工业机器人单目视觉装配系统研究

Research on Monocular Vision Assembly System of Industrial Robot

为了解决传统机器人只能通过示教或离线编程固定执行点到点装配任务的问题,进行了视觉引导的工业机器人装配方面的研究。首先对工业机器人视觉装配过程建立数学模型并标定,然后研究了基于机器视觉的零件位姿估计方法,针对法兰盘类金属零件弱纹理及无角点特征的位姿估计问题,提出了一种基于自身特征和外部标记辅助的单目视觉位姿计算方法。最后搭建了工业机器人单目视觉装配实验平台验证该位姿计算方法的可行性。实验结果表明,该系统能够实现对法兰盘类零件的定位和装配,最小位置误差为2.07mm,最小姿态误差0.13。

In order to solve the problem that traditional robots can only perform point-to-point assembly tasks by teaching or off-line programming,the research on vision-guided industrial robot assembly was conducted.Firstly,a mathematical model was es-tablished and calibrated for the visual assembly process of industrial robots,then a method of position and attitude estimation of parts based on machine vision was studied.Further,aiming at the problem of position and attitude estimation of weak texture and corner-free features of flange-type metal parts,a method of position and attitude calculation of monocular vision based on its own characteristics and external marking assistance was proposed.Finally,an experiment platform for monocular visual assembly of in-dustrial robots was built to verify the feasibility of the pose calculation method.The experimental results show that the system can re-alize the positioning and assembly of flange parts,the minimum position error is 2.07mm,and the minimum attitude error is 0.13°.