What can I do here? A Theory of Affordances in Reinforcement Learning
This addresses a fundamental limitation in reinforcement learning for embodied agents, offering a novel framework to enhance efficiency and generalization, though it is incremental in applying psychological concepts to existing models.
The paper tackles the problem of reinforcement learning agents assuming all actions are always available, by developing a theory of affordances in Markov Decision Processes to reduce available actions per situation and improve model learning. It shows that affordances enable faster planning and more efficient, precise transition model learning, supported by theoretical results and examples.
Reinforcement learning algorithms usually assume that all actions are always available to an agent. However, both people and animals understand the general link between the features of their environment and the actions that are feasible. Gibson (1977) coined the term "affordances" to describe the fact that certain states enable an agent to do certain actions, in the context of embodied agents. In this paper, we develop a theory of affordances for agents who learn and plan in Markov Decision Processes. Affordances play a dual role in this case. On one hand, they allow faster planning, by reducing the number of actions available in any given situation. On the other hand, they facilitate more efficient and precise learning of transition models from data, especially when such models require function approximation. We establish these properties through theoretical results as well as illustrative examples. We also propose an approach to learn affordances and use it to estimate transition models that are simpler and generalize better.