自动泊车系统中的实用路径规划算法研究
2025,33(1):226-234
摘要：为了解决在自动驾驶过程中短距离自动泊车场景下，受环境复杂性影响导致传统的A*算法和RS曲线加速搜索算法难以应用的问题，提出了一种加入反向搜索算法的改进混合A*算法。利用地图栅格法和A*算法计算启发值，通过检测车身轮廓线与简化后的障碍物线是否相交判断二者能否相撞，以节省搜索时间；通过控制RS曲线的扩展方向数量，保证路径的平滑性；经MATLAB仿真垂直入库和侧方泊车场景，对改进算法与传统算法进行了对比分析，验证了同等条件下改进的混合A*算法在两种仿真场景的平均搜索时间上分别减少8.18%和20.53%，且能产生更短、更平滑的路径，从而验证了所提基于反向搜索算法的混合A*算法的优越性。
关键词：自动泊车;路径规划;反向搜索;混合A*算法;障碍物距离代价
Research on practical path planning algorithm in automatic parking system
Abstract：In order to solve the problem that the traditional A* algorithm and RS curve accelerated search algorithm are difficult to apply in the short-distance automatic parking scene during autonomous driving due to the complexity of the environment, an improved hybrid A* algorithm with a reverse search algorithm is proposed. The heuristic value is calculated using the map grid method and the A* algorithm. By detecting whether the body contour line and the simplified obstacle line intersect, it is determined whether the two can collide, so as to save search time; by controlling the number of extended directions of the RS curve, the smoothness of the path is guaranteed; through MATLAB simulation of vertical parking and side parking scenes, the improved algorithm is compared with the traditional algorithm. It is verified that under the same conditions, the improved hybrid A* algorithm reduces the average search time of the two simulation scenes by 8.18% and 20.53% respectively, and can produce a shorter and smoother path, thus verifying the superiority of the proposed hybrid A* algorithm based on the reverse search algorithm.
Key words：Automatic parking;? path planning;? reverse search;? hybrid A* algorithm;? distance cost of obstacle
收稿日期：2024-07-15
基金项目：西安石油大学研究生创新与实践能力培养计划(YCS23214304)
