Vehicle Routing through Deadlock-free Policy for Cooperative Traffic Control in a Network of Intersections: Reservation and Congestion

This paper considers a network of intersections, traversed by autonomous vehicles. Vehicles negotiate the right-of-way before traversing each intersection. The cooperative traffic management at intersections allows high traffic performance at a single intersection. This paper extends the issue of the traffic control efficiency to a network of intersections. It proposes a hierarchical architecture of the traffic control for preventing traffic from deadlock into a grid of intersections. Then, it discusses policies for routing vehicles in the network. Three policies are considered: shortest path, reservation and congestion avoidance. Furthermore, three traffic contexts are considered and simulated. Since, the traffic management at intersections is adapted to the currently observed vehicles, then the simulation shows that it is required to define new policy that considers the overall network performance rather than optimizing the route of each vehicle according to a given criteria. Indeed, reorienting vehicles according to a given policy can increase the number of conflicts and risks of cycles.