Design and Implementation of a Real Time Control System for a 2DOF Robot Based on Recurrent High Order Neural Network Using a Hardware in the Loop Architecture

In this paper, a real-time implementation of sliding-mode control (SMC) in hardware-in-loop architecture is presented for robot with two degrees freedom (2DOF). It based on discrete-time recurrent neural identification method, as well the high performance obtained from advantages architecture. The process uses high-order network (RHONN) trained modified extended Kalman filter (EKF) algorithm. This method calculating covariance matrices EKF algorithm, using dynamic model associated and measurement noises, it increases proposed methodology. On other hand, decentralized SMC technique used to minimize motion error. A Virtex 7 field programmable gate array (FPGA) configured validate controller. series several experiments demonstrates robustness its immunity noise inherent external perturbation, are typically found input reference signals 2DOF manipulator robot.