Adaptive gain sliding observer based sliding controller for uncertain parameters nonlinear systems. Application to flexible joint robots

An adaptive gain smooth sliding controller, based on a smooth sliding observer, is developed to control nonlinear SISO affine systems with uncertain parameters and state functions. Furthermore, an adaptively updated parameter term is introduced in the steady state space model of the controlled system in order to obtain useful information despite fault detection. Using a sliding observer with smooth switching function and adaptive gain increases the robustness w.r.t. uncertainties. The adaptive gains smooth sliding observer and controller are designed to fulfill the attractiveness condition on the corresponding switching surfaces. An application to a single arm, flexible joint robot is presented. In order to alleviate chattering in the observer and the controller, a parameterized tangent hyperbolic is used as a switching function, instead of a pure relay one. The gain of the switching function is adaptively updated, depending on the estimation error or on tracking error.