高级检索

黄佳德,刘 勇,邓穆坤,等. 露天矿场无人驾驶自卸车路径规划方法研究[J]. 煤炭科学技术,2024,52(8):182−191

. DOI: 10.12438/cst.2023-1593
引用本文:

黄佳德,刘 勇,邓穆坤,等. 露天矿场无人驾驶自卸车路径规划方法研究[J]. 煤炭科学技术,2024,52(8):182−191

. DOI: 10.12438/cst.2023-1593

HUANG Jiade,LIU Yong,DENG Mukun,et al. Research on path planning methods for autonomous dump trucks in open-pit mines[J]. Coal Science and Technology,2024,52(8):182−191

. DOI: 10.12438/cst.2023-1593
Citation:

HUANG Jiade,LIU Yong,DENG Mukun,et al. Research on path planning methods for autonomous dump trucks in open-pit mines[J]. Coal Science and Technology,2024,52(8):182−191

. DOI: 10.12438/cst.2023-1593

露天矿场无人驾驶自卸车路径规划方法研究

Research on path planning methods for autonomous dump trucks in open-pit mines

  • 摘要: 针对矿用自卸车在露天矿场环境下的路径左行要求以及长距离运输路径的平滑效率问题,提出了基于Clothoid曲线拓展的左行混合A*算法和离散点对角向量模最小化平滑算法相结合的路径规划方法。首先,在混合A*算法节点拓展搜索过程中,采用Clothoid曲线代替传统圆弧进行拓展,以保证混合A*搜索路径的曲率连续性和曲率变化率限制要求。然后,利用左行规则改进混合A*算法累加代价和启发代价,在累加代价中加入左转代价和碰撞吸引代价保证路径向左和沿地图边界左侧进行规划,在启发代价中根据拓展方向与地图边界最近点的左右位置关系调整拓展代价得到利于左行的启发代价图,以生成左行全局粗略路径。最后,结合二次规划技术,以离散点对角向量模最小化为目标函数,坐标轴方向移动量为约束,构建平滑优化模型,对全局路径进行平滑,同时为防止平滑过程中曲率超限超出车辆转向响应能力,利用大曲率点可行域隧道缩减技术,限制大曲率点平滑可移动量。试验结果表明:所提方法可以生成适应矿区左行规则的沿地图左侧边界行驶的全局路径;启发代价的改进有效地减少了混合A*算法拓展的节点数量,有利于更好实现左行;离散点路径平滑方法有效提高了全局路径的平滑度,利于车辆控制跟踪;Clothoid曲线拓展与可行域隧道缩减技术可有效应对装载、卸载等大曲率路径,不存在曲率超限情况;对比不同长度路径规划耗时数据,路径平滑算法有效降低了路径规划的总耗时,4 km路径平滑耗时只需76 ms;左行规划的实现和总耗时的减少提高了路径对于矿区场景的适应性。

     

    Abstract: To address the requirements for left-side driving and the issue of smoothing efficiency for long-distance transportation routes of mining dump trucks in open-pit mine environments, a path planning method that combines a left-side driving hybrid A* algorithm based on Clothoid curve expansion with a smoothing algorithm that minimizes the diagonal vector norm of discrete points is proposed. Initially, Clothoid curves are used instead of traditional circular arcs in the hybrid A* node expansion search process, ensuring the continuity of curvature and meeting the requirements for curvature change rate limitations in the hybrid A* search path. The left-side driving rules are then applied to improve both the cumulative cost and heuristic cost within the hybrid A* framework, incorporating left-turn costs and collision attraction costs to ensure the path veers leftward and follows along the left side of the map border. The heuristic cost is adjusted based on the relation of the expansion direction to the nearest point on the map border, facilitating a heuristic cost map that favors left-side driving and generates a rough global path for leftward travel. Finally, by integrating quadratic programming techniques with the objective of minimizing the diagonal vector norm of discrete points under movement constraints along the coordinate axes, a smoothing optimization model is constructed to smooth the global path. To prevent the curvature from exceeding the vehicle's steering response capability during the smoothing process, a tunnel reduction technique for the feasible region of high-curvature points is employed, limiting the smooth movement of high-curvature points. The results show that the proposed method can generate a global path suitable for left-side driving rules in mining areas, which follows the left side boundary of the map. The heuristic cost improvement significantly reduces the number of nodes expanded by the hybrid A*, favoring the implementation of left-side driving. The discrete point path smoothing method effectively enhances the smoothness of the global path, aiding in vehicle control and tracking. Moreover, thanks to Clothoid curve expansion and the feasible region tunnel reduction technique, the method can effectively handle high-curvature paths, such as loading and unloading, without curvature exceeding limits. Comparing the planning times for paths of different lengths, the path smoothing algorithm significantly reduces the total planning time, with a 4-kilometer path smoothing process requiring only 76 milliseconds. The execution of left-side planning and the reduction in planning time improve the adaptability of the path planning for mining area scenarios.

     

/

返回文章
返回