A sliding mode based control of 2dof robot manipulator using neural network

Abstract

A sliding mode control has been a great interest in the control engineering community, with many applications particularly in the robot manipulators control. This paper presents investigations into the development of Sliding Mode control approach based neural network, where the model parameters are used in the equivalent control law. A neural model of robot parameters is calculated. The first one, to estimate the inertia matrix while the second, is dedicated to estimate the parameters of the matrix of the Coriolis/centripetal terms. The last one estimates the gravity vector. To demonstrate the applicability of the methods, a simulated two degrees of freedom robot manipulator is considered in order to evaluate the tracking properties and robustness capacities of neural sliding mode control technique.