面向铲装作业场景的自动驾驶矿车路径规划方法研究

Path Planning Method for Autonomous Driving Mining Vehicles in Shovel Loading Operation Scenarios

针对铲装作业场景复杂多变、矿车路径规划实时性及安全性要求高的特点,提出一种基于引导式变步长混合A^(*)算法的路径规划方法。通过建立维诺图获取矿区路网并提取关键点作为方向引导,提升探索效率的同时避免车辆在U形障碍物处陷入局部最优;引入自适应变步长算法并改进启发函数,进一步提高规划效率及路径安全性;通过矿区实车场景试验验证算法有效性。试验结果表明,本路径规划方法满足矿车复杂场景要求,规划时间相比原引导式算法降低68%,路径到障碍物平均距离增加了11%,路径曲率变动次数减少45%,显著提高了计算效率与路径质量。

The automated shovel loading of mining vehicles is one of the core operations of intelligent mining in open pit mines.However,in some complex mining scenarios,the traditional path planning methods for autonomous driving are time-consuming and the output path has many unnecessary curvature changes without sufficient consideration of safety,which leads to the reduction of operational efficiency and the safety factor.Considering the complex operational scenarios for automated mining vehicles and aiming at the high real-time and safety requirements of path planning,this paper proposes a path planning method based on the guided variable-step-length hybrid A^(*)algorithm.Initially established on the Voronoi diagram,the mine road network is obtained and the key points are extracted as directional guidance in order to improve the planning efficiency and prevent the vehicle from falling into local optimum at U-shaped obstacles.And then the adaptive variable-step-length algorithm is introduced and the heuristic function is reset to further improve the planning efficiency and path safety.Finally the effectiveness of the algorithm is verified by the application in the real mine scenes.The experimental results show that the path planning method proposed in this paper meets the requirements of complex scenarios for mining vehicles.The planning time is reduced by 68%compared with the original guided algorithm,the average path-to-obstacle distance is increased by 11%and the number of path curvature changes is reduced by 45%,which significantly show the improvement in computational efficiency and path quality.