Testing and Validation of PrePre-Crash Sensing-Crash Sensing · 2008. 7. 23. · Title: Microsoft...

31
Copyright MIRA 2008 Testing and Validation of Pre-Crash Sensing Testing and Validation of Testing and Validation of Pre Pre - - Crash Sensing Crash Sensing Jonathan Moore and Tim Edwards MIRA Ltd, UK Pre-Crash-Sensorik im Automobil 22 April, 2008. Munich

Transcript of Testing and Validation of PrePre-Crash Sensing-Crash Sensing · 2008. 7. 23. · Title: Microsoft...

  • Copyright MIRA 2008

    Testing and Validation ofPre-Crash Sensing

    Testing and Validation ofTesting and Validation ofPrePre--Crash SensingCrash Sensing

    Jonathan Moore and Tim EdwardsMIRA Ltd, UK

    Pre-Crash-Sensorik im Automobil22 April, 2008. Munich

  • Copyright MIRA 2008

    IntroductionIntroductionIntroduction

    An overview of test and validation approaches for pre-crash systemsThe sensor technologies and example test cases will largely be focused on systems to detect and respond to vulnerable road users (VRUs)

    PedestriansCyclistsMotorcyclists

    This ties in to our work on a number of EU funded research projects, such as WATCH-OVER and SAVE-U

  • Copyright MIRA 2008

    Why do we need pre-crash test facilities?

    Why do we need preWhy do we need pre--crash test crash test facilities?facilities?

    A safe environment for:Research and development

    Currently new tests are being designed for each projectComponents and sub-systems often tested in isolationFuture refinement activities would benefit from established and repeatable test procedures/environments

    Demonstration activitiesProof-of-concept tests to showcase technologies, encourage public acceptance and develop commercial support.

    Certification and ValidationCommercial systems will require validation against a set of specific requirements, and therefore a defined set of tests.

  • Copyright MIRA 2008

    Image from the United Nations Population Fund (UNPFA) web site: http://www.unfpa.org/swp/2007/presskit/images/japan.jpg

    Tokyo, Japan

  • Copyright MIRA 2008

    Recreating realistic scenariosRecreating realistic scenariosRecreating realistic scenarios

    The requirements for test scenarios are linked to the systems under test. In most cases it is desirable to have minimal test equipment in view of the vehicle sensors

    Current automotive tests such as crash, EMC, and emissions are very established, and specialist facilities exist. The body and safety tests are the closest to replicating pre-crash scenarios

    However, in these tests there are often systems such as rails and winches which are not acceptable for some pre-crash sensor tests

  • Copyright MIRA 2008

    Body and Safety TestingBody and Safety TestingBody and Safety Testing

  • Copyright MIRA 2008

    Controlling test conditionsControlling test conditionsControlling test conditions

    The test environment should be controlled to allow for safe, repeatable testsOutdoor tests introduce a number of challenges in this area. We can’t control nature, but we can…

    Automate as much as possibleTest vehicle controlData synchronisation

    Monitor environmental conditionsProvide clear signals to co-ordinate “actors”

  • Copyright MIRA 2008

    Testing safelyTesting safelyTesting safely

    Risk assessment procedure for any proving ground testDetailed FMEA (Failure Mode and Effects Analysis) for test equipmentRigorous maintenance procedures for test equipmentStrict operating procedures and checksRedundancy of safety mechanismsFail safe modes in control electronics

  • Copyright MIRA 2008

    Current/Future ADAS SensorsCurrent/Future ADAS SensorsCurrent/Future ADAS Sensors

    24GHz RadarMono and Stereo vision camerasPassive Infra-RedWireless communications

    2.4Ghz Chirp Spread Spectrum (CSS)

    LIDARUltra-wide band (UWB)

  • Copyright MIRA 2008

    Data Acquisition RequirementsData Acquisition RequirementsData Acquisition Requirements

    Quiet wireless environmentTherefore no wireless data streams during the test

    On-board storageDurableLightHigh capacityFast, shared, accessQuick “data dump” or drive switch between tests

    Interfacing with proprietary hardware / softwareAccurate time stamping

    Required for synchronising all logged data (on and off-board the test vehicle)

  • Copyright MIRA 2008

    Ground TruthWhere was everything?

    Ground TruthGround TruthWhere was everything?Where was everything?

    Very important that we can compare real events with sensor readingsKey element is timing

    External ground truth monitoring must support synchronisation with on-vehicle data

    Options include:Positioning devices on every “actor” in the scenario

    High accuracy required and therefore high costVideo footage and image processing

    Suitable observation positions can be hard to findDistorting lens increase coverage but complicate image processing

  • Copyright MIRA 2008

    Case Studies: SAVE- U and WATCH-OVER

    Case Studies: Case Studies: SAVESAVE-- U and WATCHU and WATCH--OVEROVER

    BackgroundTest conceptResults / VideoConclusions http://www.watchover-eu.org/

    http://www.save-u.org/

  • Copyright MIRA 2008

    SAVE-USAVESAVE--UU

    SAVE-U was a large European project that ended in 2005.

    It was aimed at developing an integrated safety concept for VRUs such as pedestrians and cyclists.

    An innovative sensor platform was developed with three different technologies (24 GHz radar, IR and video cameras).

    The maximum vehicle speed considered was 40 km/h

  • Copyright MIRA 2008

    WATCH-OVERWATCHWATCH--OVEROVER

    WATCH-OVER started in January 2006. It is a collaborative project co-funded by the European Commission under FP6.

    The innovative concept is represented by an on board platform and by a vulnerable user module. The system is based on short range communication and vision sensors.

    Maximum vehicle speedincreased to 50 km/h

  • Copyright MIRA 2008

    Project RequirementsProject RequirementsProject Requirements

    MIRA is responsible for dynamic evaluation of the system – especially at close range (highest risk)

    Target must:Look like a moving VRU (video)Look like a moving VRU (IR)Reflect like a moving VRU (radar)

    Hence target must BE a VRU – not a picture or a dummy…but absolutely no risk of actual impact

    MIRA solution is to mount the entire system on a tethered dynamic test rig, not a car

  • Copyright MIRA 2008

    MIRAMIRA’’ss HeadquartersHeadquarters

  • Copyright MIRA 2008

    Test site configurationTest site configurationTest site configuration

    105m

    5.5m

    27m

    25m

    10m

    50m

    19m29m

    3.2m

    1.6m OK

    No2 Circuit

    No2 Bend No3North Straight

    Observation RoadNo15 CorrugationsNo13 Resonance Road

    Parking

    Entrance

    ExitConcrete

    wall

    Crash barrier

    Grass

    T2

    Created:20070601Revised:20070611

    To Control Tower

    N

    Tarmac

    Store

    Tether point

  • Copyright MIRA 2008

    Pre-crash test vehiclePrePre--crash crash

    test vehicletest vehicle

  • Copyright MIRA 2008

    Restraint MechanismsRestraint MechanismsRestraint Mechanisms

    Test vehicle

    Falling weight

    Braking Drum

    Energy absorbers

    Deformed steel bar

  • Copyright MIRA 2008

    Braking system in use during SAVE-U tests

  • Copyright MIRA 2008

    SAVE-U test with pedestrians crossing

  • Copyright MIRA 2008

    Ground TruthGround TruthGround TruthTo allow accurate post-test data analysis, onboard sensors logged

    rig velocity and yaw, and “ground truth” information is recorded using an overhead camera.

    On board cameras

    Overhead “ground truth” camera

  • Copyright MIRA 2008

    Overhead ground truth camera capturing a SAVE-U test scenario

  • Copyright MIRA 2008

    Data acquisitionData acquisitionData acquisition

  • Copyright MIRA 2008

    Un-tethered speed measurement

    UnUn--tethered speed tethered speed measurementmeasurement

    Max: 50.4km/h

    Target: 50km/h

    8 seconds60m

    Braking took 3 seconds.With tether ropes this is reduced to 1 second.

    0

    2

    4

    6

    8

    10

    12

    14

    16

    0 5 10 15

    Time (s)

    Vel

    ocity

    (m/s

    )

  • Copyright MIRA 2008

    Test scenariosTest scenariosTest scenarios

    A variety of scenarios were run in SAVE-U with vehicles, pedestrians and cyclists

    In WATCH-OVER the scenario complexity will be increased, with more actors and night tests

  • Copyright MIRA 2008

    ConclusionsConclusionsConclusionsIncreasing scenario complexity desired by system designersSafety issues need to be tightly monitored and controlledTest vehicle positioning

    Manual/Semi-Autonomous/Fully-Autonomous vehicles all require accurate monitoring and control

    Actor control is very importantRepeatable test scenarios Safety

    Ground truthThe capture and synchronisation of position information for all actors in a test is non-trivial.This must be considered as a design requirement at an early stage of the test design

  • Copyright MIRA 2008

    Specifically the future of pre-crash testing…

    Autonomous vehiclesAutonomous VRUs?Purpose built test sites

    with protective barriers and fail safe restraintsComplex test scenariosEnvironmental conditions – Fog/Rain etcCornering, urban/built up areas

    The future of pre-crash sensing

    The future of preThe future of pre--crash crash sensingsensing

  • Copyright MIRA 2008

    Improved test scenarioImproved test scenarioImproved test scenario

    Test Scenario

    Test Vehicle

    Restraint Mechanism

  • Copyright MIRA 2008

    Autonomous Test VehicleAutonomous Test VehicleAutonomous Test Vehicle

    + Complex and repeatable test scenarios- High levels of reliability need to be assured

    for use around VRUs

  • Copyright MIRA 2008

    http://www.mira.co.uk

    MIRA LtdWatling Street

    NuneatonWarwickshire

    CV10 0TUUnited Kingdom