GENETIC PROCESSING OF THE SENSORY INFORMATION

A general method for optimizing the behavior of fuzzy control systems that receive information coming from a sensorial system is presented. This optimization is achieved by the genetic processing of the entry d ata to the system. Fuzzy control systems work with a set of fuzzy vari ables. The fuzzy membership functions that define these-variables perf orm a kind of packing over the information coming from the sensors. Th ese fuzzy membership functions have an unfixed shape and a set of unfi xed anchor points that may be adjusted by several methods in order to obtain a good performance of the control system. Due to the trial and error method being time consuming, an alternative method, based on gen etic algorithms, for adjusting these parameters is proposed. Genetic a lgorithms are a search technique analogous to natural genetics. Geneti c information encoding and the implemented genetic algorithms are used to adjust the fuzzy membership functions associated with the linguist ic labels that define the fuzzy variables of a rule-based control syst em. The designed control system allows the local navigation of a mobil e autonomous robot avoiding unexpected obstacles in a partially unknow n environment.