基于无迹卡尔曼滤波的机器人手眼标定

Robot Hand-Eye Calibration Using Unscented Kalman Filtering

  • 摘要: 提出一种基于无迹卡尔曼滤波(UKF)的机器人在线手眼标定算法来求解齐次变换矩阵方程AX=XB.建立手眼标定的隐式马尔可夫模型(HMM),并对它进行无迹卡尔曼滤波,从而对标定参数的状态进行递归贝叶斯估计和实时可视化处理,蒙特卡洛仿真结果表明,在小高斯噪声、较大高斯噪声以及非等方向性高斯噪声模型下,本文算法估计结果的精确度优于传统的最小二乘标定算法.通过实际机器人手眼标定实验验证了本文算法的稳定收敛性,标定结果的误差小于最小二乘法.

     

    Abstract: An online robot hand-eye calibration method using unscented Kalman filtering(UKF) is presented to solve the homogeneous transformation equations of the form AX=XB.The hidden Markov model(HMM) of the hand-eye calibration is constructed,based on which UKF is performed to estimate calibration parameters recursively according to the Bayesian theory,and the evolution of calibration parameters can be visualized in real time.Monte Carlo simulation shows that the proposed algorithm possesses better accuracy than the traditional least squares(LS) based method under low isotropic,high isotropic and anisotropic Gaussian noises.Real robot hand-eye calibration experiment is also performed and the result shows stable convergence of the proposed algorithm with better accuracy comparing with the LS based method.

     

/

返回文章
返回