arXiv cs.AI· Bingao Chen, Lihan Zha, C. Karen Liu, Guy Tevet·· 13 小时前AI 评分44
FoveaVLA：用注视动作替代腕部相机的 VLA 模型
FoveaVLA: Foveation Actions Replace the Wrist Camera in VLAs
AI 导读
FoveaVLA 提出用学习到的注视动作替代腕部相机，在每次策略查询时预测下一动作块需观察的交互区域，并将原始分辨率裁剪图与全景视图一并输入策略。基于 π₀.₅ 构建 Single 和 Dual 两种配置，在 DROID 上用 RoboInter 交互标注后训练。
正文 · 原文
Abstract:We present FoveaVLA, a Vision-Language-Action (VLA) model that replaces the wrist camera with learned foveation actions. Wrist cameras provide close-up detail but restrict the view to the end-effector's surroundings. External cameras see the whole scene, but the policy receives them at low resolution, so the fine detail needed for manipulation is lost. FoveaVLA recovers this detail by predicting, alongside each action chunk, the interaction region to observe at the next policy query. The policy then receives crops of these regions, taken from the original-resolution images, together with the full views. We build FoveaVLA on $\pi_{0.5}$ in two configurations, Single and Dual, using one or two external stereo cameras with full and foveal RGB and depth inputs, and post-train both on DROID with RoboInter interaction annotations. Across four real-world task suites, FoveaVLA-Single achieves 70.8% success with one external RGB-D view, and FoveaVLA-Dual achieves 77.1%, versus 52.1% for the wrist-equipped $\pi_{0.5}$. Both match the baseline on plain pick-and-place and improve success with distractors, cloth manipulation, and far-away objects.
| Comments: | Project website: this https URL |
| Subjects: | Robotics (cs.RO); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.20064 [cs.RO] |
| (or arXiv:2610.20064v2 [cs.RO] for this version) | |
| https://doi.org/10.48550/arXiv.2610.20064 arXiv-issued DOI via DataCite |
Submission history
From: Bingao Chen [view email]
[v1]
Wed, 14 Oct 2026 12:04:21 UTC (4,110 KB)
[v2]
Wed, 21 Oct 2026 14:34:52 UTC (4,122 KB)
来源：arXiv cs.AI · arxiv.org