摘要
机器人要完成零件的准确抓取,关键问题之一是需知道零件的精确位置。传统的机器人抓取零件多采用的是示教编程方法,当装配区零件的摆放位置发生改变,需重新编制控制程序,缺乏一定的柔性。为改善装配机器人的零件定位,出现了各种各样的定位方法,其中基于视觉的定位方法应用广泛。为此,本文引入了基于形状匹配的装配定位方法。首先,标定了机器视觉系统的相机变形,并将其应用到图像采集中;其次,通过图像金字塔的搜索策略和最小二乘法的优化方法,实现了形状匹配的亚像素定位。最后,由5个零件的装配实验,验证了定位方法具有较好的实时性和识别率。
One of the key issues for a robot to accurately grasp a part is to know its exact location. Traditional robots grasp a part mostly by teaching programming. In this way, when people alter the location of a part in the assembly area, the robots＇ control program should be generated again, showing poor flexibility. To im- prove the part positioning of assembly robots, there have been a variety of positioning methods, among which the vision - based positioning method is universally applied. To this end, this paper introduces the assembly location method grounded on shape matching. Firstly, the camera distortion of the machine vi- sion system is standardized and applied in image acquisition. Secondly, through the search strategy of image pyramid and the optimization method of least squares, sub - pixel positioning of shape matching is realized. Eventually, the assembly experiment of five parts is conducted, thus verifying that the location method has good recognition rate and can be realized in real -time.
出处
《制造技术与机床》
北大核心
2015年第5期97-100,共4页
Manufacturing Technology & Machine Tool
基金
福建省龙岩市重点科技计划项目(2012LY01)
关键词
机器人
形状匹配
相机标定
定位方法
robot
shape matching
camera calibration
location method
