RADAR-BASED IDENTIFICATION OF VEHICLES' CLASSES FOR MICROSCOPIC SIMULATION IN TRAFFIC PREDICTION

Modelling and simulation of motorway traffic by means of the so-called microscopic models (or even by hybrid models, which introduce discrete events in a continuous simulation) requires the definition of vehicle and driver behavioural profiles, which reflect some statistical distribution. Even though the generation of events in the simulator takes place on an individual basis, the aggregated characteristics of the traffic flow, which are produced by the global interaction of the individual trajectories, reflect the choices made in the configuration of traffic parameters. If such a simulation tool could be fed with perturbations observed in real time with great precision, it might be capable of predicting the effects on the general traffic parameters in advance, by running at higher speed than the physical phenomenon. The paper highlights the possibilities offered in this sense by the presence on the motorway of a radar-based tracking system, capable of providing a continuous picture of traffic in space and time.