# VeriVLA：面向基于流的 VLA 的带潜在世界模型的自适应测试时验证

> 原文链接：https://robotics.hellobike.com/papers/2026-10-15-2610-21985
> 频道：papers | 更新日期：2026-10-15 | 标签：arxiv, cs.RO
> 核心摘要：VeriVLA针对基于流的VLA测试时样本质量难估问题，用双轴触发与JEPA潜在世界模型排序，在真实任务和仿真中验证提升。

# VeriVLA：面向基于流的 VLA 的带潜在世界模型的自适应测试时验证

::: v-pre
> 原文链接：[VeriVLA: Adaptive Test-Time Verification for Flow-Based VLAs with Latent World Models](https://arxiv.org/abs/2610.21985)（arXiv 2610.21985 · 发布日期：2026-10-15）

**VeriVLA: Adaptive Test-Time Verification for Flow-Based VLAs with Latent World Models**

VeriVLA针对基于流的VLA测试时样本质量难估问题，用双轴触发与JEPA潜在世界模型排序，在真实任务和仿真中验证提升。

## 核心要点

- 基于流的VLA同一观测下会因噪声种子产生不同质量动作，细粒度操作中次优动作可能导致失败。
- 方法无需训练，用两个初始样本度量片段内时间变化和跨样本方差，以检测关键状态。
- 触发后，JEPA潜在世界模型预测动作条件未来潜在状态及任务进展变化，对候选动作片段排序。
- 三个真实任务平均完成分数从50.8%升至65.5%，在RoboTwin 2.0和LIBERO仿真中也观察到一致提升。

**验证范围与局限**：验证范围限于摘要所述三个真实世界任务以及RoboTwin 2.0、LIBERO仿真设置。

## 英文摘要 (Abstract)

Flow-based Vision-Language-Action (VLA) policies are inherently stochastic: different noise seeds can produce action chunks of varying quality from the same observation. At test time, however, the policy typically commits to a single sample without an explicit estimate of its quality. In fine-grained manipulation, executing a suboptimal action can irreversibly disrupt task progress and lead to failure. To exploit policy stochasticity, we introduce a training-free two-axis trigger that uses two initial samples to measure within-chunk temporal variation and cross-sample variance, detecting the critical states. Once triggered, a JEPA-based latent world model ranks candidate action chunks by predicting action-conditioned future latent states and the corresponding changes in task progress. By operating in latent space, the verifier avoids pixel reconstruction and enables efficient and more accurate ranking. Across three real-world tasks, $\textit{i.e.,}$ pen insertion, pipetting, and apple placement), our method raises the average completion score from 50.8% to 65.5%, while adding 326 ms for sampling and verification per triggered decision on a single RTX 4090. Consistent gains are also observed in simulation, where the average success rate rises from 66.4% to 73.6% in the clean setting of RoboTwin 2.0, from 62.8% to 69.6% in its randomized setting, and from 96.9% to 97.6% on LIBERO.

作者：Xueli Sun, Yuxiang Lu, Yantong Liu, Xianzhe Fan, Xi Chen, Zhe Liu, Hengshuang Zhao
:::

## 原始参考来源 (Sources)

- [arXiv 2610.21985：VeriVLA: Adaptive Test-Time Verification for Flow-Based VLAs with Latent World Models](https://arxiv.org/abs/2610.21985)