AN IN-DEPTH REVIEW OF THE SCIENTIFIC RESEARCH ON DEEP REINFORCEMENT LEARNING IN PRODUCTION SYSTEMS

Abstract

Production systems face significant problems as a result of shorter product development cycles and fully customizable 
goods. These must not only handle a greater variety of products, but also facilitate high throughputs and offer a high 
degree of flexibility and resilience to process changes and unanticipated events. Deep Reinforcement Learning (RL) has 
being used more and more for production system optimization in order to overcome these obstacles. Deep RL, in contrast 
to other machine learning techniques, uses freshly gathered sensor data in close contact with its surroundings to allow for 
real-time reactions to system modifications. A thorough review of the outcomes has not yet been established, despite the 
fact that deep RL is now being implemented in production systems. This paper's primary contribution is to give 
practitioners and researchers a summary of applications.