# VLA-$Δ$：为 Vision-Language-Action 模型学习物理状态转移

> 原文链接：https://robotics.hellobike.com/papers/2026-10-15-2610-21306
> 频道：papers | 更新日期：2026-10-15 | 标签：arxiv, cs.RO
> 核心摘要：VLA-$Δ$以当前与未来状态的物理变化为显式学习目标，并在LIBERO、RoboTwin和真机实验中验证。

# VLA-$Δ$：为 Vision-Language-Action 模型学习物理状态转移

::: v-pre
> 原文链接：[VLA-$Δ$: Learning Physical State Transition for Vision-Language-Action Models](https://arxiv.org/abs/2610.21306)（arXiv 2610.21306 · 发布日期：2026-10-15）

**VLA-$Δ$: Learning Physical State Transition for Vision-Language-Action Models**

VLA-$Δ$以当前与未来状态的物理变化为显式学习目标，并在LIBERO、RoboTwin和真机实验中验证。

## 核心要点

- 现有VLA多直接从观测映射到示教动作，缺少中间物理状态变化监督，易过拟合对应关系。
- 该框架将当前与未来状态之间的物理变化作为显式学习目标，监督连续动作生成。
- 作者构建大规模预训练数据集，并生成涵盖任务语义和视觉运动变化的状态转移标注。
- VLA-$Δ$在LIBERO和RoboTwin上达到与先进方法竞争的性能，真机实验优于$π_{0.5}$，代码已开源。

**验证范围与局限**：验证包括LIBERO、RoboTwin基准和真机实验，摘要未说明具体机器人数量或任务细节。

## 英文摘要 (Abstract)

Vision-language-action (VLA) models transfer semantic knowledge acquired through vision-language pretraining to robot control, achieving strong performance across various benchmarks. However, most VLAs directly map observations to demonstrated actions without explicit supervision of intermediate physical state changes, making them prone to overfitting observation-action correspondences. Some methods introduce intermediate reasoning or future-state prediction in language, visual, or latent spaces to guide action generation, but do not explicitly supervise the physical state transition itself. We therefore propose VLA-$Δ$, a general framework that uses physical changes between current and future states as explicit learning targets. To support this objective, we construct a large-scale pretraining dataset and derive physical state-transition annotations covering task-relevant semantics and visual motion changes, providing more direct supervision for continuous action generation. During inference, the model generates actions directly from current observations without additional inference overhead. VLA-$Δ$ achieves performance competitive with the state of the art on LIBERO and RoboTwin, and outperforms $π_{0.5}$ in real-robot experiments. Our code is available at https://github.com/ZJU-OmniAI/VLA-Delta

作者：Qi Lu, Jiarun Li, Ce Zhang, Miao Pan, Yi Pan, Xu Wang, Wenqi Zhang, Xuhong Zhang
:::

## 原始参考来源 (Sources)

- [arXiv 2610.21306：VLA-$Δ$: Learning Physical State Transition for Vision-Language-Action Models](https://arxiv.org/abs/2610.21306)