An Efficient Autonomous Driving Planning Method Based on Bidirectional Spatiotemporal Joint Search and Optimization

Abstract

Trajectory planning is a critical component of autonomous driving systems. A planning system requires the algorithm to compute a drivable trajectory within a set time while ensuring safety, comfort, and efficiency. However, the spatial complexity of spatiotemporal planning presents a major challenge in balancing model feasibility with planning efficiency. To address this issue, a two‐stage spatiotemporal joint path planning algorithm is proposed. In the first stage, a spatiotemporal driving space is constructed by using a three‐dimensional directed graph, and then the spatiotemporal bidirectional concurrent search is employed to find feasible trajectories. In the second stage, a parallel quadratic optimization framework is used to enhance the trajectory, enabling effective handling of dynamic and static obstacles in complex traffic scenarios. Furthermore, with an average planning time of 97 ms for a 5s trajectory on a PC equipped with an Intel Core i9 processor, the developed algorithm has high time efficiency.