Application of fuzzy logic for detecting incidents at signalized highway intersections

Traffic incidents - i.e., accidents, cargo spills, and vehicles stalled - are a major cause of urban congestion. Research in incident detection for signalized arterial streets is at a very initial stage. Existing algorithms are far from being robust in dealing with the difficulties related to data availability and the multi-dimensional nature of the incident detection problem. This paper discusses a fuzzy-logic based incident detection model for real-time implementation at signalized diamond interchanges. The model overcomes the boundary condition problem inherent in conventional threshold-based methods. The model captures system-wide incident effects utilizing multiple measures for more accurate and reliable detection, and is incorporated into a real-time traffic adaptive diamond interchange control system. The model is scaleable and expandable for larger systems of arterial streets. Simulation results suggest that the fuzzy-logic based approach to incident detection at signalized diamond interchanges is promising.