This paper presents a novel approach to combine three powerful methods of control theory and fuzzy c-means clustering algorithm to build a robust learning control for nonlinear uncertain systems such as advance robot manipulators. It is worth noting that the combination not only encompasses the features and capabilities of its components but also the limitations attributed to these techniques (e.g, stability issues of fuzzy controller and poor performance of PID controller in the presence of time-varying uncertainties) may be remedied by each other. To date different combinations of the above mentioned methods presented in the literature each having its own merits and limitations. But, for advance robotics applications such as medical robot, and space applications and with increasing complexity of the robot’s tasks, there is a pressing need for the control systems that are able to learn systematically and efficiently during the course of operation, from its own experience, from the demonstration and also in an unsupervised fashion. The controller proposed in this paper aims at building such a control system using a novel approach to combine adaptive fuzzy modeling algorithm, fuzzy c-means clustering algorithm, adaptive sliding mode control and PID controller. Based on the simulations and experimental results, the proposed controller performs remarkably well in terms of the tracking error convergence and robustness against uncertainties.

This content is only available via PDF.
You do not currently have access to this content.