Static model analysis and identification for serial articulated manipulators

摘要

Many applications of industrial manipulators, such as robot tracking tasks and lead-through programming, require the availability of a static model, which can compensate the gravity and joint friction accurately. Compared with dynamic model, static model identification is more attainable in practice, as it only requires measurement of joint position instead of joint velocity and acceleration. However, there is still no systematical approach for static model identification for general serial articulated manipulators. In this paper, we propose a static model for serial articulated manipulators based on the analysis with standard Denavit-Hartenberg (DH) convention. Then, a theorem is given to obtain a non-singular regression matrix for static model identification, where the ordinary least squares can be applied directly. With well conditioned regression matrix, a static model identification procedure is proposed. Both simulation and experimental results on a 6 DoF manipulator show that the proposed procedure is powerful and practical. Finally, an application of the identified static model in lead-through programming is presented for the validation of the proposed methods.