Fuzzy control: cloning and Kalman-based learning

A different perspective of fuzzy control is introduced based on evolutionary concepts and Kalman gain properties. No explicit verbal (if-then) rules are needed. The fuzzy controller emerges directly from state space design through a cloning process. State feedback gains are cloned into proportional fuzzy controllers. Each state variable is associated with a fuzzy rule. Kalman techniques are the basis for rule learning, reshaping the cloned fuzzy rules. Simulations with a rotary inverted pendulum are presented to test the method.