Near-optimal fuzzy systems using polar clustering: Application to control of vision-based arm-robot

Abstract

This paper presents a design algorithm to near-optimal fuzzy systems using polar clustering method for vision-based robot control systems. The complexity of the optimal fuzzy system for a vision-based control system is so great that it can not be applied to real systems or can not be useful. Therefore we generally use clustering method, to reduce the complexity of optimal fuzzy systems. In the class of near-optimal fuzzy systems, for more efficient use of clustering, we propose the polar clustering method using polar quantization. In order to verify the effectiveness of the proposed method, experimentally, it is applied to a vision-based arm robot control system. © Springer-Verlag Berlin Heidelberg 2005.