作　　者： ; ; ;

机构地区： 华南理工大学机械与汽车工程学院

出　　处： 《华南理工大学学报（自然科学版）》 2009年第8期65-70,共6页

摘　　要： 针对机器人姿态的视觉测量方式,建立了适用于机器人全标定的运动学模型,在此基础上推导了以世界坐标系为标定的基准坐标系且包含所有参数的误差模型,指明了误差模型的参考坐标系,解决了仿真上的关键技术,研究了验证误差模型正确性和精度的公式,并对误差模型的正确性和误差辨识精度进行了仿真对比.结果表明:误差模型有一定的精度限制;随着测量次数的增加,误差辨识的精度会提高,但当测量次数达到一定数量时,辨识精度提高不明显. A kinematic model suitable for the complete calibration of robot is built according to the vision-based pose measuring method, and an error model considering all the parameters is deduced, with the world coordinates as the reference coordinates. Then, the reference coordinates of the error model is clearly pointed out, and prob- lems relating to the key technology of simulation are solved. Moreover, some equations to validate the error model are derived, and the correctness and accuracy of the proposed error model are compared by simulation. The results indicate that the accuracy of the error model is limited, and that the identification accuracy improves; however, the identification accuracy does not exhibit obvious increase when the number of measurement times reaches a relatively constant value.

关 键 词： 工业机器人 机器人标定 误差建模 视觉测量 绝对精度

领　　域： [自动化与计算机技术] [自动化与计算机技术]