Representation of the Agent Environment for Traffic Behavioral Simulation.
Résumé
The aim of this paper is to improve the validity of traffic simulations in (sub-)urban context, with a better consideration of driver behavior in terms of anticipation of positioning on the lanes and occupation of space. We introduce a model based on a multi-agent approach and the emergence concept. This model considers that each driver perceives the situation in an ego-centered way and readapts the road space using the virtual lane concept. We implement the model with the traffic simulation tool ArchiSim. The so obtained simulator intends to reproduce the observed behavior such as filtering between vehicles (two-wheels, emergency vehicles), repositioning on lanes when approaching the road intersections and “exceptional” situations (stranded vehicle or improperly parked, etc.).