ReVLA: Reverting Visual Domain Limitation of Robotic Foundation Models

TL;DR

This paper investigates the limitations of current robotic foundation models in visual out-of-domain generalization, identifies catastrophic forgetting as a key issue, and proposes a model merging technique to enhance visual robustness, significantly improving task performance.

Contribution

It introduces ReVLA, a novel approach that mitigates visual catastrophic forgetting in robotic models through model merging, boosting out-of-domain task performance.

Findings

ReVLA improves grasping and lifting in visual OOD tasks by 77% and 66%.

Existing models lack robustness to visual out-of-domain scenarios.

Proposed model merging technique effectively restores visual generalization.

Abstract

Recent progress in large language models and access to large-scale robotic datasets has sparked a paradigm shift in robotics models transforming them into generalists able to adapt to various tasks, scenes, and robot modalities. A large step for the community are open Vision Language Action models which showcase strong performance in a wide variety of tasks. In this work, we study the visual generalization capabilities of three existing robotic foundation models, and propose a corresponding evaluation framework. Our study shows that the existing models do not exhibit robustness to visual out-of-domain scenarios. This is potentially caused by limited variations in the training data and/or catastrophic forgetting, leading to domain limitations in the vision foundation models. We further explore OpenVLA, which uses two pre-trained vision foundation models and is, therefore, expected to generalize to out-of-domain experiments. However, we showcase catastrophic forgetting by DINO-v2 in OpenVLA through its failure to fulfill the task of depth regression. To overcome the aforementioned issue of visual catastrophic forgetting, we propose a gradual backbone reversal approach founded on model merging. This enables OpenVLA -- which requires the adaptation of the visual backbones during initial training -- to regain its visual generalization ability. Regaining this capability enables our ReVLA model to improve over OpenVLA by a factor of 77\% and 66\% for grasping and lifting in visual OOD tasks. Comprehensive evaluations, episode rollouts and model weights are available on the ReVLA Page