Intelligent Traffic Management Systems

1
Intelligent Traffic Management Systems
Khaled AlMejalli
Supervisors Dr. Keshav Dahal Dr. Almgir Hossain
August 2006
2
Outline

• Introduction.
• Intelligent Transport Systems.
• General Structure of the Research.
• Literature Review.
• Decision Support Model.
• Next Step.

3
Introduction

• Problem
• The growth of the number of vehicles The
  increase of the need for transportation More
  traffic problems ( Congestions, Accidents , and
  Pollution)
• Solution
• S1. Extending the road network (adding lanes,
  creating new freeways).
• S2. Using Intelligent Transport Systems to mange
  the existing traffic network efficiently and
  safely .

4
Traffic Control Centres

• TCC are connected on-line to devices ( detectors,
  cameras, traffic lights, etc).

• TCC receives recent traffic status ? Traffic
  data (speed, flow, occupancy, etc) ?
  Environmental conditions (air, ground tem., etc)
  ? Information about current state of control
  devices
• TCC operators should ? Detect the presence of
  problems and their possible causes. ?
  Determine control actions to solve or
  reduce the severity of the problem.

5
Intelligent Transport Systems
New technology to mange the traffic network using
intelligent computing, communications
technologies and real-time data.
Using intelligent systems to process the
information
Collecting information about the current state
of the transport network using real-time data
Managing the network - directly (traffic
signals, VMS). - indirectly (travel news)

• Improve the decision making process.
• Make services more reliable.
• Provide accurate real-time information.
• Reduce accidents.
• Reduce pollution.
• Help drivers to find the best rout to their
  destination.

6
General Structure of the Research
Real Time Traffic Management
Control Evaluation Mode
Problem Detection Model
Decision Support Model
Historical Data
Guidance
Long Term Traffic Management DSS
Recommendation
High Level Management
7
Literature Review

• Several authors have described traffic controls
  and decision support systems for traffic
  management, such as TRYS (Cuena, Hernandez et al.
  1995 Molina, Hern A et al. 1998). TRYS is a
  knowledge representation environment for building
  intelligent traffic management systems
  applications for urban motorway control.
• Traffic incidents are a major cause of traffic
  congestion. In order to solve this problem,
  different algorithms have been developed for
  detection traffic problem using various
  intelligent techniques, such as the neural
  network technique (Khan and Ritchie 1998 Wen,
  Yang et al. 2001 Srinivasan, Jin et al. 2004),
  Fuzzy logic technique (Weil, Garcia-Ortiz et al.
  1998 Daehyon, Seungjae et al. 2005), .
• Also fuzzy logic technique has been used to
  control traffic signals, road junctions, ramp
  metering , and to assist the operators of the
  traffic control system to efficiently manage
  non-recurrent congestion (Heung and Ho 1998 Wei,
  Zhang et al. 2001 Niittymaki and Nevala 2001
  Hegyi, De Schutter et al. 2001).
• De et al (De Schutter, Hoogendoorn et al. 2003)
  have used a case base and fuzzy interpolation to
  develop a case-based traffic control scenario
  evaluation system that can be used by traffic
  operators to asses the approximate performance of
  several control scenarios. For similar propose
  Chai Quek el al (Chai Quek el al 2006) have used
  the fuzzy neural technuqe.

8
Proposed DSS
Problem DetectionModel (Incident Detection)
ControlStatus
Traffic situation
Decision Support Model
DatabaseHistorical Data
ControlObjectives
Control Evaluation Model(Traffic Simulation)
Best controlmeasures
Traffic Operator
9
Decision Support Model
Input Layer(Crisp inputs)
Condition Layer(Input Membership Fun.)
Hidden Layer(Fuzzy Rules)
Consequence Layer(Output Membership Fun)
Output Layer(Defuzzification)
M
Time
E
R1
O
H
L
R2
M
Density
M
Speed Ave.
L
H
R3
L
H
Incident S
M
Travel Time
M
R4
H
L
C1
Rn
C2
Control P
C3
Cn
R1 If Time is Morning and Density is Medium and
Incident_Severity is Low and Control_ Plan is C1
Then Speed Ave is High and Travel Time is
High. R2 If Time is Evening and Density is High
and Incident_Severity is High and Control_ Plan
is C2 Then Speed Ave is Medium and
Travel Time is Low. R3 If Time is Off_Peak and
Density is Low and Incident_Severity is Medium
and Control_ Plan is C1 Then Speed Ave
is High and Travel Time is Medium.
10
The Next Step

• Developing the proposed decision support model.
  ( Neuro-Fuzzy system ).
• Getting some real traffic data. ( or using a
  traffic simulation model).
• Using different learning algorithms to train the
  proposed system.
• Comparing the result with other existing systems
  ( e.g. Case-Based system developed by Schutter
  et al).

11
Thank you very much