Optimal Path Planning for Connected and Automated Vehicles in Lane-free Traffic.

This paper develops a path planning algorithm for Connected and Automated Vehicles (CAVs) driving on a lane-free highway, according to a recently proposed novel paradigm for vehicular traffic in the era of CAVs. The approach considers a simple model of vehicle kinematics, along with appropriate constraints for control variables and road boundaries. Appropriate, partly competitive sub-objectives are designed to enable efficient vehicle advancement, while avoiding collisions with other vehicles and infeasible vehicle maneuvers. Based on these elements, a nonlinear Optimal Control Problem (OCP) is formulated for each ego vehicle, and a Feasible Direction Algorithm (FDA) is employed for its computationally efficient numerical solution. The OCP is solved repeatedly for short time horizons within a Model Predictive Control (MPC) framework, while the vehicle advances. It is demonstrated via traffic simulation, involving many such vehicles, on a lane-free ring-road that the proposed approach delivers promising results and can be considered as a candidate for use in further developments related to lane-free CAV traffic.