基于深度学习的工业机器人位姿误差自动补偿方法  

Automatic compensation method of position and attitude error of industrial robot based on deep learning

作　　者：葛勇 曾劲松[1] GE Yong;ZENG Jinsong(School of electrical engineering,Anhui Institute of mechanical and electrical technology,Wuhu 241002,Anhui,China)

机构地区：[1]安徽机电职业技术学院电气工程学院,安徽芜湖市241002

出　　处：《惠州学院学报》2023年第6期6-9,共4页Journal of Huizhou University

基　　金：安徽省高校自然科学研究重点项目(KJ2020A1111)。

摘　　要：为提高工业机器人在运行过程中的位姿精度,引入深度学习,开展对其位姿误差自动补偿方法的设计研究。结合深度学习,建立机器人位姿目标识别定位模型,并在RPN网络中完成模型的迭代和学习;为实现对工业机器人位姿误差的自动补偿控制,导入误差自动补偿模糊控制器;最后将位姿误差及补偿结果输出,实现对工业机器人位姿误差的补偿。通过对比实验的方式证明,新的补偿方法可以有效提高工业机器人运行时的位姿精度,进而促进工业机器人运行质量的提升。To improve the accuracy of the attitude position of industrial robots during operation,deep learning is introduced to carry out the design and research on the automatic compensation method of position and attitude error.Combined with deep learning,the robot pose target recognition and positioning model is established,and the iteration and learning of the model are completed in RPN network;In order to realize the automatic compensation control of the position and attitude error of industrial robot,the fuzzy controller with automatic error compensation is introduced;Finally,the position and attitude error and the compensation result are output to realize the compensation of the position and attitude error of the industrial robot.Through comparative experiments,it is proved that the new compensation method can effectively improve the position and attitude accuracy of industrial robots,and then promote the operation quality of industrial robots.

关 键 词：深度学习 机器人 自动补偿 误差 位姿 工业 

分 类 号：TP242.2[自动化与计算机技术—检测技术与自动化装置]