非线性最优机器人手眼标定  

Nonlinear Optimal Robot Hand-Eye Calibration

作　　者：王君臣[1] 王田苗[1] 杨艳[1] 胡磊[1] 

机构地区：[1]北京航空航天大学机器人研究所,北京100191

出　　处：《西安交通大学学报》2011年第9期15-20,89,共7页Journal of Xi'an Jiaotong University

基　　金：国家高技术研究发展计划资助项目(2008AA040205);国家科技支撑计划资助项目(2011BAF01B02)

摘　　要：提出一种基于最大似然估计的非线性最优手眼标定算法,并采用该方法对机器人进行了手眼标定.由于机器人系统的运动学正解及相机的外参数推定都包含误差,因此不能满足手眼标定矩阵方程,从而导致现有机器人手眼标定算法对观测噪声较为敏感.新算法根据包含噪声的测量值来估计它们的真值,使得测量真值满足手眼标定方程,并且在欧拉刚性变换矩阵空间中可使真值和对应的测量值的距离最小.通过数值仿真和真实机器人手眼标定实验,将新算法与现有2种经典的手眼标定算法进行了比较,结果表明,新算法在估计误差、对位姿变换次数的敏感度和稳定性方面均优于经典算法.Nonlinear optimal robot hand-eye calibration based on maximum likelihood estimation is proposed.For the errors in both robot forward position kinematics and camera extrinsic calibration,the classical hand-eye transformation matrix equation often gets violated,thus the existing robot hand-eye calibration algorithms derived from the equation are sensitive to the measurement noise.The proposed strategy estimates the hand-eye transformation and the true measurements according to the noisy measurements to guarantee that the true measurements satisfy the hand-eye equation and the distance to the noisy measurements is minimal.Two classical hand-eye calibration methods are also implemented for comparison.Numerical simulations and real robot hand-eye calibration experiment are performed and the results show the proposed strategy has higher accuracy,sensitivity to motion numbers,and stability.

关 键 词：机器人 手眼标定 最大似然估计 

分 类 号：TP24[自动化与计算机技术—检测技术与自动化装置]