工业机器人参数辨识奇异性问题研究

Singularity study on parameters identification of industrial robot

在工业机器人两步辨识过程中都存在结构参数辨识矩阵奇异的问题。在转动参数辨识过程中通过编码器获得与辨识矩阵奇异列相对应的结构参数,利用部分矩阵处理法避免了转动参数辨识过程中的奇异问题;在其余参数辨识过程中,通过提前测得机器人某一特征参数,解决了由于辨识矩阵本身特点引起的参数不可辨识问题。在解决结构参数辨识奇异性问题基础之上,可以准确得到工业机器人结构参数的线性解。转动结构参数辨识过程中引入牛顿迭代法,辨识结果精度提高,其余参数辨识过程中通过分离与结构参数无关项,得到关于结构参数的线性方程,从而得到其线性解。

Matrix singularity problem always exists in the process of identifying structure parameters of the industrial robot based on two-step method. Using encoder, the structure parameters which correspond to singularity column of identification matrix were obtained when identify rotate parameters. Part of the matrix method was used to avoid the singularity problem of rotate parameters. When identity the rest parameters, the characteristic parameter tested in advance made the parameter unidentified problem can be solved. On the basis of solving the singularity problem, the accurate linear solution of industrial robots can be gained. Identification results accuracy will be improved introducing the newton-interpolation to the process of rotate parameters identification. When identifying other parameters, the irrelevant factors were separated and the linear equations were obtained, as well as linear solutions.