改进RRT算法的机械臂路径规划

Improved RRT Algorithm for Manipulator Path Planning

为解决快速拓展随机树(RRT)算法随机性强,收敛速度慢和结果可行性差等问题,提出了一种基于全局自适应步长与目标偏置采样的改进型RRT算法。首先,利用环境信息自适应的计算初始步长,同时在拓展过程中利用节点周围障碍物信息调整当前步长,以增强对地图的探索能力;其次,通过目标偏置采样结合改进的最近点选取策略提高算法的搜索效率,快速生成一条从起始点到目标点的路径;随后,对生成路径进行两次冗余点移除并结合最大曲率约束减少路径代价和转折角;最后,利用基于最小二乘法的五次多项式拟合对其进行优化,进一步提高路径的可行性。在机械臂上进行仿真实验,结果表明在三维空间下改进RRT算法相较于传统RRT算法,路径代价减少38.1%,规划时间减少68.4%,节点个数减少77.4%,验证了算法的有效性。

To solve the problems of low adaptability and poor feasibility of results of the RRT algorithm in manipulator path planning,an improved RRT path planning algorithm based on adaptive step with target bias sampling is proposed.First,the initial step size is calculated adaptively using the environmental information,while the current step size is adjusted using the obstacle information around the nodes during the expansion process to enhance the exploration of the map.Secondly,the search efficiency of the algorithm is improved by target bias sampling combined with an improved nearest point selection strategy to quickly generate a path from the starting point to the target point.After that,the generated path is removed twice for redundant points and combined with the maximum curvature constraint to reduce the path cost and turning angle.Finally,the path is optimized using a least-squares-based five-polynomial fit to further improve the feasibility of the path.Simulation experiments are carried out on the manipulator and the results show that the improved RRT algorithm reduces the path cost by 38.1%,the planning time by 68.4%,and the number of nodes by 77.4%compared with the traditional RRT algorithm in three-dimensional space,which verifies the effectiveness of the algorithm.