Robustness issues in the control of high performance robots

A methodology is presented to achieve effective on-line tracking control of robot manipulators in the presence of model uncertainty and torque limitations. A suction controller is used in the main feedback loop, allowing to combine robust tracking with simple controller design. Pointwise optimization techniques are incorporated into a ´slow´ feedback loop in order to exploit joint coupling and generate feasible desired trajectories optimally close to the original (perhaps unfeasible) desired trajectories. The algorithm is computationally cheap and can easily be implemented on-line. Further, the scheme facilitates higher-level programming by only requiring reduced information about the robot model and the task structure.