Adaptive Smith Predictor Fractional Control of a Tele-operated Flexible Link Robot ^*

Abstract

This work addresses the adaptive Smith predictor control of a tele-operated robot composed of a flexible link. Measurements of the angle of the motor that moves the arm by using an encoder and the moment at the base of the arm by using two strain gauges are fed back. These strain gauges normally present noticeable offset and high-frequency noise. In order to implement such control, the simultaneous real-time characterizations of the main vibration mode - which changes with the carried payload - and the delay of the tele-operated robot - which is time-varying - are carried out using a new algorithm based on an algebraic identification technique, which is robust to the previous strain gauge disturbances. This algorithm is faster than others previously developed, being especially suited to implement an adaptive version of the Smith predictor. The control of the tip position of the robot is closed using a fractional order controller, which has the advantage of removing the steady state error introduced by the strain gauges offset on the tip position. This adaptive control system is subsequently evaluated on a prototype. Simulated and experimental results are presented demonstrating the speed, accuracy, and robustness of the performed control system.