Modular Reinforcement Learning Framework for Learners and Educators

Modular Reinforcement Learning Framework for Learners and Educators

연구 분야: Artificial Intelligence

학회: FDG '21: Proceedings of the 16th International Conference on the Foundations of Digital Games

Reinforcement learning algorithms have been applied to many research areas in gaming and non-gaming applications. However, in gaming artificial intelligence (AI), existing reinforcement learning tools are aimed at experienced developers and not readily accessible to learners. The unpredictable nature of online learning is also a barrier for casual users. This paper proposes the EasyGameRL framework, a novel approach to the education of reinforcement learning in games using modular visual design patterns. The EasyGameRL framework and its software implementation in Unreal Engine are modular, reusable, and applicable to multiple game scenarios. The pattern-based approach allows users to effectively utilize reinforcement learning in their games and visualize the components of the process. This would be helpful to AI learners, educators, designers and casual users alike.