RBFNN-HOMS Nonsingular Terminal Sliding Control of n-DOF Robotic Manipulator

Abstract

Hybridization of the Sliding Mode Control schemes and the Artificial Intelligence techniques is relatively a new way in the control domain. This paper proposes a robust control scheme by the adoption of Non-singular Terminal Sliding Mode Control (NTSMC), Higher Order Sliding Mode (HOSM) and Neural Network (NN) structure for n-DOF robotic manipulator. The NTSMC is used with time delay estimation method where the equivalent control term is synthesized without requirement of the robot model. In order to overcome the chattering drawback of the NTSMC, the discontinuous term is replaced by an adaptive HOSM controller. The adaptive HOSM controller consists of the Super Twisting algorithm which is estimated adaptively using Radial Based Function Neural Network (RBFNN) structure. The used RBFNN is learned online without requirement of a prior knowledge of training data. The stability is proved using a candidate Lyapunov function and the controller parameters are adjusted adaptively. The superiority and the effectiveness of the proposed approach are tested under 2-DOF robot manipulator in trajectory tracking mode and compared with NTSMC.