English

Run-time Observation Interventions Make Vision-Language-Action Models More Visually Robust

Robotics 2024-10-04 v1 Machine Learning

Abstract

Vision-language-action (VLA) models trained on large-scale internet data and robot demonstrations have the potential to serve as generalist robot policies. However, despite their large-scale training, VLAs are often brittle to task-irrelevant visual details such as distractor objects or background colors. We introduce Bring Your Own VLA (BYOVLA): a run-time intervention scheme that (1) dynamically identifies regions of the input image that the model is sensitive to, and (2) minimally alters task-irrelevant regions to reduce the model's sensitivity using automated image editing tools. Our approach is compatible with any off the shelf VLA without model fine-tuning or access to the model's weights. Hardware experiments on language-instructed manipulation tasks demonstrate that BYOVLA enables state-of-the-art VLA models to nearly retain their nominal performance in the presence of distractor objects and backgrounds, which otherwise degrade task success rates by up to 40%. Website with additional information, videos, and code: https://aasherh.github.io/byovla/ .

Keywords

Cite

@article{arxiv.2410.01971,
  title  = {Run-time Observation Interventions Make Vision-Language-Action Models More Visually Robust},
  author = {Asher J. Hancock and Allen Z. Ren and Anirudha Majumdar},
  journal= {arXiv preprint arXiv:2410.01971},
  year   = {2024}
}

Comments

Website: https://aasherh.github.io/byovla/

R2 v1 2026-06-28T19:05:58.376Z