SCARA robot control using neural networks

Abstract

A SCARA industrial robot model is identified based on a 4-axis structure using Lagrangian mechanics, also the dynamic model for the electromechanical actuator and motion transmission systems is identified. A conventional PD controller is implemented and compared to neural networks control system to achieve precise position control of SCARA manipulator. The performance of the modeled system is simulated using several desired tracking motion for each joint. Neural networks control method has shown a remarkable improvement of tracking capabilities for the SCARA robot over conventional PD controller. The proposed neural network controller has the potential to accurately control real-time manipulator applications.