StableVLA: Towards Robust Vision-Language-Action Models without Extra Data

Yiyang Fu, Chubin Zhang, Shukai Gong, Yufan Deng, Kaiwei Sun, Qiyang Min, Qibin Hou, Yansong Tang, Jianan Wang, Daquan Zhou
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:32264-32283, 2026.

Abstract

It is infeasible to encompass all possible disturbances within the training dataset. This raises a critical question regarding the robustness of Vision-Language-Action (VLA) models when encountering unseen real-world visual disturbances, particularly under imperfect visual conditions. In this work, we conduct a systematic study based on recent state-of-the-art VLA models and reveal a significant performance drop when visual disturbances absent from the training data are introduced. To mitigate this issue, we propose a lightweight adapter module grounded in information theory, termed the Information Bottleneck Adapter (IB-Adapter), which selectively filters potential noise from visual inputs. Without requiring any extra data or augmentation strategies, IB-Adapter consistently improves over the baseline by an average of 30%, while adding fewer than 10M parameters, demonstrating notable efficiency and effectiveness. Furthermore, even with a 14x smaller backbone (0.5B parameters) and no pre-training on the Open X-Embodiment dataset, our model StableVLA achieves robustness competitive with 7B-scale state-of-the-art VLAs. With negligible parameter overhead ($<$10M), our approach maintains accuracy on long-horizon tasks and surpasses OpenPi under real-world visual disturbances. The code will be made publicly available.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-fu26x, title = {{S}table{VLA}: Towards Robust Vision-Language-Action Models without Extra Data}, author = {Fu, Yiyang and Zhang, Chubin and Gong, Shukai and Deng, Yufan and Sun, Kaiwei and Min, Qiyang and Hou, Qibin and Tang, Yansong and Wang, Jianan and Zhou, Daquan}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {32264--32283}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/fu26x/fu26x.pdf}, url = {https://proceedings.mlr.press/v306/fu26x.html}, abstract = {It is infeasible to encompass all possible disturbances within the training dataset. This raises a critical question regarding the robustness of Vision-Language-Action (VLA) models when encountering unseen real-world visual disturbances, particularly under imperfect visual conditions. In this work, we conduct a systematic study based on recent state-of-the-art VLA models and reveal a significant performance drop when visual disturbances absent from the training data are introduced. To mitigate this issue, we propose a lightweight adapter module grounded in information theory, termed the Information Bottleneck Adapter (IB-Adapter), which selectively filters potential noise from visual inputs. Without requiring any extra data or augmentation strategies, IB-Adapter consistently improves over the baseline by an average of 30%, while adding fewer than 10M parameters, demonstrating notable efficiency and effectiveness. Furthermore, even with a 14x smaller backbone (0.5B parameters) and no pre-training on the Open X-Embodiment dataset, our model StableVLA achieves robustness competitive with 7B-scale state-of-the-art VLAs. With negligible parameter overhead ($<$10M), our approach maintains accuracy on long-horizon tasks and surpasses OpenPi under real-world visual disturbances. The code will be made publicly available.} }
Endnote
%0 Conference Paper %T StableVLA: Towards Robust Vision-Language-Action Models without Extra Data %A Yiyang Fu %A Chubin Zhang %A Shukai Gong %A Yufan Deng %A Kaiwei Sun %A Qiyang Min %A Qibin Hou %A Yansong Tang %A Jianan Wang %A Daquan Zhou %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-fu26x %I PMLR %P 32264--32283 %U https://proceedings.mlr.press/v306/fu26x.html %V 306 %X It is infeasible to encompass all possible disturbances within the training dataset. This raises a critical question regarding the robustness of Vision-Language-Action (VLA) models when encountering unseen real-world visual disturbances, particularly under imperfect visual conditions. In this work, we conduct a systematic study based on recent state-of-the-art VLA models and reveal a significant performance drop when visual disturbances absent from the training data are introduced. To mitigate this issue, we propose a lightweight adapter module grounded in information theory, termed the Information Bottleneck Adapter (IB-Adapter), which selectively filters potential noise from visual inputs. Without requiring any extra data or augmentation strategies, IB-Adapter consistently improves over the baseline by an average of 30%, while adding fewer than 10M parameters, demonstrating notable efficiency and effectiveness. Furthermore, even with a 14x smaller backbone (0.5B parameters) and no pre-training on the Open X-Embodiment dataset, our model StableVLA achieves robustness competitive with 7B-scale state-of-the-art VLAs. With negligible parameter overhead ($<$10M), our approach maintains accuracy on long-horizon tasks and surpasses OpenPi under real-world visual disturbances. The code will be made publicly available.
APA
Fu, Y., Zhang, C., Gong, S., Deng, Y., Sun, K., Min, Q., Hou, Q., Tang, Y., Wang, J. & Zhou, D.. (2026). StableVLA: Towards Robust Vision-Language-Action Models without Extra Data. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:32264-32283 Available from https://proceedings.mlr.press/v306/fu26x.html.

Related Material