基于遗传算法的机器人动态视觉检测系统  

Robot dynamic visual detection system based on genetic algorithm

作　　者：吴勇[1] 郝矿荣[1] 丁永生[1,2] 张淑平[1] 

机构地区：[1]东华大学信息科学与技术学院,上海201620 [2]数字化纺织服装技术教育部工程研究中心,上海201620

出　　处：《电子测量技术》2009年第4期108-112,共5页Electronic Measurement Technology

基　　金：国家自然科学基金(60775052);国家自然科学基金重点项目(60534020);上海市国际科技合作基金项目(061307041);上海市科学技术委员会重点基础研究项目(08JC1400100);上海市人才发展资金(001);上海市领军人才后备人选专项资金

摘　　要：建立了一种基于二自由度机器人视觉伺服系统,该系统通过实时监控和跟踪运动目标,实现对运动目标的相对距离和瞬时速度的动态检测。本系统为双目视觉系统,采用遗传算法进行特征点匹配,利用其随机搜索的特点,更快更好地找到全局最优值,从而得到较高匹配精度的点。最后通过三维重建和建立速度模型,实现了对机器人动平台运动参数的动态检测,并给出测量精度及改进方法。此方法还适用于多自由度并联机器人和空间复杂运动的检测。A visual servo system based on a 2-DOF robot is presented in this paper. This system can dynamically detect the relative distance and the instantaneous velocity of a moving object. It is a binocular vision system and has already calibrated. It uses genetic algorithm to match the feature points. Base on GA＇s random searching method, we can find the global optimization value faster and better, and then get higher accuracy of matching points. Finally, through 3-D reconstruction and the establishment of velocity model, we reach the goal of dynamic detection about robot＇s moving platform motion parameters. We also give the measurement accuracy and improved method of the system. This improved method can also applicable to multi-DOF parallel robot and space complex motion detection.

关 键 词：立体视觉 遗传算法 动态检测 二自由度机器人 

分 类 号：TP2[自动化与计算机技术—检测技术与自动化装置]