基于双臂机器人和三维视觉的物体抓取和轴孔装配  

Rasping and Shaft Hole Assembly Based on Dual Arm Robot and 3D Vision

作　　者：徐友法[1] 王卫军[1] 王兆广 陈宝[1] XU Youfa;WANG Weijun;WANG Zhaoguang;CHEN Bao(No.21 Research Institute of CETC,Shagnhai 200233,China)

机构地区：[1]中国电子科技集团公司第二十一研究所,上海200233

出　　处：《微特电机》2022年第1期27-31,共5页Small & Special Electrical Machines

摘　　要：三维视觉的机器人抓取能够将通过点云获取的匹配矩阵应用到智能抓取中。基于随机抽样一致性(RANSAC)和近点迭代法(ICP)模板匹配算法获取的矩阵,设计了一种机器人抓取点确定和机器人末端抓取姿态获取的方法。同时为机器人设计了一个预抓取位置,机器人在抓取时先调整抓取姿态到达预抓取位置,随后保持抓取姿态不变运行到抓取点对物体进行抓取。实验结果表明,将模板匹配算法获得的矩阵与机器人的实际抓取结合的方法,能够通过匹配矩阵对任意摆放物体进行抓取,使点云理论研究在实际抓取应用中得到落地。Robot grasping based on 3 D vision can apply the matching matrix acquired based on point cloud to intelligent actual grasping.Based on the matrix obtained by random sample consensus(RANSAC)and iterative closest points(ICP)template matching algorithm,a method for determining the grasping point of the robot and acquiring the end pose of the robot was studied.A pre-grabbing position was designed for the robot,and the robot first adjusted the grasping posture when grasping reached the pre-grabbing position and then kept the grasping posture unchanged to grasp the object.The experimental results show that the proposed method that combines the matrix obtained by the template matching algorithm with the actual grasping of the robot,can grasp arbitrarily placed objects throughthe matching matrix,so that the research of point cloud theory can be implemented in the actual grasping application.

关 键 词：粗匹配 精匹配 模板 位姿 抓取点 

分 类 号：TM383[电气工程—电机]