Reinforcement Learning from Delayed Observations via World Models

TL;DR

This paper introduces a method using world models to effectively handle delayed observations in reinforcement learning, improving performance in partially observable environments with visual data.

Contribution

It proposes a novel approach to address observation delays by leveraging world models, extending reinforcement learning to delay-aware visual control tasks.

Findings

Outperforms naive model-based methods by up to 250%.

Successfully applies delay-aware RL to visual environments.

Enhances performance in partially observable, delayed settings.

Abstract

In standard reinforcement learning settings, agents typically assume immediate feedback about the effects of their actions after taking them. However, in practice, this assumption may not hold true due to physical constraints and can significantly impact the performance of learning algorithms. In this paper, we address observation delays in partially observable environments. We propose leveraging world models, which have shown success in integrating past observations and learning dynamics, to handle observation delays. By reducing delayed POMDPs to delayed MDPs with world models, our methods can effectively handle partial observability, where existing approaches achieve sub-optimal performance or degrade quickly as observability decreases. Experiments suggest that one of our methods can outperform a naive model-based approach by up to 250%. Moreover, we evaluate our methods on visual delayed environments, for the first time showcasing delay-aware reinforcement learning continuous control with visual observations.