Authors

Abstract

To address the problems of low search efficiency, long time-consuming planning and poor adaptability to narrow passages of the original Hybrid A* algorithm in path planning for intelligent grain transfer vehicles in grain depot scenarios, a Hybrid A* algorithm with variable resolution and variable step size is proposed. First, the dual-heuristic function and the cost function with steering and reversing penalties are reasonably designed to ensure that the algorithm can search for a derivable path. Second, the distance cost based on the KD-Tree algorithm is combined into the node expansion, and the node expansion is performed using variable resolution and variable step size according to the change of the distance cost to improve the node search efficiency. Then, Reeds-Shepp curve search fitting is used to pinpoint the target point pose. The simulation verification results show that the number of search nodes of the improved Hybrid A* algorithm is reduced by 80.6%, and the planning elapsed time is shortened by 56.6%, which improves the path search efficiency. After the real vehicle test, the effectiveness of the improved algorithm was confirmed, which enhances the operational efficiency of the grain transfer vehicle.

Abstract in Chinese

针对粮库场景中智能粮食转运车辆在路径规划时原始Hybrid A*算法存在搜索效率低、规划耗时长及狭窄通道适应性差等问题，提出一种变分辨率和变步长的Hybrid A*算法。首先，合理设计双启发函数与含有转向和倒车惩罚的代价函数，保证算法能够搜索到一条可行驶的路径。其次，将基于KD-Tree算法的距离代价结合到节点扩展中，根据距离代价的变化使用变分辨率和变步长进行节点扩展，提高节点搜索效率。然后，采用Reeds-Shepp曲线搜索拟合精确定位目标点位姿。仿真验证结果表明，改进Hybrid A*算法搜索节点数量减少了80.6%，规划耗时缩短了56.6%，提高了路径搜索效率。经过实车试验，证实了改进算法的有效性，从而提升了粮食转运车辆的运行效率。