Spatio-Temporal Motion Planning for Autonomous Vehicle in Dynamic Urban Environment

Due to the influence of dynamic objects and traffic rules, most of the existing methods on motion planning when applying to the dynamic urban environment show limited performance. Here, we propose a new motion planning method for autonomous vehicles by integrating the spatio-temporal map into motion planning. The proposed method contains three stages: spatio-temporal map, path planning, and trajectory optimization. First, a spatio-temporal map is constructed based on the static environments and predicted trajectories of dynamic objects. Next, an improved hybrid A star algorithm integrating the obstacle avoidance, vehicle kinematics, and traffic rules is developed to efficiently search collision-free feasible paths. Finally, a B-spline curve combined with vehicle dynamics is proposed to further optimize the trajectory, thus the autonomous vehicle can move safely and smoothly. Simulation experiments are conducted and compared with existing methods. Results show the proposed motion planning method achieves promising performance in urban environments for autonomous driving.