Enhanced lumped stiffness model for industrial robots under multi-axial loading: Trade-off between the model complexity and positional accuracy

Abstract

The elastostatic calibration of industrial robots using reduced order stiffness models predominantly relies on joint compliance. The contribution of link compliance ignored in these models assumes significance, especially when the end-effector employs external loads. The presented work overcomes the limitation by proposing a simpler reduced order model encompassing joint and link compliance. The proposed model has only nine parameters compared to an existing model with twenty-six parameters. A novel top-down approach is adopted for identifying these nine parameters, wherein the parameters are lumped to accurately model the end-effector deflection under multi-axial loading. A dedicated experimental setup involving two ABB IRB 7600-500 robots was used to validate the model. The performance of the nine parameter model is akin to other sophisticated models involving many parameters. The calibration resulted in a 91% reduction in position error. The model and identification strategies are generic and can be adapted to any similar serial robot. The compatibility of the identified parameters was tested using another robot of the same (make) specification. The error predictions were in a similar order, confirming the robustness of the approach.