: Engell, S.; Kleiner, A.

Abstract
Despite the fact that a robot can be linearized and decoupled by suitable non-linear transformations, current industrial practice is independent joint control with fixed linear control laws. This is mainly due to the computational complexity and the lack of robustness results for such nonlinear schemes. As the joint dynamics vary with the position and speed of the controlled joint and the other joints as well as with changes in the mass of the payload, the design of fixed linear joint controllers is a robust control problem. In recent years, frequency domain methods have been proposed as the most suitable tools for the analysis and synthesis of robust control systems (1-4). In this paper, we report of the identification of the dynamics of one joint of an industrial robot using frequency domain methods. This approach yields not only a nominal model for controller design purposes but also information about the uncertainty in the system dynamics. From a frequency domain description of the deviations of the dynamic behaviour from the nominal model, the attainable performance of the controlled joint can be estimated.