Sampling-Based Optimal Trajectory Generation for Autonomous Vehicles Using Reachable Sets

Abstract

Motion planners for autonomous vehicles must obtain feasible trajectories in real-time regardless of the complexity of traffic conditions. Planning approaches that discretize the search space may perform sufficiently in general driving situations, however, they inherently struggle in critical situations with small solution spaces. To address this problem, we prune the search space of a sampling-based motion planner using reachable sets, i.e., sets of states that the ego vehicle can reach without collision. By only creating samples within the collision-free reachable sets, we can drastically reduce the number of required samples and thus the computation time of the planner to find a feasible trajectory, especially in critical situations. The benefits of our novel concept are demonstrated using scenarios from the CommonRoad benchmark suite.