Abstract: Aiming at the problems of weak exploration ability,slow convergence speed and poor path quality in the manipulator motion planning of rapidly-exploring random tree algorithm,an improved Rapidly-exploring Random Tree (RRT) algorithm based on map pixelation was proposed.A uniform node sampling strategy was adopted,which significantly enhanced the exploration ability of RRT through central sampling and avoiding repeated sampling.Besides,a node rejection strategy was proposed to effectively reduce the number of collision detections and improve the algorithm efficiency utilizing pixel map to record obstacle information.At the same time,a dynamic path strategy was proposed to form several paths based on whether the nodes could connect to the target point without collision or not,and set the optimal path was set as the final path to accelerate the convergence speed of the algorithm.After searching,redundant nodes in the path were eliminated,and the triangle inequality was used to optimize the path.Finally,the simulation results of various two-dimensional map showed that the improved RRT algorithm could generate a more optimal path in less time.Furthermore,the superiority and practicability of the algorithm were verified by the Robotics Toolbox simulation experiment and obstacle avoidance experiment of the real manipulator.

Key words: manipulator, motion planning, rapidly-exploring random tree, uniform node sampling

摘要： 针对快速搜索随机树(RRT)应用于机械臂运动规划存在的探索能力弱、收敛速度慢、路径质量差等问题,提出一种基于地图像素化的改进RRT算法。首先采用均匀节点采样策略,通过中心采样和避免重复采样提高了算法探索能力;其次提出节点拒绝策略,利用像素地图记录障碍物信息,可有效减少碰撞检测次数,提高算法效率;同时,提出动态路径策略,根据节点能否与目标点无碰撞连接形成若干条路径,并将最优路径设置为最终路径,加快了算法收敛速度;搜索完成后,剔除路径中的冗余节点,并借助三角不等式优化路径,提高了路径质量。最后,在MATLAB中进行不同二维地图仿真实验,结果表明改进RRT算法可以在更短时间内规划出更优路径。进一步通过Robotics Toolbox仿真实验,以及真实机械臂避障实验,验证了该算法的优越性和实用性。

关键词: 机械臂, 运动规划, 快速搜索随机树, 均匀节点采样