Agent-Controller Representations: Principled Offline RL with Rich Exogenous Information

TL;DR

This paper introduces a new offline RL benchmark for environments with rich, irrelevant exogenous visual information and proposes multi-step inverse models to learn effective agent-controller representations, outperforming existing methods.

Contribution

The paper presents a novel offline RL benchmark for exogenous visual information and demonstrates that multi-step inverse models improve representation learning in noisy, complex environments.

Findings

Multi-step inverse models outperform baselines in noisy environments.

Representations learned with ACRO are more robust to irrelevant visual noise.

The benchmark enables studying exogenous information in offline RL.

Abstract

Learning to control an agent from data collected offline in a rich pixel-based visual observation space is vital for real-world applications of reinforcement learning (RL). A major challenge in this setting is the presence of input information that is hard to model and irrelevant to controlling the agent. This problem has been approached by the theoretical RL community through the lens of exogenous information, i.e, any control-irrelevant information contained in observations. For example, a robot navigating in busy streets needs to ignore irrelevant information, such as other people walking in the background, textures of objects, or birds in the sky. In this paper, we focus on the setting with visually detailed exogenous information, and introduce new offline RL benchmarks offering the ability to study this problem. We find that contemporary representation learning techniques can fail on datasets where the noise is a complex and time dependent process, which is prevalent in practical applications. To address these, we propose to use multi-step inverse models, which have seen a great deal of interest in the RL theory community, to learn Agent-Controller Representations for Offline-RL (ACRO). Despite being simple and requiring no reward, we show theoretically and empirically that the representation created by this objective greatly outperforms baselines.