基于PLS的拾放机械手位姿分析

Position and Attitude Analysis Based on PLS for Pick & Place Manipulator

为提高拾放机械手位姿的精确度,对采集到的末端执行器位移信号进行了研究。主要采用多分辨率小波变换法提取机械手运动时产生的位移信号的小波系数标准偏差作为特征矢量,将特征矢量输入PLS进行特征提取和动作模式分类,最终实现了对拾放机械手位移信号的四种不同动作模式的分类。试验结果表明,采用多分辨率小波变换法进行特征提取的PLS分类方法,具有更高的识别率,运算速度更快,对于拾放机械手的位姿分析具有一定的借鉴意义。

To improve the accuracy of the posture of manipulator,the displacement signals collected from end effector are researched. The standard deviation of wavelet coefficients of the displacement signal generated by manipulator movement is extracted by adopting multi-resolution wavelet transform,and then used as characteristic vector for calculation. The characteristic vector is input partial least square( PLS),for feature extraction and motion pattern classification,finally,classification for four of the motion patterns of displacement signals of pick & place manipulator is implemented. The test results show that the PLS classification method using multiresolution wavelet transform for feature extraction possesses higher recognition rate,faster operation speed,this provide reference significance for position and attitude analysis for pick & place manipulators.