Industrial insights of practical deep reinforcement learning projects

Antoine Plissonneau-Duquène, Hugo Massonnat, Elie KADOCHE, Louis Verny, Florence Carton

Keywords: Reinforcement Learning, Industrialization, Energy

Abstract:

Reinforcement learning is a category of artificial intelligence algorithms that has seen significant growth in recent years. Despite impressive results in a wide range of domains such as robotics, game playing, and autonomous systems, its deployment in real-world systems remains very limited.
In this paper, we present two case studies on the deployment of RL agents in real-world systems within an industrial context. The paper is structured chronologically around the life cycle of a project, in order to highlight the different stages, key considerations, and challenges involved in deploying an RL agent in a real-world system. This paper aims at providing insights into the following questions: What should be considered when selecting a reinforcement learning project? How can an RL project be successfully implemented in the industry? What challenges arise when deploying RL in production?

Submission Number: 5