A conceptual framework of online collision prediction and avoidance system for signalized intersections is proposed. A fuzzy logic inference system is developed to perform online collision prediction, which is the core for the implementation of intersection collision alerting/avoidance technologies. The risk is quantified by three fuzzy inputs to the model, the time to collision, severity of the crash and the local perception of the drivers. Perception Index and Severity Index are integrated to reflect driver-vehicle system safety characteristics. An example at a hypothetical intersection is presented to demonstrate the proposed methodology for the red light running vehicle. Discussion of the results and some extensions of this research are recommended. (Author/publisher).
Samenvatting