A practical decentralized access protocol for autonomous vehicles at isolated under-saturated intersections

Abstract

The majority of research efforts in the field of access control of autonomous vehicles at intersections are geared towards fully connected vehicles. The underlying assumptions for such efforts are active vehicle-to-vehicle (V2V), vehicle-to-infrastructure (V2I), infrastructure-to- vehicle (I2V) communications, and/or presence of a central controller. Though efficiency is proven to be significantly enhanced, the assumptions face inherent security and privacy obstacles and require high infrastructure costs. In previous work, the authors designed and demonstrated a simpler, less costly, and more secure approach to autonomous vehicle management at intersections. The approach allows vehicles to make autonomous decisions at intersections based solely on sensing and/or beacon information with no V2V or V2I communications required. This article extends our model to account for various vehicle classes, all possible turns at the intersection with corresponding safe turning speeds, and various intersection geometries. Compared to a fully actuated signal controller, the proposed and improved model is again proven operationally more efficient, as it reduced the average delay per vehicle by at least 21% and up to 51% for the various simulated scenarios. After 40 million seconds of simulation, the proposed model proved collision free operations.