露天煤矿小曲率弯道自卸矿车路径规划方法

Path planning method for autonomous haul trucks on small-curvature curves in open-pit mines

  • 摘要: 面对露天煤矿小曲率弯道非结构化特征显著、时空动态扰动强、几何约束严苛等复杂工况,现有路径规划算法存在路径平滑性差、转向频繁且曲率突变等问题。针对上述问题,提出了一种面向露天煤矿小曲率弯道自卸矿车路径规划的改进混合A*算法。对估值函数进行改进:引入距离惩罚函数,通过划分碰撞危险区、碰撞潜在区与碰撞安全区,动态调整节点扩展优先级;引入代价函数,对前进行驶与航向角稳定状态赋予权重倾斜,引导自卸矿车沿弯道中心线行驶,降低无效转向频次,规避曲率突变导致的规划失效。构建了包含平滑代价项、距离代价项、偏离代价项及曲率变化率代价项的多目标非线性优化模型,通过位置与曲率约束确保轨迹符合车辆运动学要求,并结合二次规划与内点法求解器对路径进行平滑优化。仿真实验与现场应用结果表明:在典型小曲率弯道场景下,与Dijkstra、快速扩展随机树和混合A*算法相比,改进混合A*算法规划的路径更贴近弯道中心线且连续平滑,最小安全裕度更高,曲率和曲率变化率更小,且在动态避障实时性与可靠性方面均显著优于对比算法,提升了自卸矿车在露天煤矿复杂小曲率弯道场景下的通行安全性和行驶稳定性。

     

    Abstract: Existing path planning algorithms for open-pit coal mine haul trucks often suffer from poor path smoothness, frequent steering, and abrupt curvature variations when operating in small-curvature curves characterized by unstructured environments, strong spatiotemporal disturbances, and stringent geometric constraints. To address these problems, this study proposed an improved Hybrid A* algorithm for path planning of haul trucks in small-curvature curves of open-pit coal mines. The heuristic function was improved by introducing a distance penalty function, in which collision-risk, collision-potential, and collision-safe zones were defined to dynamically adjust node expansion priorities. A cost function was further introduced to assign greater weights to forward driving and stable heading states, guiding the haul truck to travel along the centerline of the curve, reducing unnecessary steering maneuvers, and avoiding planning failure caused by abrupt curvature changes. A multi-objective nonlinear optimization model incorporating a smoothing cost, a distance cost, a deviation cost, and a curvature variation rate cost was constructed. Position and curvature constraints were imposed to ensure compliance with vehicle kinematic requirements, and the planned path was further smoothed using quadratic programming and an interior-point solver. Simulation and field application results showed that, in typical small-curvature curve scenarios, the improved Hybrid A* algorithm generated paths that were closer to the curve centerline and exhibited better continuity and smoothness than those generated by the Dijkstra, Rapidly-Exploring Random Tree (RRT), and Hybrid A* algorithms. In addition, the proposed algorithm achieved a larger minimum safety margin, lower curvature and curvature variation rate, and significantly outperformed the comparison algorithms in terms of real-time dynamic obstacle avoidance and reliability, thereby improving the driving safety and operational stability of haul trucks in complex small-curvature curve scenarios of open-pit coal mines.

     

/

返回文章
返回