Fault Recovery Using Evolvable Fuzzy Systems

Autonomous systems must survive for long periods without relying on humans for fault recovery. Faults may be impossible to analyze remotely and redundant hardware is frequently not allowed. The only recourse may be to replace the existing control strategy. This paper proposes a fuzzy logic controller architecture that executes a replacement control strategy for autonomous systems. This paper shows this architecture is ideally suited for fault recovery in autonomous systems with imprecisely defined faults. The controller is evolved in-situ and evaluated intrinsically