Policy Decorator: Model-Agnostic Online Refinement for Large Policy Model

TL;DR

This paper introduces Policy Decorator, a model-agnostic online refinement method that enhances large imitation learning policies during online interactions, improving stability and sample efficiency across multiple tasks.

Contribution

It presents a novel residual policy approach for online refinement of large imitation models, maintaining smooth behaviors and improving performance without destabilizing the learned policies.

Findings

Effective policy improvement across eight tasks

Preserves smooth motion of policies

Avoids erratic behaviors typical of pure RL methods

Abstract

Recent advancements in robot learning have used imitation learning with large models and extensive demonstrations to develop effective policies. However, these models are often limited by the quantity, quality, and diversity of demonstrations. This paper explores improving offline-trained imitation learning models through online interactions with the environment. We introduce Policy Decorator, which uses a model-agnostic residual policy to refine large imitation learning models during online interactions. By implementing controlled exploration strategies, Policy Decorator enables stable, sample-efficient online learning. Our evaluation spans eight tasks across two benchmarks-ManiSkill and Adroit-and involves two state-of-the-art imitation learning models (Behavior Transformer and Diffusion Policy). The results show Policy Decorator effectively improves the offline-trained policies and preserves the smooth motion of imitation learning models, avoiding the erratic behaviors of pure RL policies. See our project page (https://policydecorator.github.io) for videos.