Adaptive robot control based on multiple incremental fuzzy neural networks

Abstract

An adaptive control for robot manipulators based on multiple incremental fuzzy neural networks (FNNs) is proposed in this paper. The overall controller is comprised of a feedback controller and multiple FNNs which learn inverse dynamics of the robot manipulator for different tasks. The multiple FNNs are switched or blended to improve the transient response when manipulating objects are changed. The structure and parameters of the FNNs are determined dynamically using an incremental learning algorithm which reduces complexity and computation induced by the use of multiple models considerably. The parameters are refined online to compensate for uncertainties. The closed-loop system with a switching or blending law is proven to be stable in Lyapunov sense. The proposed scheme is applied to control a two-link robot manipulator with varying payloads.