Look Before Acting: Enhancing Vision Foundation Representations for Vision-Language-Action Models
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866911521656799232 |
|---|---|
| author | Luo, Yulin Chen, Hao Wu, Zhuangzhe Sui, Bowen Liu, Jiaming Gu, Chenyang Liu, Zhuoyang Feng, Qiuxuan Yu, Jiale Gu, Shuo Jia, Peng Heng, Pheng-Ann Zhang, Shanghang |
| author_facet | Luo, Yulin Chen, Hao Wu, Zhuangzhe Sui, Bowen Liu, Jiaming Gu, Chenyang Liu, Zhuoyang Feng, Qiuxuan Yu, Jiale Gu, Shuo Jia, Peng Heng, Pheng-Ann Zhang, Shanghang |
| contents | Vision-Language-Action (VLA) models have recently emerged as a promising paradigm for robotic manipulation, in which reliable action prediction critically depends on accurately interpreting and integrating visual observations conditioned on language instructions. Although recent works have sought to enhance the visual capabilities of VLA models, most approaches treat the LLM backbone as a black box, providing limited insight into how visual information is grounded into action generation. Therefore, we perform a systematic analysis of multiple VLA models across different action-generation paradigms and observe that sensitivity to visual tokens progressively decreases in deeper layers during action generation. Motivated by this observation, we propose \textbf{DeepVision-VLA}, built on a \textbf{Vision-Language Mixture-of-Transformers (VL-MoT)} framework. This framework enables shared attention between the vision foundation model and the VLA backbone, injecting multi-level visual features from the vision expert into deeper layers of the VLA backbone to enhance visual representations for precise and complex manipulation. In addition, we introduce \textbf{Action-Guided Visual Pruning (AGVP)}, which leverages shallow-layer attention to prune irrelevant visual tokens while preserving task-relevant ones, reinforcing critical visual cues for manipulation with minimal computational overhead. DeepVision-VLA outperforms prior state-of-the-art methods by 9.0\% and 7.5\% on simulated and real-world tasks, respectively, providing new insights for the design of visually enhanced VLA models. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_15618 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Look Before Acting: Enhancing Vision Foundation Representations for Vision-Language-Action Models Luo, Yulin Chen, Hao Wu, Zhuangzhe Sui, Bowen Liu, Jiaming Gu, Chenyang Liu, Zhuoyang Feng, Qiuxuan Yu, Jiale Gu, Shuo Jia, Peng Heng, Pheng-Ann Zhang, Shanghang Computer Vision and Pattern Recognition Vision-Language-Action (VLA) models have recently emerged as a promising paradigm for robotic manipulation, in which reliable action prediction critically depends on accurately interpreting and integrating visual observations conditioned on language instructions. Although recent works have sought to enhance the visual capabilities of VLA models, most approaches treat the LLM backbone as a black box, providing limited insight into how visual information is grounded into action generation. Therefore, we perform a systematic analysis of multiple VLA models across different action-generation paradigms and observe that sensitivity to visual tokens progressively decreases in deeper layers during action generation. Motivated by this observation, we propose \textbf{DeepVision-VLA}, built on a \textbf{Vision-Language Mixture-of-Transformers (VL-MoT)} framework. This framework enables shared attention between the vision foundation model and the VLA backbone, injecting multi-level visual features from the vision expert into deeper layers of the VLA backbone to enhance visual representations for precise and complex manipulation. In addition, we introduce \textbf{Action-Guided Visual Pruning (AGVP)}, which leverages shallow-layer attention to prune irrelevant visual tokens while preserving task-relevant ones, reinforcing critical visual cues for manipulation with minimal computational overhead. DeepVision-VLA outperforms prior state-of-the-art methods by 9.0\% and 7.5\% on simulated and real-world tasks, respectively, providing new insights for the design of visually enhanced VLA models. |
| title | Look Before Acting: Enhancing Vision Foundation Representations for Vision-Language-Action Models |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2603.15618 |