Autonomous Vehicle Modelingonlinepubs.trb.org/onlinepubs/Conferences/2018/ITM/SRuegg.pdf · —...
Transcript of Autonomous Vehicle Modelingonlinepubs.trb.org/onlinepubs/Conferences/2018/ITM/SRuegg.pdf · —...
Development and Application of a Model to Estimate Driverless Autonomous
Vehicle Trips
Innovations in Travel Modeling, June 2017
1. Scenarios2. Twin Cities Model Application3. Estimating Owned/Shared Mix4. Assignment of AV trips5. Feedback and integration with the planning model6. Examples of Types of Results7. Additional Research
Presentation Outline
—Ownership Model, All AVs are privately Owned—Sharing Model, All AVs are publicly-available—Mixed, Some owned, some shared AVs—Partial Implementation, Mixed AVs with traditional
non-autonomous vehicles
1. Scenarios for AV Use
1. Cost2. Auto Availability3. Capacity and Flow Model4. Driverless Vehicle Movements
1. Ownership Scenario2. Sharing Scenario
2. Model Adjustments for AV Scenarios
— Parking Costs— Auto Operating Costs— Value of Time
— What are these costs?— Relatively easy to implement within
a model.— May need to stratify costs between
traditional and AVs, Driverless and Occupied.
— Some policy assumptions needed, e.g., tolling
2.1 Cost Assumptions
2.2 Auto Availability Adjustment for AVs
— AVs will allow access to autos for populations that previously did not have access:—Elderly and disabled—Children—Low income (partially)—Auto-deficient households
— Model Adjustments—Adjust inputs so that 95% of Households above lowest Income
(>25k) have sufficient autos to serve adult population. Adjust to 50% for lowest income group.
7
2.3 Capacity Adjustment
— AV use will increase capacity by—Ability to maintain shorter headways on freeways and express
ways—AV’s have the ability to mitigate the effects of congestion on
travel time
— Model Adjustments – Owned & Shared Scenarios—Increase capacity by 50% for freeways and expressways—Increase capacity by 10% for Arterials—Modify the relationship between volume and speed to be more
“forgiving” with regard to demand
8
AVO
NLY
2.3 Capacity Adjustment for AVs
9
0
500
1,000
1,500
2,000
2,500
3,000
Cap
acity
(Veh
/HR
/Lan
e)
BASE AV
10
2.3 Adjusting Volume-Delay Functions
2.4.1 Driverless Vehicle Movements for the Ownership Scenario, Using Activity-Based Model Outputs— Consider all model-estimated vehicle trips for each household,
including origin, destination, start and end times— Create an AV, and connect household vehicle trips sequentially
through the day— Consider time necessary for each driverless trip, and compare with
available time— In some cases consider intermediate parking— Continue to create new AVs until all household trips are served
11
8am 10am NOON 2pm 4pm 6pm 8pm 10pm MIDNIGHT
Homer
Marge
Bart
Lisa
Maggie
2.4.1 Driverless Vehicle Movements for the Ownership Scenario
12
AV1AV2
2.4.1 Service Algorithms for AVs, Ownership Scenario
— Household Members availability based on location and time— Choice of intermediate parking location compared with home
location if there is more than 30 min wait.— A score is computed for trips to home and the “best” intermediate
parking location. Based on total time.— Parking availability based on a user-supplied share of undeveloped
land— Remote parking demand constrained by capacity
13
2.4.1 Example:Owned Vehicle, Household 195302Home Zone 2881
26 Occupied Trips3 vehiclesVehicle 1: 9 DL tripsVehicle 2: 4 DL tripsVehicle 3: 2 DL trips
14
2.4.1 Outputs for Ownership Model
— Selected Households – may be a subset of region— Driverless trip records – includes
— Household ID— Vehicle ID— Origin and Destination Zones— Start and End times
— Number of AVs required by household — Number of AVs in intermediate parking, by zone and by time of day
15
2.4.2 Driverless Vehicle Movements for the Shared Vehicle Scenario
— Same principal as used for ownership scenario – except all occupied vehicle trips are open to being served
— Search pattern for next available trip seeks to minimize driverless trip time and dwell time between services
— User specifies a minimum and maximum allowable dwell times— User specifies maximum allowable driverless trip time— Result is a set of driverless vehicle trip records, and a log of each
vehicle’s movements throughout the day— Segmentation of input is permitted to allow for parallel processing
16
2.4.2 Example:Shared Vehicle 6316
30 Occupied Trips29 Driverless Trips
292 Occupied Miles115 Driverless Miles
17
3. Estimating Ownership or Sharing by Household for the Mixed scenario
— Using the 100% shared scenario, Identify vehicles that are used 7 or fewer times/day
— Compute for each household the average number of trips by shared autos used
— For households served inefficiently by shared autos, tag these as “ownership” households.
— This resulted in about 45% of households owning AVs, 55% of households using shared AVs
18
0
20,000
40,000
60,000
80,000
100,000
120,000
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Num
ber o
f Veh
icle
s
Trips/Day
Distribution of Autonomous Vehicles by Number of Daily Trips
Average Driverless trips/veh=10.0
3. Mixed Scenario: Map of Zones by Share of Households Owning AVs
19
4. Assignment of Driverless Vehicles
— Added driverless vehicles as an additional class— Model information available to plot where AVs would dwell when not
in use.— End of Day re-positioning— Wealth of MOE’s available for both occupied and driverless vehicles— Feedback ensures that congestion imposed by driverless vehicles
influences other behavior
20
5. Autonomous Vehicle Model Flowchart – Twin Cities ABM
TOURCAST
11 PERIOD HIGHWAY ASSIGNMENT
AVDL ESTIMATION
4 PERIOD HIGHWAY AND TRANSIT ASSIGNMENT
HIGHWAY AND TRANSIT SKIMS
AVDL TRIPS
OCCUPIED VEHICLE TRIPS
HIGHWAY SKIMS
Model Feedback
21
6. Examples of Results that are Available
1. Vehicle Fleet Size Estimates2. Trip Length Frequency Distribution3. Efficiency of Use by Shared AVs4. VMT by Level of Service by Scenario5. End of Day Vehicle Re-positioning Map
22
6.1 Vehicle Fleet Requirements
23
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
NoBuild Owned AVs Shared AVs Mixed AVs Partial Mixed
VEH
ICLE
S/H
OU
SEH
OLD
AV NON-AV
6.2 Ownership Scenario Driverless Trips by Vehicle
0
100,000
200,000
300,000
400,000
500,000
600,000
700,000
800,000
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49
TRIP
S
TIME (MIN)
Driverless AV Trip Length Freq Distributions by time -- OWNERSHIP
6-Midnight5-6pm4-5pm3-4pm2-3pm9am-2pm8-9am7-8am6-7amMidnight to 6AMPerson-Veh Trips
24
0
100,000
200,000
300,000
400,000
500,000
600,000
700,000
800,000
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49
TRIP
S
TIME (MIN)
Driverless AV Trip Length Freq Distributions by time -- SHARED
6-Midnight5-6pm4-5pm3-4pm2-3pm9am-2pm8-9am7-8am6-7amMidnight to 6AMPerson-Veh Trips
6.2 Shared Scenario Driverless Trips by Vehicle
25
6.3 Efficiency of Use: Shared Scenario Driverless Trips by Vehicle
26
0%2%4%6%8%
10%12%14%16%18%20%
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44
SHAR
E O
F D
RIV
ERLE
SS T
RIP
S
DRIVERLESS TRIPS PER AV
Driverless Trips Per AV: Shared and Mixed ScenariosSHARED MIXED
Change Average Trips/Vehicle from 10.0 to 18.1
6.4 VMT by Level Of Service
27
0
2,000,000
4,000,000
6,000,000
8,000,000
10,000,000
12,000,000
14,000,000
16,000,000
18,000,000
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 > 1.55VOLUME/CAPACITY RATIO
VMT by LOSNoAVs SHARED OWNERSHIP MIXED MIXED PARTIAL
6.5 Shared Vehicle Repositioning – Shared Scenario3.6M VMT, 64K VHT
6.5 Shared Vehicle Repositioning – Mixed Scenario0.9M VMT, 15K VHT
— AV Driving Characteristics— Vehicle Capital Cost for
each scenario— Vehicle Operating Cost for
each scenario— Behavioral Changes for
— Former non-drivers— Activity pattern changes as
a result of AVs
7. Additional Research
Isaac, Lauren, “Driving Towards Driverless: A Guide For Government Agencies. WSP|Parsons Brinckerhoff. 2016. Page 13.
Questions?