Fuzzy Stochastic Automata for Reactive Learning and Hybrid Control

Fuzzy Stochastic Automata (FSA) are suitable for the modelling of the reactive (memoryless) learning and for the control of hybrid systems. The concept of FSA is to switch between a fuzzy increase and a fuzzy decrease of the control action according to the sign of the product e e, where e = x - xd is the error of the system's output and is its first derivative. The learning in FSA has stochastic features. The applications of FSA concern mainly autonomous systems and intelligent robots.