基于3D相机的拆垛机器人手眼标定方法研究
Research on Hand-Eye Calibration Method of Bin-Packing Robot Based on 3D Camera
郭源 1张爱军1
作者信息
- 1. 南京理工大学机械工程学院,南京 210094
- 折叠
摘要
为了提高拆垛机器人手眼标定的准确度,设计了一种结合有序点云和对偶四元数的改进粒子群优化算法的手眼标定方法.首先,采用三层梯状采样法获取不同姿态的手眼标定数据;然后,对圆形标定板进行圆心提取,采用加权奇异值分解的方案求解定位点的旋转平移矩阵,减少了噪声干扰;最后,将求得的刚性变换矩阵转换为对偶四元数,采用改进的自适应粒子群优化算法求解出手眼矩阵的旋转和平移部分.实验结果表明,手眼标定方法可稳定地处理标定数据,且平均误差小于2.5 mm,优于Tsai法和传统对偶四元数法.
Abstract
In order to improve the accuracy of hand-eye calibration of bin-packing robot,an improved parti-cle swarm optimization algorithm combining ordered-point cloud and dual quaternion was designed.Firstly,the hand-eye calibration data with different gestures were obtained by the three-layer ladder sampling meth-od,and then the center of the circular calibration plate was extracted.The rotation and translation matrix of the calibration point was solved by the weighted singular value decomposition scheme,which reduced the noise interference.Finally,the rigid transformation matrix is converted into dual quaternion,and the rotation and shift parts of the hand-eye matrix are solved by an improved adaptive particle swarm optimization algo-rithm.The experimental results show that the proposed method can process the calibration data stably and the av-erage error is less than 2.5 mm,which is better than Tsai's method and the traditional dual quaternion method.
关键词
手眼标定/机器视觉/粒子群算法/3D相机/SVD算法Key words
hand-eye calibration/machine vision/particle swarm algorithm/3D camera/SVD algorithm引用本文复制引用
出版年
2024
NETL
NSTL
维普
万方数据