工业机器人装配中基于相机位姿估计算法的单目视觉定位研究  

Research on Monocular Vision Positioning Based on Camera Pose Estimation Algorithm in Industrial Robot Assembly

作　　者：余娜 何国荣[1] 晁阳[1] 李培东 YU Na;HE Guorong;CHAO Yang;LI Peidong(School of Mechanical and Electrical Engineering,Yangling Vocational and Technical College,Yangling 712100,China)

机构地区：[1]杨凌职业技术学院,机电工程学院,陕西杨凌712100

出　　处：《微型电脑应用》2024年第4期85-88,共4页Microcomputer Applications

基　　金：杨凌职业技术学院2020年自然科学基金项目(ZK20-43)。

摘　　要：为了提升工业机器人在装配作业中的效率和精准度,结合软指派算法进行研究,构建基于相机位姿估计算法的单目标视觉定位系统,并对其进行仿真验证。实验结果显示,在参数对成功率的影响实验中实验成功率最低为92.5%,在图像噪声对成功率的影响实验中成功率最高达到了97.5%,位置估计误差最大值不超过4 mm,旋转误差在伪特征点的数量为最大时得到了最大值7°。In order to improve the efficiency and accuracy of industrial robots in assembly operations,a monocular vision positioning system based on camera pose estimation algorithm is constructed by combining soft assignment algorithm,and simulation verification is conducted.The experimental results show that the lowest success rate is 92.5%in the experiment on the influence of parameters on success rate,and the highest success rate is 97.5%in the experiment on the influence of image noise on success rate.The maximum position estimation error does not exceed 4 mm,and the maximum error of rotation is 7°when the number of pseudo-feature points is maximum.

关 键 词：单目视觉 工业机器人 位姿估计 视觉定位 软指派算法 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]