A robustness analysis of feedback linearization or nonlinear decoupling is proposed for a class of elastic-joint robot manipulator control loops. It is shown that by splitting the modeling-error nonlinearities of incomplete feedback linearization or nonlinear decoupling by a suitable transformation, a robustness analysis becomes feasible under weak assumptions without the necessity of estimating bounds on derivatives of robot nonlinearities. In principle, systems other than elastic-joint robots can be investigated using the proposed methodology, which makes any robustness analysis and synthesis of feedback linearization or nonlinear decoupling more transparent.>

A methodology for the design of robust manipulator control loops is presented. Generally applicable stability conditions for a class of complex, nonlinear, time-variant control systems are applied to robot manipulators with rigid body dynamics. The robust stability design method allows the model uncertainties of various nonlinear or pure linear controllers to be taken into account, and may be viewed as a framework for robust manipulator control. It is shown that for a big class of robot manipulators, robust control can be achieved by frequently used linear time-invariant state-feedback controllers. Graphs are developed to facilitate the controller design.>