Automatic incident detection: Methods and practices
COHEN
Type de document
ARTICLE A COMITE DE LECTURE NON REPERTORIE DANS BDI (ACLN)
Langue
anglais
Auteur
COHEN
Résumé / Abstract
This paper describes the research work carried out on aid methods as well as the practices currently in use. It examines first the technological evolution of the traffic monitoring systems. Among the different approaches considered today, the video technology appears to be the most efficient. Nevertheless new developments and improvements of the traditional approaches are also of interest. The main categories of algorithms using point-based data are described in particular the California algorithm which remains a reference. The calibration of the thresholds of these algorithms is based on an optimization process which is described and illustrated for a concrete example. General trends concerning the average of the potential performances of the different systems also are given. Sensitivity to a large number of factors, like weather, lighting and visibility, road geometrics and traffic level is underlined. The conclusion suggests some possible new directions for the practical implementation of these systems.
