Increasing the accuracy of the robot by using neural networks (neural compensators and nonlinear dynamics)

The subject of this article is a programmable control system for a robotic arm. The complex nonlinear dynamics associated with the practical application of systems and manipulators is considered. The traditional control method is replaced by the developed Elma system and the adaptive radial neural network core function, which improves system stability and response speed. With the help of software associated with MATLAB, the corresponding controllers and compensators are developed. The results of training a neural network controller for programming robot trajectories are presented. Dynamic errors of various types of neural network controllers and two control methods are analyzed.