Development of a fuzzy-expert system for incident detection and classification

This research presents the exploratory results of using fuzzy-expert systems for incident detection and classification. The proposed system functions to detect not only the occurrence of incidents, but also their located lanes, and the resulting type of severity. With such information, the traffic control center can better advise drivers to take necessary lane changes and take timely actions for minimizing the incident impacts on traffic conditions. Although the current research remains exploratory in nature, as both the fuzzy membership functions and key parameters were determined empirically rather than fine-tuned with a neural network or genetic algorithm, the preliminary results have confirmed the promising properties of the proposed system.