基于图像的强化学习机械臂无标定视觉伺服系统  

The Uncalibrated Visual Servo System for Manipulators Basedon the Reinforcement Learning of Images

作　　者：陈砆兴 朱亚军 CHEN Fuxing;ZHU Yajun(Lanzhou Petrochemical University of Vocational Technology,Lanzhou,Gansu Province,730060 China)

机构地区：[1]兰州石化职业技术大学,甘肃兰州730060

出　　处：《大众科学》2024年第4期1-4,共4页China Public Science

基　　金：甘肃省教育厅2022年度高等学校创新基金项目“苹果采摘机器人视觉定位与检测系统研究”(项目编号:2022A-225)。

摘　　要：为了解决传统机械臂视觉伺服受外界环境影响较大的问题,提出一种基于图像的无标定视觉伺服控制方法。该方法对采集的每一帧图像参数进行处理,无需繁琐的摄像头校准定位等操作,从而在复杂环境下获得了较高的抗干扰工作性能。通过动态的迭代学习过程实现机械臂的最优轨迹控制。在自由度机器臂测试平台上的实验结果表明:所提方法既可以减少模型参数,提高计算速度,又可以有效地对机械臂进行控制,可以对位置信息实施精确调整。In order to solve the problem that the traditional visual servo of manipulators is greatly affected by the external environment,this paper proposes an image-based uncalibrated visual servo control method.This method processes the parameters of each frame of captured image without complicated camera calibration and positioning and other operations,so as to obtain high anti-interference performance in the complex environment,and realizes the optimal trajectory control of the manipulator through a dynamic iterative learning process.Experimental results on the freedom testing platform of the manipulator show that the proposed method can not only reduce model parameters and increase the calculation speed,but also effectively control the manipulator and accurately adjust its position information.

关 键 词：机械臂 无标定视觉 非线性矩阵 轨迹控制 

分 类 号：TP24[自动化与计算机技术—检测技术与自动化装置]