基于PSO优化神经网络和空间网格的机器人位姿标定方法.pdfVIP

第42卷第8期 中国测试 Vol.42 No.8 2016年8月 CHINA MEASUREMENT TEST August，2016 doi院10.11857/j.issn.1674-5124.2016.08.020 基于PSO优化神经网络和空间网格的 机器人位姿标定方法 王 一袁 宋志伟袁 王祎泽袁 张湧涛 渊华北理工大学电气工程学院袁河北 唐山 063009冤 摘 要院该文提出一种将机器人的位置和姿态拆分开袁分别进行标定的机器人位姿标定方法遥采用空间精度控制网格 标定机器人定位误差袁粒子群优化算法渊particle swarm optimization袁PSO冤优化神经网络标定机器人定姿误差遥该方法 以指数积公式 渊product of exponentials袁POE冤为基础建立机器人正向运动学模型袁用映射法建立空间网格袁用三坐标 测量臂测量机器人位姿袁用空间网格精度标定定位误差袁用PSO优化的神经网络标定定姿误差遥其优点在于既标定机 器人工具中心点渊TCP冤的定位误差袁又标定机器人工具坐标系的姿态误差袁使得机器人定位尧定姿误差都得到补偿遥实 验结果表明机器人的定位尧定姿均方根误差减小接近一个数量级遥 关键词院空间网格精度曰粒子群算法曰机器人标定曰神经网络 文献标志码院A 文章编号院1674-5124渊2016冤08-0098-05 Robot calibration method based on spatial mesh and PSO optimal neural network WANG Yi袁 SONG Zhiwei袁 WANG Yize袁 ZHANG Yongtao 渊College of Electrical Engineering袁North China University of Science and Technology袁 Tangshan 063009袁China冤 Abstract: A robot爷s posture calibration method which separate the rotation and location of robot to separate calibrate is introduced. Uses spatial mesh precision to calibrate the robot爷s position error袁 and applys PSO neural network to calibrate the robot爷s rotation error. This paper use POE formula to establish the forward kinematics model of robot. A coordinate measuring machine has been used to measure the robot working space. Apply reflection method to set up spatial mesh. Position error of the robot has been calibrated by spatial mesh precision and the rotation error has been calibrated by neural network. The robot calibrate method in this paper is different from the generals. Advantage of this method is that it not only cal