METRANS UTC

Analysis and Prediction of Spatiotemporal Impact of Traffic Incidents for Better Mobility and Safety in Transportation Systems

Project Number

14-04

Project Summary

In this research, we study a machine learning approach to predict the spatiotemporal impact of traffic accidents on the upstream traffic and the surrounding region.  The main objective of our research is to forecast how and when the travel-time delays - caused by road accidents - will occur on the transportation network in both time and space.  Towards this end, we will conduct fundamental research in mining and correlation of traffic incidents and sensor datasets that we have been collecting and archiving in the last past three years.  Furthermore, to demonstrate the benefits of our research, we will develop a novel proof-of-concept mobile application and extend our existing web based application to enable monitoring and querying of the incident impacts on real-time and historical datasets.  This research will exploit the real-world Los Angeles traffic sensor data and California Highway Patrol (CHP) accident logs collected from Regional Integration of Intelligent Transportation Systems (RIITS) under Archived Traffic Data Management System (ADMS) project.

Project Status

Complete

Year

2014

Topic Area

Integrated Freight and Passenger Systems

P.I. Name & Address

Professor and Director of the Information Laboratory (InfoLAB), Department of Computer Science; USC Viterbi School of Engineering
University of Southern California
3737 Watts Way
Charles Lee Powell Hall (PHE) 306A
Los Angeles, CA 90089-0781
United States
shahabi@usc.edu

Co-P.I.

Research Scientist, Integrated Media Systems Center; USC Viterbi School of Engineering
University of Southern California
3737 Watts Way
Charles Lee Powell Hall (PHE) 335
Los Angeles, CA 90089-1211
United States
demiryur@usc.edu

Funding Source(s) and
Amounts Provided (by each agency or organization)

Caltrans

$99,999

Total Project Cost

$99,999

Agency ID or Contract Number

Grant No: 65A0533

Start and End Dates

1/1/2015 to 12/31/2015

Brief Description of
Research Project

In this research, we study a machine learning approach to predict the spatiotemporal impact of traffic accidents on the upstream traffic and the surrounding region.  The main objective of our research is to forecast how and when the travel-time delays - caused by road accidents - will occur on the transportation network in both time and space.  Towards this end, we will conduct fundamental research in mining and correlation of traffic incidents and sensor datasets that we have been collecting and archiving in the last past three years.  Furthermore, to demonstrate the benefits of our research, we will develop a novel proof-of-concept mobile application and extend our existing web based application to enable monitoring and querying of the incident impacts on real-time and historical datasets.  This research will exploit the real-world Los Angeles traffic sensor data and California Highway Patrol (CHP) accident logs collected from Regional Integration of Intelligent Transportation Systems (RIITS) under Archived Traffic Data Management System (ADMS) project.

Describe Implementation of Research Outcomes (or why not implemented)

image

Our research resulted in efficient machine learning algorithms and software to accurately model the current and future state of travel-times in the presence of events (e.g., accidents, weather conditions and concerts).

Impacts/Benefits of Implementation (actual, not anticipated)

The outcome of our research has been used in a route planning prototype application developed in our lab. Our experiments with real-world data showed that such impact quantification helps to alleviate traffic congestion caused by accidents by providing alternative paths to drivers. We have shared the preliminary results of research with transportation research community and professionals at Annual Transportation Research Forum 2015 in Atlanta, GA.

Web Links, Reports, Project website

http://infolab.usc.edu/research.php#traffic_prediction

http://www.metrans.org/research-projects/metrans-utc