Observer-based fault detection for robot manipulators

The adoption of efficient online fault detection and isolation (FDI) tools is becoming of the utmost importance for robots, especially for those operating in remote or hazardous environments, where a high degree of safety and self-diagnostics capabilities is required. In this paper a new observer-based approach to fault detection and isolation for robot manipulators is proposed. A nonlinear observer of the system's outputs, joint position and velocities, is designed directly in the discrete-time domain; in order to improve the robustness of the observer to unknown dynamics and discretization errors, a linear feedback of the observation error and a delayed nonlinear compensation action are added. The residuals for the detection of both sensor and actuator failures are then generated from the observation errors. Simulation results show the effectiveness of the proposed approach, even if a relatively low sampling rate is adopted and in the presence of large modeling error