Membership-based Manoeuvre Negotiation in Autonomous and Safety-critical Vehicular Systems

Abstract

This thesis presents an extension to a communication protocol proposed by Casimiro and Schiller [1] for manoeuvre negotiation amongst autonomous vehicles. Local traffic rules establish a priority ordering amongst vehicles sharing a section of road. The manoeuvre negotiation protocol establishes procedures for how lowpriority vehicles can seek permission from high-priority vehicles to enter shared areas of road, in order to avoid collisions even in the presence of some common network failures. The extensions proposed in this thesis adapt the protocol to be used in intersections and increase safety in traffic situations where vehicles do not maintain an ideal speed. This protocol is run alongside a risk estimator that continuously estimates the likelihood of a collision between vehicles. When a high risk situation is detected, the risk estimator will trigger an emergency breaking event. The combined system was tested in simulations where two vehicles approached an intersection from opposite directions and the intended path of one vehicle turns left across the intended path of the other vehicle. Variables to model communication loss, noisy positional data and violations of traffic rules by one vehicle were incorporated into the simulations in order to evaluate the rigidity of the system. The combined system resulted in fewer collisions and near collisions than were observed when running the same simulations using either the manoeuvre negotiation protocol or risk estimator separately. Furthermore, the combined system was shown to adhere to the established local traffic rules. The results indicate that a combined risk estimator and manoeuvre negotiation protocol can create a more robust safety system than one that relies solely on either component. A correctness proof is also provided in this thesis to demonstrate analytically that the protocol increases safety by both reducing the risk collision between vehicles and preventing violations of the established local traffic rules.