Fault diagnosis for robot manipulators via vision servoing based suboptimal second order sliding mode

This paper deals with the formulation of a fault diagnosis strategy for industrial robotic manipulators. The core of the proposed scheme is the inverse dynamics-based feedback linearized robotic MIMO system, that is a set of linearized decoupled SISO systems affected by uncertain terms. Relying on this set of systems, in the paper the problem of detecting and isolating both sensor and actuator faults is considered. The proposed fault diagnosis strategy consists of a vision based logic, to detect possible malfunctions of the sensors of the robot, and a set of Unknown Input Observers (UIOs) of second order sliding mode type, to perform the Fault Detection and Isolation (FDI) on the actuators. The sliding mode approach provides good performance in terms of stability and robustness, while guaranteeing a satisfactory estimate of the faults. To give the possibility to the fault diagnosis system to distinguish between sensor and actuator faults, a vision sensor is used in the scheme. This sensor also allows to design a fault tolerant control strategy in case of sensor faults. The verification and the validation of the present proposal have been carried out, both in simulation and experimentally, on an industrial robotic manipulator.