Using fuzzy logic for bounded recovery of autonomous agents

The solution to the problem of application-independent fault recovery of autonomous agents requires a specification method for the agent´s capacity to act outside of its normal operational limits. This paper presents a recovery method based upon the theory of a fuzzy finite state machine. A fuzzy specification is given for the bounds within which an autonomous agent is capable to recover after an unexpected situation has occurred in its environment. It has been shown that the three main components of the recovery problem: fault detection, fault recovery, and the properties of the actuator/sensor gear of an autonomous agent are interrelated. The suggested method can be implemented either by an application-independent software algorithm, or by fuzzy logic hardware