MOTO: Offline Pre-training to Online Fine-tuning for Model-based Robot Learning
This addresses the challenge of adapting model-based RL to offline-to-online fine-tuning for realistic robot manipulation, though it is incremental in improving upon existing methods.
The authors tackled the problem of offline pre-training and online fine-tuning for model-based reinforcement learning in high-dimensional robot tasks, proposing MOTO, which successfully solved tasks from MetaWorld and the Franka Kitchen environment from images, achieving a first in solving that environment from pixels.
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.