Evaluating Policies for Reservation-based Intersection Control

Autonomous vehicle navigation in urban road networks will be possible in the future, making traveling yet another activity that does not need the human intervention. In this paper we empirically evaluate different policies to manage a reservation-based intersection, an infrastructure facility proposed by Dresner and Stone to regulate the transit of vehicles through intersections. We evaluate two different scenarios, a single intersection and a network of intersections, comparing the original policy employed by Dresner and Stone in their work with 4 policies, inspired by the adversarial queueing theory (AQT).