velocity vectors

not aligned

Track Results -Slip Measurements

Comparative Lap Results

Acceleration vs Position

ldquoHuman Centeredrdquo InformationRequirements Analysis

bull Integrated Human Centered Systems Approachbull ldquoSemi-Structuredrdquo Decision Theorybull Driver Distraction Analysis

Driving Task Definition

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Vehicle Operation Tasks

bull Vehicle control tasks [skill based]ndash Lateral control (steering)ndash Longitudinal control (accel braking)

bull Tactical decisions [rule based]ndash Maneuveringndash Systems management

bull Strategic decisions [knowledge based]ndash Route selectionndash Goal management

bull Monitoring [skill rule knowledge]ndash Situation awareness

[Rasmussen Skill Rule Knowledge Hierarchy]

Lateral Tracking Loop

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

Driver InputOutput Modes

Visual- External Scene - Forward - Roadway - Traffic - Lights - Signage - Environment - Peripheral - Rear- Internal - Instrument Cluster

TactileProprioceptive- Lateral Accel- Longitudinal Accel- Road Surface- Control Force Feedback- Vibration- Switch Feedback

Audio- Roadway- Alerts - Horns (ext) - ChimesClicks- Velocity (Wind)- Engine- Radio- Passenger

Olfactory- Gasoline- SmokeFire- Passenger

Other

Manual (Hand)- Wheel- Steering Column- Gear Shift- Switch (Panel)- Switch (Other)

Manual (Foot)- Throttle- Brake- Other

Voice

Other- Eye Tracking- Blink- Gesture- Thought

Driver

ldquoSituation Awarenessrdquo

bull Term originally defined for air combatbull Working Definition Sufficiently detailed mental picture of the

vehicle and environment (ie world model) to allow the operator to make well-informed (ie conditionally correct) decisions

Driver Situation AwarenessComponents

PersonalFactors

EquipmentStatus

VehiclePerformance

ControlSettings

Passengers

Non-drivingElements

Internal

SituationAwareness

Weather

AdjacentEnvironment

Signage

RoadwaySurface

LocationRoute

AdjacentTraffic

External

Endsley Situation Awareness Model

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Interaction Metaphors

bull Car as homebull Car as kitchenbull Car as bathroombull Car as bedroombull Car as music roombull Car as playroombull Car as entertainment ctrbull Car as toolboxbull Car as closet

bull Car as officebull Car as comm center

bull Car as image statementbull Car as clothingbull Car as jewelrybull Car as sports equipmentbull Car as safe spacebull Car as cocoon

Trends in Driver Attentional Demand

Driv

er A

tten

tion

al D

emDr

iver

Att

enti

onal

Dem

and

and

VehicleVehicleOperation OperationOperation Operation

SecondarySecondaryTasks Tasks

TimeTime

Time Spent in Vehicles (US)Time Spent in Vehicles (US)- Average 350 hrsyrperson Average 350 hrsyrperson - 500 Million hrsweek- 500 Million hrsweek

76 of Drivers Report Activities HaveldquoCausedNearly Causedrdquo an Accident

TALKING ON CELLPHONE

TURNING TO SPEAK

USING COMPUTER

CIGARETTE ASHES

BREAKING UP FIGHTBETWEEN KIDS

SPILLING COFFEE

Source Opinion Research Corp Interviews Time5800 (N=1016)

Concerns Regarding High SecondaryTask Loads

bull Growing Evidence and Public Perception of Safety Problem

bull Cell Phone use in USndash 115-120 Million Active Cell Phonesndash 50-70 Use in Vehiclesndash 39 of Drivers Using (Daylight Hours)

bull NE Journal of Medicine Estimatendash 4 fold increase in collision risk using cell phone

bull NHTSA Estimatesndash 12 Million Accidents (25-30) caused by Distracted

Driverbull Not limited to cell phones

bull Note These effects may be latent in normal operations and may only manifest in non-normal or emergency situations

Typical Performance vs TaskLoad Curve

Per

form

ance

Task Load

Distraction Components

bull Manualndash Inability or delay in operation of vehicle controlndash Hands occupied or out of position (cell phone)

bull Visualndash Head Down Problemsndash Visual Accommodationndash Visual Clutterndash Visual Compulsion

bull Cognitivendash Lack of cognitive engagement with primary Taskndash Latency in mental context shiftsndash Multi-tasking capabilitiesndash Prioritization

bull Emotionalbull High Individual Variability in Multi-Tasking Capability

Distraction Data Sources

bull Controlled Experimentsndash Vary Secondary Task Load (Independent Variable)

bull Visualbull Cognitive

ndash Measurebull Performancebull Response to Disturbancebull Situation Awareness

ndash Need to Increase Task Loading to Saturationndash Need to Include Unanticipated Eventsndash Need Lowest Common Denominator Population

bull Field Datandash Event Recordersndash Cell Phone Triggered

bull Subjective Survey Data

Controlled Experiment Issues (1)

bull Simulator Testingndash Controlled Scenarios +ndash Safe to go to Task Saturationndash Face and Cue Validity Issues ndashndash Simulator Sickness ndashndash Cost -

bull Dual Control Vehicle Testing (Test Track)ndash Good Validity +ndash Safety Issues at Task Saturation ndashndash Cost +

Controlled Experiment Issues (2)

bull Variability in Primary and Secondary Taskingndash How do you measure Secondary Task Loadndash How do you control Secondary Task Load

bull Performance Measuresndash Trackingndash Reaction Timendash Side Task Performance

bull Subjective Workload Measuresbull Situation Awareness Measures

ndash Testable Response Method

Simulator Studies of Driver CognitiveDistraction Caused by Cell Phone Use

bull Independent Variablesndash Cognitive Loadingndash Hands freeHands Fixed

bull Dependant Variablesndash Situation Awarenessndash Reaction Timendash Tracking MIT Age Lab Simulator

Cognitive Loading Levels

bull Lowndash Small talk

bull bullMediumndash Discuss movieoperaballet plot-line

bull bullHighndash Edit document by phone

Results Situational Awareness

bull Significant reduction in SA with Cell Phone Use

bull No significant difference between HFamp HH

Ave

rag

e F

ract

ion

Co

rrec

t

Normal HF HHPhone Setup

Stop Sign Response Time Resultsby Age Group

bull Significant effect of Age

bull No significant effect of Cell Phone Use

Re

sp

on

se

Tim

e [

se

c]

Normal1

Normal2

HFSmall

HFMovie

HF Edit HHSmall

HHMovie

HH Edit

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 8: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Track Hardware Layout

bull Two 2 GPS antennas were mounted on the car to form a single baseline

bull Data-Linc Group (SRM6000) Modem antenna also attached to roll barndash Real-time communication with ground station

MIT Run Results

Typical Performance

bull Precise state determination ndash 2 - 5 cm position error ndash 1 - 2 cms velocity error ndash 1 - 2 degrees heading ndash 5 - 10 Hz

Track Results -Slip Measurements

bull Car heading and

velocity vectors

not aligned

Track Results -Slip Measurements

Comparative Lap Results

Acceleration vs Position

ldquoHuman Centeredrdquo InformationRequirements Analysis

bull Integrated Human Centered Systems Approachbull ldquoSemi-Structuredrdquo Decision Theorybull Driver Distraction Analysis

Driving Task Definition

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Vehicle Operation Tasks

bull Vehicle control tasks [skill based]ndash Lateral control (steering)ndash Longitudinal control (accel braking)

bull Tactical decisions [rule based]ndash Maneuveringndash Systems management

bull Strategic decisions [knowledge based]ndash Route selectionndash Goal management

bull Monitoring [skill rule knowledge]ndash Situation awareness

[Rasmussen Skill Rule Knowledge Hierarchy]

Lateral Tracking Loop

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

Driver InputOutput Modes

Visual- External Scene - Forward - Roadway - Traffic - Lights - Signage - Environment - Peripheral - Rear- Internal - Instrument Cluster

TactileProprioceptive- Lateral Accel- Longitudinal Accel- Road Surface- Control Force Feedback- Vibration- Switch Feedback

Audio- Roadway- Alerts - Horns (ext) - ChimesClicks- Velocity (Wind)- Engine- Radio- Passenger

Olfactory- Gasoline- SmokeFire- Passenger

Other

Manual (Hand)- Wheel- Steering Column- Gear Shift- Switch (Panel)- Switch (Other)

Manual (Foot)- Throttle- Brake- Other

Voice

Other- Eye Tracking- Blink- Gesture- Thought

Driver

ldquoSituation Awarenessrdquo

bull Term originally defined for air combatbull Working Definition Sufficiently detailed mental picture of the

vehicle and environment (ie world model) to allow the operator to make well-informed (ie conditionally correct) decisions

Driver Situation AwarenessComponents

PersonalFactors

EquipmentStatus

VehiclePerformance

ControlSettings

Passengers

Non-drivingElements

Internal

SituationAwareness

Weather

AdjacentEnvironment

Signage

RoadwaySurface

LocationRoute

AdjacentTraffic

External

Endsley Situation Awareness Model

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Interaction Metaphors

bull Car as homebull Car as kitchenbull Car as bathroombull Car as bedroombull Car as music roombull Car as playroombull Car as entertainment ctrbull Car as toolboxbull Car as closet

bull Car as officebull Car as comm center

bull Car as image statementbull Car as clothingbull Car as jewelrybull Car as sports equipmentbull Car as safe spacebull Car as cocoon

Trends in Driver Attentional Demand

Driv

er A

tten

tion

al D

emDr

iver

Att

enti

onal

Dem

and

and

VehicleVehicleOperation OperationOperation Operation

SecondarySecondaryTasks Tasks

TimeTime

Time Spent in Vehicles (US)Time Spent in Vehicles (US)- Average 350 hrsyrperson Average 350 hrsyrperson - 500 Million hrsweek- 500 Million hrsweek

76 of Drivers Report Activities HaveldquoCausedNearly Causedrdquo an Accident

TALKING ON CELLPHONE

TURNING TO SPEAK

USING COMPUTER

CIGARETTE ASHES

BREAKING UP FIGHTBETWEEN KIDS

SPILLING COFFEE

Source Opinion Research Corp Interviews Time5800 (N=1016)

Concerns Regarding High SecondaryTask Loads

bull Growing Evidence and Public Perception of Safety Problem

bull Cell Phone use in USndash 115-120 Million Active Cell Phonesndash 50-70 Use in Vehiclesndash 39 of Drivers Using (Daylight Hours)

bull NE Journal of Medicine Estimatendash 4 fold increase in collision risk using cell phone

bull NHTSA Estimatesndash 12 Million Accidents (25-30) caused by Distracted

Driverbull Not limited to cell phones

bull Note These effects may be latent in normal operations and may only manifest in non-normal or emergency situations

Typical Performance vs TaskLoad Curve

Per

form

ance

Task Load

Distraction Components

bull Manualndash Inability or delay in operation of vehicle controlndash Hands occupied or out of position (cell phone)

bull Visualndash Head Down Problemsndash Visual Accommodationndash Visual Clutterndash Visual Compulsion

bull Cognitivendash Lack of cognitive engagement with primary Taskndash Latency in mental context shiftsndash Multi-tasking capabilitiesndash Prioritization

bull Emotionalbull High Individual Variability in Multi-Tasking Capability

Distraction Data Sources

bull Controlled Experimentsndash Vary Secondary Task Load (Independent Variable)

bull Visualbull Cognitive

ndash Measurebull Performancebull Response to Disturbancebull Situation Awareness

ndash Need to Increase Task Loading to Saturationndash Need to Include Unanticipated Eventsndash Need Lowest Common Denominator Population

bull Field Datandash Event Recordersndash Cell Phone Triggered

bull Subjective Survey Data

Controlled Experiment Issues (1)

bull Simulator Testingndash Controlled Scenarios +ndash Safe to go to Task Saturationndash Face and Cue Validity Issues ndashndash Simulator Sickness ndashndash Cost -

bull Dual Control Vehicle Testing (Test Track)ndash Good Validity +ndash Safety Issues at Task Saturation ndashndash Cost +

Controlled Experiment Issues (2)

bull Variability in Primary and Secondary Taskingndash How do you measure Secondary Task Loadndash How do you control Secondary Task Load

bull Performance Measuresndash Trackingndash Reaction Timendash Side Task Performance

bull Subjective Workload Measuresbull Situation Awareness Measures

ndash Testable Response Method

Simulator Studies of Driver CognitiveDistraction Caused by Cell Phone Use

bull Independent Variablesndash Cognitive Loadingndash Hands freeHands Fixed

bull Dependant Variablesndash Situation Awarenessndash Reaction Timendash Tracking MIT Age Lab Simulator

Cognitive Loading Levels

bull Lowndash Small talk

bull bullMediumndash Discuss movieoperaballet plot-line

bull bullHighndash Edit document by phone

Results Situational Awareness

bull Significant reduction in SA with Cell Phone Use

bull No significant difference between HFamp HH

Ave

rag

e F

ract

ion

Co

rrec

t

Normal HF HHPhone Setup

Stop Sign Response Time Resultsby Age Group

bull Significant effect of Age

bull No significant effect of Cell Phone Use

Re

sp

on

se

Tim

e [

se

c]

Normal1

Normal2

HFSmall

HFMovie

HF Edit HHSmall

HHMovie

HH Edit

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 9: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

MIT Run Results

Typical Performance

bull Precise state determination ndash 2 - 5 cm position error ndash 1 - 2 cms velocity error ndash 1 - 2 degrees heading ndash 5 - 10 Hz

Track Results -Slip Measurements

bull Car heading and

velocity vectors

not aligned

Track Results -Slip Measurements

Comparative Lap Results

Acceleration vs Position

ldquoHuman Centeredrdquo InformationRequirements Analysis

bull Integrated Human Centered Systems Approachbull ldquoSemi-Structuredrdquo Decision Theorybull Driver Distraction Analysis

Driving Task Definition

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Vehicle Operation Tasks

bull Vehicle control tasks [skill based]ndash Lateral control (steering)ndash Longitudinal control (accel braking)

bull Tactical decisions [rule based]ndash Maneuveringndash Systems management

bull Strategic decisions [knowledge based]ndash Route selectionndash Goal management

bull Monitoring [skill rule knowledge]ndash Situation awareness

[Rasmussen Skill Rule Knowledge Hierarchy]

Lateral Tracking Loop

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

Driver InputOutput Modes

Visual- External Scene - Forward - Roadway - Traffic - Lights - Signage - Environment - Peripheral - Rear- Internal - Instrument Cluster

TactileProprioceptive- Lateral Accel- Longitudinal Accel- Road Surface- Control Force Feedback- Vibration- Switch Feedback

Audio- Roadway- Alerts - Horns (ext) - ChimesClicks- Velocity (Wind)- Engine- Radio- Passenger

Olfactory- Gasoline- SmokeFire- Passenger

Other

Manual (Hand)- Wheel- Steering Column- Gear Shift- Switch (Panel)- Switch (Other)

Manual (Foot)- Throttle- Brake- Other

Voice

Other- Eye Tracking- Blink- Gesture- Thought

Driver

ldquoSituation Awarenessrdquo

bull Term originally defined for air combatbull Working Definition Sufficiently detailed mental picture of the

vehicle and environment (ie world model) to allow the operator to make well-informed (ie conditionally correct) decisions

Driver Situation AwarenessComponents

PersonalFactors

EquipmentStatus

VehiclePerformance

ControlSettings

Passengers

Non-drivingElements

Internal

SituationAwareness

Weather

AdjacentEnvironment

Signage

RoadwaySurface

LocationRoute

AdjacentTraffic

External

Endsley Situation Awareness Model

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Interaction Metaphors

bull Car as homebull Car as kitchenbull Car as bathroombull Car as bedroombull Car as music roombull Car as playroombull Car as entertainment ctrbull Car as toolboxbull Car as closet

bull Car as officebull Car as comm center

bull Car as image statementbull Car as clothingbull Car as jewelrybull Car as sports equipmentbull Car as safe spacebull Car as cocoon

Trends in Driver Attentional Demand

Driv

er A

tten

tion

al D

emDr

iver

Att

enti

onal

Dem

and

and

VehicleVehicleOperation OperationOperation Operation

SecondarySecondaryTasks Tasks

TimeTime

Time Spent in Vehicles (US)Time Spent in Vehicles (US)- Average 350 hrsyrperson Average 350 hrsyrperson - 500 Million hrsweek- 500 Million hrsweek

76 of Drivers Report Activities HaveldquoCausedNearly Causedrdquo an Accident

TALKING ON CELLPHONE

TURNING TO SPEAK

USING COMPUTER

CIGARETTE ASHES

BREAKING UP FIGHTBETWEEN KIDS

SPILLING COFFEE

Source Opinion Research Corp Interviews Time5800 (N=1016)

Concerns Regarding High SecondaryTask Loads

bull Growing Evidence and Public Perception of Safety Problem

bull Cell Phone use in USndash 115-120 Million Active Cell Phonesndash 50-70 Use in Vehiclesndash 39 of Drivers Using (Daylight Hours)

bull NE Journal of Medicine Estimatendash 4 fold increase in collision risk using cell phone

bull NHTSA Estimatesndash 12 Million Accidents (25-30) caused by Distracted

Driverbull Not limited to cell phones

bull Note These effects may be latent in normal operations and may only manifest in non-normal or emergency situations

Typical Performance vs TaskLoad Curve

Per

form

ance

Task Load

Distraction Components

bull Manualndash Inability or delay in operation of vehicle controlndash Hands occupied or out of position (cell phone)

bull Visualndash Head Down Problemsndash Visual Accommodationndash Visual Clutterndash Visual Compulsion

bull Cognitivendash Lack of cognitive engagement with primary Taskndash Latency in mental context shiftsndash Multi-tasking capabilitiesndash Prioritization

bull Emotionalbull High Individual Variability in Multi-Tasking Capability

Distraction Data Sources

bull Controlled Experimentsndash Vary Secondary Task Load (Independent Variable)

bull Visualbull Cognitive

ndash Measurebull Performancebull Response to Disturbancebull Situation Awareness

ndash Need to Increase Task Loading to Saturationndash Need to Include Unanticipated Eventsndash Need Lowest Common Denominator Population

bull Field Datandash Event Recordersndash Cell Phone Triggered

bull Subjective Survey Data

Controlled Experiment Issues (1)

bull Simulator Testingndash Controlled Scenarios +ndash Safe to go to Task Saturationndash Face and Cue Validity Issues ndashndash Simulator Sickness ndashndash Cost -

bull Dual Control Vehicle Testing (Test Track)ndash Good Validity +ndash Safety Issues at Task Saturation ndashndash Cost +

Controlled Experiment Issues (2)

bull Variability in Primary and Secondary Taskingndash How do you measure Secondary Task Loadndash How do you control Secondary Task Load

bull Performance Measuresndash Trackingndash Reaction Timendash Side Task Performance

bull Subjective Workload Measuresbull Situation Awareness Measures

ndash Testable Response Method

Simulator Studies of Driver CognitiveDistraction Caused by Cell Phone Use

bull Independent Variablesndash Cognitive Loadingndash Hands freeHands Fixed

bull Dependant Variablesndash Situation Awarenessndash Reaction Timendash Tracking MIT Age Lab Simulator

Cognitive Loading Levels

bull Lowndash Small talk

bull bullMediumndash Discuss movieoperaballet plot-line

bull bullHighndash Edit document by phone

Results Situational Awareness

bull Significant reduction in SA with Cell Phone Use

bull No significant difference between HFamp HH

Ave

rag

e F

ract

ion

Co

rrec

t

Normal HF HHPhone Setup

Stop Sign Response Time Resultsby Age Group

bull Significant effect of Age

bull No significant effect of Cell Phone Use

Re

sp

on

se

Tim

e [

se

c]

Normal1

Normal2

HFSmall

HFMovie

HF Edit HHSmall

HHMovie

HH Edit

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 10: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Typical Performance

bull Precise state determination ndash 2 - 5 cm position error ndash 1 - 2 cms velocity error ndash 1 - 2 degrees heading ndash 5 - 10 Hz

Track Results -Slip Measurements

bull Car heading and

velocity vectors

not aligned

Track Results -Slip Measurements

Comparative Lap Results

Acceleration vs Position

ldquoHuman Centeredrdquo InformationRequirements Analysis

bull Integrated Human Centered Systems Approachbull ldquoSemi-Structuredrdquo Decision Theorybull Driver Distraction Analysis

Driving Task Definition

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Vehicle Operation Tasks

bull Vehicle control tasks [skill based]ndash Lateral control (steering)ndash Longitudinal control (accel braking)

bull Tactical decisions [rule based]ndash Maneuveringndash Systems management

bull Strategic decisions [knowledge based]ndash Route selectionndash Goal management

bull Monitoring [skill rule knowledge]ndash Situation awareness

[Rasmussen Skill Rule Knowledge Hierarchy]

Lateral Tracking Loop

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

Driver InputOutput Modes

Visual- External Scene - Forward - Roadway - Traffic - Lights - Signage - Environment - Peripheral - Rear- Internal - Instrument Cluster

TactileProprioceptive- Lateral Accel- Longitudinal Accel- Road Surface- Control Force Feedback- Vibration- Switch Feedback

Audio- Roadway- Alerts - Horns (ext) - ChimesClicks- Velocity (Wind)- Engine- Radio- Passenger

Olfactory- Gasoline- SmokeFire- Passenger

Other

Manual (Hand)- Wheel- Steering Column- Gear Shift- Switch (Panel)- Switch (Other)

Manual (Foot)- Throttle- Brake- Other

Voice

Other- Eye Tracking- Blink- Gesture- Thought

Driver

ldquoSituation Awarenessrdquo

bull Term originally defined for air combatbull Working Definition Sufficiently detailed mental picture of the

vehicle and environment (ie world model) to allow the operator to make well-informed (ie conditionally correct) decisions

Driver Situation AwarenessComponents

PersonalFactors

EquipmentStatus

VehiclePerformance

ControlSettings

Passengers

Non-drivingElements

Internal

SituationAwareness

Weather

AdjacentEnvironment

Signage

RoadwaySurface

LocationRoute

AdjacentTraffic

External

Endsley Situation Awareness Model

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Interaction Metaphors

bull Car as homebull Car as kitchenbull Car as bathroombull Car as bedroombull Car as music roombull Car as playroombull Car as entertainment ctrbull Car as toolboxbull Car as closet

bull Car as officebull Car as comm center

bull Car as image statementbull Car as clothingbull Car as jewelrybull Car as sports equipmentbull Car as safe spacebull Car as cocoon

Trends in Driver Attentional Demand

Driv

er A

tten

tion

al D

emDr

iver

Att

enti

onal

Dem

and

and

VehicleVehicleOperation OperationOperation Operation

SecondarySecondaryTasks Tasks

TimeTime

Time Spent in Vehicles (US)Time Spent in Vehicles (US)- Average 350 hrsyrperson Average 350 hrsyrperson - 500 Million hrsweek- 500 Million hrsweek

76 of Drivers Report Activities HaveldquoCausedNearly Causedrdquo an Accident

TALKING ON CELLPHONE

TURNING TO SPEAK

USING COMPUTER

CIGARETTE ASHES

BREAKING UP FIGHTBETWEEN KIDS

SPILLING COFFEE

Source Opinion Research Corp Interviews Time5800 (N=1016)

Concerns Regarding High SecondaryTask Loads

bull Growing Evidence and Public Perception of Safety Problem

bull Cell Phone use in USndash 115-120 Million Active Cell Phonesndash 50-70 Use in Vehiclesndash 39 of Drivers Using (Daylight Hours)

bull NE Journal of Medicine Estimatendash 4 fold increase in collision risk using cell phone

bull NHTSA Estimatesndash 12 Million Accidents (25-30) caused by Distracted

Driverbull Not limited to cell phones

bull Note These effects may be latent in normal operations and may only manifest in non-normal or emergency situations

Typical Performance vs TaskLoad Curve

Per

form

ance

Task Load

Distraction Components

bull Manualndash Inability or delay in operation of vehicle controlndash Hands occupied or out of position (cell phone)

bull Visualndash Head Down Problemsndash Visual Accommodationndash Visual Clutterndash Visual Compulsion

bull Cognitivendash Lack of cognitive engagement with primary Taskndash Latency in mental context shiftsndash Multi-tasking capabilitiesndash Prioritization

bull Emotionalbull High Individual Variability in Multi-Tasking Capability

Distraction Data Sources

bull Controlled Experimentsndash Vary Secondary Task Load (Independent Variable)

bull Visualbull Cognitive

ndash Measurebull Performancebull Response to Disturbancebull Situation Awareness

ndash Need to Increase Task Loading to Saturationndash Need to Include Unanticipated Eventsndash Need Lowest Common Denominator Population

bull Field Datandash Event Recordersndash Cell Phone Triggered

bull Subjective Survey Data

Controlled Experiment Issues (1)

bull Simulator Testingndash Controlled Scenarios +ndash Safe to go to Task Saturationndash Face and Cue Validity Issues ndashndash Simulator Sickness ndashndash Cost -

bull Dual Control Vehicle Testing (Test Track)ndash Good Validity +ndash Safety Issues at Task Saturation ndashndash Cost +

Controlled Experiment Issues (2)

bull Variability in Primary and Secondary Taskingndash How do you measure Secondary Task Loadndash How do you control Secondary Task Load

bull Performance Measuresndash Trackingndash Reaction Timendash Side Task Performance

bull Subjective Workload Measuresbull Situation Awareness Measures

ndash Testable Response Method

Simulator Studies of Driver CognitiveDistraction Caused by Cell Phone Use

bull Independent Variablesndash Cognitive Loadingndash Hands freeHands Fixed

bull Dependant Variablesndash Situation Awarenessndash Reaction Timendash Tracking MIT Age Lab Simulator

Cognitive Loading Levels

bull Lowndash Small talk

bull bullMediumndash Discuss movieoperaballet plot-line

bull bullHighndash Edit document by phone

Results Situational Awareness

bull Significant reduction in SA with Cell Phone Use

bull No significant difference between HFamp HH

Ave

rag

e F

ract

ion

Co

rrec

t

Normal HF HHPhone Setup

Stop Sign Response Time Resultsby Age Group

bull Significant effect of Age

bull No significant effect of Cell Phone Use

Re

sp

on

se

Tim

e [

se

c]

Normal1

Normal2

HFSmall

HFMovie

HF Edit HHSmall

HHMovie

HH Edit

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 11: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Track Results -Slip Measurements

bull Car heading and

velocity vectors

not aligned

Track Results -Slip Measurements

Comparative Lap Results

Acceleration vs Position

ldquoHuman Centeredrdquo InformationRequirements Analysis

bull Integrated Human Centered Systems Approachbull ldquoSemi-Structuredrdquo Decision Theorybull Driver Distraction Analysis

Driving Task Definition

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Vehicle Operation Tasks

bull Vehicle control tasks [skill based]ndash Lateral control (steering)ndash Longitudinal control (accel braking)

bull Tactical decisions [rule based]ndash Maneuveringndash Systems management

bull Strategic decisions [knowledge based]ndash Route selectionndash Goal management

bull Monitoring [skill rule knowledge]ndash Situation awareness

[Rasmussen Skill Rule Knowledge Hierarchy]

Lateral Tracking Loop

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

Driver InputOutput Modes

Visual- External Scene - Forward - Roadway - Traffic - Lights - Signage - Environment - Peripheral - Rear- Internal - Instrument Cluster

TactileProprioceptive- Lateral Accel- Longitudinal Accel- Road Surface- Control Force Feedback- Vibration- Switch Feedback

Audio- Roadway- Alerts - Horns (ext) - ChimesClicks- Velocity (Wind)- Engine- Radio- Passenger

Olfactory- Gasoline- SmokeFire- Passenger

Other

Manual (Hand)- Wheel- Steering Column- Gear Shift- Switch (Panel)- Switch (Other)

Manual (Foot)- Throttle- Brake- Other

Voice

Other- Eye Tracking- Blink- Gesture- Thought

Driver

ldquoSituation Awarenessrdquo

bull Term originally defined for air combatbull Working Definition Sufficiently detailed mental picture of the

vehicle and environment (ie world model) to allow the operator to make well-informed (ie conditionally correct) decisions

Driver Situation AwarenessComponents

PersonalFactors

EquipmentStatus

VehiclePerformance

ControlSettings

Passengers

Non-drivingElements

Internal

SituationAwareness

Weather

AdjacentEnvironment

Signage

RoadwaySurface

LocationRoute

AdjacentTraffic

External

Endsley Situation Awareness Model

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Interaction Metaphors

bull Car as homebull Car as kitchenbull Car as bathroombull Car as bedroombull Car as music roombull Car as playroombull Car as entertainment ctrbull Car as toolboxbull Car as closet

bull Car as officebull Car as comm center

bull Car as image statementbull Car as clothingbull Car as jewelrybull Car as sports equipmentbull Car as safe spacebull Car as cocoon

Trends in Driver Attentional Demand

Driv

er A

tten

tion

al D

emDr

iver

Att

enti

onal

Dem

and

and

VehicleVehicleOperation OperationOperation Operation

SecondarySecondaryTasks Tasks

TimeTime

Time Spent in Vehicles (US)Time Spent in Vehicles (US)- Average 350 hrsyrperson Average 350 hrsyrperson - 500 Million hrsweek- 500 Million hrsweek

76 of Drivers Report Activities HaveldquoCausedNearly Causedrdquo an Accident

TALKING ON CELLPHONE

TURNING TO SPEAK

USING COMPUTER

CIGARETTE ASHES

BREAKING UP FIGHTBETWEEN KIDS

SPILLING COFFEE

Source Opinion Research Corp Interviews Time5800 (N=1016)

Concerns Regarding High SecondaryTask Loads

bull Growing Evidence and Public Perception of Safety Problem

bull Cell Phone use in USndash 115-120 Million Active Cell Phonesndash 50-70 Use in Vehiclesndash 39 of Drivers Using (Daylight Hours)

bull NE Journal of Medicine Estimatendash 4 fold increase in collision risk using cell phone

bull NHTSA Estimatesndash 12 Million Accidents (25-30) caused by Distracted

Driverbull Not limited to cell phones

bull Note These effects may be latent in normal operations and may only manifest in non-normal or emergency situations

Typical Performance vs TaskLoad Curve

Per

form

ance

Task Load

Distraction Components

bull Manualndash Inability or delay in operation of vehicle controlndash Hands occupied or out of position (cell phone)

bull Visualndash Head Down Problemsndash Visual Accommodationndash Visual Clutterndash Visual Compulsion

bull Cognitivendash Lack of cognitive engagement with primary Taskndash Latency in mental context shiftsndash Multi-tasking capabilitiesndash Prioritization

bull Emotionalbull High Individual Variability in Multi-Tasking Capability

Distraction Data Sources

bull Controlled Experimentsndash Vary Secondary Task Load (Independent Variable)

bull Visualbull Cognitive

ndash Measurebull Performancebull Response to Disturbancebull Situation Awareness

ndash Need to Increase Task Loading to Saturationndash Need to Include Unanticipated Eventsndash Need Lowest Common Denominator Population

bull Field Datandash Event Recordersndash Cell Phone Triggered

bull Subjective Survey Data

Controlled Experiment Issues (1)

bull Simulator Testingndash Controlled Scenarios +ndash Safe to go to Task Saturationndash Face and Cue Validity Issues ndashndash Simulator Sickness ndashndash Cost -

bull Dual Control Vehicle Testing (Test Track)ndash Good Validity +ndash Safety Issues at Task Saturation ndashndash Cost +

Controlled Experiment Issues (2)

bull Variability in Primary and Secondary Taskingndash How do you measure Secondary Task Loadndash How do you control Secondary Task Load

bull Performance Measuresndash Trackingndash Reaction Timendash Side Task Performance

bull Subjective Workload Measuresbull Situation Awareness Measures

ndash Testable Response Method

Simulator Studies of Driver CognitiveDistraction Caused by Cell Phone Use

bull Independent Variablesndash Cognitive Loadingndash Hands freeHands Fixed

bull Dependant Variablesndash Situation Awarenessndash Reaction Timendash Tracking MIT Age Lab Simulator

Cognitive Loading Levels

bull Lowndash Small talk

bull bullMediumndash Discuss movieoperaballet plot-line

bull bullHighndash Edit document by phone

Results Situational Awareness

bull Significant reduction in SA with Cell Phone Use

bull No significant difference between HFamp HH

Ave

rag

e F

ract

ion

Co

rrec

t

Normal HF HHPhone Setup

Stop Sign Response Time Resultsby Age Group

bull Significant effect of Age

bull No significant effect of Cell Phone Use

Re

sp

on

se

Tim

e [

se

c]

Normal1

Normal2

HFSmall

HFMovie

HF Edit HHSmall

HHMovie

HH Edit

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 12: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Track Results -Slip Measurements

Comparative Lap Results

Acceleration vs Position

ldquoHuman Centeredrdquo InformationRequirements Analysis

bull Integrated Human Centered Systems Approachbull ldquoSemi-Structuredrdquo Decision Theorybull Driver Distraction Analysis

Driving Task Definition

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Vehicle Operation Tasks

bull Vehicle control tasks [skill based]ndash Lateral control (steering)ndash Longitudinal control (accel braking)

bull Tactical decisions [rule based]ndash Maneuveringndash Systems management

bull Strategic decisions [knowledge based]ndash Route selectionndash Goal management

bull Monitoring [skill rule knowledge]ndash Situation awareness

[Rasmussen Skill Rule Knowledge Hierarchy]

Lateral Tracking Loop

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

Driver InputOutput Modes

Visual- External Scene - Forward - Roadway - Traffic - Lights - Signage - Environment - Peripheral - Rear- Internal - Instrument Cluster

TactileProprioceptive- Lateral Accel- Longitudinal Accel- Road Surface- Control Force Feedback- Vibration- Switch Feedback

Audio- Roadway- Alerts - Horns (ext) - ChimesClicks- Velocity (Wind)- Engine- Radio- Passenger

Olfactory- Gasoline- SmokeFire- Passenger

Other

Manual (Hand)- Wheel- Steering Column- Gear Shift- Switch (Panel)- Switch (Other)

Manual (Foot)- Throttle- Brake- Other

Voice

Other- Eye Tracking- Blink- Gesture- Thought

Driver

ldquoSituation Awarenessrdquo

bull Term originally defined for air combatbull Working Definition Sufficiently detailed mental picture of the

vehicle and environment (ie world model) to allow the operator to make well-informed (ie conditionally correct) decisions

Driver Situation AwarenessComponents

PersonalFactors

EquipmentStatus

VehiclePerformance

ControlSettings

Passengers

Non-drivingElements

Internal

SituationAwareness

Weather

AdjacentEnvironment

Signage

RoadwaySurface

LocationRoute

AdjacentTraffic

External

Endsley Situation Awareness Model

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Interaction Metaphors

bull Car as homebull Car as kitchenbull Car as bathroombull Car as bedroombull Car as music roombull Car as playroombull Car as entertainment ctrbull Car as toolboxbull Car as closet

bull Car as officebull Car as comm center

bull Car as image statementbull Car as clothingbull Car as jewelrybull Car as sports equipmentbull Car as safe spacebull Car as cocoon

Trends in Driver Attentional Demand

Driv

er A

tten

tion

al D

emDr

iver

Att

enti

onal

Dem

and

and

VehicleVehicleOperation OperationOperation Operation

SecondarySecondaryTasks Tasks

TimeTime

Time Spent in Vehicles (US)Time Spent in Vehicles (US)- Average 350 hrsyrperson Average 350 hrsyrperson - 500 Million hrsweek- 500 Million hrsweek

76 of Drivers Report Activities HaveldquoCausedNearly Causedrdquo an Accident

TALKING ON CELLPHONE

TURNING TO SPEAK

USING COMPUTER

CIGARETTE ASHES

BREAKING UP FIGHTBETWEEN KIDS

SPILLING COFFEE

Source Opinion Research Corp Interviews Time5800 (N=1016)

Concerns Regarding High SecondaryTask Loads

bull Growing Evidence and Public Perception of Safety Problem

bull Cell Phone use in USndash 115-120 Million Active Cell Phonesndash 50-70 Use in Vehiclesndash 39 of Drivers Using (Daylight Hours)

bull NE Journal of Medicine Estimatendash 4 fold increase in collision risk using cell phone

bull NHTSA Estimatesndash 12 Million Accidents (25-30) caused by Distracted

Driverbull Not limited to cell phones

bull Note These effects may be latent in normal operations and may only manifest in non-normal or emergency situations

Typical Performance vs TaskLoad Curve

Per

form

ance

Task Load

Distraction Components

bull Manualndash Inability or delay in operation of vehicle controlndash Hands occupied or out of position (cell phone)

bull Visualndash Head Down Problemsndash Visual Accommodationndash Visual Clutterndash Visual Compulsion

bull Cognitivendash Lack of cognitive engagement with primary Taskndash Latency in mental context shiftsndash Multi-tasking capabilitiesndash Prioritization

bull Emotionalbull High Individual Variability in Multi-Tasking Capability

Distraction Data Sources

bull Controlled Experimentsndash Vary Secondary Task Load (Independent Variable)

bull Visualbull Cognitive

ndash Measurebull Performancebull Response to Disturbancebull Situation Awareness

ndash Need to Increase Task Loading to Saturationndash Need to Include Unanticipated Eventsndash Need Lowest Common Denominator Population

bull Field Datandash Event Recordersndash Cell Phone Triggered

bull Subjective Survey Data

Controlled Experiment Issues (1)

bull Simulator Testingndash Controlled Scenarios +ndash Safe to go to Task Saturationndash Face and Cue Validity Issues ndashndash Simulator Sickness ndashndash Cost -

bull Dual Control Vehicle Testing (Test Track)ndash Good Validity +ndash Safety Issues at Task Saturation ndashndash Cost +

Controlled Experiment Issues (2)

bull Variability in Primary and Secondary Taskingndash How do you measure Secondary Task Loadndash How do you control Secondary Task Load

bull Performance Measuresndash Trackingndash Reaction Timendash Side Task Performance

bull Subjective Workload Measuresbull Situation Awareness Measures

ndash Testable Response Method

Simulator Studies of Driver CognitiveDistraction Caused by Cell Phone Use

bull Independent Variablesndash Cognitive Loadingndash Hands freeHands Fixed

bull Dependant Variablesndash Situation Awarenessndash Reaction Timendash Tracking MIT Age Lab Simulator

Cognitive Loading Levels

bull Lowndash Small talk

bull bullMediumndash Discuss movieoperaballet plot-line

bull bullHighndash Edit document by phone

Results Situational Awareness

bull Significant reduction in SA with Cell Phone Use

bull No significant difference between HFamp HH

Ave

rag

e F

ract

ion

Co

rrec

t

Normal HF HHPhone Setup

Stop Sign Response Time Resultsby Age Group

bull Significant effect of Age

bull No significant effect of Cell Phone Use

Re

sp

on

se

Tim

e [

se

c]

Normal1

Normal2

HFSmall

HFMovie

HF Edit HHSmall

HHMovie

HH Edit

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 13: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Comparative Lap Results

Acceleration vs Position

ldquoHuman Centeredrdquo InformationRequirements Analysis

bull Integrated Human Centered Systems Approachbull ldquoSemi-Structuredrdquo Decision Theorybull Driver Distraction Analysis

Driving Task Definition

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Vehicle Operation Tasks

bull Vehicle control tasks [skill based]ndash Lateral control (steering)ndash Longitudinal control (accel braking)

bull Tactical decisions [rule based]ndash Maneuveringndash Systems management

bull Strategic decisions [knowledge based]ndash Route selectionndash Goal management

bull Monitoring [skill rule knowledge]ndash Situation awareness

[Rasmussen Skill Rule Knowledge Hierarchy]

Lateral Tracking Loop

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

Driver InputOutput Modes

Visual- External Scene - Forward - Roadway - Traffic - Lights - Signage - Environment - Peripheral - Rear- Internal - Instrument Cluster

TactileProprioceptive- Lateral Accel- Longitudinal Accel- Road Surface- Control Force Feedback- Vibration- Switch Feedback

Audio- Roadway- Alerts - Horns (ext) - ChimesClicks- Velocity (Wind)- Engine- Radio- Passenger

Olfactory- Gasoline- SmokeFire- Passenger

Other

Manual (Hand)- Wheel- Steering Column- Gear Shift- Switch (Panel)- Switch (Other)

Manual (Foot)- Throttle- Brake- Other

Voice

Other- Eye Tracking- Blink- Gesture- Thought

Driver

ldquoSituation Awarenessrdquo

bull Term originally defined for air combatbull Working Definition Sufficiently detailed mental picture of the

vehicle and environment (ie world model) to allow the operator to make well-informed (ie conditionally correct) decisions

Driver Situation AwarenessComponents

PersonalFactors

EquipmentStatus

VehiclePerformance

ControlSettings

Passengers

Non-drivingElements

Internal

SituationAwareness

Weather

AdjacentEnvironment

Signage

RoadwaySurface

LocationRoute

AdjacentTraffic

External

Endsley Situation Awareness Model

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Interaction Metaphors

bull Car as homebull Car as kitchenbull Car as bathroombull Car as bedroombull Car as music roombull Car as playroombull Car as entertainment ctrbull Car as toolboxbull Car as closet

bull Car as officebull Car as comm center

bull Car as image statementbull Car as clothingbull Car as jewelrybull Car as sports equipmentbull Car as safe spacebull Car as cocoon

Trends in Driver Attentional Demand

Driv

er A

tten

tion

al D

emDr

iver

Att

enti

onal

Dem

and

and

VehicleVehicleOperation OperationOperation Operation

SecondarySecondaryTasks Tasks

TimeTime

Time Spent in Vehicles (US)Time Spent in Vehicles (US)- Average 350 hrsyrperson Average 350 hrsyrperson - 500 Million hrsweek- 500 Million hrsweek

76 of Drivers Report Activities HaveldquoCausedNearly Causedrdquo an Accident

TALKING ON CELLPHONE

TURNING TO SPEAK

USING COMPUTER

CIGARETTE ASHES

BREAKING UP FIGHTBETWEEN KIDS

SPILLING COFFEE

Source Opinion Research Corp Interviews Time5800 (N=1016)

Concerns Regarding High SecondaryTask Loads

bull Growing Evidence and Public Perception of Safety Problem

bull Cell Phone use in USndash 115-120 Million Active Cell Phonesndash 50-70 Use in Vehiclesndash 39 of Drivers Using (Daylight Hours)

bull NE Journal of Medicine Estimatendash 4 fold increase in collision risk using cell phone

bull NHTSA Estimatesndash 12 Million Accidents (25-30) caused by Distracted

Driverbull Not limited to cell phones

bull Note These effects may be latent in normal operations and may only manifest in non-normal or emergency situations

Typical Performance vs TaskLoad Curve

Per

form

ance

Task Load

Distraction Components

bull Manualndash Inability or delay in operation of vehicle controlndash Hands occupied or out of position (cell phone)

bull Visualndash Head Down Problemsndash Visual Accommodationndash Visual Clutterndash Visual Compulsion

bull Cognitivendash Lack of cognitive engagement with primary Taskndash Latency in mental context shiftsndash Multi-tasking capabilitiesndash Prioritization

bull Emotionalbull High Individual Variability in Multi-Tasking Capability

Distraction Data Sources

bull Controlled Experimentsndash Vary Secondary Task Load (Independent Variable)

bull Visualbull Cognitive

ndash Measurebull Performancebull Response to Disturbancebull Situation Awareness

ndash Need to Increase Task Loading to Saturationndash Need to Include Unanticipated Eventsndash Need Lowest Common Denominator Population

bull Field Datandash Event Recordersndash Cell Phone Triggered

bull Subjective Survey Data

Controlled Experiment Issues (1)

bull Simulator Testingndash Controlled Scenarios +ndash Safe to go to Task Saturationndash Face and Cue Validity Issues ndashndash Simulator Sickness ndashndash Cost -

bull Dual Control Vehicle Testing (Test Track)ndash Good Validity +ndash Safety Issues at Task Saturation ndashndash Cost +

Controlled Experiment Issues (2)

bull Variability in Primary and Secondary Taskingndash How do you measure Secondary Task Loadndash How do you control Secondary Task Load

bull Performance Measuresndash Trackingndash Reaction Timendash Side Task Performance

bull Subjective Workload Measuresbull Situation Awareness Measures

ndash Testable Response Method

Simulator Studies of Driver CognitiveDistraction Caused by Cell Phone Use

bull Independent Variablesndash Cognitive Loadingndash Hands freeHands Fixed

bull Dependant Variablesndash Situation Awarenessndash Reaction Timendash Tracking MIT Age Lab Simulator

Cognitive Loading Levels

bull Lowndash Small talk

bull bullMediumndash Discuss movieoperaballet plot-line

bull bullHighndash Edit document by phone

Results Situational Awareness

bull Significant reduction in SA with Cell Phone Use

bull No significant difference between HFamp HH

Ave

rag

e F

ract

ion

Co

rrec

t

Normal HF HHPhone Setup

Stop Sign Response Time Resultsby Age Group

bull Significant effect of Age

bull No significant effect of Cell Phone Use

Re

sp

on

se

Tim

e [

se

c]

Normal1

Normal2

HFSmall

HFMovie

HF Edit HHSmall

HHMovie

HH Edit

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 14: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Acceleration vs Position

ldquoHuman Centeredrdquo InformationRequirements Analysis

bull Integrated Human Centered Systems Approachbull ldquoSemi-Structuredrdquo Decision Theorybull Driver Distraction Analysis

Driving Task Definition

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Vehicle Operation Tasks

bull Vehicle control tasks [skill based]ndash Lateral control (steering)ndash Longitudinal control (accel braking)

bull Tactical decisions [rule based]ndash Maneuveringndash Systems management

bull Strategic decisions [knowledge based]ndash Route selectionndash Goal management

bull Monitoring [skill rule knowledge]ndash Situation awareness

[Rasmussen Skill Rule Knowledge Hierarchy]

Lateral Tracking Loop

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

Driver InputOutput Modes

Visual- External Scene - Forward - Roadway - Traffic - Lights - Signage - Environment - Peripheral - Rear- Internal - Instrument Cluster

TactileProprioceptive- Lateral Accel- Longitudinal Accel- Road Surface- Control Force Feedback- Vibration- Switch Feedback

Audio- Roadway- Alerts - Horns (ext) - ChimesClicks- Velocity (Wind)- Engine- Radio- Passenger

Olfactory- Gasoline- SmokeFire- Passenger

Other

Manual (Hand)- Wheel- Steering Column- Gear Shift- Switch (Panel)- Switch (Other)

Manual (Foot)- Throttle- Brake- Other

Voice

Other- Eye Tracking- Blink- Gesture- Thought

Driver

ldquoSituation Awarenessrdquo

bull Term originally defined for air combatbull Working Definition Sufficiently detailed mental picture of the

vehicle and environment (ie world model) to allow the operator to make well-informed (ie conditionally correct) decisions

Driver Situation AwarenessComponents

PersonalFactors

EquipmentStatus

VehiclePerformance

ControlSettings

Passengers

Non-drivingElements

Internal

SituationAwareness

Weather

AdjacentEnvironment

Signage

RoadwaySurface

LocationRoute

AdjacentTraffic

External

Endsley Situation Awareness Model

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Interaction Metaphors

bull Car as homebull Car as kitchenbull Car as bathroombull Car as bedroombull Car as music roombull Car as playroombull Car as entertainment ctrbull Car as toolboxbull Car as closet

bull Car as officebull Car as comm center

bull Car as image statementbull Car as clothingbull Car as jewelrybull Car as sports equipmentbull Car as safe spacebull Car as cocoon

Trends in Driver Attentional Demand

Driv

er A

tten

tion

al D

emDr

iver

Att

enti

onal

Dem

and

and

VehicleVehicleOperation OperationOperation Operation

SecondarySecondaryTasks Tasks

TimeTime

Time Spent in Vehicles (US)Time Spent in Vehicles (US)- Average 350 hrsyrperson Average 350 hrsyrperson - 500 Million hrsweek- 500 Million hrsweek

76 of Drivers Report Activities HaveldquoCausedNearly Causedrdquo an Accident

TALKING ON CELLPHONE

TURNING TO SPEAK

USING COMPUTER

CIGARETTE ASHES

BREAKING UP FIGHTBETWEEN KIDS

SPILLING COFFEE

Source Opinion Research Corp Interviews Time5800 (N=1016)

Concerns Regarding High SecondaryTask Loads

bull Growing Evidence and Public Perception of Safety Problem

bull Cell Phone use in USndash 115-120 Million Active Cell Phonesndash 50-70 Use in Vehiclesndash 39 of Drivers Using (Daylight Hours)

bull NE Journal of Medicine Estimatendash 4 fold increase in collision risk using cell phone

bull NHTSA Estimatesndash 12 Million Accidents (25-30) caused by Distracted

Driverbull Not limited to cell phones

bull Note These effects may be latent in normal operations and may only manifest in non-normal or emergency situations

Typical Performance vs TaskLoad Curve

Per

form

ance

Task Load

Distraction Components

bull Manualndash Inability or delay in operation of vehicle controlndash Hands occupied or out of position (cell phone)

bull Visualndash Head Down Problemsndash Visual Accommodationndash Visual Clutterndash Visual Compulsion

bull Cognitivendash Lack of cognitive engagement with primary Taskndash Latency in mental context shiftsndash Multi-tasking capabilitiesndash Prioritization

bull Emotionalbull High Individual Variability in Multi-Tasking Capability

Distraction Data Sources

bull Controlled Experimentsndash Vary Secondary Task Load (Independent Variable)

bull Visualbull Cognitive

ndash Measurebull Performancebull Response to Disturbancebull Situation Awareness

ndash Need to Increase Task Loading to Saturationndash Need to Include Unanticipated Eventsndash Need Lowest Common Denominator Population

bull Field Datandash Event Recordersndash Cell Phone Triggered

bull Subjective Survey Data

Controlled Experiment Issues (1)

bull Simulator Testingndash Controlled Scenarios +ndash Safe to go to Task Saturationndash Face and Cue Validity Issues ndashndash Simulator Sickness ndashndash Cost -

bull Dual Control Vehicle Testing (Test Track)ndash Good Validity +ndash Safety Issues at Task Saturation ndashndash Cost +

Controlled Experiment Issues (2)

bull Variability in Primary and Secondary Taskingndash How do you measure Secondary Task Loadndash How do you control Secondary Task Load

bull Performance Measuresndash Trackingndash Reaction Timendash Side Task Performance

bull Subjective Workload Measuresbull Situation Awareness Measures

ndash Testable Response Method

Simulator Studies of Driver CognitiveDistraction Caused by Cell Phone Use

bull Independent Variablesndash Cognitive Loadingndash Hands freeHands Fixed

bull Dependant Variablesndash Situation Awarenessndash Reaction Timendash Tracking MIT Age Lab Simulator

Cognitive Loading Levels

bull Lowndash Small talk

bull bullMediumndash Discuss movieoperaballet plot-line

bull bullHighndash Edit document by phone

Results Situational Awareness

bull Significant reduction in SA with Cell Phone Use

bull No significant difference between HFamp HH

Ave

rag

e F

ract

ion

Co

rrec

t

Normal HF HHPhone Setup

Stop Sign Response Time Resultsby Age Group

bull Significant effect of Age

bull No significant effect of Cell Phone Use

Re

sp

on

se

Tim

e [

se

c]

Normal1

Normal2

HFSmall

HFMovie

HF Edit HHSmall

HHMovie

HH Edit

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 15: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

ldquoHuman Centeredrdquo InformationRequirements Analysis

bull Integrated Human Centered Systems Approachbull ldquoSemi-Structuredrdquo Decision Theorybull Driver Distraction Analysis

Driving Task Definition

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Vehicle Operation Tasks

bull Vehicle control tasks [skill based]ndash Lateral control (steering)ndash Longitudinal control (accel braking)

bull Tactical decisions [rule based]ndash Maneuveringndash Systems management

bull Strategic decisions [knowledge based]ndash Route selectionndash Goal management

bull Monitoring [skill rule knowledge]ndash Situation awareness

[Rasmussen Skill Rule Knowledge Hierarchy]

Lateral Tracking Loop

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

Driver InputOutput Modes

Visual- External Scene - Forward - Roadway - Traffic - Lights - Signage - Environment - Peripheral - Rear- Internal - Instrument Cluster

TactileProprioceptive- Lateral Accel- Longitudinal Accel- Road Surface- Control Force Feedback- Vibration- Switch Feedback

Audio- Roadway- Alerts - Horns (ext) - ChimesClicks- Velocity (Wind)- Engine- Radio- Passenger

Olfactory- Gasoline- SmokeFire- Passenger

Other

Manual (Hand)- Wheel- Steering Column- Gear Shift- Switch (Panel)- Switch (Other)

Manual (Foot)- Throttle- Brake- Other

Voice

Other- Eye Tracking- Blink- Gesture- Thought

Driver

ldquoSituation Awarenessrdquo

bull Term originally defined for air combatbull Working Definition Sufficiently detailed mental picture of the

vehicle and environment (ie world model) to allow the operator to make well-informed (ie conditionally correct) decisions

Driver Situation AwarenessComponents

PersonalFactors

EquipmentStatus

VehiclePerformance

ControlSettings

Passengers

Non-drivingElements

Internal

SituationAwareness

Weather

AdjacentEnvironment

Signage

RoadwaySurface

LocationRoute

AdjacentTraffic

External

Endsley Situation Awareness Model

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Interaction Metaphors

bull Car as homebull Car as kitchenbull Car as bathroombull Car as bedroombull Car as music roombull Car as playroombull Car as entertainment ctrbull Car as toolboxbull Car as closet

bull Car as officebull Car as comm center

bull Car as image statementbull Car as clothingbull Car as jewelrybull Car as sports equipmentbull Car as safe spacebull Car as cocoon

Trends in Driver Attentional Demand

Driv

er A

tten

tion

al D

emDr

iver

Att

enti

onal

Dem

and

and

VehicleVehicleOperation OperationOperation Operation

SecondarySecondaryTasks Tasks

TimeTime

Time Spent in Vehicles (US)Time Spent in Vehicles (US)- Average 350 hrsyrperson Average 350 hrsyrperson - 500 Million hrsweek- 500 Million hrsweek

76 of Drivers Report Activities HaveldquoCausedNearly Causedrdquo an Accident

TALKING ON CELLPHONE

TURNING TO SPEAK

USING COMPUTER

CIGARETTE ASHES

BREAKING UP FIGHTBETWEEN KIDS

SPILLING COFFEE

Source Opinion Research Corp Interviews Time5800 (N=1016)

Concerns Regarding High SecondaryTask Loads

bull Growing Evidence and Public Perception of Safety Problem

bull Cell Phone use in USndash 115-120 Million Active Cell Phonesndash 50-70 Use in Vehiclesndash 39 of Drivers Using (Daylight Hours)

bull NE Journal of Medicine Estimatendash 4 fold increase in collision risk using cell phone

bull NHTSA Estimatesndash 12 Million Accidents (25-30) caused by Distracted

Driverbull Not limited to cell phones

bull Note These effects may be latent in normal operations and may only manifest in non-normal or emergency situations

Typical Performance vs TaskLoad Curve

Per

form

ance

Task Load

Distraction Components

bull Manualndash Inability or delay in operation of vehicle controlndash Hands occupied or out of position (cell phone)

bull Visualndash Head Down Problemsndash Visual Accommodationndash Visual Clutterndash Visual Compulsion

bull Cognitivendash Lack of cognitive engagement with primary Taskndash Latency in mental context shiftsndash Multi-tasking capabilitiesndash Prioritization

bull Emotionalbull High Individual Variability in Multi-Tasking Capability

Distraction Data Sources

bull Controlled Experimentsndash Vary Secondary Task Load (Independent Variable)

bull Visualbull Cognitive

ndash Measurebull Performancebull Response to Disturbancebull Situation Awareness

ndash Need to Increase Task Loading to Saturationndash Need to Include Unanticipated Eventsndash Need Lowest Common Denominator Population

bull Field Datandash Event Recordersndash Cell Phone Triggered

bull Subjective Survey Data

Controlled Experiment Issues (1)

bull Simulator Testingndash Controlled Scenarios +ndash Safe to go to Task Saturationndash Face and Cue Validity Issues ndashndash Simulator Sickness ndashndash Cost -

bull Dual Control Vehicle Testing (Test Track)ndash Good Validity +ndash Safety Issues at Task Saturation ndashndash Cost +

Controlled Experiment Issues (2)

bull Variability in Primary and Secondary Taskingndash How do you measure Secondary Task Loadndash How do you control Secondary Task Load

bull Performance Measuresndash Trackingndash Reaction Timendash Side Task Performance

bull Subjective Workload Measuresbull Situation Awareness Measures

ndash Testable Response Method

Simulator Studies of Driver CognitiveDistraction Caused by Cell Phone Use

bull Independent Variablesndash Cognitive Loadingndash Hands freeHands Fixed

bull Dependant Variablesndash Situation Awarenessndash Reaction Timendash Tracking MIT Age Lab Simulator

Cognitive Loading Levels

bull Lowndash Small talk

bull bullMediumndash Discuss movieoperaballet plot-line

bull bullHighndash Edit document by phone

Results Situational Awareness

bull Significant reduction in SA with Cell Phone Use

bull No significant difference between HFamp HH

Ave

rag

e F

ract

ion

Co

rrec

t

Normal HF HHPhone Setup

Stop Sign Response Time Resultsby Age Group

bull Significant effect of Age

bull No significant effect of Cell Phone Use

Re

sp

on

se

Tim

e [

se

c]

Normal1

Normal2

HFSmall

HFMovie

HF Edit HHSmall

HHMovie

HH Edit

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 16: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Driving Task Definition

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Vehicle Operation Tasks

bull Vehicle control tasks [skill based]ndash Lateral control (steering)ndash Longitudinal control (accel braking)

bull Tactical decisions [rule based]ndash Maneuveringndash Systems management

bull Strategic decisions [knowledge based]ndash Route selectionndash Goal management

bull Monitoring [skill rule knowledge]ndash Situation awareness

[Rasmussen Skill Rule Knowledge Hierarchy]

Lateral Tracking Loop

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

Driver InputOutput Modes

Visual- External Scene - Forward - Roadway - Traffic - Lights - Signage - Environment - Peripheral - Rear- Internal - Instrument Cluster

TactileProprioceptive- Lateral Accel- Longitudinal Accel- Road Surface- Control Force Feedback- Vibration- Switch Feedback

Audio- Roadway- Alerts - Horns (ext) - ChimesClicks- Velocity (Wind)- Engine- Radio- Passenger

Olfactory- Gasoline- SmokeFire- Passenger

Other

Manual (Hand)- Wheel- Steering Column- Gear Shift- Switch (Panel)- Switch (Other)

Manual (Foot)- Throttle- Brake- Other

Voice

Other- Eye Tracking- Blink- Gesture- Thought

Driver

ldquoSituation Awarenessrdquo

bull Term originally defined for air combatbull Working Definition Sufficiently detailed mental picture of the

vehicle and environment (ie world model) to allow the operator to make well-informed (ie conditionally correct) decisions

Driver Situation AwarenessComponents

PersonalFactors

EquipmentStatus

VehiclePerformance

ControlSettings

Passengers

Non-drivingElements

Internal

SituationAwareness

Weather

AdjacentEnvironment

Signage

RoadwaySurface

LocationRoute

AdjacentTraffic

External

Endsley Situation Awareness Model

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Interaction Metaphors

bull Car as homebull Car as kitchenbull Car as bathroombull Car as bedroombull Car as music roombull Car as playroombull Car as entertainment ctrbull Car as toolboxbull Car as closet

bull Car as officebull Car as comm center

bull Car as image statementbull Car as clothingbull Car as jewelrybull Car as sports equipmentbull Car as safe spacebull Car as cocoon

Trends in Driver Attentional Demand

Driv

er A

tten

tion

al D

emDr

iver

Att

enti

onal

Dem

and

and

VehicleVehicleOperation OperationOperation Operation

SecondarySecondaryTasks Tasks

TimeTime

Time Spent in Vehicles (US)Time Spent in Vehicles (US)- Average 350 hrsyrperson Average 350 hrsyrperson - 500 Million hrsweek- 500 Million hrsweek

76 of Drivers Report Activities HaveldquoCausedNearly Causedrdquo an Accident

TALKING ON CELLPHONE

TURNING TO SPEAK

USING COMPUTER

CIGARETTE ASHES

BREAKING UP FIGHTBETWEEN KIDS

SPILLING COFFEE

Source Opinion Research Corp Interviews Time5800 (N=1016)

Concerns Regarding High SecondaryTask Loads

bull Growing Evidence and Public Perception of Safety Problem

bull Cell Phone use in USndash 115-120 Million Active Cell Phonesndash 50-70 Use in Vehiclesndash 39 of Drivers Using (Daylight Hours)

bull NE Journal of Medicine Estimatendash 4 fold increase in collision risk using cell phone

bull NHTSA Estimatesndash 12 Million Accidents (25-30) caused by Distracted

Driverbull Not limited to cell phones

bull Note These effects may be latent in normal operations and may only manifest in non-normal or emergency situations

Typical Performance vs TaskLoad Curve

Per

form

ance

Task Load

Distraction Components

bull Manualndash Inability or delay in operation of vehicle controlndash Hands occupied or out of position (cell phone)

bull Visualndash Head Down Problemsndash Visual Accommodationndash Visual Clutterndash Visual Compulsion

bull Cognitivendash Lack of cognitive engagement with primary Taskndash Latency in mental context shiftsndash Multi-tasking capabilitiesndash Prioritization

bull Emotionalbull High Individual Variability in Multi-Tasking Capability

Distraction Data Sources

bull Controlled Experimentsndash Vary Secondary Task Load (Independent Variable)

bull Visualbull Cognitive

ndash Measurebull Performancebull Response to Disturbancebull Situation Awareness

ndash Need to Increase Task Loading to Saturationndash Need to Include Unanticipated Eventsndash Need Lowest Common Denominator Population

bull Field Datandash Event Recordersndash Cell Phone Triggered

bull Subjective Survey Data

Controlled Experiment Issues (1)

bull Simulator Testingndash Controlled Scenarios +ndash Safe to go to Task Saturationndash Face and Cue Validity Issues ndashndash Simulator Sickness ndashndash Cost -

bull Dual Control Vehicle Testing (Test Track)ndash Good Validity +ndash Safety Issues at Task Saturation ndashndash Cost +

Controlled Experiment Issues (2)

bull Variability in Primary and Secondary Taskingndash How do you measure Secondary Task Loadndash How do you control Secondary Task Load

bull Performance Measuresndash Trackingndash Reaction Timendash Side Task Performance

bull Subjective Workload Measuresbull Situation Awareness Measures

ndash Testable Response Method

Simulator Studies of Driver CognitiveDistraction Caused by Cell Phone Use

bull Independent Variablesndash Cognitive Loadingndash Hands freeHands Fixed

bull Dependant Variablesndash Situation Awarenessndash Reaction Timendash Tracking MIT Age Lab Simulator

Cognitive Loading Levels

bull Lowndash Small talk

bull bullMediumndash Discuss movieoperaballet plot-line

bull bullHighndash Edit document by phone

Results Situational Awareness

bull Significant reduction in SA with Cell Phone Use

bull No significant difference between HFamp HH

Ave

rag

e F

ract

ion

Co

rrec

t

Normal HF HHPhone Setup

Stop Sign Response Time Resultsby Age Group

bull Significant effect of Age

bull No significant effect of Cell Phone Use

Re

sp

on

se

Tim

e [

se

c]

Normal1

Normal2

HFSmall

HFMovie

HF Edit HHSmall

HHMovie

HH Edit

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 17: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Vehicle Operation Tasks

bull Vehicle control tasks [skill based]ndash Lateral control (steering)ndash Longitudinal control (accel braking)

bull Tactical decisions [rule based]ndash Maneuveringndash Systems management

bull Strategic decisions [knowledge based]ndash Route selectionndash Goal management

bull Monitoring [skill rule knowledge]ndash Situation awareness

[Rasmussen Skill Rule Knowledge Hierarchy]

Lateral Tracking Loop

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

Driver InputOutput Modes

Visual- External Scene - Forward - Roadway - Traffic - Lights - Signage - Environment - Peripheral - Rear- Internal - Instrument Cluster

TactileProprioceptive- Lateral Accel- Longitudinal Accel- Road Surface- Control Force Feedback- Vibration- Switch Feedback

Audio- Roadway- Alerts - Horns (ext) - ChimesClicks- Velocity (Wind)- Engine- Radio- Passenger

Olfactory- Gasoline- SmokeFire- Passenger

Other

Manual (Hand)- Wheel- Steering Column- Gear Shift- Switch (Panel)- Switch (Other)

Manual (Foot)- Throttle- Brake- Other

Voice

Other- Eye Tracking- Blink- Gesture- Thought

Driver

ldquoSituation Awarenessrdquo

bull Term originally defined for air combatbull Working Definition Sufficiently detailed mental picture of the

vehicle and environment (ie world model) to allow the operator to make well-informed (ie conditionally correct) decisions

Driver Situation AwarenessComponents

PersonalFactors

EquipmentStatus

VehiclePerformance

ControlSettings

Passengers

Non-drivingElements

Internal

SituationAwareness

Weather

AdjacentEnvironment

Signage

RoadwaySurface

LocationRoute

AdjacentTraffic

External

Endsley Situation Awareness Model

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Interaction Metaphors

bull Car as homebull Car as kitchenbull Car as bathroombull Car as bedroombull Car as music roombull Car as playroombull Car as entertainment ctrbull Car as toolboxbull Car as closet

bull Car as officebull Car as comm center

bull Car as image statementbull Car as clothingbull Car as jewelrybull Car as sports equipmentbull Car as safe spacebull Car as cocoon

Trends in Driver Attentional Demand

Driv

er A

tten

tion

al D

emDr

iver

Att

enti

onal

Dem

and

and

VehicleVehicleOperation OperationOperation Operation

SecondarySecondaryTasks Tasks

TimeTime

Time Spent in Vehicles (US)Time Spent in Vehicles (US)- Average 350 hrsyrperson Average 350 hrsyrperson - 500 Million hrsweek- 500 Million hrsweek

76 of Drivers Report Activities HaveldquoCausedNearly Causedrdquo an Accident

TALKING ON CELLPHONE

TURNING TO SPEAK

USING COMPUTER

CIGARETTE ASHES

BREAKING UP FIGHTBETWEEN KIDS

SPILLING COFFEE

Source Opinion Research Corp Interviews Time5800 (N=1016)

Concerns Regarding High SecondaryTask Loads

bull Growing Evidence and Public Perception of Safety Problem

bull Cell Phone use in USndash 115-120 Million Active Cell Phonesndash 50-70 Use in Vehiclesndash 39 of Drivers Using (Daylight Hours)

bull NE Journal of Medicine Estimatendash 4 fold increase in collision risk using cell phone

bull NHTSA Estimatesndash 12 Million Accidents (25-30) caused by Distracted

Driverbull Not limited to cell phones

bull Note These effects may be latent in normal operations and may only manifest in non-normal or emergency situations

Typical Performance vs TaskLoad Curve

Per

form

ance

Task Load

Distraction Components

bull Manualndash Inability or delay in operation of vehicle controlndash Hands occupied or out of position (cell phone)

bull Visualndash Head Down Problemsndash Visual Accommodationndash Visual Clutterndash Visual Compulsion

bull Cognitivendash Lack of cognitive engagement with primary Taskndash Latency in mental context shiftsndash Multi-tasking capabilitiesndash Prioritization

bull Emotionalbull High Individual Variability in Multi-Tasking Capability

Distraction Data Sources

bull Controlled Experimentsndash Vary Secondary Task Load (Independent Variable)

bull Visualbull Cognitive

ndash Measurebull Performancebull Response to Disturbancebull Situation Awareness

ndash Need to Increase Task Loading to Saturationndash Need to Include Unanticipated Eventsndash Need Lowest Common Denominator Population

bull Field Datandash Event Recordersndash Cell Phone Triggered

bull Subjective Survey Data

Controlled Experiment Issues (1)

bull Simulator Testingndash Controlled Scenarios +ndash Safe to go to Task Saturationndash Face and Cue Validity Issues ndashndash Simulator Sickness ndashndash Cost -

bull Dual Control Vehicle Testing (Test Track)ndash Good Validity +ndash Safety Issues at Task Saturation ndashndash Cost +

Controlled Experiment Issues (2)

bull Variability in Primary and Secondary Taskingndash How do you measure Secondary Task Loadndash How do you control Secondary Task Load

bull Performance Measuresndash Trackingndash Reaction Timendash Side Task Performance

bull Subjective Workload Measuresbull Situation Awareness Measures

ndash Testable Response Method

Simulator Studies of Driver CognitiveDistraction Caused by Cell Phone Use

bull Independent Variablesndash Cognitive Loadingndash Hands freeHands Fixed

bull Dependant Variablesndash Situation Awarenessndash Reaction Timendash Tracking MIT Age Lab Simulator

Cognitive Loading Levels

bull Lowndash Small talk

bull bullMediumndash Discuss movieoperaballet plot-line

bull bullHighndash Edit document by phone

Results Situational Awareness

bull Significant reduction in SA with Cell Phone Use

bull No significant difference between HFamp HH

Ave

rag

e F

ract

ion

Co

rrec

t

Normal HF HHPhone Setup

Stop Sign Response Time Resultsby Age Group

bull Significant effect of Age

bull No significant effect of Cell Phone Use

Re

sp

on

se

Tim

e [

se

c]

Normal1

Normal2

HFSmall

HFMovie

HF Edit HHSmall

HHMovie

HH Edit

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 18: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Vehicle Operation Tasks

bull Vehicle control tasks [skill based]ndash Lateral control (steering)ndash Longitudinal control (accel braking)

bull Tactical decisions [rule based]ndash Maneuveringndash Systems management

bull Strategic decisions [knowledge based]ndash Route selectionndash Goal management

bull Monitoring [skill rule knowledge]ndash Situation awareness

[Rasmussen Skill Rule Knowledge Hierarchy]

Lateral Tracking Loop

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

Driver InputOutput Modes

Visual- External Scene - Forward - Roadway - Traffic - Lights - Signage - Environment - Peripheral - Rear- Internal - Instrument Cluster

TactileProprioceptive- Lateral Accel- Longitudinal Accel- Road Surface- Control Force Feedback- Vibration- Switch Feedback

Audio- Roadway- Alerts - Horns (ext) - ChimesClicks- Velocity (Wind)- Engine- Radio- Passenger

Olfactory- Gasoline- SmokeFire- Passenger

Other

Manual (Hand)- Wheel- Steering Column- Gear Shift- Switch (Panel)- Switch (Other)

Manual (Foot)- Throttle- Brake- Other

Voice

Other- Eye Tracking- Blink- Gesture- Thought

Driver

ldquoSituation Awarenessrdquo

bull Term originally defined for air combatbull Working Definition Sufficiently detailed mental picture of the

vehicle and environment (ie world model) to allow the operator to make well-informed (ie conditionally correct) decisions

Driver Situation AwarenessComponents

PersonalFactors

EquipmentStatus

VehiclePerformance

ControlSettings

Passengers

Non-drivingElements

Internal

SituationAwareness

Weather

AdjacentEnvironment

Signage

RoadwaySurface

LocationRoute

AdjacentTraffic

External

Endsley Situation Awareness Model

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Interaction Metaphors

bull Car as homebull Car as kitchenbull Car as bathroombull Car as bedroombull Car as music roombull Car as playroombull Car as entertainment ctrbull Car as toolboxbull Car as closet

bull Car as officebull Car as comm center

bull Car as image statementbull Car as clothingbull Car as jewelrybull Car as sports equipmentbull Car as safe spacebull Car as cocoon

Trends in Driver Attentional Demand

Driv

er A

tten

tion

al D

emDr

iver

Att

enti

onal

Dem

and

and

VehicleVehicleOperation OperationOperation Operation

SecondarySecondaryTasks Tasks

TimeTime

Time Spent in Vehicles (US)Time Spent in Vehicles (US)- Average 350 hrsyrperson Average 350 hrsyrperson - 500 Million hrsweek- 500 Million hrsweek

76 of Drivers Report Activities HaveldquoCausedNearly Causedrdquo an Accident

TALKING ON CELLPHONE

TURNING TO SPEAK

USING COMPUTER

CIGARETTE ASHES

BREAKING UP FIGHTBETWEEN KIDS

SPILLING COFFEE

Source Opinion Research Corp Interviews Time5800 (N=1016)

Concerns Regarding High SecondaryTask Loads

bull Growing Evidence and Public Perception of Safety Problem

bull Cell Phone use in USndash 115-120 Million Active Cell Phonesndash 50-70 Use in Vehiclesndash 39 of Drivers Using (Daylight Hours)

bull NE Journal of Medicine Estimatendash 4 fold increase in collision risk using cell phone

bull NHTSA Estimatesndash 12 Million Accidents (25-30) caused by Distracted

Driverbull Not limited to cell phones

bull Note These effects may be latent in normal operations and may only manifest in non-normal or emergency situations

Typical Performance vs TaskLoad Curve

Per

form

ance

Task Load

Distraction Components

bull Manualndash Inability or delay in operation of vehicle controlndash Hands occupied or out of position (cell phone)

bull Visualndash Head Down Problemsndash Visual Accommodationndash Visual Clutterndash Visual Compulsion

bull Cognitivendash Lack of cognitive engagement with primary Taskndash Latency in mental context shiftsndash Multi-tasking capabilitiesndash Prioritization

bull Emotionalbull High Individual Variability in Multi-Tasking Capability

Distraction Data Sources

bull Controlled Experimentsndash Vary Secondary Task Load (Independent Variable)

bull Visualbull Cognitive

ndash Measurebull Performancebull Response to Disturbancebull Situation Awareness

ndash Need to Increase Task Loading to Saturationndash Need to Include Unanticipated Eventsndash Need Lowest Common Denominator Population

bull Field Datandash Event Recordersndash Cell Phone Triggered

bull Subjective Survey Data

Controlled Experiment Issues (1)

bull Simulator Testingndash Controlled Scenarios +ndash Safe to go to Task Saturationndash Face and Cue Validity Issues ndashndash Simulator Sickness ndashndash Cost -

bull Dual Control Vehicle Testing (Test Track)ndash Good Validity +ndash Safety Issues at Task Saturation ndashndash Cost +

Controlled Experiment Issues (2)

bull Variability in Primary and Secondary Taskingndash How do you measure Secondary Task Loadndash How do you control Secondary Task Load

bull Performance Measuresndash Trackingndash Reaction Timendash Side Task Performance

bull Subjective Workload Measuresbull Situation Awareness Measures

ndash Testable Response Method

Simulator Studies of Driver CognitiveDistraction Caused by Cell Phone Use

bull Independent Variablesndash Cognitive Loadingndash Hands freeHands Fixed

bull Dependant Variablesndash Situation Awarenessndash Reaction Timendash Tracking MIT Age Lab Simulator

Cognitive Loading Levels

bull Lowndash Small talk

bull bullMediumndash Discuss movieoperaballet plot-line

bull bullHighndash Edit document by phone

Results Situational Awareness

bull Significant reduction in SA with Cell Phone Use

bull No significant difference between HFamp HH

Ave

rag

e F

ract

ion

Co

rrec

t

Normal HF HHPhone Setup

Stop Sign Response Time Resultsby Age Group

bull Significant effect of Age

bull No significant effect of Cell Phone Use

Re

sp

on

se

Tim

e [

se

c]

Normal1

Normal2

HFSmall

HFMovie

HF Edit HHSmall

HHMovie

HH Edit

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 19: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Lateral Tracking Loop

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

Driver InputOutput Modes

Visual- External Scene - Forward - Roadway - Traffic - Lights - Signage - Environment - Peripheral - Rear- Internal - Instrument Cluster

TactileProprioceptive- Lateral Accel- Longitudinal Accel- Road Surface- Control Force Feedback- Vibration- Switch Feedback

Audio- Roadway- Alerts - Horns (ext) - ChimesClicks- Velocity (Wind)- Engine- Radio- Passenger

Olfactory- Gasoline- SmokeFire- Passenger

Other

Manual (Hand)- Wheel- Steering Column- Gear Shift- Switch (Panel)- Switch (Other)

Manual (Foot)- Throttle- Brake- Other

Voice

Other- Eye Tracking- Blink- Gesture- Thought

Driver

ldquoSituation Awarenessrdquo

bull Term originally defined for air combatbull Working Definition Sufficiently detailed mental picture of the

vehicle and environment (ie world model) to allow the operator to make well-informed (ie conditionally correct) decisions

Driver Situation AwarenessComponents

PersonalFactors

EquipmentStatus

VehiclePerformance

ControlSettings

Passengers

Non-drivingElements

Internal

SituationAwareness

Weather

AdjacentEnvironment

Signage

RoadwaySurface

LocationRoute

AdjacentTraffic

External

Endsley Situation Awareness Model

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Interaction Metaphors

bull Car as homebull Car as kitchenbull Car as bathroombull Car as bedroombull Car as music roombull Car as playroombull Car as entertainment ctrbull Car as toolboxbull Car as closet

bull Car as officebull Car as comm center

bull Car as image statementbull Car as clothingbull Car as jewelrybull Car as sports equipmentbull Car as safe spacebull Car as cocoon

Trends in Driver Attentional Demand

Driv

er A

tten

tion

al D

emDr

iver

Att

enti

onal

Dem

and

and

VehicleVehicleOperation OperationOperation Operation

SecondarySecondaryTasks Tasks

TimeTime

Time Spent in Vehicles (US)Time Spent in Vehicles (US)- Average 350 hrsyrperson Average 350 hrsyrperson - 500 Million hrsweek- 500 Million hrsweek

76 of Drivers Report Activities HaveldquoCausedNearly Causedrdquo an Accident

TALKING ON CELLPHONE

TURNING TO SPEAK

USING COMPUTER

CIGARETTE ASHES

BREAKING UP FIGHTBETWEEN KIDS

SPILLING COFFEE

Source Opinion Research Corp Interviews Time5800 (N=1016)

Concerns Regarding High SecondaryTask Loads

bull Growing Evidence and Public Perception of Safety Problem

bull Cell Phone use in USndash 115-120 Million Active Cell Phonesndash 50-70 Use in Vehiclesndash 39 of Drivers Using (Daylight Hours)

bull NE Journal of Medicine Estimatendash 4 fold increase in collision risk using cell phone

bull NHTSA Estimatesndash 12 Million Accidents (25-30) caused by Distracted

Driverbull Not limited to cell phones

bull Note These effects may be latent in normal operations and may only manifest in non-normal or emergency situations

Typical Performance vs TaskLoad Curve

Per

form

ance

Task Load

Distraction Components

bull Manualndash Inability or delay in operation of vehicle controlndash Hands occupied or out of position (cell phone)

bull Visualndash Head Down Problemsndash Visual Accommodationndash Visual Clutterndash Visual Compulsion

bull Cognitivendash Lack of cognitive engagement with primary Taskndash Latency in mental context shiftsndash Multi-tasking capabilitiesndash Prioritization

bull Emotionalbull High Individual Variability in Multi-Tasking Capability

Distraction Data Sources

bull Controlled Experimentsndash Vary Secondary Task Load (Independent Variable)

bull Visualbull Cognitive

ndash Measurebull Performancebull Response to Disturbancebull Situation Awareness

ndash Need to Increase Task Loading to Saturationndash Need to Include Unanticipated Eventsndash Need Lowest Common Denominator Population

bull Field Datandash Event Recordersndash Cell Phone Triggered

bull Subjective Survey Data

Controlled Experiment Issues (1)

bull Simulator Testingndash Controlled Scenarios +ndash Safe to go to Task Saturationndash Face and Cue Validity Issues ndashndash Simulator Sickness ndashndash Cost -

bull Dual Control Vehicle Testing (Test Track)ndash Good Validity +ndash Safety Issues at Task Saturation ndashndash Cost +

Controlled Experiment Issues (2)

bull Variability in Primary and Secondary Taskingndash How do you measure Secondary Task Loadndash How do you control Secondary Task Load

bull Performance Measuresndash Trackingndash Reaction Timendash Side Task Performance

bull Subjective Workload Measuresbull Situation Awareness Measures

ndash Testable Response Method

Simulator Studies of Driver CognitiveDistraction Caused by Cell Phone Use

bull Independent Variablesndash Cognitive Loadingndash Hands freeHands Fixed

bull Dependant Variablesndash Situation Awarenessndash Reaction Timendash Tracking MIT Age Lab Simulator

Cognitive Loading Levels

bull Lowndash Small talk

bull bullMediumndash Discuss movieoperaballet plot-line

bull bullHighndash Edit document by phone

Results Situational Awareness

bull Significant reduction in SA with Cell Phone Use

bull No significant difference between HFamp HH

Ave

rag

e F

ract

ion

Co

rrec

t

Normal HF HHPhone Setup

Stop Sign Response Time Resultsby Age Group

bull Significant effect of Age

bull No significant effect of Cell Phone Use

Re

sp

on

se

Tim

e [

se

c]

Normal1

Normal2

HFSmall

HFMovie

HF Edit HHSmall

HHMovie

HH Edit

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 20: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Driver InputOutput Modes

Visual- External Scene - Forward - Roadway - Traffic - Lights - Signage - Environment - Peripheral - Rear- Internal - Instrument Cluster

TactileProprioceptive- Lateral Accel- Longitudinal Accel- Road Surface- Control Force Feedback- Vibration- Switch Feedback

Audio- Roadway- Alerts - Horns (ext) - ChimesClicks- Velocity (Wind)- Engine- Radio- Passenger

Olfactory- Gasoline- SmokeFire- Passenger

Other

Manual (Hand)- Wheel- Steering Column- Gear Shift- Switch (Panel)- Switch (Other)

Manual (Foot)- Throttle- Brake- Other

Voice

Other- Eye Tracking- Blink- Gesture- Thought

Driver

ldquoSituation Awarenessrdquo

bull Term originally defined for air combatbull Working Definition Sufficiently detailed mental picture of the

vehicle and environment (ie world model) to allow the operator to make well-informed (ie conditionally correct) decisions

Driver Situation AwarenessComponents

PersonalFactors

EquipmentStatus

VehiclePerformance

ControlSettings

Passengers

Non-drivingElements

Internal

SituationAwareness

Weather

AdjacentEnvironment

Signage

RoadwaySurface

LocationRoute

AdjacentTraffic

External

Endsley Situation Awareness Model

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Interaction Metaphors

bull Car as homebull Car as kitchenbull Car as bathroombull Car as bedroombull Car as music roombull Car as playroombull Car as entertainment ctrbull Car as toolboxbull Car as closet

bull Car as officebull Car as comm center

bull Car as image statementbull Car as clothingbull Car as jewelrybull Car as sports equipmentbull Car as safe spacebull Car as cocoon

Trends in Driver Attentional Demand

Driv

er A

tten

tion

al D

emDr

iver

Att

enti

onal

Dem

and

and

VehicleVehicleOperation OperationOperation Operation

SecondarySecondaryTasks Tasks

TimeTime

Time Spent in Vehicles (US)Time Spent in Vehicles (US)- Average 350 hrsyrperson Average 350 hrsyrperson - 500 Million hrsweek- 500 Million hrsweek

76 of Drivers Report Activities HaveldquoCausedNearly Causedrdquo an Accident

TALKING ON CELLPHONE

TURNING TO SPEAK

USING COMPUTER

CIGARETTE ASHES

BREAKING UP FIGHTBETWEEN KIDS

SPILLING COFFEE

Source Opinion Research Corp Interviews Time5800 (N=1016)

Concerns Regarding High SecondaryTask Loads

bull Growing Evidence and Public Perception of Safety Problem

bull Cell Phone use in USndash 115-120 Million Active Cell Phonesndash 50-70 Use in Vehiclesndash 39 of Drivers Using (Daylight Hours)

bull NE Journal of Medicine Estimatendash 4 fold increase in collision risk using cell phone

bull NHTSA Estimatesndash 12 Million Accidents (25-30) caused by Distracted

Driverbull Not limited to cell phones

bull Note These effects may be latent in normal operations and may only manifest in non-normal or emergency situations

Typical Performance vs TaskLoad Curve

Per

form

ance

Task Load

Distraction Components

bull Manualndash Inability or delay in operation of vehicle controlndash Hands occupied or out of position (cell phone)

bull Visualndash Head Down Problemsndash Visual Accommodationndash Visual Clutterndash Visual Compulsion

bull Cognitivendash Lack of cognitive engagement with primary Taskndash Latency in mental context shiftsndash Multi-tasking capabilitiesndash Prioritization

bull Emotionalbull High Individual Variability in Multi-Tasking Capability

Distraction Data Sources

bull Controlled Experimentsndash Vary Secondary Task Load (Independent Variable)

bull Visualbull Cognitive

ndash Measurebull Performancebull Response to Disturbancebull Situation Awareness

ndash Need to Increase Task Loading to Saturationndash Need to Include Unanticipated Eventsndash Need Lowest Common Denominator Population

bull Field Datandash Event Recordersndash Cell Phone Triggered

bull Subjective Survey Data

Controlled Experiment Issues (1)

bull Simulator Testingndash Controlled Scenarios +ndash Safe to go to Task Saturationndash Face and Cue Validity Issues ndashndash Simulator Sickness ndashndash Cost -

bull Dual Control Vehicle Testing (Test Track)ndash Good Validity +ndash Safety Issues at Task Saturation ndashndash Cost +

Controlled Experiment Issues (2)

bull Variability in Primary and Secondary Taskingndash How do you measure Secondary Task Loadndash How do you control Secondary Task Load

bull Performance Measuresndash Trackingndash Reaction Timendash Side Task Performance

bull Subjective Workload Measuresbull Situation Awareness Measures

ndash Testable Response Method

Simulator Studies of Driver CognitiveDistraction Caused by Cell Phone Use

bull Independent Variablesndash Cognitive Loadingndash Hands freeHands Fixed

bull Dependant Variablesndash Situation Awarenessndash Reaction Timendash Tracking MIT Age Lab Simulator

Cognitive Loading Levels

bull Lowndash Small talk

bull bullMediumndash Discuss movieoperaballet plot-line

bull bullHighndash Edit document by phone

Results Situational Awareness

bull Significant reduction in SA with Cell Phone Use

bull No significant difference between HFamp HH

Ave

rag

e F

ract

ion

Co

rrec

t

Normal HF HHPhone Setup

Stop Sign Response Time Resultsby Age Group

bull Significant effect of Age

bull No significant effect of Cell Phone Use

Re

sp

on

se

Tim

e [

se

c]

Normal1

Normal2

HFSmall

HFMovie

HF Edit HHSmall

HHMovie

HH Edit

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 21: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

ldquoSituation Awarenessrdquo

bull Term originally defined for air combatbull Working Definition Sufficiently detailed mental picture of the

vehicle and environment (ie world model) to allow the operator to make well-informed (ie conditionally correct) decisions

Driver Situation AwarenessComponents

PersonalFactors

EquipmentStatus

VehiclePerformance

ControlSettings

Passengers

Non-drivingElements

Internal

SituationAwareness

Weather

AdjacentEnvironment

Signage

RoadwaySurface

LocationRoute

AdjacentTraffic

External

Endsley Situation Awareness Model

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Interaction Metaphors

bull Car as homebull Car as kitchenbull Car as bathroombull Car as bedroombull Car as music roombull Car as playroombull Car as entertainment ctrbull Car as toolboxbull Car as closet

bull Car as officebull Car as comm center

bull Car as image statementbull Car as clothingbull Car as jewelrybull Car as sports equipmentbull Car as safe spacebull Car as cocoon

Trends in Driver Attentional Demand

Driv

er A

tten

tion

al D

emDr

iver

Att

enti

onal

Dem

and

and

VehicleVehicleOperation OperationOperation Operation

SecondarySecondaryTasks Tasks

TimeTime

Time Spent in Vehicles (US)Time Spent in Vehicles (US)- Average 350 hrsyrperson Average 350 hrsyrperson - 500 Million hrsweek- 500 Million hrsweek

76 of Drivers Report Activities HaveldquoCausedNearly Causedrdquo an Accident

TALKING ON CELLPHONE

TURNING TO SPEAK

USING COMPUTER

CIGARETTE ASHES

BREAKING UP FIGHTBETWEEN KIDS

SPILLING COFFEE

Source Opinion Research Corp Interviews Time5800 (N=1016)

Concerns Regarding High SecondaryTask Loads

bull Growing Evidence and Public Perception of Safety Problem

bull Cell Phone use in USndash 115-120 Million Active Cell Phonesndash 50-70 Use in Vehiclesndash 39 of Drivers Using (Daylight Hours)

bull NE Journal of Medicine Estimatendash 4 fold increase in collision risk using cell phone

bull NHTSA Estimatesndash 12 Million Accidents (25-30) caused by Distracted

Driverbull Not limited to cell phones

bull Note These effects may be latent in normal operations and may only manifest in non-normal or emergency situations

Typical Performance vs TaskLoad Curve

Per

form

ance

Task Load

Distraction Components

bull Manualndash Inability or delay in operation of vehicle controlndash Hands occupied or out of position (cell phone)

bull Visualndash Head Down Problemsndash Visual Accommodationndash Visual Clutterndash Visual Compulsion

bull Cognitivendash Lack of cognitive engagement with primary Taskndash Latency in mental context shiftsndash Multi-tasking capabilitiesndash Prioritization

bull Emotionalbull High Individual Variability in Multi-Tasking Capability

Distraction Data Sources

bull Controlled Experimentsndash Vary Secondary Task Load (Independent Variable)

bull Visualbull Cognitive

ndash Measurebull Performancebull Response to Disturbancebull Situation Awareness

ndash Need to Increase Task Loading to Saturationndash Need to Include Unanticipated Eventsndash Need Lowest Common Denominator Population

bull Field Datandash Event Recordersndash Cell Phone Triggered

bull Subjective Survey Data

Controlled Experiment Issues (1)

bull Simulator Testingndash Controlled Scenarios +ndash Safe to go to Task Saturationndash Face and Cue Validity Issues ndashndash Simulator Sickness ndashndash Cost -

bull Dual Control Vehicle Testing (Test Track)ndash Good Validity +ndash Safety Issues at Task Saturation ndashndash Cost +

Controlled Experiment Issues (2)

bull Variability in Primary and Secondary Taskingndash How do you measure Secondary Task Loadndash How do you control Secondary Task Load

bull Performance Measuresndash Trackingndash Reaction Timendash Side Task Performance

bull Subjective Workload Measuresbull Situation Awareness Measures

ndash Testable Response Method

Simulator Studies of Driver CognitiveDistraction Caused by Cell Phone Use

bull Independent Variablesndash Cognitive Loadingndash Hands freeHands Fixed

bull Dependant Variablesndash Situation Awarenessndash Reaction Timendash Tracking MIT Age Lab Simulator

Cognitive Loading Levels

bull Lowndash Small talk

bull bullMediumndash Discuss movieoperaballet plot-line

bull bullHighndash Edit document by phone

Results Situational Awareness

bull Significant reduction in SA with Cell Phone Use

bull No significant difference between HFamp HH

Ave

rag

e F

ract

ion

Co

rrec

t

Normal HF HHPhone Setup

Stop Sign Response Time Resultsby Age Group

bull Significant effect of Age

bull No significant effect of Cell Phone Use

Re

sp

on

se

Tim

e [

se

c]

Normal1

Normal2

HFSmall

HFMovie

HF Edit HHSmall

HHMovie

HH Edit

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 22: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Driver Situation AwarenessComponents

PersonalFactors

EquipmentStatus

VehiclePerformance

ControlSettings

Passengers

Non-drivingElements

Internal

SituationAwareness

Weather

AdjacentEnvironment

Signage

RoadwaySurface

LocationRoute

AdjacentTraffic

External

Endsley Situation Awareness Model

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Interaction Metaphors

bull Car as homebull Car as kitchenbull Car as bathroombull Car as bedroombull Car as music roombull Car as playroombull Car as entertainment ctrbull Car as toolboxbull Car as closet

bull Car as officebull Car as comm center

bull Car as image statementbull Car as clothingbull Car as jewelrybull Car as sports equipmentbull Car as safe spacebull Car as cocoon

Trends in Driver Attentional Demand

Driv

er A

tten

tion

al D

emDr

iver

Att

enti

onal

Dem

and

and

VehicleVehicleOperation OperationOperation Operation

SecondarySecondaryTasks Tasks

TimeTime

Time Spent in Vehicles (US)Time Spent in Vehicles (US)- Average 350 hrsyrperson Average 350 hrsyrperson - 500 Million hrsweek- 500 Million hrsweek

76 of Drivers Report Activities HaveldquoCausedNearly Causedrdquo an Accident

TALKING ON CELLPHONE

TURNING TO SPEAK

USING COMPUTER

CIGARETTE ASHES

BREAKING UP FIGHTBETWEEN KIDS

SPILLING COFFEE

Source Opinion Research Corp Interviews Time5800 (N=1016)

Concerns Regarding High SecondaryTask Loads

bull Growing Evidence and Public Perception of Safety Problem

bull Cell Phone use in USndash 115-120 Million Active Cell Phonesndash 50-70 Use in Vehiclesndash 39 of Drivers Using (Daylight Hours)

bull NE Journal of Medicine Estimatendash 4 fold increase in collision risk using cell phone

bull NHTSA Estimatesndash 12 Million Accidents (25-30) caused by Distracted

Driverbull Not limited to cell phones

bull Note These effects may be latent in normal operations and may only manifest in non-normal or emergency situations

Typical Performance vs TaskLoad Curve

Per

form

ance

Task Load

Distraction Components

bull Manualndash Inability or delay in operation of vehicle controlndash Hands occupied or out of position (cell phone)

bull Visualndash Head Down Problemsndash Visual Accommodationndash Visual Clutterndash Visual Compulsion

bull Cognitivendash Lack of cognitive engagement with primary Taskndash Latency in mental context shiftsndash Multi-tasking capabilitiesndash Prioritization

bull Emotionalbull High Individual Variability in Multi-Tasking Capability

Distraction Data Sources

bull Controlled Experimentsndash Vary Secondary Task Load (Independent Variable)

bull Visualbull Cognitive

ndash Measurebull Performancebull Response to Disturbancebull Situation Awareness

ndash Need to Increase Task Loading to Saturationndash Need to Include Unanticipated Eventsndash Need Lowest Common Denominator Population

bull Field Datandash Event Recordersndash Cell Phone Triggered

bull Subjective Survey Data

Controlled Experiment Issues (1)

bull Simulator Testingndash Controlled Scenarios +ndash Safe to go to Task Saturationndash Face and Cue Validity Issues ndashndash Simulator Sickness ndashndash Cost -

bull Dual Control Vehicle Testing (Test Track)ndash Good Validity +ndash Safety Issues at Task Saturation ndashndash Cost +

Controlled Experiment Issues (2)

bull Variability in Primary and Secondary Taskingndash How do you measure Secondary Task Loadndash How do you control Secondary Task Load

bull Performance Measuresndash Trackingndash Reaction Timendash Side Task Performance

bull Subjective Workload Measuresbull Situation Awareness Measures

ndash Testable Response Method

Simulator Studies of Driver CognitiveDistraction Caused by Cell Phone Use

bull Independent Variablesndash Cognitive Loadingndash Hands freeHands Fixed

bull Dependant Variablesndash Situation Awarenessndash Reaction Timendash Tracking MIT Age Lab Simulator

Cognitive Loading Levels

bull Lowndash Small talk

bull bullMediumndash Discuss movieoperaballet plot-line

bull bullHighndash Edit document by phone

Results Situational Awareness

bull Significant reduction in SA with Cell Phone Use

bull No significant difference between HFamp HH

Ave

rag

e F

ract

ion

Co

rrec

t

Normal HF HHPhone Setup

Stop Sign Response Time Resultsby Age Group

bull Significant effect of Age

bull No significant effect of Cell Phone Use

Re

sp

on

se

Tim

e [

se

c]

Normal1

Normal2

HFSmall

HFMovie

HF Edit HHSmall

HHMovie

HH Edit

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 23: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Endsley Situation Awareness Model

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Interaction Metaphors

bull Car as homebull Car as kitchenbull Car as bathroombull Car as bedroombull Car as music roombull Car as playroombull Car as entertainment ctrbull Car as toolboxbull Car as closet

bull Car as officebull Car as comm center

bull Car as image statementbull Car as clothingbull Car as jewelrybull Car as sports equipmentbull Car as safe spacebull Car as cocoon

Trends in Driver Attentional Demand

Driv

er A

tten

tion

al D

emDr

iver

Att

enti

onal

Dem

and

and

VehicleVehicleOperation OperationOperation Operation

SecondarySecondaryTasks Tasks

TimeTime

Time Spent in Vehicles (US)Time Spent in Vehicles (US)- Average 350 hrsyrperson Average 350 hrsyrperson - 500 Million hrsweek- 500 Million hrsweek

76 of Drivers Report Activities HaveldquoCausedNearly Causedrdquo an Accident

TALKING ON CELLPHONE

TURNING TO SPEAK

USING COMPUTER

CIGARETTE ASHES

BREAKING UP FIGHTBETWEEN KIDS

SPILLING COFFEE

Source Opinion Research Corp Interviews Time5800 (N=1016)

Concerns Regarding High SecondaryTask Loads

bull Growing Evidence and Public Perception of Safety Problem

bull Cell Phone use in USndash 115-120 Million Active Cell Phonesndash 50-70 Use in Vehiclesndash 39 of Drivers Using (Daylight Hours)

bull NE Journal of Medicine Estimatendash 4 fold increase in collision risk using cell phone

bull NHTSA Estimatesndash 12 Million Accidents (25-30) caused by Distracted

Driverbull Not limited to cell phones

bull Note These effects may be latent in normal operations and may only manifest in non-normal or emergency situations

Typical Performance vs TaskLoad Curve

Per

form

ance

Task Load

Distraction Components

bull Manualndash Inability or delay in operation of vehicle controlndash Hands occupied or out of position (cell phone)

bull Visualndash Head Down Problemsndash Visual Accommodationndash Visual Clutterndash Visual Compulsion

bull Cognitivendash Lack of cognitive engagement with primary Taskndash Latency in mental context shiftsndash Multi-tasking capabilitiesndash Prioritization

bull Emotionalbull High Individual Variability in Multi-Tasking Capability

Distraction Data Sources

bull Controlled Experimentsndash Vary Secondary Task Load (Independent Variable)

bull Visualbull Cognitive

ndash Measurebull Performancebull Response to Disturbancebull Situation Awareness

ndash Need to Increase Task Loading to Saturationndash Need to Include Unanticipated Eventsndash Need Lowest Common Denominator Population

bull Field Datandash Event Recordersndash Cell Phone Triggered

bull Subjective Survey Data

Controlled Experiment Issues (1)

bull Simulator Testingndash Controlled Scenarios +ndash Safe to go to Task Saturationndash Face and Cue Validity Issues ndashndash Simulator Sickness ndashndash Cost -

bull Dual Control Vehicle Testing (Test Track)ndash Good Validity +ndash Safety Issues at Task Saturation ndashndash Cost +

Controlled Experiment Issues (2)

bull Variability in Primary and Secondary Taskingndash How do you measure Secondary Task Loadndash How do you control Secondary Task Load

bull Performance Measuresndash Trackingndash Reaction Timendash Side Task Performance

bull Subjective Workload Measuresbull Situation Awareness Measures

ndash Testable Response Method

Simulator Studies of Driver CognitiveDistraction Caused by Cell Phone Use

bull Independent Variablesndash Cognitive Loadingndash Hands freeHands Fixed

bull Dependant Variablesndash Situation Awarenessndash Reaction Timendash Tracking MIT Age Lab Simulator

Cognitive Loading Levels

bull Lowndash Small talk

bull bullMediumndash Discuss movieoperaballet plot-line

bull bullHighndash Edit document by phone

Results Situational Awareness

bull Significant reduction in SA with Cell Phone Use

bull No significant difference between HFamp HH

Ave

rag

e F

ract

ion

Co

rrec

t

Normal HF HHPhone Setup

Stop Sign Response Time Resultsby Age Group

bull Significant effect of Age

bull No significant effect of Cell Phone Use

Re

sp

on

se

Tim

e [

se

c]

Normal1

Normal2

HFSmall

HFMovie

HF Edit HHSmall

HHMovie

HH Edit

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 24: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Primary Task Analysis

DistractionPotential

DriverDriver

Everything Else(Secondary Tasks)

SecondarySystem

(eg cell phone)

VehicleOperation Task(Primary Task)

Vehicle

Interaction Metaphors

bull Car as homebull Car as kitchenbull Car as bathroombull Car as bedroombull Car as music roombull Car as playroombull Car as entertainment ctrbull Car as toolboxbull Car as closet

bull Car as officebull Car as comm center

bull Car as image statementbull Car as clothingbull Car as jewelrybull Car as sports equipmentbull Car as safe spacebull Car as cocoon

Trends in Driver Attentional Demand

Driv

er A

tten

tion

al D

emDr

iver

Att

enti

onal

Dem

and

and

VehicleVehicleOperation OperationOperation Operation

SecondarySecondaryTasks Tasks

TimeTime

Time Spent in Vehicles (US)Time Spent in Vehicles (US)- Average 350 hrsyrperson Average 350 hrsyrperson - 500 Million hrsweek- 500 Million hrsweek

76 of Drivers Report Activities HaveldquoCausedNearly Causedrdquo an Accident

TALKING ON CELLPHONE

TURNING TO SPEAK

USING COMPUTER

CIGARETTE ASHES

BREAKING UP FIGHTBETWEEN KIDS

SPILLING COFFEE

Source Opinion Research Corp Interviews Time5800 (N=1016)

Concerns Regarding High SecondaryTask Loads

bull Growing Evidence and Public Perception of Safety Problem

bull Cell Phone use in USndash 115-120 Million Active Cell Phonesndash 50-70 Use in Vehiclesndash 39 of Drivers Using (Daylight Hours)

bull NE Journal of Medicine Estimatendash 4 fold increase in collision risk using cell phone

bull NHTSA Estimatesndash 12 Million Accidents (25-30) caused by Distracted

Driverbull Not limited to cell phones

bull Note These effects may be latent in normal operations and may only manifest in non-normal or emergency situations

Typical Performance vs TaskLoad Curve

Per

form

ance

Task Load

Distraction Components

bull Manualndash Inability or delay in operation of vehicle controlndash Hands occupied or out of position (cell phone)

bull Visualndash Head Down Problemsndash Visual Accommodationndash Visual Clutterndash Visual Compulsion

bull Cognitivendash Lack of cognitive engagement with primary Taskndash Latency in mental context shiftsndash Multi-tasking capabilitiesndash Prioritization

bull Emotionalbull High Individual Variability in Multi-Tasking Capability

Distraction Data Sources

bull Controlled Experimentsndash Vary Secondary Task Load (Independent Variable)

bull Visualbull Cognitive

ndash Measurebull Performancebull Response to Disturbancebull Situation Awareness

ndash Need to Increase Task Loading to Saturationndash Need to Include Unanticipated Eventsndash Need Lowest Common Denominator Population

bull Field Datandash Event Recordersndash Cell Phone Triggered

bull Subjective Survey Data

Controlled Experiment Issues (1)

bull Simulator Testingndash Controlled Scenarios +ndash Safe to go to Task Saturationndash Face and Cue Validity Issues ndashndash Simulator Sickness ndashndash Cost -

bull Dual Control Vehicle Testing (Test Track)ndash Good Validity +ndash Safety Issues at Task Saturation ndashndash Cost +

Controlled Experiment Issues (2)

bull Variability in Primary and Secondary Taskingndash How do you measure Secondary Task Loadndash How do you control Secondary Task Load

bull Performance Measuresndash Trackingndash Reaction Timendash Side Task Performance

bull Subjective Workload Measuresbull Situation Awareness Measures

ndash Testable Response Method

Simulator Studies of Driver CognitiveDistraction Caused by Cell Phone Use

bull Independent Variablesndash Cognitive Loadingndash Hands freeHands Fixed

bull Dependant Variablesndash Situation Awarenessndash Reaction Timendash Tracking MIT Age Lab Simulator

Cognitive Loading Levels

bull Lowndash Small talk

bull bullMediumndash Discuss movieoperaballet plot-line

bull bullHighndash Edit document by phone

Results Situational Awareness

bull Significant reduction in SA with Cell Phone Use

bull No significant difference between HFamp HH

Ave

rag

e F

ract

ion

Co

rrec

t

Normal HF HHPhone Setup

Stop Sign Response Time Resultsby Age Group

bull Significant effect of Age

bull No significant effect of Cell Phone Use

Re

sp

on

se

Tim

e [

se

c]

Normal1

Normal2

HFSmall

HFMovie

HF Edit HHSmall

HHMovie

HH Edit

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 25: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Interaction Metaphors

bull Car as homebull Car as kitchenbull Car as bathroombull Car as bedroombull Car as music roombull Car as playroombull Car as entertainment ctrbull Car as toolboxbull Car as closet

bull Car as officebull Car as comm center

bull Car as image statementbull Car as clothingbull Car as jewelrybull Car as sports equipmentbull Car as safe spacebull Car as cocoon

Trends in Driver Attentional Demand

Driv

er A

tten

tion

al D

emDr

iver

Att

enti

onal

Dem

and

and

VehicleVehicleOperation OperationOperation Operation

SecondarySecondaryTasks Tasks

TimeTime

Time Spent in Vehicles (US)Time Spent in Vehicles (US)- Average 350 hrsyrperson Average 350 hrsyrperson - 500 Million hrsweek- 500 Million hrsweek

76 of Drivers Report Activities HaveldquoCausedNearly Causedrdquo an Accident

TALKING ON CELLPHONE

TURNING TO SPEAK

USING COMPUTER

CIGARETTE ASHES

BREAKING UP FIGHTBETWEEN KIDS

SPILLING COFFEE

Source Opinion Research Corp Interviews Time5800 (N=1016)

Concerns Regarding High SecondaryTask Loads

bull Growing Evidence and Public Perception of Safety Problem

bull Cell Phone use in USndash 115-120 Million Active Cell Phonesndash 50-70 Use in Vehiclesndash 39 of Drivers Using (Daylight Hours)

bull NE Journal of Medicine Estimatendash 4 fold increase in collision risk using cell phone

bull NHTSA Estimatesndash 12 Million Accidents (25-30) caused by Distracted

Driverbull Not limited to cell phones

bull Note These effects may be latent in normal operations and may only manifest in non-normal or emergency situations

Typical Performance vs TaskLoad Curve

Per

form

ance

Task Load

Distraction Components

bull Manualndash Inability or delay in operation of vehicle controlndash Hands occupied or out of position (cell phone)

bull Visualndash Head Down Problemsndash Visual Accommodationndash Visual Clutterndash Visual Compulsion

bull Cognitivendash Lack of cognitive engagement with primary Taskndash Latency in mental context shiftsndash Multi-tasking capabilitiesndash Prioritization

bull Emotionalbull High Individual Variability in Multi-Tasking Capability

Distraction Data Sources

bull Controlled Experimentsndash Vary Secondary Task Load (Independent Variable)

bull Visualbull Cognitive

ndash Measurebull Performancebull Response to Disturbancebull Situation Awareness

ndash Need to Increase Task Loading to Saturationndash Need to Include Unanticipated Eventsndash Need Lowest Common Denominator Population

bull Field Datandash Event Recordersndash Cell Phone Triggered

bull Subjective Survey Data

Controlled Experiment Issues (1)

bull Simulator Testingndash Controlled Scenarios +ndash Safe to go to Task Saturationndash Face and Cue Validity Issues ndashndash Simulator Sickness ndashndash Cost -

bull Dual Control Vehicle Testing (Test Track)ndash Good Validity +ndash Safety Issues at Task Saturation ndashndash Cost +

Controlled Experiment Issues (2)

bull Variability in Primary and Secondary Taskingndash How do you measure Secondary Task Loadndash How do you control Secondary Task Load

bull Performance Measuresndash Trackingndash Reaction Timendash Side Task Performance

bull Subjective Workload Measuresbull Situation Awareness Measures

ndash Testable Response Method

Simulator Studies of Driver CognitiveDistraction Caused by Cell Phone Use

bull Independent Variablesndash Cognitive Loadingndash Hands freeHands Fixed

bull Dependant Variablesndash Situation Awarenessndash Reaction Timendash Tracking MIT Age Lab Simulator

Cognitive Loading Levels

bull Lowndash Small talk

bull bullMediumndash Discuss movieoperaballet plot-line

bull bullHighndash Edit document by phone

Results Situational Awareness

bull Significant reduction in SA with Cell Phone Use

bull No significant difference between HFamp HH

Ave

rag

e F

ract

ion

Co

rrec

t

Normal HF HHPhone Setup

Stop Sign Response Time Resultsby Age Group

bull Significant effect of Age

bull No significant effect of Cell Phone Use

Re

sp

on

se

Tim

e [

se

c]

Normal1

Normal2

HFSmall

HFMovie

HF Edit HHSmall

HHMovie

HH Edit

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 26: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Trends in Driver Attentional Demand

Driv

er A

tten

tion

al D

emDr

iver

Att

enti

onal

Dem

and

and

VehicleVehicleOperation OperationOperation Operation

SecondarySecondaryTasks Tasks

TimeTime

Time Spent in Vehicles (US)Time Spent in Vehicles (US)- Average 350 hrsyrperson Average 350 hrsyrperson - 500 Million hrsweek- 500 Million hrsweek

76 of Drivers Report Activities HaveldquoCausedNearly Causedrdquo an Accident

TALKING ON CELLPHONE

TURNING TO SPEAK

USING COMPUTER

CIGARETTE ASHES

BREAKING UP FIGHTBETWEEN KIDS

SPILLING COFFEE

Source Opinion Research Corp Interviews Time5800 (N=1016)

Concerns Regarding High SecondaryTask Loads

bull Growing Evidence and Public Perception of Safety Problem

bull Cell Phone use in USndash 115-120 Million Active Cell Phonesndash 50-70 Use in Vehiclesndash 39 of Drivers Using (Daylight Hours)

bull NE Journal of Medicine Estimatendash 4 fold increase in collision risk using cell phone

bull NHTSA Estimatesndash 12 Million Accidents (25-30) caused by Distracted

Driverbull Not limited to cell phones

bull Note These effects may be latent in normal operations and may only manifest in non-normal or emergency situations

Typical Performance vs TaskLoad Curve

Per

form

ance

Task Load

Distraction Components

bull Manualndash Inability or delay in operation of vehicle controlndash Hands occupied or out of position (cell phone)

bull Visualndash Head Down Problemsndash Visual Accommodationndash Visual Clutterndash Visual Compulsion

bull Cognitivendash Lack of cognitive engagement with primary Taskndash Latency in mental context shiftsndash Multi-tasking capabilitiesndash Prioritization

bull Emotionalbull High Individual Variability in Multi-Tasking Capability

Distraction Data Sources

bull Controlled Experimentsndash Vary Secondary Task Load (Independent Variable)

bull Visualbull Cognitive

ndash Measurebull Performancebull Response to Disturbancebull Situation Awareness

ndash Need to Increase Task Loading to Saturationndash Need to Include Unanticipated Eventsndash Need Lowest Common Denominator Population

bull Field Datandash Event Recordersndash Cell Phone Triggered

bull Subjective Survey Data

Controlled Experiment Issues (1)

bull Simulator Testingndash Controlled Scenarios +ndash Safe to go to Task Saturationndash Face and Cue Validity Issues ndashndash Simulator Sickness ndashndash Cost -

bull Dual Control Vehicle Testing (Test Track)ndash Good Validity +ndash Safety Issues at Task Saturation ndashndash Cost +

Controlled Experiment Issues (2)

bull Variability in Primary and Secondary Taskingndash How do you measure Secondary Task Loadndash How do you control Secondary Task Load

bull Performance Measuresndash Trackingndash Reaction Timendash Side Task Performance

bull Subjective Workload Measuresbull Situation Awareness Measures

ndash Testable Response Method

Simulator Studies of Driver CognitiveDistraction Caused by Cell Phone Use

bull Independent Variablesndash Cognitive Loadingndash Hands freeHands Fixed

bull Dependant Variablesndash Situation Awarenessndash Reaction Timendash Tracking MIT Age Lab Simulator

Cognitive Loading Levels

bull Lowndash Small talk

bull bullMediumndash Discuss movieoperaballet plot-line

bull bullHighndash Edit document by phone

Results Situational Awareness

bull Significant reduction in SA with Cell Phone Use

bull No significant difference between HFamp HH

Ave

rag

e F

ract

ion

Co

rrec

t

Normal HF HHPhone Setup

Stop Sign Response Time Resultsby Age Group

bull Significant effect of Age

bull No significant effect of Cell Phone Use

Re

sp

on

se

Tim

e [

se

c]

Normal1

Normal2

HFSmall

HFMovie

HF Edit HHSmall

HHMovie

HH Edit

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 27: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

76 of Drivers Report Activities HaveldquoCausedNearly Causedrdquo an Accident

TALKING ON CELLPHONE

TURNING TO SPEAK

USING COMPUTER

CIGARETTE ASHES

BREAKING UP FIGHTBETWEEN KIDS

SPILLING COFFEE

Source Opinion Research Corp Interviews Time5800 (N=1016)

Concerns Regarding High SecondaryTask Loads

bull Growing Evidence and Public Perception of Safety Problem

bull Cell Phone use in USndash 115-120 Million Active Cell Phonesndash 50-70 Use in Vehiclesndash 39 of Drivers Using (Daylight Hours)

bull NE Journal of Medicine Estimatendash 4 fold increase in collision risk using cell phone

bull NHTSA Estimatesndash 12 Million Accidents (25-30) caused by Distracted

Driverbull Not limited to cell phones

bull Note These effects may be latent in normal operations and may only manifest in non-normal or emergency situations

Typical Performance vs TaskLoad Curve

Per

form

ance

Task Load

Distraction Components

bull Manualndash Inability or delay in operation of vehicle controlndash Hands occupied or out of position (cell phone)

bull Visualndash Head Down Problemsndash Visual Accommodationndash Visual Clutterndash Visual Compulsion

bull Cognitivendash Lack of cognitive engagement with primary Taskndash Latency in mental context shiftsndash Multi-tasking capabilitiesndash Prioritization

bull Emotionalbull High Individual Variability in Multi-Tasking Capability

Distraction Data Sources

bull Controlled Experimentsndash Vary Secondary Task Load (Independent Variable)

bull Visualbull Cognitive

ndash Measurebull Performancebull Response to Disturbancebull Situation Awareness

ndash Need to Increase Task Loading to Saturationndash Need to Include Unanticipated Eventsndash Need Lowest Common Denominator Population

bull Field Datandash Event Recordersndash Cell Phone Triggered

bull Subjective Survey Data

Controlled Experiment Issues (1)

bull Simulator Testingndash Controlled Scenarios +ndash Safe to go to Task Saturationndash Face and Cue Validity Issues ndashndash Simulator Sickness ndashndash Cost -

bull Dual Control Vehicle Testing (Test Track)ndash Good Validity +ndash Safety Issues at Task Saturation ndashndash Cost +

Controlled Experiment Issues (2)

bull Variability in Primary and Secondary Taskingndash How do you measure Secondary Task Loadndash How do you control Secondary Task Load

bull Performance Measuresndash Trackingndash Reaction Timendash Side Task Performance

bull Subjective Workload Measuresbull Situation Awareness Measures

ndash Testable Response Method

Simulator Studies of Driver CognitiveDistraction Caused by Cell Phone Use

bull Independent Variablesndash Cognitive Loadingndash Hands freeHands Fixed

bull Dependant Variablesndash Situation Awarenessndash Reaction Timendash Tracking MIT Age Lab Simulator

Cognitive Loading Levels

bull Lowndash Small talk

bull bullMediumndash Discuss movieoperaballet plot-line

bull bullHighndash Edit document by phone

Results Situational Awareness

bull Significant reduction in SA with Cell Phone Use

bull No significant difference between HFamp HH

Ave

rag

e F

ract

ion

Co

rrec

t

Normal HF HHPhone Setup

Stop Sign Response Time Resultsby Age Group

bull Significant effect of Age

bull No significant effect of Cell Phone Use

Re

sp

on

se

Tim

e [

se

c]

Normal1

Normal2

HFSmall

HFMovie

HF Edit HHSmall

HHMovie

HH Edit

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 28: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Concerns Regarding High SecondaryTask Loads

bull Growing Evidence and Public Perception of Safety Problem

bull Cell Phone use in USndash 115-120 Million Active Cell Phonesndash 50-70 Use in Vehiclesndash 39 of Drivers Using (Daylight Hours)

bull NE Journal of Medicine Estimatendash 4 fold increase in collision risk using cell phone

bull NHTSA Estimatesndash 12 Million Accidents (25-30) caused by Distracted

Driverbull Not limited to cell phones

bull Note These effects may be latent in normal operations and may only manifest in non-normal or emergency situations

Typical Performance vs TaskLoad Curve

Per

form

ance

Task Load

Distraction Components

bull Manualndash Inability or delay in operation of vehicle controlndash Hands occupied or out of position (cell phone)

bull Visualndash Head Down Problemsndash Visual Accommodationndash Visual Clutterndash Visual Compulsion

bull Cognitivendash Lack of cognitive engagement with primary Taskndash Latency in mental context shiftsndash Multi-tasking capabilitiesndash Prioritization

bull Emotionalbull High Individual Variability in Multi-Tasking Capability

Distraction Data Sources

bull Controlled Experimentsndash Vary Secondary Task Load (Independent Variable)

bull Visualbull Cognitive

ndash Measurebull Performancebull Response to Disturbancebull Situation Awareness

ndash Need to Increase Task Loading to Saturationndash Need to Include Unanticipated Eventsndash Need Lowest Common Denominator Population

bull Field Datandash Event Recordersndash Cell Phone Triggered

bull Subjective Survey Data

Controlled Experiment Issues (1)

bull Simulator Testingndash Controlled Scenarios +ndash Safe to go to Task Saturationndash Face and Cue Validity Issues ndashndash Simulator Sickness ndashndash Cost -

bull Dual Control Vehicle Testing (Test Track)ndash Good Validity +ndash Safety Issues at Task Saturation ndashndash Cost +

Controlled Experiment Issues (2)

bull Variability in Primary and Secondary Taskingndash How do you measure Secondary Task Loadndash How do you control Secondary Task Load

bull Performance Measuresndash Trackingndash Reaction Timendash Side Task Performance

bull Subjective Workload Measuresbull Situation Awareness Measures

ndash Testable Response Method

Simulator Studies of Driver CognitiveDistraction Caused by Cell Phone Use

bull Independent Variablesndash Cognitive Loadingndash Hands freeHands Fixed

bull Dependant Variablesndash Situation Awarenessndash Reaction Timendash Tracking MIT Age Lab Simulator

Cognitive Loading Levels

bull Lowndash Small talk

bull bullMediumndash Discuss movieoperaballet plot-line

bull bullHighndash Edit document by phone

Results Situational Awareness

bull Significant reduction in SA with Cell Phone Use

bull No significant difference between HFamp HH

Ave

rag

e F

ract

ion

Co

rrec

t

Normal HF HHPhone Setup

Stop Sign Response Time Resultsby Age Group

bull Significant effect of Age

bull No significant effect of Cell Phone Use

Re

sp

on

se

Tim

e [

se

c]

Normal1

Normal2

HFSmall

HFMovie

HF Edit HHSmall

HHMovie

HH Edit

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 29: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Typical Performance vs TaskLoad Curve

Per

form

ance

Task Load

Distraction Components

bull Manualndash Inability or delay in operation of vehicle controlndash Hands occupied or out of position (cell phone)

bull Visualndash Head Down Problemsndash Visual Accommodationndash Visual Clutterndash Visual Compulsion

bull Cognitivendash Lack of cognitive engagement with primary Taskndash Latency in mental context shiftsndash Multi-tasking capabilitiesndash Prioritization

bull Emotionalbull High Individual Variability in Multi-Tasking Capability

Distraction Data Sources

bull Controlled Experimentsndash Vary Secondary Task Load (Independent Variable)

bull Visualbull Cognitive

ndash Measurebull Performancebull Response to Disturbancebull Situation Awareness

ndash Need to Increase Task Loading to Saturationndash Need to Include Unanticipated Eventsndash Need Lowest Common Denominator Population

bull Field Datandash Event Recordersndash Cell Phone Triggered

bull Subjective Survey Data

Controlled Experiment Issues (1)

bull Simulator Testingndash Controlled Scenarios +ndash Safe to go to Task Saturationndash Face and Cue Validity Issues ndashndash Simulator Sickness ndashndash Cost -

bull Dual Control Vehicle Testing (Test Track)ndash Good Validity +ndash Safety Issues at Task Saturation ndashndash Cost +

Controlled Experiment Issues (2)

bull Variability in Primary and Secondary Taskingndash How do you measure Secondary Task Loadndash How do you control Secondary Task Load

bull Performance Measuresndash Trackingndash Reaction Timendash Side Task Performance

bull Subjective Workload Measuresbull Situation Awareness Measures

ndash Testable Response Method

Simulator Studies of Driver CognitiveDistraction Caused by Cell Phone Use

bull Independent Variablesndash Cognitive Loadingndash Hands freeHands Fixed

bull Dependant Variablesndash Situation Awarenessndash Reaction Timendash Tracking MIT Age Lab Simulator

Cognitive Loading Levels

bull Lowndash Small talk

bull bullMediumndash Discuss movieoperaballet plot-line

bull bullHighndash Edit document by phone

Results Situational Awareness

bull Significant reduction in SA with Cell Phone Use

bull No significant difference between HFamp HH

Ave

rag

e F

ract

ion

Co

rrec

t

Normal HF HHPhone Setup

Stop Sign Response Time Resultsby Age Group

bull Significant effect of Age

bull No significant effect of Cell Phone Use

Re

sp

on

se

Tim

e [

se

c]

Normal1

Normal2

HFSmall

HFMovie

HF Edit HHSmall

HHMovie

HH Edit

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 30: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Distraction Components

bull Manualndash Inability or delay in operation of vehicle controlndash Hands occupied or out of position (cell phone)

bull Visualndash Head Down Problemsndash Visual Accommodationndash Visual Clutterndash Visual Compulsion

bull Cognitivendash Lack of cognitive engagement with primary Taskndash Latency in mental context shiftsndash Multi-tasking capabilitiesndash Prioritization

bull Emotionalbull High Individual Variability in Multi-Tasking Capability

Distraction Data Sources

bull Controlled Experimentsndash Vary Secondary Task Load (Independent Variable)

bull Visualbull Cognitive

ndash Measurebull Performancebull Response to Disturbancebull Situation Awareness

ndash Need to Increase Task Loading to Saturationndash Need to Include Unanticipated Eventsndash Need Lowest Common Denominator Population

bull Field Datandash Event Recordersndash Cell Phone Triggered

bull Subjective Survey Data

Controlled Experiment Issues (1)

bull Simulator Testingndash Controlled Scenarios +ndash Safe to go to Task Saturationndash Face and Cue Validity Issues ndashndash Simulator Sickness ndashndash Cost -

bull Dual Control Vehicle Testing (Test Track)ndash Good Validity +ndash Safety Issues at Task Saturation ndashndash Cost +

Controlled Experiment Issues (2)

bull Variability in Primary and Secondary Taskingndash How do you measure Secondary Task Loadndash How do you control Secondary Task Load

bull Performance Measuresndash Trackingndash Reaction Timendash Side Task Performance

bull Subjective Workload Measuresbull Situation Awareness Measures

ndash Testable Response Method

Simulator Studies of Driver CognitiveDistraction Caused by Cell Phone Use

bull Independent Variablesndash Cognitive Loadingndash Hands freeHands Fixed

bull Dependant Variablesndash Situation Awarenessndash Reaction Timendash Tracking MIT Age Lab Simulator

Cognitive Loading Levels

bull Lowndash Small talk

bull bullMediumndash Discuss movieoperaballet plot-line

bull bullHighndash Edit document by phone

Results Situational Awareness

bull Significant reduction in SA with Cell Phone Use

bull No significant difference between HFamp HH

Ave

rag

e F

ract

ion

Co

rrec

t

Normal HF HHPhone Setup

Stop Sign Response Time Resultsby Age Group

bull Significant effect of Age

bull No significant effect of Cell Phone Use

Re

sp

on

se

Tim

e [

se

c]

Normal1

Normal2

HFSmall

HFMovie

HF Edit HHSmall

HHMovie

HH Edit

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 31: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Distraction Data Sources

bull Controlled Experimentsndash Vary Secondary Task Load (Independent Variable)

bull Visualbull Cognitive

ndash Measurebull Performancebull Response to Disturbancebull Situation Awareness

ndash Need to Increase Task Loading to Saturationndash Need to Include Unanticipated Eventsndash Need Lowest Common Denominator Population

bull Field Datandash Event Recordersndash Cell Phone Triggered

bull Subjective Survey Data

Controlled Experiment Issues (1)

bull Simulator Testingndash Controlled Scenarios +ndash Safe to go to Task Saturationndash Face and Cue Validity Issues ndashndash Simulator Sickness ndashndash Cost -

bull Dual Control Vehicle Testing (Test Track)ndash Good Validity +ndash Safety Issues at Task Saturation ndashndash Cost +

Controlled Experiment Issues (2)

bull Variability in Primary and Secondary Taskingndash How do you measure Secondary Task Loadndash How do you control Secondary Task Load

bull Performance Measuresndash Trackingndash Reaction Timendash Side Task Performance

bull Subjective Workload Measuresbull Situation Awareness Measures

ndash Testable Response Method

Simulator Studies of Driver CognitiveDistraction Caused by Cell Phone Use

bull Independent Variablesndash Cognitive Loadingndash Hands freeHands Fixed

bull Dependant Variablesndash Situation Awarenessndash Reaction Timendash Tracking MIT Age Lab Simulator

Cognitive Loading Levels

bull Lowndash Small talk

bull bullMediumndash Discuss movieoperaballet plot-line

bull bullHighndash Edit document by phone

Results Situational Awareness

bull Significant reduction in SA with Cell Phone Use

bull No significant difference between HFamp HH

Ave

rag

e F

ract

ion

Co

rrec

t

Normal HF HHPhone Setup

Stop Sign Response Time Resultsby Age Group

bull Significant effect of Age

bull No significant effect of Cell Phone Use

Re

sp

on

se

Tim

e [

se

c]

Normal1

Normal2

HFSmall

HFMovie

HF Edit HHSmall

HHMovie

HH Edit

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 32: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Controlled Experiment Issues (1)

bull Simulator Testingndash Controlled Scenarios +ndash Safe to go to Task Saturationndash Face and Cue Validity Issues ndashndash Simulator Sickness ndashndash Cost -

bull Dual Control Vehicle Testing (Test Track)ndash Good Validity +ndash Safety Issues at Task Saturation ndashndash Cost +

Controlled Experiment Issues (2)

bull Variability in Primary and Secondary Taskingndash How do you measure Secondary Task Loadndash How do you control Secondary Task Load

bull Performance Measuresndash Trackingndash Reaction Timendash Side Task Performance

bull Subjective Workload Measuresbull Situation Awareness Measures

ndash Testable Response Method

Simulator Studies of Driver CognitiveDistraction Caused by Cell Phone Use

bull Independent Variablesndash Cognitive Loadingndash Hands freeHands Fixed

bull Dependant Variablesndash Situation Awarenessndash Reaction Timendash Tracking MIT Age Lab Simulator

Cognitive Loading Levels

bull Lowndash Small talk

bull bullMediumndash Discuss movieoperaballet plot-line

bull bullHighndash Edit document by phone

Results Situational Awareness

bull Significant reduction in SA with Cell Phone Use

bull No significant difference between HFamp HH

Ave

rag

e F

ract

ion

Co

rrec

t

Normal HF HHPhone Setup

Stop Sign Response Time Resultsby Age Group

bull Significant effect of Age

bull No significant effect of Cell Phone Use

Re

sp

on

se

Tim

e [

se

c]

Normal1

Normal2

HFSmall

HFMovie

HF Edit HHSmall

HHMovie

HH Edit

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 33: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Controlled Experiment Issues (2)

bull Variability in Primary and Secondary Taskingndash How do you measure Secondary Task Loadndash How do you control Secondary Task Load

bull Performance Measuresndash Trackingndash Reaction Timendash Side Task Performance

bull Subjective Workload Measuresbull Situation Awareness Measures

ndash Testable Response Method

Simulator Studies of Driver CognitiveDistraction Caused by Cell Phone Use

bull Independent Variablesndash Cognitive Loadingndash Hands freeHands Fixed

bull Dependant Variablesndash Situation Awarenessndash Reaction Timendash Tracking MIT Age Lab Simulator

Cognitive Loading Levels

bull Lowndash Small talk

bull bullMediumndash Discuss movieoperaballet plot-line

bull bullHighndash Edit document by phone

Results Situational Awareness

bull Significant reduction in SA with Cell Phone Use

bull No significant difference between HFamp HH

Ave

rag

e F

ract

ion

Co

rrec

t

Normal HF HHPhone Setup

Stop Sign Response Time Resultsby Age Group

bull Significant effect of Age

bull No significant effect of Cell Phone Use

Re

sp

on

se

Tim

e [

se

c]

Normal1

Normal2

HFSmall

HFMovie

HF Edit HHSmall

HHMovie

HH Edit

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 34: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Simulator Studies of Driver CognitiveDistraction Caused by Cell Phone Use

bull Independent Variablesndash Cognitive Loadingndash Hands freeHands Fixed

bull Dependant Variablesndash Situation Awarenessndash Reaction Timendash Tracking MIT Age Lab Simulator

Cognitive Loading Levels

bull Lowndash Small talk

bull bullMediumndash Discuss movieoperaballet plot-line

bull bullHighndash Edit document by phone

Results Situational Awareness

bull Significant reduction in SA with Cell Phone Use

bull No significant difference between HFamp HH

Ave

rag

e F

ract

ion

Co

rrec

t

Normal HF HHPhone Setup

Stop Sign Response Time Resultsby Age Group

bull Significant effect of Age

bull No significant effect of Cell Phone Use

Re

sp

on

se

Tim

e [

se

c]

Normal1

Normal2

HFSmall

HFMovie

HF Edit HHSmall

HHMovie

HH Edit

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 35: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Cognitive Loading Levels

bull Lowndash Small talk

bull bullMediumndash Discuss movieoperaballet plot-line

bull bullHighndash Edit document by phone

Results Situational Awareness

bull Significant reduction in SA with Cell Phone Use

bull No significant difference between HFamp HH

Ave

rag

e F

ract

ion

Co

rrec

t

Normal HF HHPhone Setup

Stop Sign Response Time Resultsby Age Group

bull Significant effect of Age

bull No significant effect of Cell Phone Use

Re

sp

on

se

Tim

e [

se

c]

Normal1

Normal2

HFSmall

HFMovie

HF Edit HHSmall

HHMovie

HH Edit

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 36: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Results Situational Awareness

bull Significant reduction in SA with Cell Phone Use

bull No significant difference between HFamp HH

Ave

rag

e F

ract

ion

Co

rrec

t

Normal HF HHPhone Setup

Stop Sign Response Time Resultsby Age Group

bull Significant effect of Age

bull No significant effect of Cell Phone Use

Re

sp

on

se

Tim

e [

se

c]

Normal1

Normal2

HFSmall

HFMovie

HF Edit HHSmall

HHMovie

HH Edit

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 37: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Stop Sign Response Time Resultsby Age Group

bull Significant effect of Age

bull No significant effect of Cell Phone Use

Re

sp

on

se

Tim

e [

se

c]

Normal1

Normal2

HFSmall

HFMovie

HF Edit HHSmall

HHMovie

HH Edit

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 38: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Approaches to Enhancing Focus in theInformation Rich Cockpit

bull Heads-Up Operationndash Parsing operating logicndash Glance Display Designs (lt 1 sec)ndash Tactile input devices

bull Voice Input-Outputndash not a Panacea

bull Prioritization Systemsndash Intelligent Situation Assessmentndash Interruption ldquoStand-byrdquo Architecture Example

bull Communicationsbull Information Systemsbull Non Critical Warnings

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 39: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

300-MIT Testbed

bull Collaborationndash MIT Media Lab CC++ndash Motorolandash DiamlerChrysler

bull Highly Instrumented Platformndash External Environmentndash Internal Environmentndash Vehicle Statesndash Driver Cognitive and Emotional States

bull Prototype Platformbull Platform - Chrysler 300M

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 40: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

ldquoStandbyrdquo System

Issue Criteria for automatic ldquoStandby Status

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 41: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

300M Experiment to DetermineIndicators of Driver ldquoBusyrdquo States

bull 20+ Subjects Drove Challenging Trajectory in Boston-Cambridge Area in 300M Instrumented Vehicle

bull Subjects indicated transitions between ldquoBusyrdquo and ldquonon-Busyrdquo States

bull Attempted Correlation with Observable States

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 42: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Example ldquoBusyrdquo State Data

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 43: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Good Correlation with Specific Locations

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 44: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Merge onto Major Roadway

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 45: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Complicated Intersection With Merging Traffic

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 46: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Rotary

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 47: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Parking Lot with Pedestrians

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 48: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Narrow Side Street

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 49: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Complex Urban Location ldquoHarvard Squarerdquo

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 50: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Weak Correlation with SimpleDynamic States

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 51: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Results Consistent With SubjectiveReporting of High Workload Tasks

bull Merge points

bull Pedestrians

bull Rotaries

bull Narrow Streets

bull Busy Intersections

bull Unfamiliar Locations

ndash Searching for Locations

bull Construction Zones

bull Poor Weather Conditions

bull Potential for Adaptive Learning Algorithims with Multi-Attribute Correlations

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 52: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 53: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Enhanced VisionSynthetic Vision

bull Goal is to increase safety and capacitybull Challenge is to ensure no adverse effects are created

Boeing is investigating these technologies including evaluating prototype systems on the 737 Technology Demonstrator in early 2002 While these technologies hold promisefor increasing safety and potentially improving airport capacity the designs must be approached carefully to ensure no harmful side effects are induced

Enhanced Vision Synthetic Vision

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 54: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Enhanced Vision

Picture of the outside world created by real-time weather and darkness penetrating on-board sensors (eg Cameras FLIR MMW radar and weather radar)

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 55: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 56: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 57: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash GPSndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 58: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

GPS Progressive RouteGuidance

bull Progressive Turn Guidancebull Current Limitations

ndash Complex Intersectionsndash Database Resolutionndash Database Structurendash CA Code Precisionndash Limited command set

bull Potential to degrade SAndash Dependencyndash Not robust to interruptionserrorsndash Head Downndash Lack of ldquoNaturalistic Interfacerdquo

bull Kamla Topsey amp Kate Zimmerman Expt

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 59: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Naturalistic Direction Study

bull 13 Subjects Directions categorizedbull Most common types

ndash Street names and route signs 26ndash Leftright turn indication 23ndash Distance by Reference point

bull Stoplights Stop signs 21bull Landmarks 11

bull Least common typesndash Distance by measurement 4ndash Heading 1

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 60: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

K2 Navigation System Prototype

bull No visual demand voice-basedbull Syntax

ndash Reference + Action + Target

bull Examplesndash ldquoAt the next light make a left onto the street betwe

en UNOrsquos Pizzeria and Fleet Bankrdquondash ldquoJust after the Star Market bear right onto Belmont

Strdquo

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 61: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Testing

bull Systems Map Carin K2bull Routes

ndash Start from MITndash Use 15 commandsndash All include a rotary

bull Subjectsndash 2 males 4 femalesndash 20-21 years oldndash 4-5 years driving experiencendash Unfamiliar with driving in Boston

area

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 62: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Results Navigation Errors

An error is defined as any deviation from the intended path

Map Carin K2

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 63: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Results Comparative Ratings

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 64: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Recently Developed Weather Datalink Products

ARNAV

Avidyne

BendixKing FAA FISDL

Control Vision

Digital Cyclone

Echo Flight

Garmin

UPS ndash AirCell

Vigyan

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 65: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision Syste

m)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architecture

sndash ldquoMaster Cautionrdquo

bull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 66: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Fundamental Tradeoff in AlertingDecisions

bull When to alertndash Too early 10487621048762Unnecessary Alert

bull Operator would have avoided hazard without alertbull Leads to distrust of system delayed response

ndash Too late 1048762 Missed Detectionbull Incident occurs even with the alerting systembull Must balance Unnecessary Alerts and Missed Detec

tions

Uncertaincurrent state Hazard

Un

cert

ain

Fu

ture

Tra

ject

ory

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 67: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

System Operating Characteristic Curve

Ideal Alerting System

Example Alerting

Threshold Locations

Pro

bab

ility

of

Suc

cess

ful A

lert

P(S

A)

Probability of UnnecessaryAlertP(FA)

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 68: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Kinematics

bull Alert time talert = (r - d)vtalert = 0 rarr braking must begin immediatelytalert = rarr alert is issued 1048762 seconds before braking is required

bull Determine P(UA) and P(SA) as function of talert

bull V = 35 mph in following example

Alert Issued

Vehicle

Total Braking Distance

ResponseLatency

Braking Distance

Hazard

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 69: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Case 1 Perfect Sensors

Deterministic probabilities are 0 or 1Ideal performance achievable for talert between 10 - 25 s

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 70: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Case 2 Add Sensor Uncertainty

Nearly ideal performance for talert between 145 - 175 s

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 71: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Example Response TimeDistribution

Lognormal distribution (mode = 107 s dispersion = 049) [Najm et al]

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 72: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Case 3 Add Response DelayUncertainty

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 73: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Case 4 Add DecelerationUncertainty

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 74: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Kinematics Sensors (eg RADAR) Limited byVehicle Dynamics and Response Time

Need Intent States

Courtesy of Prof Jim Kuchar MIT Dept of AeroAstro

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 75: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Intent States in the Lateral Tracking

External Environment

GoalSelection

RouteSelection

LaneLineSelection

LaneLineTracking Vehicle

Goal RouteDesired

LineSteering

CommandVehicleStates

StrategicFactors

HazardMonitoring

Threats

Disturbances

- Default- Open Loop- Optimized- Commanded- Prior History- Instructed- Wander

- Best Line- Lane Switching- Traffic- Speed

Wheel Position(force)

Acceleration

Velocity

Position

X = (Goal Subsequent Planned Trajectory Current Target State Acceleration Velocity Position)

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 76: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Intent Observability States

bull bull Roadwaybull Indicator Lights

ndash Break Lightsndash Turn Signalsndash Stop Lights

bull Acceleration Statesbull GPS Routingbull Head Positionbull Dynamic Historybull Tracking Behavior

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 77: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 78: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

ldquoMaster Cautionrdquo System

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 79: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

ldquoMaster Cautionrdquo Architecture

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 80: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 81: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

300M Face Analysis

bull Driver Internal Statendash Vigilancendash Stress

bull Driver Habitsndash Scan Patterns

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 82: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

300M Pupil Tracking

IBM BlueEyes Camera

Image withOn-Axis LEDs off

Image withOn-Axis LEDs on

Difference Image

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 83: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Approaches to Enhancing Focus in theInformation Rich Cockpit (cont)

bull Enhanced Perceptionndash IRMMW Radar (eg Cadillac Night Vision System)ndash Multidimensional video (ldquosuper mirrorrdquo)ndash Prioritized audio

bull Situation Awareness Displaysndash Datalink

bull Alerting Systemsbull Advanced Internal Diagnostics Architectures

ndash ldquoMaster Cautionrdquobull Driver Condition Monitoringbull External Visual Systems

ndash Active Signage

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 84: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Discussion

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 85: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Aerospace Experiencebull Drive by Wire

ndash Criticality - Fault Tolerancebull B-777 example

bull Collision alert criteriandash Alert vs autobrake

bull F-16 examplendash Complex threat fieldndash False alarm issue

bull System Operations Curvesbull HUD applications

ndash Limited FOVbull Runway alignmentbull Gunsight applications

ndash Visual Accommodationbull Infinity Opticsbull Visual anomalies

ndash egLack of Fusionbull Situation Awareness (SA) displays

ndash Testing Methods

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 86: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Technology Migrating intoAutomobile Cockpits

bull Mobile Communications (Voice Data)bull Portable Devices (Cell Phone PDA Wireles

s)ndash Not Controlled by Automobile Industry

bull Entertainment Info Systems (CD Web)bull Navigation and Guidance (GPS DGPS)bull Advanced Displays (Flat Panel HUD)bull Sensors (Radar IR MEMS)bull Databus Architectures (CANAIRINC)bull On-board Processors (Embedded Auto-PC)bull Control augmentation (ABS Cruise C)bull

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 87: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Background Current Systems

bull Information Structurendash Distance to turnndash Street namesndash Heading

bull Interfacendash Moving mapsndash Iconsndash Voice instructions

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 88: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Current System Phillips Carin

bull Uses maps icons and voicecommands to guide the driver

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 89: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Direction Study Conclusions

bull Humans and navigation systems both use street name and direction of turn to describe the action

bull They differ in the method used to warn the driver of the upcoming actionndash Humans rely on external reference points la

ndmarks stoplightsndash Navigation systems use distance amp heading

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 90: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Data

bull Subjective feedbackndash Cooper-Harper rating scale evaluationndash Best amp worst features evaluationndash Comparative rating of systems

bull Observationsndash Errors comments body language

bull Measurementsndash Position velocity

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 91: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Error ExampleJamaica Pond Trajectory

Lo

ng

itu

de

Latitude

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 92: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Results System RatingsN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

sN

umbe

r of

Sub

ject

s

Rating

Rating

Rating

K2

Carin

Map

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 93: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Obstacle Collision AlertingSystem Example

bull Examine effect of design parameters on performancendash sensor accuracyndash operator responsendash braking deceleration

bull Performance shown using SOC curves

bull Monte Carlo simulation used to estimate probabilitiesndash v = 35 mph (56 kmhr)ndash Safety evaluation

bull Avoidance trajectory

ndash variable delay 16 fts2 deceleration (05 g)

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 94: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

False Alarm Estimation

bull Was an alert unnecessarybull What would have happened without the aler

tndash Require Nominal trajectoryndash Definition of false alarm is situation- and operato

r-specificndash Baseline

bull If collision would occur after

15 s delay 10 fts2 deceleration (13 g) then an alert is necessarybull Otherwise an alert is a false alarm

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 95: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Implications

bull A single decision threshold for all users will not be acceptablendash uncertainties in response time and braking dominatendash some users will experience apparent false alarmsndash some users will experience apparent late alarms

bull Automating the braking response could improve safetyndash less uncertainty in response delay and deceleration pr

ofilendash may still be prone to perceived false alarmsndash may encourage complacency over-reliance

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 96: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

What Will Ultimately ControlSecondary Task Levels

bull Market Forces tend to increase complexityndash Market Values functionally gtgt complexity

bull Industry Practicebull Regulatory Actionbull Litigationbull Insurancebull Public Awareness

bull Pressure for actionbull We donrsquot have sufficient data to support rational action

at this point

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 97: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Transportation Systems Level

bull Fleet managementndash Monitoringndash Dispatchndash Reportingndash Support

bull Personal vehicle managementndash Teenage driver monitoringndash Transponders ldquofast passrdquondash Enforcement

bull Passenger vehicle as part of distribution networkndash Low end e-commerce

bull Drive through pick-upndash Foodndash Retailndash Services

bull Active ride share matching

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 98: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Careful Formatting Turns Data IntoInformation

Information distinguished by

bull Location

bull Labeling

bull Shape coding

bull NOT color

To help pilots identify information

bull Consistent use of shading

bull Consistent use of color

Every pixel earns its way on the display

Formatting not color used to distinguish information hellipSurprising thing is NOT colorOriginally in the early days of CRTs we did not use color because one failure mode of the CRT is reversion to BampW But we liked the human factors benefit we received fromthe additional clarity so we kept the philosophy on the LCD We donrsquot use color as the only means to distinguish information Color helps the pilot locate information but not todistinguish itShading is also used to help pilots identify certain informationSo on these altitude tapes notice- Boeing tape does not use white outlines because they are not necessary- The box shape is as simple as possible- Gaps between numbers and lines are intentional to separate the information- color is used sparingly and consistentlyI have a couple slides on shadingAnd a couple slides on color

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 99: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

External Vision ndash AC 25773-1

bull Vision Polar

bull 3-Second ldquoRulerdquo

Alsobull Precipitation clearingbull Post widthsbull other details

Vision polar and 3 second ruledrive airplane configuration(approach attitude and speed)

VrefMax landing weightForward cg10 kt crosswind

Down-vision angle

1200rsquo Runway Visual Range

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 100: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Synthetic Vision

Picture of the outside world created by combining precise navigationposition with databases of comprehensive geographic cultural andtactical information

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 101: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

Chrysler 300M Research Platform

DaimlerChrysler

Motorola CC++ MIT

Cooperative

Effort

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103
Page 102: The Coming Transition in Automobile Cockpits Insights from Aerospace Prof. R. John Hansman Department of Aeronautics & Astronautics.

CC++ 300M Driver StudyApply lab experience in Media and Human

Interface technologies

Build vehicle platform to develop and test driver behavior

Develop information workload managerIdentifyVehicleMotion

(stop turnaccel)

Location Aware(speed

position hellip)

Develop Tangible

Interfaces(transfer info

tohuman)

Chrysler 300M

Vehicle Thinks

Controls Flow of Information

(warnings phone etchellip)Monitor In-

vehicle

Situation

(mood cognitive load)

Identify Driver

Behavior and Stress

Level

  • Slide 1
  • Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10
  • Slide 11
  • Slide 12
  • Slide 13
  • Slide 14
  • Slide 15
  • Slide 16
  • Slide 17
  • Slide 18
  • Slide 19
  • Slide 20
  • Slide 21
  • Slide 22
  • Slide 23
  • Slide 24
  • Slide 25
  • Slide 26
  • Slide 27
  • Slide 28
  • Slide 29
  • Slide 30
  • Slide 31
  • Slide 32
  • Slide 33
  • Slide 34
  • Slide 35
  • Slide 36
  • Slide 37
  • Slide 38
  • Slide 39
  • Slide 40
  • Slide 41
  • Slide 42
  • Slide 43
  • Slide 44
  • Slide 45
  • Slide 46
  • Slide 47
  • Slide 48
  • Slide 49
  • Slide 50
  • Slide 51
  • Slide 52
  • Slide 53
  • Slide 54
  • Slide 55
  • Slide 56
  • Slide 57
  • Slide 58
  • Slide 59
  • Slide 60
  • Slide 61
  • Slide 62
  • Slide 63
  • Slide 64
  • Slide 65
  • Slide 66
  • Slide 67
  • Slide 68
  • Slide 69
  • Slide 70
  • Slide 71
  • Slide 72
  • Slide 73
  • Slide 74
  • Slide 75
  • Slide 76
  • Slide 77
  • Slide 78
  • Slide 79
  • Slide 80
  • Slide 81
  • Slide 82
  • Slide 83
  • Slide 84
  • Slide 85
  • Slide 86
  • Slide 87
  • Slide 88
  • Slide 89
  • Slide 90
  • Slide 91
  • Slide 92
  • Slide 93
  • Slide 94
  • Slide 95
  • Slide 96
  • Slide 97
  • Slide 98
  • Slide 99
  • Slide 100
  • Slide 101
  • Slide 102
  • Slide 103