Methodology for Evaluating Collision Risks Using Vehicle Trajectory Data

An innovative feature of this study is to propose a methodology for evaluating safety performance in real time based on vehicle trajectory data extracted from video images. The essence of evaluating safety performance is to capture unsafe car-following and lane-changing events generated by individual vehicles traveling within video surveillance area. The proposed methodology derived three indices including real-time safety index(RSI) based on the concept of safe stopping distance, time-to-collision(TTC), and the collision energy based on the conservation of momentum. It is believed that outcomes would be greatly utilized in developing a new generation of video images processing(VIP) based traffic detection systems capable of producing safety performance measurements. Relevant technical challenges for such detection systems are also discussed.