Vehicular Trajectory Optimization for Cooperative Collision Avoidance at High Speeds

Abstract

Traffic safety is a key aspect in new-generation intelligent transportation systems. Among other areas, an active field of research is cooperative collision avoidance, where vehicles cooperatively calculate trajectories under tight time constraints to avoid colliding under specific road-traffic domains (overtaking, intersections, etc.). In this paper, we particularly analyze the problem of collision avoidance in scenarios in which high-speed vehicles need to generate evasive maneuvers within very short time intervals to avoid or at least mitigate a hypothetical (multiple) collision. We pose this as a multiobjective optimization problem and simplify it by considering only lateral motion for the optimization process, thus having to solve a 1-D trajectory generation problem. The routes of vehicles are optimized according to a weighted aggregation functional that: 1) maximizes the lateral distances between vehicle-vehicle and vehicle-obstacle pairs at the time of overcoming the obstacles; 2) minimizes the lateral speeds at the end of the path; and 3) minimizes the instantaneous lateral acceleration (inertia) along the maneuver. In addition, we compute trajectories by following an optimization strategy that divides the problem into a set of independent subproblems, which are optimized in parallel by using a gradient-descent-based methodology. From this set of solutions, the most suitable option, according to our selected criteria, is chosen. Results show the utility of our approach and its flexibility to compute evasive trajectories adapted to different requirements. Additionally, a simulation of the mechanical response of the vehicles during the evasive maneuvers is conducted.