Cooperative highway traffic: multi-agent modelling and robustness to local perturbations.

As cooperative systems, a.k.a. connected vehicles, enable the communication and exchange of information between vehicles and infrastructure, it is expected that their communication capabilities can lead to a better active traffic management on urban motorways. In such a context, technological constraints must be the basis for any management strategy. If it has been analytically proven that communication can help stabilize traffic flow at a microscopic level, it is interesting to evaluate realistic communication strategies taking into consideration multiple perturbations such as sensors faults or driver cooperation. In this paper, a three-layer multi-agent framework is used to model and control the homogenization of traffic flow. The physical layer coordinates the vehicles dynamics based on a cooperative car following model. This layer includes cooperation derived from the communication and trust layers that respectively manage information and its reliability. Simulation results highlight the positive impacts of communication and control on traffic flow stability.