Travel Simulator Laboratory

1
Travel Simulator Laboratory

• TSL An Introduction

Dr. Ir. S. P. Hoogendoorn (s.hoogendoorn_at_ct.tudelf
t.nl)
Faculty of Civil Engineering and Geosciences
2
Presentation overview

• Project objectives
• Features of TSL
• Functional architecture
• Prototyping examples

3
Example research questions

• What are determinant choice factors in hybrid
  traffic networks?
• How does uncertainty influence choice behavior?
• How does paying for infrastructure affects its
  use?
• Do travelers anticipate onbehavior of other
  travelers?
• What is the impact of different types of
  information (road-side system, in-car navigation
  system,graphical user-interface, etc.) and
  control on travel choice?

4
Travel behaviour studies (SAS!)
5
Research objective

• Interactive simulator (hybrid approach combining
  benefits SP / RP) to study different types of
  adaptive travel choices in a controlled dynamic
  environment, e.g.
• Decision to acquire and use information, decision
  to pay for use of infrastructure, departure time
  choice, en-route route choice, mode-choice,
  destination choice, etc.
• Respondents can be subjected repeatedly to either
• Similar choice conditions over time or
• Completely different choice conditions
• Unlike traditional SP approaches, choice context
  is not hypothetical in sense that travelers will
  face (simulated) setting as well as consequences
  of decision-making

6
Pathdemandgenerator
Choicessynthetics
Day-to-day
Trafficsimulation
Participants
En-route
ITS measuressimulator
Networkperformance
Choicealternatives
Choice-setgeneration
7
Main features of laboratory

• Multi-user capabilities for simultaneous data
  collection from interacting respondents, to study
  behavioral differences between (different types
  of) travelers. System behavior can in part be
  derived from collective responses of involved
  actors
• Internet-based architecture ensures flexibility
• Ability to investigate within day / day-to-day
  behavioral dynamics (learning system behavior,
  information characteristics) for different
  uncertainty levels (daily fluctuations OD
  patterns, randomness choice behavior,
  fluctuations traffic conditions)
• Consideration transport mode alternatives joint
  consideration of urban and motorway networks
• Inclusion of pricing measures / mechanisms

8
Application examples

• Prototyping

9
Assignment for information lecture

• Route-choice simulator to study learning with and
  without information several scenarios
• Odd study number scenario 1
• Even study number scenario 2

10
Route choice simulator

• Make optimal route-choice (in terms of travel
  time)
• Best performance wins a bottle of cheap wine

11
Route choice simulator example

• Test travelers learning abilities using TSL
• Tree routes with different travel time
  distributions
• Route 1 uniform distribution 15,35 mean 25
  min, Pr(Tgt35)0
• Route 2 uniform distribution 27,29 mean 28
  min, Pr(Tgt35)0
• Route 3 Pr(T20)0.8. Pr(T50)0.2 mean 26 min,
  Pr(Tgt35)0.2
• Scenario 1 only information on performance
  chosen route
• Scenario 2 information on all routes

12
Test experiment

• To test set-up and gain some experience with the
  system, 20 respondents were asked to participate
• Using collected data, adapted Horowitz model was
  estimated
• Adapted Horowitz model assumes MNL model (Gumbel
  e)

13
Scenario 2

• Estimation of parameters Maximum Likelihood
  method
• Parameters are estimated using ALOGIT package
• Discussion of results
• Learning effects (parameter values)
• Statistical significance of results

Modeling effects ATIS
14
Scenario 1

• Scenario 1 only provide feedback on chosen route
• Use same model, but let
• (i.e. Sk(t-1) denotes the last experienced
  travel time of route k)

Modeling effects ATIS
15
Scenario 1

• Estimation of parameters Maximum Likelihood
  method
• Discussion of results
• Learning effects (parameter values)
• Statistical significance of results
• Comparison between scenario 1 and 2

Modeling effects ATIS
16
Beneficiaries

• TU Delft received grant from NWO to build TSL
• Potential beneficiaries TSL MD-PIT, PITA, AMICI
• Commonalities between programmes
• Study behavioral aspects / dynamic responses to
  information, control, financial stimuli (road
  pricing, pay lanes, information acquisition),
  etc.
• Share need for research tool to (experimentally)
  study within-day and day-to-day dynamic
  behavioral adaptations influence various
  stimuli and external conditions, e.g. traffic
  conditions (congestion), control, information
  (public / private, free / commercial), and
  pricing strategies, in controlled laboratory
  environment

17
State-of-affairs

• Project will be finished September 2003
• Achievements
• Inventory user-needs and requirements
• Functional specification
• 2 Prototypes
• Simulation software design
• User-interface review
• ICT architecture design

18
Questions and discussion