Residual Learning and Context Encoding for Adaptive Offline-to-Online Reinforcement Learning

TL;DR

This paper introduces a residual learning and context encoding method to adapt offline reinforcement learning models to environments with changing dynamics during online fine-tuning, demonstrating improved adaptability and sample efficiency.

Contribution

The paper proposes a novel residual learning approach combined with a context encoder to handle changing dynamics in offline-to-online RL scenarios.

Findings

Effective adaptation to dynamic changes in MuJoCo environments

Outperforms comparison methods in unseen perturbations

Sample-efficient learning in variable environments

Abstract

Offline reinforcement learning (RL) allows learning sequential behavior from fixed datasets. Since offline datasets do not cover all possible situations, many methods collect additional data during online fine-tuning to improve performance. In general, these methods assume that the transition dynamics remain the same during both the offline and online phases of training. However, in many real-world applications, such as outdoor construction and navigation over rough terrain, it is common for the transition dynamics to vary between the offline and online phases. Moreover, the dynamics may vary during the online fine-tuning. To address this problem of changing dynamics from offline to online RL we propose a residual learning approach that infers dynamics changes to correct the outputs of the offline solution. At the online fine-tuning phase, we train a context encoder to learn a representation that is consistent inside the current online learning environment while being able to predict dynamic transitions. Experiments in D4RL MuJoCo environments, modified to support dynamics' changes upon environment resets, show that our approach can adapt to these dynamic changes and generalize to unseen perturbations in a sample-efficient way, whilst comparison methods cannot.