Extended Kalman Filter Based Modelled Predictor for Fusion of Accelerometer and Camera Signal to Estimate the Vibration of a Mobile Flexible Link Manipulator

Abstract

This paper presents the fusion of accelerometer and camera for active vibration prediction for a mobile flexible link manipulator based on Extended Kalman filter-based modelled predictor. The tip position of the manipulator is unpredictable due to the singularity of the mobile flexible manipulator, as well as the phase lag in the control system due to the time delay between the sensor feedback and the control input. The purpose is thus to improve the prediction accuracy of the tip position. The time delayed in camera data estimates is used to correct the drifting accelerometer’s signal. The dynamic model of the mobile flexible link manipulator is derived and is used to feed to the prediction stage of the Extended Kalman filter, which is used for vibration prediction. In order to investigate the efficiency of the proposed method, simulation and experimental studies are performed considering a single link flexible manipulator on a wheeled base. Experimental verifications showed that the proposed method produced good vibration prediction of the mobile manipulator compared to other model based predictor.