Paramics User Group at the University of Toronto

1
Paramics User Group at the University of Toronto
2008 Paramics North American User Group
Meeting Rutgers University Busch Campus,
Piscataway, New Jersey July 14th15th
Mohamed Wahba, Ph.D. Candidate Researcher and
Lecturer Civil Engineering Department University
of Toronto
2
Outline

• Motivation and Scope
• Paramics Applications
• Mixed-Reality Driver-Simulator for Travel
  Behaviour Modelling with Information Provision
• MILATRAS MIcrosimulation Learning-based Approach
  to TRansit ASsignment

3
Motivation and Scope
ITS Supply Representation
Challenges of ITS Technologies
Traveler Response Modeling
Supply and Demand Interaction
4
Mixed-Reality Driver-Simulator
The objective is to enable credible evaluation of
route choice behavior under various ITS
applications
5
PARAMICS Microscopic Traffic Simulator

• Offers virtual reproduction of full-scale traffic
  networks
• But, lacks behavioral realism

6
Driver Simulators

• Allow for the direct testing of real subjects
• But, lack network realism

7
Mixed Reality Traffic Analysis Platform - IDEA
Integration
Driving simulator
Microscopic traffic simulation
Mixed Reality System
8
Mixed Reality Traffic Analysis Platform - Features

• Real driver
• Full scale realistic traffic network model
• Ability to execute routing decisions through a
  steering device

Mixed Reality System
9
Mixed Reality Traffic Analysis Platform - Features

• External control of the lateral and routing
  movements of a single vehicle
• a simulated traffic network,
• a PC steering input device connected to the
  Paramics micro-simulator

Mixed Reality System

• Longitudinal control of the driven vehicle and
  surrounding vehicles through Paramics.

10
Mixed Reality Traffic Analysis Platform - System
Architecture

• Translates external acts on the input device
  into numerical values

Input Capturing

• Translates numerical values from IC into
  movement directions

Communicator (C)
Shared Memory
IPC
Communicator (C)

• Allows for external control of routing decisions

Paramics API Plugins

• Allows for information dissemination

• Controls simulator/driver interface

11
Mixed Reality Traffic Analysis Platform - In
Action

• Simulator/Driver Interface

Driven Vehicle identified by a circle drawn on top
Attached Steering wheel
12
Mixed Reality Traffic Analysis Platform -
Application
13
Mixed Reality Traffic Analysis Platform -
Application
14
Mixed Reality Traffic Analysis Platform -
Application
Gardiner Expressway en-route divergence node
Pre-trip Decision
Lake Shore Blvd. en-route divergence node
15
Mixed Reality Traffic Analysis Platform -
Application

• Pre-trip Decision
• ? Anticipated Congestion States (Incident
  Probabilities)

Pre-trip Decision Node
No Incident (L)
Incident (H)
Incident (H)
No Incident (L)
Incident (H)
No Incident (L)
L
Divergence node
M
H
16
Mixed Reality Traffic Analysis Platform -
Application

• En-route Decision
• ? Anticipated Congestion States (Incident
  Probabilities)

Pre-trip Decision Node
No Incident (L)
Incident (H)
L
Divergence node
M
L
H
H
17
Mixed Reality Traffic Analysis Platform -
Application

• Choice Variables
• Travel Time
• No Incident, based only on pre-specified demand
• Incident, based on pre-specified demand
  incident timing and duration
• Travel Distance, Fixed for each route
• Freeway usage, Fixed for each route

18
Mixed Reality Traffic Analysis Platform -
Application

• Information Provision
• Information Type
• No Info
• Descriptive Info, Gardiner Moving Slowly
• Prescriptive Info, Gardiner Travel Time is
  22mins
• Information Reliability
• Reliability level, Probability of providing
  correct info
• For each subject there is a pre-specified
  reliability level

19
Mixed Reality Traffic Analysis Platform - Tools
Mixed Reality
Map-based

• Pre-trip Choices with no information
• En-route choices with a random information form
  and stochastically reliable content

• Pre-trip Choices, with/without information
• Information forms are randomly generated with
  stochastically reliable content

20
Mixed Reality Traffic Analysis Platform -
Application Results

• Recorded data (for each subject, for each trip)
• Chosen Route
• Captured by continuously tracking the driven
  vehicles current link.
• Deliberation Time (time taken to make a decision)
• Captured by recording the time elapsed from info
  display until an action is executed. (Subjects
  are asked to press a decision button once a
  decision is reached)

21
Mixed Reality Traffic Analysis Platform -
Application Results

• Collected data for 30 subjects are used in the
• Statistical analysis of route choice behavioral
  patterns under different situational/personal
  factors
• Development of a dynamic route choice
  deliberation model based on the theoretical
  ground of a state-of-the-art behavioral decision
  theory

22
Mixed Reality Traffic Analysis Platform -
Application Results
Map-based ? Increased Gardiner Expressway Choice
Percentage (G) reflects a tendency to stick to
the same choice without actually deliberating
23
Mixed Reality Traffic Analysis Platform -
Application
Mixed Reality ? Longer deliberation time frames
reflect a more cautious or serious choice attitude
24
Mixed Reality Traffic Analysis Platform -
Application

• Main Conclusion
• Mixed reality platform has the potential to
  enhance the realism of in-lab simulated route
  choice experiments and hence improve the
  credibility of collected data
• Future Extensions
• Incorporation of external longitudinal control,
  such as speed and acceleration control, using a
  game type pedal as an input device.

Please contact Hoda Talaat (hoda.talaat_at_utoronto.c
a) or Baher Abdulhai (baher.abdulhai_at_utoronto.ca)
for more details
25
Motivation and Scope
ITS Supply Representation
Challenges of ITS Technologies
Travelers Responses Modeling
Supply and Demand Interaction
26
Transit Planning Operations
The objective is to advance the
state-of-the-practice of transit service
microsimulation for transit-ITS modelling
Another objective is to enhance the transit
modelling capabilities of Paramics
27
Traffic Improvement Projects
Impacts on link flows and bus loads
28
Transit-ITS Projects
more efficient, more reliable
modelling framework for with and without
Service improvement technologies
(Automated) Traveller Information Systems
(Dynamics) Service Characteristics
Travellers Behavior (decisions and responses)
29
Transit-ITS Projects
TSP
ATIS
modelling of Information Provision
modelling of Transit Signal Priority
impact on speed (and travel time)
impact on passengers response
impact on (of) transit service (passengers
response)
impact on general traffic (e.g. G/C)
Generation Evaluation
30
Transit Modelling in Paramics

• Paramics, like many microsimulation tools, has a
  principal focus on auto traffic simulation.
• For transit demand modeling
• Paramics does not support transfers between
  transit routes
• Paramics does not support tracking individual
  passenger identities
• The number of alighting passengers at any transit
  stop is determined as a percentage of the
  stopping transit vehicle occupancy
• The input for transit demand can only be entered
  at the stop level, not the zone level
• Paramics does not perform transit assignment

31
MILATRASMIcrosimulation Learning-based Approach
to TRansit ASsignment
Schedule delay
Scheduled Arrival Time
in-vehicle time
egress walking time
transfer waiting time
origin waiting time
The passenger adjusts travel decisions to arrive
at SAT, while minimizing travel cost
access walking time
Origin Departure Time
32
MILATRASMIcrosimulation Learning-based Approach
to TRansit ASsignment

• Individual Passengers are expected to adjust
  their travel behavior (i.e. trip choices)
  according to their knowledge and experience with
  the transit system conditions

Passengers decide about their choices for a trip
on consecutive days based on a mental model of
the transit system network conditions, which is
built over time through experience with the
transit system
33
MILATRASMIcrosimulation Learning-based Approach
to TRansit ASsignment
Scheduled Arrival Time
e.g. 900am
Network Structure
715
710
705
700
Origin Departure Time
Network Conditions
34
MILATRASMIcrosimulation Learning-based Approach
to TRansit ASsignment
NODE, LINK, ZONE, VEHICLE, BUSSTOP, DETECTOR,
CARPARK, BEACON, LOOP, SLIP, RAMP.
PASSENGER
35
MILATRASMIcrosimulation Learning-based Approach
to TRansit ASsignment
36
MILATRASMIcrosimulation Learning-based Approach
to TRansit ASsignment

1. represents each passenger as an agent/object,
   with attributes
2. accepts transit demand at the zone level, not the
   stop level
3. is capable of tracing every passenger-agent
   through the transit network
4. supports transfers between routes
5. deals with boarding and alighting at the
   passenger level
6. provides passenger-specific measurements and
7. models the (adaptive) behavior of transit riders
   when faced with a departure time choice, stop
   choice or run choice

37
MILATRASMIcrosimulation Learning-based Approach
to TRansit ASsignment
38
MILATRAS ApplicationMIcrosimulation
Learning-based Approach to TRansit ASsignment

• Base-Case Scenario No Information Provision
• Scenario I, Information at Boarding Stops
• Scenario III, Information On-Board
• Scenario IV, Pre-Trip Information

• Travel Time Prediction Model (supply side)
• Information and Experience Integration Model
  (demand side)

39
MILATRAS ApplicationMIcrosimulation
Learning-based Approach to TRansit ASsignment
40
MILATRAS ApplicationMIcrosimulation
Learning-based Approach to TRansit ASsignment
41
Advancing MILATRASMIcrosimulation Learning-based
Approach to TRansit ASsignment

• Ongoing Research
• Experimenting with different path choice
  mechanisms - e.g. learning methods
• Incorporating mode choice/switch into framework
• MILATRAS provides a consistent way of combining
  traffic and transit in a simultaneous modeling
  framework therefore, it is able to represent the
  impact of roadway congestion on transit service
  and vice versa.
• Future Research
• Extension of MILATRAS for optimal service design
  e.g. transit network route design, timetable
  design, etc.

42
Thank You