A Mixed Gaussian Membership Function Fuzzy CMAC for a Three-Link Robot

Abstract

This research produces a mixed Gaussian membership function (GMF) fuzzy cerebellar model articulation controller (CMAC) for a three-link robot. A mixed GMF is created using the current and the previous GMFs on each layer of CMAC to detect errors efficiently, so a mixed GMF fuzzy CMAC (MGMFFC) is able to train parameters efficiently and constructs the MGMFFC structure automatically. A Lyapunov cost function and the gradient descent techniques are utilized to get the adaptive with guaranteed system's stable. Simulation studies for a three-link robot show that the MGMFFC attains favorable tracking performance.