Reinforcement Learning for Extended Reality: Designing Self-Play Scenarios

Abstract

A common problem for deep reinforcement learning networks is a lack of training data to learn specific tasks through generalization. In this review, we look at extended reality, a promising but often overlooked field, for training agents using reinforcement learning. We review several techniques from the literature and then synthesize the information in order to propose a recommended design. Meta learning offers an important way forward, but the agents ability to perform self-play is considered crucial for achieving successful AI. Therefore, we focus on improving self-play scenarios for teaching self-learning agents, by providing a supportive environment for improved agent-environment interaction.