# NarrativeFlow：基于流匹配的语言条件机器人速度场生成模型

> 原文链接：https://robotics.hellobike.com/papers/2026-10-01-2610-00981
> 频道：papers | 更新日期：2026-10-01 | 标签：auto, arxiv
> 核心摘要：针对跨具身数据复用难题，提出NarrativeFlow模型，将机器人轨迹建模为语言条件连续速度场。该方法在标准数据集及真实操作任务中成功率均优于基线，实现具身无关的运动表征学习。

# NarrativeFlow：基于流匹配的语言条件机器人速度场生成模型

::: v-pre
> 本文由 AI 根据公开内容整理，可能有误，请以原文为准：[NarrativeFlow: Flow-Based Vision-Language-Action Model Using Robot Velocity Fields](https://arxiv.org/abs/2610.00981)（arXiv 2610.00981）

## 核心内容

针对语言条件操作中具身特定数据采集成本高、跨平台复用难的问题，研究团队提出NarrativeFlow模型。该工作摒弃稀疏关键点近似方法，采用流匹配公式将机器人运动建模为语言条件下的连续速度场，旨在构建具身无关且物理一致的运动表征。实验表明，该模型在标准评测指标及真实世界多项操作任务中的成功率均显著优于现有代表性基线方法，项目代码与资源已开源。

技术上，NarrativeFlow通过流匹配机制直接生成连续机器人速度场，克服了传统方法对离散关键点的依赖，确保生成轨迹符合真实物理约束。作为具身无关的运动表示，该架构有效整合多平台异构数据，解决了基础模型训练中数据孤岛问题。相比仅能处理单一模态或粗粒度近似的旧方案，NarrativeFlow实现了语言指令到连续动作空间的精准映射，提升了泛化能力与部署鲁棒性。

## 关键看点

- **核心架构**：采用流匹配公式建模语言条件连续速度场，替代稀疏关键点位移近似，保证生成轨迹物理一致性
- **数据策略**：构建具身无关运动表征，支持跨机器人平台数据复用，降低基础模型对特定硬件数据的依赖
- **性能验证**：在标准数据集与真实操作任务中，成功率与评估指标均超越现有代表性基线方法
- **开源资产**：项目主页已发布，提供模型权重与复现代码，支持社区验证与二次开发

## 工程与产业观察

该工作为具身智能基础模型提供了关键的跨具身数据对齐范式。通过将轨迹抽象为连续速度场，有效解耦了语义理解与底层动力学差异，大幅降低多源异构数据清洗成本。对于算法工程师而言，这种流匹配驱动的VLA架构是解决Sim2Real鸿沟与提升零样本泛化能力的可行技术路径，值得在通用操作策略训练中优先验证。

## 英文摘要 (Abstract)

We focus on language-conditioned flow-based manipulation, where robot flows (robot velocity fields) serve as embodiment-agnostic, motion-centric representations for leveraging data collected from multiple robot platforms. This task is crucial because language-conditioned manipulation is essential for practical robotic systems, yet scaling robot foundation models remains limited by the labor-intensive collection of embodiment-specific data. Existing methods either coarsely approximate robot flows with sparse keypoint displacements, or cannot handle language-conditioned manipulation. To address this limitation, we propose NarrativeFlow, which models robot flows as continuous velocity fields using a flow-matching formulation conditioned on language. Accordingly, NarrativeFlow generates robot flows that are physically consistent with real-world manipulation. To validate NarrativeFlow, we have conducted experiments on standard datasets for language-conditioned manipulation. The experimental results show that NarrativeFlow outperforms representative baseline methods on standard evaluation metrics. Furthermore, through real-world experiments, we show that NarrativeFlow achieves higher success rates than baseline methods across multiple manipulation tasks. The project page is available at https://shota0520.github.io/NarrativeFlow-project-page/

作者：Shota Kobayashi, Koki Seno, Daichi Yashima, Komei Sugiura
:::

## 原始参考来源 (Sources)

- [arXiv 2610.00981：NarrativeFlow: Flow-Based Vision-Language-Action Model Using Robot Velocity Fields](https://arxiv.org/abs/2610.00981)