Robust feedback linearization using higher order sliding mode observer

A novel robust feedback linearization scheme is proposed in this paper based on a modified robust exact differentiator. The states and drift terms in the system are estimated simultaneously by the observer using back injection of the control effort. The estimated drift term is used in the feedback loop to compensate the disturbances and observed states are used for feedback linearization. Finite time convergence of the complete closed-loop system is proved and thus a form of separation principle is satisfied, i.e., the controller and observer can be separately designed. The design is verified through simulations and by experiments on a DC motor rig.