Signal-free Path-free Intersection Control for Connected Vehicles under Automated Driving

Abstract

We present a method for the signal-free path-free intersection control (SPIC) problem which involves coordinating connected vehicles under automated driving (CVAD) at urban intersections. A receding horizon planner is formulated as an extension of the nonlinear model predictive contouring control (NMPCC), referred to as ENMPCC. The ENMPCC method involves the inclusion of additional cost terms, traffic safety constraints, and the generation of multiple time-independent reference paths. The proposed method drops the constraints on the allowed paths within the intersection. This allows CVAD to move freely along optimal trajectories and paths, maximizing efficiency without compromising safety. Utilizing a polytopic representation of each CVAD and incorporating duality theory to formulate collision avoidance constraints, our method enables optimal and collision-free traversal of CVAD within the intersections. We demonstrate the proposed method in an urban driving scenario involving both straight and turning movements at an intersection. In comparison to the conventional path-based (lane-based) driving of the CVAD, our method shows a better use of intersection space resulting in lower travel time and fuel consumption as demonstrated through our simulation results.