Look Before Acting: Enhancing Vision Foundation Representations for Vision-Language-Action Models
2026-03-16 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors studied how robots understand visual information to decide actions based on language instructions. They found that deeper model layers pay less attention to visual details during action prediction. To fix this, they created DeepVision-VLA, a system that shares vision data throughout the model layers, improving robot manipulation. They also introduced a way to remove irrelevant visual info early on to focus on important cues without slowing down the system. Their approach performed better than previous methods in both simulations and real-world tests.
Vision-Language-Action modelsrobotic manipulationlarge language models (LLM)visual tokenstransformersmulti-level visual featuresattention mechanismVision-Language Mixture-of-TransformersAction-Guided Visual Pruningvisual representation
Authors
Yulin Luo, Hao Chen, Zhuangzhe Wu, Bowen Sui, Jiaming Liu, Chenyang Gu, Zhuoyang Liu, Qiuxuan Feng, Jiale Yu, Shuo Gu, Peng Jia, Pheng-Ann Heng, Shanghang Zhang
Abstract
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.