Performance Evaluation of Travel-Time Estimation Methods for Real-Time Traffic Applications

Abstract

Travel time for an itinerary constitutes one of the most relevant roadway traffic metrics. Numerous studies have been conducted to estimate travel times on the basis of data from loop detectors. This article focuses on evaluating the performance of a set of estimation methods—algorithms that estimate route travel times using specific speed data from dual loop detectors for real-time applications, such as displaying travel times on changeable message signs. The travel-time estimates are evaluated against probe vehicle data obtained from FasTrak in the San Francisco Bay Area to determine the accuracy of the estimates. The study reveals that (a) the dispersion of travel time is often larger in off-peak periods than in peak periods for heavily congested freeways; (b) assessed against the same set of speed data, the evaluated algorithms have similar performance; (c) the accuracy of travel-time estimates on the basis of loop-detector data is better in off-peak periods than in peak periods; (d) lane-by-lane speeds from loop detectors may be used to improve travel time estimation; and (e) travel time estimation quality is more sensitive to actual sensor locations if sensors are sparse.

Keywords:

• median absolute deviation
• probe vehicle
• sensor location
• sensor spacing
• travel-time estimation
• travel-time evaluation
• travel-time representation

This study is partially supported by a grant from the California Department of Transportation to the California PATH Program. The contents of this article reflect the views of the authors who are solely responsible for the facts and the accuracy of the results presented herein.