Neural Net Control of High Nonlinear 2-DOF Nonlinear Robotic Arm

Abstract

A robust ANN system to enhance the control performance and robustness of a Two links robotic system (UoB-Arm) is presented. The UoB-Arm control architecture consists of a classical (PD) controller, independent of the high nonlinear dynamics of the two links robotic model, in parallel with a trainable ANN architecture. ANN architecture is used to identify the whole robotic arm inverse dynamics for arm control compensation. The (PD) controller is used to guarantee the stability and robustness of the whole control system, hence achieve a pré cised tracking accuracy.