Towards autonomous fuzzy control

The efficient implementation of on-line adaptation in real time is an important research problem in fuzzy control. The goal is to develop autonomous self-organizing controllers employing system-independent control meta-knowledge which enables them to adjust their control policies depending on the systems they control and the environments in which they operate. An autonomous fuzzy controller would continuously observe system behavior while implementing its control actions and would use the outcomes of these actions to refine its control policy. It could be designed to lie dormant when its control actions give rise to adequate performance characteristics but could rapidly and autonomously initiate real-time adaptation whenever its performance degrades. Such an autonomous fuzzy controller would have immense practical value. It could accommodate individual variations in system characteristics and also compensate for degradations in system characteristics caused by wear and tear. It could also potentially deal with black-box systems and control scenarios. On-going research in autonomous fuzzy control is reported. The ultimate research objective is to develop robust and relatively inexpensive autonomous fuzzy control hardware suitable for use in real time environments.