Abstract: Obstacle avoidance trajectory planning is an indispensable part of the autonomous technology, which is related to the safety of the vehicles. In this paper, the trajectory planning method based on artificial potential field is mainly studied, aiming at the inherent defects such that local optimization and unreachable target, considering that the longitudinal change ability of vehicles traveling on structured roads is much greater than that of horizontal. An improved method of potential field distribution based on Frenet coordinate is proposed, which scales the potential field proportionally along the horizontal, so as to avoid the unstable or even planning failure caused by the superposition of multiple potential fields when the host vehicle is detouring the borrowing road. Aiming at the local optimal problem, an escape trajectory planning method based on safety corridor is established. When the host vehicle falls into the unattainable target situation, the vehicle will be guided to plan the optimal scheme by dynamically adjusting the potential field. Finally, the proposed improvement scheme is simulated and verified under different traffic scenarios, and the results show that the proposed scheme can effectively improve the performance of the artificial potential field method.

Key words: Autonomous vehicle; Trajectory planning; Safe corridor; Artificial potential field method

摘要： 避障轨迹规划是无人驾驶技术栈中不可或缺的一环，关系到无人驾驶车辆在道路上行 驶时的安全性。文章主要研究基于人工势场法的轨迹规划方法，针对该方法中的局部最优及 目标不可达的固有缺陷，考虑到在结构化道路上行驶的车辆纵向的变化能力远大于横向。提 出了基于 Frenet 坐标系的势场分布改进方法，将势场沿横向按比例缩放，避免了自车在进行 借道绕行时由于多重势场的叠加而产生的路径抖动甚至规划失败。针对局部最优问题，建立 了基于安全走廊的逃逸轨迹规划方法。当自车陷入目标不可达情况时，通过动态调整势场引 导自车规划出最优方案。最后，在不同交通场景下对提出的改进方案进行了仿真验证，仿真 结果表明，提出的方案能够有效提升人工势场法的性能。

关键词: 无人驾驶汽车；轨迹规划；安全走廊；人工势场法