Fault Diagnosis of a SCARA Robot

Abstract

A new approach for fault detection of a robot manipulator was introduced in this paper. Unlike existing methods mounting varies sensors at every joints, the new technique applied only one accelerometer mounted at the tip of the robot. Based on dynamic analysis of the robot, the theoretical acceleration profile of the programmed motion was obtained, which was compared with the acceleration signals measured during working for fault diagnosis. Advanced detrending algorithm was adopted to extract programmed acceleration data from the tested noisy acceleration signals. Typical FFT-based spectrum analysis was applied for analyzing both the trended and detrended data for error detection. The effectiveness of the proposed approach was verified with experiments conducted on a four-joint SCARA manipulator.