Abstract: Under urban road conditions, road constraints are more complex, and the path planning effect of RRT and target-based Bg-RRT algorithms is not good, which is prone to cause problems such as node redundancy and low smoothness of search path. Aiming at these problems, an improved RRT algorithm based on the force field optimization sampling area was proposed. Firstly, the proposed algorithm dynamically optimized the sampling area based on the road environment and vehicle position, then adjusted the sampling area in real time based on the maximum angle constraint of the force field and the vehicle, and finally adopted a dynamic step selection strategy based on safety distance and 270-degree collision detection. On this basis, greedy thinking and curvature constraint were combined to post-process the path. Simulation experiment was carried out on the improved algorithm. The research results show that the improved algorithm reduces the number of nodes by 72.16%, improves the path search efficiency by 83.57%, and ensures the smoothness of the path meanwhile. The effectiveness and adaptability of the proposed algorithm are verified.

Key words: traffic and transportation engineering; autonomous vehicles; RRT algorithm; path planning

摘要： 在城市路况下道路约束较为复杂，RRT以及基于目标的Bg-RRT等算法的路径规划效果不佳，容易产生节点冗余、搜索的路径平滑度低等问题；针对这些问题，提出了一种基于势力场优化采样区域的改进RRT算法，该方法首先基于道路环境及车辆位置对采样区域动态优化，再基于势力场和车辆最大转角约束对采样区域进行实时调整，最后采取基于安全距离和270°碰撞检测的动态步长选择策略，并在此基础上结合贪心思想及曲率约束对路径进行后处理；将改进算法在进行仿真实验；研究结果表明：改进后的算法在节点数上减少了72.16%，路径搜索效率提高了83.57%，同时保证了路径的平滑度；验证了算法的有效性和适应性。

关键词: 交通运输工程；自动驾驶车辆；RRT算法；路径规划