As of October 1, 2016, the SWUTC concluded its 28 years of operation and is no longer an active center of the Texas A&M Transportation Institute. The archived SWUTC website remains available here.

600451-00083

SWUTC Research Project Description

Improving the Reliability of Automated Freeway Incident Detection Using Multiple Real Time Data Sources

University: University of Texas at Austin

Principal Investigator:
Randy Machemehl
Department of Civil and Environmental Engineering
(512) 471-4541

Project Monitor:
Carlos Lopez
Vice President
HNTB
Austin, TX

Funding Source: USDOT and State of Texas General Revenue Funds

Total Project Cost: $75,900

Project Number: 600451-00083

Date Started: 1/1/13

Estimated Completion Date:  12/31/13

Project Summary

Project Abstract:
Freeway incident detection and clearance activities are very important parts of all urban traffic control centers.  Traditionally, control centers have relied upon inductance loop detector technology for real time sensing of fundamental traffic parameters.  Incident detection algorithms have been developed to use loop detector data to identify incidents.  Although these systems tend to be very reliable under high speed low traffic demand operations, they tend to be much less reliable under congested variable speed and flow situations.  The lack of reliability during congested freeway conditions is highly problematic since incidents are more numerous under those conditions and delayed identification and clearance causes significant user costs.  A new generation of detection systems using video and/or radar can provide more robust real time descriptions of traffic operations and algorithms based upon area rather than point sensing can significantly improve reliability of incident detection systems.  This study will select one more off the shelf detector system(s) and develop a hybrid algorithm for incident detection.  The resulting system will be tested in a simulation environment where conditions can be controlled.

Project Objectives:

The study encompasses three objectives:

  1. Identify one or more detector systems that offer potential improvements in sensing that can be incorporated into a next generation of incident detection algorithm.
  2. Develop a next generation incident detection algorithm that will significantly improve reliability of incident detection through application of the new detector system(s).
  3. Compare the reliability of the new detector-algorithm system to existing inductance loop detector based systems using either simulation or field-testing.

Task Descriptions:

Task 1.  Review of current incident detection concepts.
An updated review of incident management algorithms and detector systems will be developed.  This effort will expand upon and update the review that has already been developed as part of this proposal.  The product will be a description of the hardware and software systems currently being used, or proposed, and a summary of the reliability and cost effectiveness of the current systems.  This review will provide the basis from which the following tasks will be launched.

Task 2.  Identification of candidate detector systems.
With the objective of identifying an “off the shelf” detector hardware system that could improve reliability and cost effectiveness of existing systems, a search of literature and experiences of user agencies has already been done.  That effort has identified several candidate systems as indicated in the background section of this proposal.  This task will include updating those findings and expanding them to include the most current findings by agencies performing incident management activities.

Task 3.  Algorithm Development.
Based upon the characteristics of the chosen detection system(s) one or more candidate algorithms will be selected or developed.  According to Parkany’s review of incident detection algorithms (Parkany, 2005), commonly recognized algorithms can be grouped into seven categories.  These algorithm categories include (1) comparative, (2) statistical, (3) time series, (4) filtering/smoothing, (5) traffic modeling, (6) artificial intelligence, (7) image processing, and a family of algorithms using probe vehicles.  All of the categories except the probe vehicle family use loop detector or loop-emulating data collected at fixed points along the subject freeway.

Since this effort is targeted at data produced by a detection system other than fixed-point loop detectors, a new or hybrid version of the existing techniques is envisioned.  Several of the seven categories appear to provide potentially useful approaches to the re-configured problem presented by more robust detector information.

Task 4.  Algorithm-detector system evaluation.
The product of Tasks 2 and 3 will be evaluated through a detailed testing process.  Testing will include the reliability of the system to detect freeway incidents during traffic conditions ranging from light flows with high consistent speeds to very heavy flows with low speeds and variable speeds.  Initial testing will be performed within a micro-simulation environment where a designed set of conditions can be produced and replicated.   Based upon discussions with officials at both the City and State level, a field test can likely be arranged but implementation of a field-testing program will depend upon the willingness of those officials to participate.

Task 5.  Preparation of comprehensive final report.
Since this work is planned as a doctoral dissertation, the dissertation will provide the basis for the final report.  Additionally, at least two refereed journal papers are planned.  The final research report will provide a complete description of the problem, approach, methodology, findings, conclusions, and recommendations, developed in the project and will completely document all data gathered, analyses performed, and results achieved.


Implementation of Research Outcomes:
Automated incident detection has been studied for over 30 years and most urban traffic control centers have some form of this technology in operation. Most of these techniques are quite accurate when freeway speeds are high and density is low. However, under high density, low speed, stop and go, conditions, undetected incidents and false alarms are common. The technology developed here extends the state of the art providing a new tool that improves reliability under the critical low speed, high density traffic conditions.

Products developed by this research:

Model Developed:  Through this work, researchers developed an incident detection model designed to have significantly better reliability than existing algorithms during very high density, stop-and-go, freeway driving conditions.  Ultimately, the technology could significantly improve the response time of emergency and enforcement activities to freeway incidents. This improvement in emergency response to incidents will reduce the number of secondary collisions and generally mean that user time lost in congestion due to incidents will be reduced.

Undergraduate and Graduate Curricula Developed:  The methodology developed by this research is being implemented in the curricula for a graduate and an undergraduate traffic engineering course within the University of Texas Transportation Engineering program.  Study results have led to new course problem sets and field tests to be performed by the students in the classes.

Planned Presentation:  A Freeway Incident Detection Algorithm Using Dynamic Time Warping, Motamed, Moggan and Randy Machemehl, paper to be offered for presentation and publication at the January 2015 Transportation Research Board Meeting, Washington, DC, January 2015.

Planned Presentation:  Pilot Testing of a Freeway Incident Detection Algorithm Using Dynamic Time Warping, Motamed, Moggan and Randy Machemehl, paper to be offered for presentation and publication at the June 2015 Canadian Society for Civil Engineering Meeting, June 2015.


Impacts/Benefits of Implementation:
Travelers using urban freeways are severely impacted by incident based congestion and travelers are critical of the lack of timely response by enforcement and emergency agencies. Delayed responses are often due to the fact that incidents, particularly minor crashes, are “detected” through cell phone reports of the involved travelers. The new technology will enable control centers to detect the problem essentially as it happens and greatly impress those involved with the responsiveness reducing user time loss, secondary collisions and positively impact the opinions of travelers.

The research team is currently working with the Texas Department of Transportation to implement the new technology in traffic control center operations.


Web Links:

Final Report