Adaptive eXogenous Kalman Filter for Actuator Fault Diagnosis in Robotics and Autonomous Systems

This paper presents an algorithm for actuator fault diagnosis in robotics and autonomous systems under random uncertainties based on a cascade of nonlinear observer and linearized Kalman filter. The two-stage estimation method assumes uniform complete observability and controllability conditions and persistent excitation condition. To this end, we consider dynamical systems of robotics and autonomous systems with one-sided Lipschitz nonlinearity. To demonstrate the effectiveness of the proposed algorithm, numerical simulations in a single-link flexible joint robot are performed.