A sensitivity analysis based geometric parameter identification for 6-DOF heavy-duty industrial robots

Different structural parameters of 6-DOF (degree-of-freedom) heavy-duty robots have varying impacts on the absolute positioning accuracy of the end-effector. In this paper, a sensitivity analysis based geometric parameter identification method is proposed for 6-DOF heavy-duty industrial robots. A total of 34 geometric error parameters were modeled, which include robot D-H (Denavit-Hartenberg) parameters, tracker transformation parameters, and target ball installation parameters. Firstly, all parameters are identified globally using the least squares method, while the accuracy of geometric parameters is limited in nature. Given that each parameter has different effects on end-effector accuracy, sensitivity analysis is introduced to identify a subset of key parameters. On this basis, high-precision calibration of the selected 14 sensitive parameters avoids the disadvantage of local calibration only for key subsets. The method was experimentally verified on a heavy-duty industrial robot, and the results showed that the absolute positioning error of the end effector is reduced by 90.59% (compared with the uncalibrated state). In addition, the proposed method achieves an improvement of 24.05% over the traditional global calibration, and also outperforms several commonly used identification algorithms.