A fuzzy classifier system using hyper-cone membership functions and its application to inverted pendulum control

This paper proposes a fuzzy classifier system (FCS) using fuzzy rules given by hyper-cone membership functions. The hyper-cone membership function is expressed by a kind of radial basis function, and its fuzzy rules can be flexibly located in input and output spaces. Therefore, the FCS can generate excellent rules which have the best location and shape of membership functions. We apply the FCS to a fuzzy rule generation for the inverted pendulum control. Also, we introduce the simplified reward acquisition method for evaluation of inverted pendulum performance.