Neuro-Fuzzy Controllers

Fuzzy controllers belong to the class of knowledge basedsystems. Their main goal is to implement human know-how orheuristic rules in the form of a computer
program. Fuzzy logic provides a mathematical formalism forthis goal. An original method for the design and analysisof a fuzzy controller is proposed in this book.
The main idea is to apply an on-line adaptive algorithm toautomatically adjust the parameters of the fuzzycontroller, depending on its inputs and desired outputs. Itis
shown that a fuzzy controller can learn to approximatenon-linear functions arbitrarily well. The author alsointroduces a new method for the extraction oflinguistic
rules from the adapted parameters. The knowledge acquiredduring learning can be represented in a clear and intuitivesyntax. This gives the designer a
comprehensive understanding of the new rule base, i.e., thecontroller features. In the last part of the book, theapplication of fuzzy and neuro-fuzzy controllers tothe
navigation of a mobile robot is investigated. Severalexperiments with the miniature mobile Khepera are reported.The proposed design method for fuzzy controller is
tested, showing that it provides the designer with atechnique to handle complex tasks: the robot cansuccessfully learn the desired behaviours, even withnoisy
sensors and large rule bases. The extraction of rulesreveals to be extremely useful for practical applicationsbecause it helps the engineer to learn more on the
complete system.