Neuro-sliding Mode Control for Robot Manipulators

This paper presents a neuro-sliding mode control method for a robot manipulator. The sliding mode controller requires to identify robot models to lessen the work of the controller although it can be used as a non model-based nonlinear controller. Selection of the gain for the nonlinear function becomes important and affects the performance. To make it easy of selecting the gain, a neural network is used to compensate for the modeling error to form a neuro-sliding control method. Stability analysis for the neuro-sliding control method is provided. Simulation studies of controlling joints of the three-link robot manipulator are presented.