Neuro-fuzzy-discrete sliding mode control of a tree-link SCARA robot

The objective of this study is to develop free model sliding mode controllers (SMC) in discrete-time domain for robot manipulators. First, an appropriate discrete sliding mode controller (DSMC) with stability proof is designed. A neuro-fuzzy discrete sliding mode (NFDSMC) is developed to remove the drawbacks of DSMC. In the NFDSMC, a two-layer neural network used to approximate the equivalent control law. In order to evaluate the switching law, a fuzzy system is used. Neural network weights are updated adaptively with back-propagation (BP) algorithm using the sliding manifold. Simulation results illustrate performances of these controllers in the control of the three link SCARA robot. Moreover, the sampling time effects on the sliding surface convergence are discussed.