The Hanshin Expressway Public Corporation (HEPC) has been making efforts to reduce multiple-vehicle accidents by installing abnormal incident detecting equipment at points frequented by accidents. However, traffic accidents may occur at other than such accident-prone points. So, the HEPC collected data at 30-second intervals from its existing ultrasonic vehicle detectors for traffic data measurement and verified three different algorithms for accident detection. As a result, the HEPC was able to make clear the characteristics of traffic flow when an accident has occurred, based on which it has succeeded in building a foundation for a new function to detect accidents on the through lanes.
Abstract