Decentralized neural-network sliding-mode robot controller

This paper develops a method for decentralized adaptive neural network control design with continuous sliding modes in which robustness is inherent. Neural network control is formulated to become a class of variable structure system control. Sliding modes are used to determine the best values for parameters in neural network learning rules; thereby, robustness in learning control can be improved. Derived equations of the decentralized neural network sliding-mode controller (DNNSMC) were verified on a real direct-drive 3-DOF PUMA mechanism. The new DNNSMC was successfully tested for adaptation capability of the algorithm for sudden changes in the manipulator dynamics (load).