arXiv cs.AI· Guangyu Zhao, Yongchang Zhang, Haoxuan Ru, Kaichen He, Haoran Wen, Kun Zhan, Zeyu Zhang, Yin Li, Yiwu Zhong·· 1 天前AI 评分31
STAR：在推理阶段引导视觉-语言-动作模型的动作表征
STAR: Steering Action Representations during Inference
AI 导读
针对视觉-语言-动作模型（VLA）动作输出附近表征缺乏任务相关视觉 grounding 的问题，研究者提出免训练推理方法 STAR，对后层残差更新做指数移动平均，让后续动作表征继承前层的物体感知更新方向，并用本体感知状态自适应调整混合强度。在机器人操作基准上，STAR 将平均任务成功率提升 4.0 个百分点，代码已开源。
正文 · 原文
Abstract:Vision-language-action models (VLAs) map language instructions and visual observations to robot actions, showing strong capabilities across robotic manipulation tasks. However, in our probing experiment, we find that action representations near the action output do not fully preserve task-relevant visual grounding. Motivated by this late-layer grounding gap, we introduce STAR, for STeering Action Representations, a training-free inference method for pre-trained VLAs. STAR applies an exponential moving average over late-layer residual updates, allowing later action representations to inherit object-aware update directions from earlier layers without model updates or architectural changes. It further uses recent proprioceptive states to adapt the mixing strength, using conservative mixing when execution dynamics change rapidly. Experiments on robot manipulation benchmarks show that STAR improves late-layer object awareness and increases average task success rates by 4.0 percentage points. Our code is available at this https URL.
| Subjects: | Robotics (cs.RO); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.41009 [cs.RO] |
| (or arXiv:2610.41009v1 [cs.RO] for this version) | |
| https://doi.org/10.48550/arXiv.2610.41009 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Guangyu Zhao [view email]
[v1]
Thu, 29 Oct 2026 21:42:17 UTC (1,866 KB)
来源：arXiv cs.AI · arxiv.org