Enhancing Automatic Incident Detection Using Vehicular Communications

One of the fundamental requirements of a traffic management system is the ability to determine when an incident has occurred so that proper responses can be initiated. Most of the existing automatic incident detection techniques suffer from many limitations including their inability to detect incidents under non dense traffic conditions and generation of many false positive alarms. In this paper, we introduce a novel Bayesian-based approach to enhance the performance of existing techniques specially under non-dense traffic flow through vehicle to readside communications. The proposed technique also offers zero false positive alarms under most situations and can be integrated with any of the current techniques.