A design of CMAC-based fuzzy logic controller with fast learning and accurate approximation (Q1602864)

Fuzzy logic controllers (FLC) have been widely applied to both consumer products and industrial process controls. Besides, there have been attempts to combine the network structure and learning capability of neural networks (NN) with FLCs. For the local weight adjustment, which exhibits faster learning speed and easier hardware implementation, the so-called cerebellar model articulation controller (CMAC) was proposed and widely used. The paper presents a new design for a CMAC-based FLC with fast learning and good approximation possibilities. The proposed controller is then applied to the truck backer-upper control problem. The results obtained are compared to performances with other types of FLCs. But a simulation of the above mentioned problem is realized by means of approximate control dynamics for the truck backer-upper control problem in the form of an iteration scheme. This seems different from a solution to the real truck backer-upper control problem and so does not require the use of the very sophisticated CMAC-based FLC.