MPC
Research Projects (1999-00)

Identifying Number

MPC-185

Project Title

Incident Detection Algorithm Evaluation

University

University of Utah

Project Investigator

Peter Martin
University of Utah
(801)581-7144
martin@civil.utah.edu

External Project Contact

None

Project Objective

• Develop a set of requirements for incident detection algorithms
• Recommend an algorithm or combination of algorithms
• Recommend any necessary algorithm improvements to meet the requirements

Project Abstract

This project will research, evaluate and recommend an incident detection algorithm to operate in the TOC's of the Salt Lake Valley. The purpose of the algorithm is to notify TOC operators of potential incidents so they can respond quickly. The value of this is that incidents can be responded to quickly by incident management teams. This, in turn, has many benefits, including the reduction in the resulting traffic congestion and faster response to serious injuries. Several incident detection algorithms are in use in North America. This project will evaluate the current literature on the different algorithms and current incident detection performance measures. This information will be used to determine acceptable performance criteria for the use of incident detection algorithms in the Salt Lake Valley, which will in turn be used to evaluate each algorithm in more detail.

Task Descriptions

The first task in the project is to perform a literature search for the documentation available on as many algorithms as possible that are currently used in North America. This documentation may consist of evaluation studies, algorithm descriptions, system performance, and vendor information from proprietary systems.

The literature should yield a wide variety of information the different measures of effectiveness (MOEs) that can be used for preliminary evaluation and algorithm selection criteria. Some of the MOE's that may be used are the number of false calls, the number of accidents not detected, or the input requirements of an algorithm. Using these criteria, several of the best algorithms will be selected for further evaluation.

Milestones, Dates

• Starting Date: July 1, 1999
• Project Milestones:
• Interim Report: March 1, 2000
• Draft Research Report: May 1, 2000
• Final Research Report: June 1, 2000
• Ending Date: June 30, 2000

Yearly and Total Budget

USDOT $35,000
University $62,647
TOTAL $97,647

Student Involvement

Two graduate students and one thesis.

Relationship to Other Research Projects

N/A

Technology Transfer Activities

Seminar for Utah Department of Transportation data preparation personnel.

Potential Benefits of the Project

Speeding the detection rate of incidents will reduce congestion impacts and the likelihood of incident-related secondary accidents.

TRB Keywords

Traffic surveillance, automatic incident detection