Real-Time Detection of Hazardous Traffic Events on Freeways Methodology and Prototypical Implementation

Abstract

A novel surveillance system to detect hazardous traffic events on freeways is presented. This study developed a methodology to evaluate freeway safety performance on the basis of vehicle trajectory data and a prototype implementation. An important feature of this study is to capture unsafe traffic situations of car-following and lane-changing events, which potentially lead to collisions. The proposed methodology used a real-time safety index based on the concept of safe stopping distance and time to collision. A video image processing-based detection system using a long-distance tracing algorithm was also developed to implement the proposed methodology. The outcomes of this study would be valuable for freeway traffic operators in evaluating traffic conditions in terms of safety in real time. The system is expected to be used to support the development of effective traffic management strategies for safety enhancement on freeways.