A case study of knowledge acquisition: From connectionist learning to an optimized fuzzy knowledge base

An automated knowledge acquisition architecture for docking a truck problem is presented. The architecture consists of a neural network controller a fuzzy rule maker, and a fuzzy controller. The neural network controller is used to learn the driving knowledge from trials. The driving knowledge is then extracted by the fuzzy rule maker to form a driving knowledge rule base. The driving knowledge rule base is further optimized using a genetic algorithm. Computer simulations are presented to show the effectiveness of the architecture.<
>