一种面向机器人抓取的铆钉视觉识别与定位测量方法  

A Rivet Visual Recognition and Positioning Measurement Method for Robot Grasping

作　　者：王菲儿 姜涛 朱磊 陈胤竹 金子阳 WANG Feier;JIANG Tao;ZHU Lei;CHEN Yinzhu;JIN Ziyang(School of Mechanical and Electrical Engineering,Suqian University,Suqian 223865,China)

机构地区：[1]宿迁学院机电工程学院,江苏宿迁223865

出　　处：《机械制造与自动化》2023年第6期164-167,共4页Machine Building & Automation

基　　金：宿迁学院大学生创新创业训练计划项目(2021XSJ140Y);宿迁学院高层次人才引进项目(校2022XRC012)。

摘　　要：视觉引导机器人定位抓取研究和应用逐渐广泛,其关键问题是基于视觉的被测对象高精度测量定位。以抽芯铆钉作为研究对象,研究一种面向机器人抓取的抽芯铆钉视觉识别与定位测量方法,研制一种融合视觉系统的抓取末端。设计一种基于多重腐蚀和膨胀算子的抽芯铆钉钉头图像分割方法,可有效实现抽芯铆钉钉头和钉杆的分离。基于PCA算法对抓取角度进行主方向分析,根据抓取要求对主方向进行校正。基于单应矩阵变换,实现图像坐标系和机器人末端夹爪坐标系标定。通过实验验证了抽芯铆钉视觉识别的效果。The research and application of vision guided robot positioning and grasping are becoming more and more extensive,and the key factor lies in the high-precision measurement and positioning of the measured object based on vision.Taking blind rivets as the research object,studies a visual recognition and positioning measurement method of blind rivets for robot grasping,and develops a grasping end of fusion vision system.An image segmentation method of blind rivet head based on multiple corrosion and expansion operators is designed,which can effectively realize the separation of blind rivet head and nail rod.The principal direction of the grasping angle is analyzed based on PCA algorithm,and the principal direction is corrected according to the grasping requirements.Based on homography matrix transformation,the image coordinate system and the robot end gripper coordinate system are calibrated.The effect of visual recognition of blind rivets is verified by experiments.

关 键 词：视觉测量 识别抓取 图像处理 抽芯铆钉 单应矩阵 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]