Applying a Generative Adversarial Approach to Build an Intelligent Control System for Robotic Systems

Abstract

In today's society, robotics systems are being widely adopted in various fields of human activity. Robots not only play a decisive role in modern intelligent industries, but are also crucial to industrial production as a whole. The use of robotic devices is increasing in the field of unmanned transport, the gaming industry and ubiquitous systems. Parallel to the processes of robotisation, there is a lag in robot behaviour relative to human intelligence. Most robotics (numerically controlled machine tool based manufacturing) operates based on rigidly-defined algorithms and is not capable of generalising experience and making decisions in a human-like manner. This paper proposes a control system architecture for a robotic device based on an generative approach. This paper deals with the problem of controlling the motion of a robot. As a solution to the problem, a deep model like generative adversarial approach is proposed for generating motion control commands. The proposed architecture of intelligent control system and methods of generating control commands can be generalized to control systems of more complex robotic complexes and technological processes.