基于PIB-RRTstar的荔枝采摘机械臂运动规划方法

Motion Planning for Lychee Picking Manipulator Based on PIB-RRTstar Algorithm

摘要:

为解决非结构环境下，采摘机械臂路径规划时存在的效率低、采摘成功率不高的问题，提出一种结合人工势场法的四向搜索RRTstar算法。首先通过人工势场法对空间进行分割，获取空间分割点xsplit，进行四向搜索;其次结合人工势场法引导随机采样，提高采样节点质量;然后基于节点历史扩展信息，添加信息因子，实现自适应动态步长扩展。最后通过贪婪回溯对生成路径进行优化。通过二维模拟实验、6自由度机械臂下的仿真实验与采摘实验验证提出算法的有效性。二维仿真对比实验表明：相比于RRTstar算法，改进算法路径成本缩短2.01%，时间代价降低98.81%，迭代次数减少92.49%。在机器人操作系统(Robot operating system，ROS)中进行6自由度机械臂下的仿真实验，相比于RRTstar算法，规划时间减少93.17%，路径成本降低36.62%，扩展节点数减少97.00%。最后使用6自由度机械臂进行采摘实验，实验结果表明本文算法有效提升采摘成功率，采摘成功率达85%，并在结合姿态约束方法后，采摘成功率达95%。所提出的路径规划方法在路径规划时间上存在一定优势，采摘实验证明本文算法可提高荔枝采摘成功率。

Abstract:

In order to solve the problems of low efficiency and poor picking success rate in the path planning of picking robotic arm in unstructured environment, a four-way search RRTstar algorithm combined with artificial potential field method was proposed. Firstly, the space was segmented by the artificial potential field method to obtain the spatial segmentation point x-split for four-way search;secondly, the random sampling was guided by the artificial potential field method to improve the quality of the sampled nodes;then, based on the node history expansion information, the information factor was added to achieve the adaptive dynamic step size expansion. Finally, the generation path was optimized by greedy backtracking. The effectiveness of the proposed algorithm was verified by two-dimensional simulation experiments, simulation experiments under 6-degree-of-freedom robotic arm and picking experiments. The 2D simulation comparison experiment showed that compared with the RRTstar algorithm, the path cost of the improved algorithm was shortened by 2.01%, the time cost was reduced by 98.81%, and the sampling nodes were reduced by 92.49%. Simulation experiments under 6-degree-of-freedom robotic arm in robot operating system (ROS) showed that compared with RRTstar algorithm, the planning time was reduced by 93.17%, the path cost was reduced by 36.62%, and the number of expansion nodes was reduced by 97.00%. Finally, the picking experiment was carried out with a 6-degree-of-freedom robotic arm, and the experimental results showed that the algorithm effectively improved the picking success rate, which reached 85%, and after combining the attitude constraint method, the picking success rate reached 95%. The proposed path planning method had certain advantages in path planning time, and the picking test proved that the algorithm improved the success rate of lychee picking, and contributed to the development of lychee picking robot.