Analysis of Freeway Accident Detection

Abstract

Instrumented traffic management can assist in the detection of freeway incidents and reduce the time required to initiate effective traffic management strategies and emergency response measures. Although instrumented freeway traffic management is concerned primarily with general incidents, reportable vehicle accidents are the focus of this research. Reportable accidents account for 20 percent of all freeway incidents and give rise to much of the nonrecurrent traffic congestion experienced on many freeways. Explored here is how the use of various accident-detection criteria, such as change in speed, vehicle occupancy, and traffic volume, affects the time to detection for a mix of factors (preaccident traffic characteristics, accident lane-blockage pattern, position and distance of detector with respect to each accident). A representative sample of Toronto freeway accidents for 1994 was analyzed using analysis of variance. The results of this analysis suggest ways in which instrumented detection of freeway accidents can be made more efficient by reducing the time to detection.