A Practical Approach to Decentralized Intersection Access Control for Autonomous Vehicles

In a shift from the current research trend which is mostly geared towards centralizing access control of autonomous vehicles at intersection of the near future, the authors extend their previously presented intersection management model, which is fully decentralized. Even though it is the most efficient, central control requires fully automated connected vehicles, active vehicle-to-vehicle and vehicle-to-infrastructure communications, which introduces inherent security, privacy and feasibility issues. We simulate autonomous intelligent vehicles making local access decisions at isolated intersections, with only sensing and/or beacon information. The simulation proves to be collision-free. The extended management model is now able to account for various vehicle classes, all possible turns at the intersection and safe turning speeds. Compared to a standard signal controller, the proposed and improved model is again proven to be operationally more efficient.