Abstract: To quickly plan a better path for the manipulator in the obstacles environment,a kind of improved RRT*-DR path planning algorithm based on RRT* was proposed.The entire planning process was divided into two steps:fast exploration of path and optimization of the initial path.A path connecting the starting point and target was found by exploring quickly with a half-goal-guiding expanding mechanism.Then,the dynamic region sampling method was used to always sample in the surrounding range of the current optimal path,and the node tree near the current optimal path was densified,which saved computing resources and made the initial path converge to the asymptotic optimal path quickly through iteration.At the same time,a variable step size mechanism for obstacle-nearing nodes was proposed,which selectively reduced the extended step size of the obstacle-nearing nodes,effectively reduced the number of collision detection failures,and improved the algorithm efficiency.The simulation results of MATLAB and the Robot Operating System(ROS) showed that the improved algorithm RRT*-DR could optimize the path in a shorter time,and effectively reducing the path cost.Furthermore,the practicability and effectiveness of the algorithm were verified by the path planning and obstacle avoidance experiment of the real manipulator.

Key words: manipulator, path planning, dynamic sampling region, variable step size

摘要： 为使机械臂在障碍物环境下快速规划较优路径,提出了一种基于动态区域采样的改进RRT*-DR路径规划算法,将整个规划过程分为快速探索路径和优化初始路径两个步骤。首先利用半目标导向扩展快速探索,找到连接起始点和目标点的路径。随后利用动态区域采样方法,始终在当前最优路径的周边范围内采样,优先密化当前最优路径附近的节点树,节省计算资源,使初始路径经过迭代快速向渐进最优路径收敛。同时,提出一种近障碍节点变步长机制,有选择性地缩短靠近障碍节点的扩展步长,可有效减少碰撞检测失败次数,提高算法效率。最后,在MATLAB和ROS系统下进行路径规划算法仿真,结果表明RRT*-DR算法可在更短时间内实现路径规划,同时有效缩小路径代价。进一步通过实体机器人路径规划避障实验,验证了该算法的实用性和有效性。

关键词: 机械臂, 路径规划, 动态采样区域, 变步长