Actuator Fault Estimation for a Class of Euler-Lagrange Systems Using Super-Twisting Observers

This paper develops an adaptive actuator fault estimation scheme for a class of Euler-Lagrange systems with relative degree two, which allows the loss of actuator effectiveness and generalised velocities to be estimated. In particular, a super-twisting observer is created for Euler-Lagrange systems with multiple degrees of freedom and the ‘equivalent output error injection’ signals, associated with the super-twisting sliding motion, are exploited to reconstruct the loss of effectiveness for all the actuators in an adaptive sense. The proposed scheme is then applied to a six-link underactuated snake robot model and the simulation results demonstrate the efficacy of the scheme.