动态环境下AGV避障轨迹规划问题研究  

Research on obstacle avoidance trajectory planning of AGV with moving obstacle

作　　者：何均锋 李鹏[1] 申艳红 邹安帮 熊威 吴婷婷 HE Junfeng;LI Peng;SHEN Yanhong;ZOU Anbang;XIONG Wei;WU Tingting(College of Automobile and Traffic Engineering,Nanjing Forestry University,Nanjing 210037,China)

机构地区：[1]南京林业大学汽车与交通工程学院,南京210037

出　　处：《重庆理工大学学报（自然科学）》2022年第3期135-143,共9页Journal of Chongqing University of Technology：Natural Science

基　　金：国家自然科学基金项目(61403204,71701099)。

摘　　要：针对自动导引车(automated guide vehicle, AGV)在非协作移动障碍物环境下避障困难的问题,将复杂的动态障碍物环境下的AGV避障轨迹规划转化成了一个最优控制问题,建立了AGV的运动约束、避碰约束以及两点边界条件,以AGV的最终停止位置与期望位置的距离作为优化目标。提出并运用了一种“改进A;+直接配点”组合优化算法对最优控制问题进行求解。该算法在时间维度上,采用A;算法搜索出了路径成本最小的路径节点,并将这些节点作为最优控制问题的初始轨迹点;由于初始轨迹存在尖点,无法满足AGV的运动学约束,需要结合直接配点法进行优化,将最优控制问题离散成带约束的非线性规划问题,运用内点法进行寻优。通过算例,分别在单障碍物和多障碍物环境下,对提出的动态环境下AGV轨迹规划方案与算法进行了验证,结果显示:在单障碍物和多障碍物环境下,AGV外接圆和障碍物外接圆的最小圆心距分别为3.19 m和2.78 m,满足预设的安全距离要求,因此利用改进A;算法能有效搜索出一条无碰的初始轨迹。利用直接配点法优化的结果表明:优化后的轨迹在满足运动学约束的情况下与初始轨迹基本一致;另外,代价函数在每一时刻的梯度都是下降的且在结束时刻趋于平缓,因此,优化后的轨迹能在避开障碍物的同时向目标点逼近。To solve the problem of obstacle avoidance of AGV with moving obstacle, firstly, the Obstacle Avoidance Trajectory Planning of AGV in complex dynamic obstacle environment is transformed into an optimal control problem.The motion constraints, collision avoidance constraints and two-point boundary conditions of AGV are established.The optimal distance between the final stop position and the expected position of AGV is taken as the optimization objective.Secondly, an optimization method combining improved A;algorithm with direct collocation algorithm is proposed and used to cope with the optimal control problem.The improved A;algorithm is adopted to search out the path nodes with the minimum path cost in the time dimension, and these nodes are taken as the initial trajectory points.Because of the existence of cusps in the initial trajectory, it can not meet the kinematic constraints of AGV.Therefore, the direct collocation method is used to optimize.Firstly, the optimal control problem is discretized into a constrained nonlinear programming problem, and then the interior point method is used to optimize.Finally, the proposed AGV trajectory planning scheme and algorithm in dynamic environment are verified by examples in single obstacle and multi obstacle environments.The results show that the minimum center distance of AGV and obstacle circumcircle are 3.19 m and 2.78 m respectively in the environment of single obstacle and multi obstacle, which meets the preset safety distance requirements in this paper.Therefore, the improved A;algorithm can effectively search out an initial track without collision;the optimization results by direct collocation method show that the optimized trajectory satisfies the kinematic constraints In addition, the gradient of the target function is decreasing at each time and tends to be gentle at the end of the time.Therefore, the optimized trajectory can approach the target point while avoiding obstacles.

关 键 词：AGV避障 改进A 算法 直接配点法 轨迹优化 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]