This report describes results of the first of two years of research on computer algorithms that make use of freeway surveillance data to detect incidents. The research approach is based upon an extensive data base obtained from loop detectors on freeways in the Los Angeles area. The results are presented in three parts: Part I: Background and Evaluation Criteria; Part II: Evaluation of Existing Incident Detection Algorithms; Part III: Preliminary Results in the Development of New Incident Detection Algorithms.
Abstract