# VLA-Fabric：面向可组合多臂 VLA 智能体的交互学习

> 原文链接：https://robotics.hellobike.com/papers/2026-10-16-2610-23018
> 频道：papers | 更新日期：2026-10-16 | 标签：arxiv, cs.RO
> 核心摘要：针对多臂 VLA 动作生成中的协同问题，VLA-Fabric 为每臂配置完整 VLA 并学习跨策略交互，在仿真和物理实验中验证。

# VLA-Fabric：面向可组合多臂 VLA 智能体的交互学习

::: v-pre
> 原文链接：[VLA-Fabric: Learning Interaction for Composable Multi-Arm VLA Agents](https://arxiv.org/abs/2610.23018)（arXiv 2610.23018 · 发布日期：2026-10-16）

**VLA-Fabric: Learning Interaction for Composable Multi-Arm VLA Agents**

针对多臂 VLA 动作生成中的协同问题，VLA-Fabric 为每臂配置完整 VLA 并学习跨策略交互，在仿真和物理实验中验证。

## 核心要点

- 多臂真实操作需要协同，完整动作生成 VLA 之间如何显式交互仍缺乏探索。
- 方法为每个机械臂组织一个完整 VLA，保留本地感知与动作所有权，并学习跨策略边界交互。
- 系统研究交互位置、同伴信息聚合与学习方式，并在结构不同的 VLA 架构中重建这些功能。
- 三臂和四臂团队在仿真与物理实验评估，四个物理任务中较独立本地策略成功率提升16至40个百分点。

**验证范围与局限**：验证范围包括仿真和物理实验中的三臂、四臂团队，以及四个物理任务。

## 英文摘要 (Abstract)

Vision-language-action (VLA) models are rapidly expanding what individual robots can accomplish, yet many real-world manipulation settings require multiple arms to operate as a coordinated system. Existing approaches organize multi-arm coordination in different ways, including centralized joint policies, high-level task or role assignment, and locally executed policies that react to teammates. However, how complete action-producing VLAs should explicitly interact with one another during action generation remains largely unexplored. We introduce VLA-Fabric, which organizes one complete VLA per arm while preserving local perception and action ownership, and enables coordination through learned interaction across policy boundaries. We first systematically study where such interaction should occur, how peer information should be aggregated, and how the interaction should be learned, revealing complementary coordination functions spanning shared context, peer-specific latent information, and action-stage interaction. We then re-instantiate these functions in a structurally different VLA architecture, showing that the coordination principle is not tied to a particular tensor layout or backbone implementation. Finally, we compose teams of three and four arms and evaluate them in simulation and physical experiments. On all four physical tasks, Full interaction improves success over independently trained local policies by 16--40 percentage points. Together, these results suggest that explicit interaction among complete VLA policies offers a scalable path toward networked embodied intelligence.

作者：Chenglong Zhang, Shuaijun Liu, Feiyang You, Shuyang Hao, Chengyu Wu, Ningxin Su
:::

## 原始参考来源 (Sources)

- [arXiv 2610.23018：VLA-Fabric: Learning Interaction for Composable Multi-Arm VLA Agents](https://arxiv.org/abs/2610.23018)