Computer Engineering and Applications ›› 2023, Vol. 59 ›› Issue (14): 285-292.DOI: 10.3778/j.issn.1002-8331.2212-0133

• Engineering and Applications • Previous Articles     Next Articles

Research on Path Planning of Terminal Operation for Grape Picking Robot

JIA Jiaoyu, YUAN Jie, LI Zhonghua, LIU Qiang, GAO Xingjian   

  1. School of Electrical Engineering, Xinjiang University, Urumqi 830017, China
  • Online:2023-07-15 Published:2023-07-15

葡萄采摘机器人终端作业路径规划方法研究

贾焦予,袁杰,李中华,刘强,高星健   

  1. 新疆大学 电气工程学院,乌鲁木齐 830017

Abstract: Grape picking has the characteristics of many obstacles and complex structure. In order to achieve efficient picking and stable obstacle avoidance, a collision free bidirectional RRT(CGB-RRT) algorithm is proposed for obstacle avoidance path planning. When the dual tree nodes expand towards each other, they expand directly to each other without collision, and when collision occurs, they expand to free nodes to speed up the expansion. When obstacle avoidance path planning fails, a flexible obstacle avoidance path planning strategy based on CGB-RRT algorithm is further proposed. By introducing marker joints, the collision shape of flexible contact between the robot arm and the obstacle is output in the terminal obstacle avoidance path, and then the flexible obstacle avoidance path planning is realized by using the step-by-step planning method of “initial position-intermediate position-target position”. In addition, a grape grid collection box with known collection point location is designed, and the collection job path can be planned in advance through CGB-RRT algorithm for direct call by the collection job. The experimental results show that CGB-RRT algorithm in obstacle avoidance path planning reduces the running time by 27.61% and the expansion node by 24.32% compared with BI-RRT algorithm. When obstacle avoidance path planning fails, the flexible obstacle avoidance path planning strategy combined with different algorithms can achieve path planning, and the performance is better when combined with CGB-RRT algorithm.

Key words: path planning, flexible obstacle avoidance, RRT algorithm, robot, grape picking

摘要: 葡萄采摘具有障碍物多、结构复杂的特点,为实现高效采摘和稳定避障,提出无碰撞双向扩展RRT(collisionless growth bidirectional RRT,CGB-RRT)算法进行避障路径规划。该算法双树节点相向扩展时,不发生碰撞直接向对方扩展,发生碰撞则转向自由节点扩展,加快扩展速度。针对复杂障碍物导致避障路径规划失败的情况,进一步提出基于CGB-RRT算法的柔性避障路径规划策略,通过引入标记关节,在终端避障路径输出机械臂与障碍物发生柔性接触的碰撞形位,采用“初始位置-中间位置-目标位置”分步规划的方式实现柔性避障路径规划。此外,设计了一种收集点位置已知的葡萄网格收集箱,通过CGB-RRT算法可提前规划得到收集作业路径,供收集作业直接调用。实验结果表明,避障路径规划中CGB-RRT算法与BI-RRT算法相比,运行时间减少27.61%,扩展节点减少24.32%;避障路径规划失败时,柔性避障路径规划策略结合不同算法均能实现路径规划,与CGB-RRT算法结合时表现出来的性能较好。

关键词: 路径规划, 柔性避障, RRT算法, 机器人, 葡萄采摘