This paper presents multiple methods for the design of a robust disturbance torque estimator. The benefit of these designs is that they ensure robust estimation in the presence of model uncertainties and/or noise. It is shown that these estimation schemes can be used to estimate both constant and nearly constant disturbance torques. In order to design the observers, the nominal plant model is expanded to incorporate uncertainties. Various cases for the design of the observer are presented. All of the cases are tested on a real system using varying degrees of model uncertainty to ensure that robust estimation is achieved. The results are validated using an in-line torque sensor and are presented accordingly.

No abstract

No abstract