NN Approaches on Fuzzy Sliding Mode Controller Design for Robot Trajectory Tracking

Abstract

The main problem of sliding mode controllers is that a whole knowledge system parameters is required to compute the equivalent control. Neural networks are used to compute the equivalent control. Standard two layer feed-forward neural network training with the backprobagation algorithm and Radial Basis Function Neural Networks (RBFNN) are the most popular methods that used on robot control. This paper applies these structures to Fuzzy Sliding Mode Control (FSMC). Methods are tested for robot trajectory tracking with computer simulations. Computer simulations of three link robot manipulator show that RBFNN is more efficient on FSMC for trajectory control applications.