基于新型RRT算法与视觉定位的机械臂工件抓取

Workpiece Grasping of Manipulator Based on Improved RRT Algorithm and Visual Positioning

针对机械臂在工业任务中与环境里的障碍发生碰撞,以及待抓取工件的位置变动等问题,提出一种采用新型快速拓展随机树(RRT)算法与视觉定位的机械臂工件抓取方案。在关节空间内采用新型RRT算法进行路径规划,算法采用基于目标偏置的拓展点选择策略以提高搜索效率,并采用三次样条插值法对路径进行平滑优化。当机械臂到达目标位置后,通过霍夫圆变换算法进行工件检测,同时通过轮廓提取及多边形逼近法检测出托盘凹槽,经过双重检测,位于凹槽内的圆型工件为目标工件。实验结果表明,所提出的新型RRT算法搜索效率高,所规划的路径平滑连续,双重检测算法可以有效地识别出待抓取的目标工件,满足工业任务的稳定性要求。

Aiming at the problems that the manipulator collides with the obstacles in the environment and the position change of the target in the industrial grasping task,this paper proposes a manipulator workpiece grasping scheme based on improved rapid expansion random tree(RRT)algorithm and visual positioning.In the joint space,a new RRT algorithm is used for path planning.The algorithm adopts the expansion point selection strategy based on target offset to improve the search efficiency,and uses cubic spline interpolation to smooth and optimize the path.When the manipulator reaches the target position,the workpiece is detected by Hough circle transformation algorithm.At the same time,the tray groove is detected by contour extraction and polygon approximation method.After double detection,the circular workpiece located in the groove is the target workpiece.The experimental results show that the proposed new RRT algorithm has high search efficiency,the planned path is smooth and continuous,and the dual detection algorithm can effectively identify the target workpiece to be captured,which meets the stability requirements of industrial tasks.