Industrial General Reinforcement Learning Control Framework System based on Intelligent Edge

Abstract

This paper is about the intelligent edge-based reinforcement learning control framework technology for manufacturing field solution and features large-scale learning, scalable edge distribution technology that can be applied to various task. In this paper, two items are proposed as features. The first proposes the General Reinforcement Learning Framework(GRLF) in manufacturing, and the second proposes the edge solution technology based on the GRLF in manufacturing. We apply the industrial solution such as grid sorter system for example. As a result of the industrial GRLF based grid sorter system, it was confirmed that when a total of 100 deliveries are randomly received into the grid sorter system by any emitter, all shipments are 100% accurate. It also classifies approximately 0.5 deliveries per step. This shows the efficiency of classifying around 30 deliveries per minute, assuming a step is performed per second.