MOTO: Offline Pre-training to Online Fine-tuning for Model-based Robot Learning

TL;DR

This paper introduces MOTO, a novel on-policy model-based reinforcement learning method that effectively leverages offline data for high-dimensional robot tasks, enabling successful pixel-based manipulation and outperforming existing approaches.

Contribution

MOTO is the first method to enable offline pre-training and online fine-tuning for pixel-based robot manipulation tasks using model-based RL.

Findings

Successfully solves MetaWorld benchmark tasks.

Achieves complete pixel-based manipulation in Franka Kitchen environment.

Outperforms existing offline and online RL methods.

Abstract

We study the problem of offline pre-training and online fine-tuning for reinforcement learning from high-dimensional observations in the context of realistic robot tasks. Recent offline model-free approaches successfully use online fine-tuning to either improve the performance of the agent over the data collection policy or adapt to novel tasks. At the same time, model-based RL algorithms have achieved significant progress in sample efficiency and the complexity of the tasks they can solve, yet remain under-utilized in the fine-tuning setting. In this work, we argue that existing model-based offline RL methods are not suitable for offline-to-online fine-tuning in high-dimensional domains due to issues with distribution shifts, off-dynamics data, and non-stationary rewards. We propose an on-policy model-based method that can efficiently reuse prior data through model-based value expansion and policy regularization, while preventing model exploitation by controlling epistemic uncertainty. We find that our approach successfully solves tasks from the MetaWorld benchmark, as well as the Franka Kitchen robot manipulation environment completely from images. To the best of our knowledge, MOTO is the first method to solve this environment from pixels.