基于强化学习微操作系统的机器视觉与无标定视觉伺服研究  

Research on Machine Vision and Uncalibrated Visual Servo Based on Reinforcement Learning Micro Operation System

作　　者：涂胜倩 TU Shengqian(Basic Experimental Teaching Center,Anhui Sanlian University,Hefei 230601,China)

机构地区：[1]安徽三联学院基础实验教学中心,合肥230601

出　　处：《长春大学学报》2021年第12期12-17,34,共7页Journal of Changchun University

基　　金：国家自然科学基金项目(21XZX006);安徽三联学院校级重点项目(KJZD2020007)。

摘　　要：建立视觉伺服系统中的雅克比矩阵,进行运动学分析。并设计一种机器视觉模型,为无标定视觉伺服系统的展开奠定了理论基础。提出了一种雅可比矩阵图像辨识法和自适策略,达到了好的效果。对系统控制的精确性验证,得到良好的系统性能。最后,设计微操作系统自动平台和四层软件构架,以提升软件维护和升级性能。用提出的算法,在平台上实行靶装配ICF实验,验证了方法的先进性和有效性。Jacobian matrix of visual servo system is established and kinematics analysis is carried out.The design of the machine vision model lays a theoretical foundation for the development of uncalibrated visual servo(UVs)system.A Jacobian matrix image recognition method and adaptive strategy are proposed,and good results are achieved.The accuracy of the system control is verified,and good system performance is obtained.Finally,the automatic platform and four-layer software architecture are designed to improve the performance of software maintenance and upgrade.The proposed algorithm is used to carry out the target assembly ICF experiment on the platform,and the method is proved to be advanced and effective.

关 键 词：微操作系统 机器视觉 视觉伺服 

分 类 号：TP242.6[自动化与计算机技术—检测技术与自动化装置]