A Novel Dynamic Lane-Changing Trajectory Planning for Autonomous Vehicles Based on Improved APF and RRT Algorithm

To satisfy multi-objective requirements of the dynamic lane-changing trajectory planning (DLTP) for autonomous vehicles, a novel DLTP method based on the improved artificial potential field (APF) and rapidly exploring random tree (RRT) algorithm is proposed. The problem of lane-changing trajectory planning can be decoupled into trajectory shape planning and speed planning. First, the Frenet coordinate system is employed to transform the planning trajectory on curved roads to that on straight roads. Second, based on sinusoidal obstacle avoidance lane-changing, the potential field of virtual obstacle points at the road boundary is established by integrating information on the position and state of surrounding vehicles. The improved APF algorithm is utilized to plan the shape of the lane-changing trajectory. Then, the motion states of surrounding vehicles are mapped to the obstacle region in the space-time graph, transforming speed planning into a path-searching problem. The efficiency of the RRT algorithm is improved by increasing the heuristic information of the lane-changing endpoint and the multi-objective constraints of the random sampling region. Finally, simulation results validate that the proposed method can plan a smooth lane-changing trajectory, effectively avoid collisions with surrounding vehicles, and ensure real-time stability of the lane-changing process.