Robot Manipulator Disturbance Observation and Servo-Fault Diagnosis without Speed Sensor

Unknown friction and disturbance have always been complex problems for the control and fault diagnosis of robotic manipulators. This paper proposed a method for robot manipulators to realize disturbance observation and servo motor faults diagnosis under an unknown environment with only the joint position sensors. The friction and disturbance observation is based on an improved second-order sliding mode observer (SMO) and a nonlinear disturbance observer (NDOB). The improved SMO can achieve the joint speed observation considering the model uncertainties. Given the indistinguishable problem of servo motor error torque and environmental disturbances. The Fourier expansion model is proposed as an observation model to capture the relationship between the disturbance and time in a specific time interval. By combining with the Particle Filter (PF) algorithm, this method can accurately compensate for robotic manipulators' fault-free friction and disturbances. Meantime, it also realizes the distinction of the servo error torque from the observation results of NDOB. The whole algorithm is verified in MATLAB environments.