modeling drivers’ route choice behavior, and traffic estimation and prediction byungkyu brian...

21
Modeling Drivers’ Route Choice Behavior, and Traffic Estimation and Prediction Byungkyu Brian Park, Ph.D. Center for Transportation Studies University of Virginia DriveSense14 Workshop, Norfolk, VA

Upload: kathlyn-martin

Post on 18-Jan-2016

215 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Modeling Drivers’ Route Choice Behavior, and Traffic Estimation and Prediction Byungkyu Brian Park, Ph.D. Center for Transportation Studies University

Modeling Drivers’ Route Choice Behavior, and Traffic Estimation and

Prediction

Byungkyu Brian Park, Ph.D.Center for Transportation Studies

University of Virginia

DriveSense14 Workshop, Norfolk, VA

Page 2: Modeling Drivers’ Route Choice Behavior, and Traffic Estimation and Prediction Byungkyu Brian Park, Ph.D. Center for Transportation Studies University

Drivers’ Route Choice Model

• Existing literature– Considers disaggregate information but ends up

with an aggregate model • Can we consider a model for each driver?

– Seems feasible with connected vehicle and smart phones and driver’s opt-in

Page 3: Modeling Drivers’ Route Choice Behavior, and Traffic Estimation and Prediction Byungkyu Brian Park, Ph.D. Center for Transportation Studies University

Traffic Estimation & Prediction

• Estimates existing network condition using probe vehicles

• Estimates origin destination matrices for next 15-30 minutes

• Predicts future traffic conditions by assigning the OD matrices

• Evaluates multiple operational strategies and recommends best strategy

Page 4: Modeling Drivers’ Route Choice Behavior, and Traffic Estimation and Prediction Byungkyu Brian Park, Ph.D. Center for Transportation Studies University

Motivation

• Weather vs. Route Guidance

Page 5: Modeling Drivers’ Route Choice Behavior, and Traffic Estimation and Prediction Byungkyu Brian Park, Ph.D. Center for Transportation Studies University

Connected Vehicle Technology

• Wireless communications among vehicles and infrastructure

Page 6: Modeling Drivers’ Route Choice Behavior, and Traffic Estimation and Prediction Byungkyu Brian Park, Ph.D. Center for Transportation Studies University

Questions on the Route Guidance

6

Will connected vehicle technology improve the quality of route guidance?

What happens if multiple route guidance strategies were implemented? Will they cancel-off benefits?

Page 7: Modeling Drivers’ Route Choice Behavior, and Traffic Estimation and Prediction Byungkyu Brian Park, Ph.D. Center for Transportation Studies University

Route Guidance System• Assumptions

– Every equipped vehicle provides its origin-destination information (opt-in)

– No Communications Loss• Perfect communications V-2-I and V-2-V

– On-Board Equipment (OBE or OBU) Vehicles• Act as probe vehicles

Page 8: Modeling Drivers’ Route Choice Behavior, and Traffic Estimation and Prediction Byungkyu Brian Park, Ph.D. Center for Transportation Studies University

Route Guidance System• Assumptions (cont’d)

– Guided Drivers• Time varying traffic assignment• A link-weighted K-Shortest Path algorithm to create

reasonable path alternatives• Time dependant minimum travel time path

– Unguided Drivers• Static assignment• Fixed shortest distance path

Page 9: Modeling Drivers’ Route Choice Behavior, and Traffic Estimation and Prediction Byungkyu Brian Park, Ph.D. Center for Transportation Studies University

Microscopic Traffic Simulation Model - VISSIM

• Microscopic, Time-step

based simulation model• Simulate traffic operations

in urban streets and freeways• Emphasize multi-modal

transportations (Bus, LRT, Heavy

Rail, etc.)

OverviewOverview

Page 10: Modeling Drivers’ Route Choice Behavior, and Traffic Estimation and Prediction Byungkyu Brian Park, Ph.D. Center for Transportation Studies University

Microscopic Traffic Simulation Model – VISSIM (cont’d)

Traffic Flow ModelTraffic Flow Model Signal Control ModelSignal Control Model

Page 11: Modeling Drivers’ Route Choice Behavior, and Traffic Estimation and Prediction Byungkyu Brian Park, Ph.D. Center for Transportation Studies University

Microscopic Traffic Simulation Model – VISSIM (cont’d)

• Various measures of effectiveness

(e.g., delay, travel time, queue

length, etc.)

• 2D & 3D animations

OutputOutput

Page 12: Modeling Drivers’ Route Choice Behavior, and Traffic Estimation and Prediction Byungkyu Brian Park, Ph.D. Center for Transportation Studies University

Route Guidance Strategies

Guidance Strat-egy

Acro-nym

Major Information from VII

Latest Travel time -based Guidance LTG The latest link travel time

Averaged Travel time-based Guidance ATG The average of link travel

times

Routing Travel time-based Guidance RTG

Individual vehicles’ travel times of directional movements

Predicted Travel time-based Guidance PTG

Individual vehicles’ origin- destination tables

Page 13: Modeling Drivers’ Route Choice Behavior, and Traffic Estimation and Prediction Byungkyu Brian Park, Ph.D. Center for Transportation Studies University

13

• Travel time of directional movements at an intersection

• Gathers all individual directional travel time through individual vehicles’ trajectory

Routing Travel time-based Guidance (RTG)

Page 14: Modeling Drivers’ Route Choice Behavior, and Traffic Estimation and Prediction Byungkyu Brian Park, Ph.D. Center for Transportation Studies University

Predicted Travel Time-Based Guidance (PTG)

Simulationfor current state estimation

OD table Extraction

Simulationfor link travel time prediction

Converged?

PredictedLink Travel Time

Generation ofPrediction-based Guidance

• Based on DynaMIT program (i.e., Traffic Estimation and Prediction)

• Travel info (origin-destination) obtained from equipped vehicles

Page 15: Modeling Drivers’ Route Choice Behavior, and Traffic Estimation and Prediction Byungkyu Brian Park, Ph.D. Center for Transportation Studies University

Route Guidance System Evaluation• Simulation Test-Bed

– Microscopic Traffic Simulator: VISSIM– A Hypothetical Urban Network

• 118 Road Segments including

- a freeway - a major arterial

• 21 Signalized Intersections• 9 All-Way-Stop Control• 25 Origin/Destination

Page 16: Modeling Drivers’ Route Choice Behavior, and Traffic Estimation and Prediction Byungkyu Brian Park, Ph.D. Center for Transportation Studies University

16

Experimental Design

• Experimental factors and levels

• Experimental setup– Single operation : Total 2175 simulation runs

and 1197 computer hours– Multiple operation : Total 150 simulation runs

and 93 computer hours– Made 5 replications for each simulation

Page 17: Modeling Drivers’ Route Choice Behavior, and Traffic Estimation and Prediction Byungkyu Brian Park, Ph.D. Center for Transportation Studies University

17

Benefits of CV-based guidance strategies

• Single operation of guidance strategies

• Multiple operation of guidance strategies

Page 18: Modeling Drivers’ Route Choice Behavior, and Traffic Estimation and Prediction Byungkyu Brian Park, Ph.D. Center for Transportation Studies University

18

Benefits of individual strategies• All guidance strategies produced

benefits – Single operation of guidance

strategies

– Multiple operation of guidance strategies

Page 19: Modeling Drivers’ Route Choice Behavior, and Traffic Estimation and Prediction Byungkyu Brian Park, Ph.D. Center for Transportation Studies University

19

Impact of Market Penetration RateVH

T (

Vehic

le-H

ours

)

gudanceMP

PTGRTGATGLTGBASE100703051007030510070305100703050

1100

1000

900

800

700

600

500

400

300

Boxplot of VHT

Page 20: Modeling Drivers’ Route Choice Behavior, and Traffic Estimation and Prediction Byungkyu Brian Park, Ph.D. Center for Transportation Studies University

Proposed Research

• Bundle drivers’ route choice behavior model and traffic estimation & prediction system

• How? – Develop each driver’s route choice behavior

model and keep model parameters on his/her smartphone or cloud

– Implement driver’s route choice behavior model in TrEP

Page 21: Modeling Drivers’ Route Choice Behavior, and Traffic Estimation and Prediction Byungkyu Brian Park, Ph.D. Center for Transportation Studies University

Where Are We?

• Just completed IRB training! • Developed survey questionnaire to

understand drivers’ characteristics and their stated preferences

• Evaluate drivers’ route choice behavior using driving simulator