Control of robotic systems by using adaptive control and learning methods

ABSTRACT CONTROL OF ROBOTIC SYSTEMS BY USING ADAPTIVE CONTROL AND LEARNING METHODS TOPALBEKIROGLU Mehmet M.S. in Mechanical Engineering Supervisor: Assist. Prof. Dr. Ali KIRECCI January 1996, 138 pages In this study, advanced control methods such as adaptive and learning control methods are aimed to investigate and apply experimentally which ensure the stability of robotic systems under every conditions. Also, performance of these control methods and conventional control method are compared. Adaptive control can be classified into two groups, namely, self-tuning and model reference adaptive control. Additionally, self-tuning adaptive control can be divided into many groups theoretically, however, two of them can be practically used, which are explicit and implicit self-tuning adaptive control methods. The general strategy for all adaptive control methods is the systematic estimation of system parameters, which minimise the output error of the system at the next step. Learning control is based upon systematic tuning the control input to reduce the error using the information of previous cycles. Adaptive control and learning control methods are applied to a hydraulic robot which has three DOF, and a DC servo system. The experimental results show that these control methods can be applied to ant robotic system which can provide very good performance with these control method under any circumstances. Keywords: Adaptive control, implicit self-tuning, explicit self-tuning, model reference adaptive control, learning control and parameter estimation methods. m