Abstract: To conquer the defect of strong randomness,poor guidance,long planning time and poor smoothness of the planned path of traditional Rapidly exploring Random Tree (RRT) algorithm,an improved RRT algorithm based on target bias strategy combined with adaptive variable step size named P-Adaptive Variable Step size-RRT (PAVS-RRT) was proposed.A target bias threshold was set on the basis of the traditional RRT algorithm,and a local expansion mechanism was introduced to avoid local optimization problems caused by changing the sampling structure;the search time was optimized by combining the adaptive step strategy;the cubic B-spline function was used to fit and optimize the planned path.The proposed algorithm in the simulation experiment ensured that the manipulator successfully avoided obstacles and reaches the target smoothly,meanwhile,its joint parameters had small fluctuations and no sudden changes,which effectively reduced the chattering of the manipulator during the motion planning process.Experimental results showed that the average path search time of the proposed algorithm was increased by 73.49% compared with the basic algorithm;the search efficiency and smoothness of the algorithm were significantly improved.

Key words: industrial manipulator, motion planning, rapidly exploring random tree algorithms, target bias strategy, adaptive variable step size, cubic B-spline

摘要： 为解决传统快速拓展随机树(RRT)算法的随机性强,导向性差,规划时间长及寻迹平滑度差等问题,提出一种基于目标偏置策略结合自适应可变步长的改进型RRT算法(PAVS-RRT)。首先,在传统RRT算法基础上设置一个目标偏置阈值,同时引入局部扩展机制避免因改变采样结构而造成的局部最优问题;其次,结合自适应步长策略优化其搜索时间;最后,采用三次B样条函数对所规划路径进行拟合优化。仿真实验中所提算法在保证机械臂成功避障且顺利抵达目标位置的同时,其各关节参数均波动较小且未发生突变,有效降低了机械臂在运动规划过程中的抖振情况。实验结果表明,所提算法较基本算法其平均路径搜索时间提高了73.49%,算法搜索效率及平滑性得到显著改善。

关键词: 工业机械臂, 运动规划, 快速拓展随机树算法, 目标偏置策略, 自适应可变步长, 三次B样条