Data-Guided Distributed Intersection Management for Connected and Automated Vehicles

In this paper, we seek a scalable method for safe and efficient coordination of a continual stream of connected and automated vehicles at an intersection without signal lights. To handle a continual stream of vehicles, we propose trajectory computation in two phases - in the first phase, vehicles are constrained to not enter the intersection; and in the second phase multiple vehicles’ trajectories are planned for coordinated use of the intersection. For computational scalability, we propose a data-guided method to obtain the intersection usage sequence through an online "classification" and obtain the vehicles’ trajectories sequentially. We show that the proposed algorithm is provably safe and can be implemented in a distributed manner. We compare the proposed algorithm against traditional methods of intersection management and against some existing literature through simulations. We also demonstrate through simulations that for the proposed algorithm, the computation time per vehicle remains constant over a wide range of traffic arrival rates.