Regressor based sliding mode control of robotic manipulators

A regressor-based sliding mode controller is presented for robotic manipulators with the consideration of computational efficiency. Three specificities distinguish the controller from the known designs. First, the regressor matrix used by the control law is calculated about the desired trajectory rather than with respect to current robot states. This results in significant simplifications since the regressor can be calculated offline, which makes the controller implementation simple and fast. Second, the authors have proved the global asymptotic stability of the controller, and its robustness with respect to a class of state-dependent uncertainties, which may result, for example, from unmodeled dynamics. This class includes uncertainties which can be bounded by a first degree polynomial of the tracking error and its first derivative. The robustness of the controller with respect to the measurement noise has also been proved. Third, an adaptive version of the controller, also presented, provides a technique for reducing magnitudes of the control signals