# 当聆听变得更简单：为无捷径VLA剔除视觉线索

> 原文链接：https://robotics.hellobike.com/papers/2026-10-07-2610-10912
> 频道：papers | 更新日期：2026-10-07 | 标签：arxiv, cs.RO
> 核心摘要：针对VLA模型的视觉捷径学习问题，提出任务 scrubbing 域对抗训练方法，在仿真和真实环境中验证了方法的分布外鲁棒性提升效果。

# 当聆听变得更简单：为无捷径VLA剔除视觉线索

::: v-pre
> 原文链接：[When Listening Becomes Easier: Scrubbing Visual Cues for Shortcut-Free VLAs](https://arxiv.org/abs/2610.10912)（arXiv 2610.10912 · 发布日期：2026-10-07）

**When Listening Becomes Easier: Scrubbing Visual Cues for Shortcut-Free VLAs**

针对VLA模型的视觉捷径学习问题，提出任务 scrubbing 域对抗训练方法，在仿真和真实环境中验证了方法的分布外鲁棒性提升效果。

## 核心要点

- 机器人演示数据集多样性有限易导致策略学习到虚假相关性，收集高多样性数据成本高效率低。
- 提出表征层面的动作裕度指标，可在无需策略部署的情况下评估模型的捷径学习易感性。
- 提出任务scrubbing域对抗训练方法，降低模型使用视觉捷径的概率，提升VLA的泛化能力。
- 仿真与真实环境的多模型多视觉线索实验显示，该方法提升分布外鲁棒性，常可消除视觉捷径学习。

**验证范围与局限**：验证范围覆盖仿真与真实环境，涉及多个VLA模型及多类视觉线索。

## 英文摘要 (Abstract)

Shortcut learning is a prevalent issue in robot learning. The limited diversity of robot demonstration datasets can mislead policies into exploiting spurious correlations between tasks and irrelevant features, such as viewpoint or background. Collecting sufficiently diverse robot demonstrations is costly and inefficient, motivating algorithmic alternatives. We focus on vision-language-action (VLA) models and discover that different vision-language model backbones exhibit substantially different levels of susceptibility to visual shortcut learning. We find that model behavior correlates with our proposed representation-level metric, action margin, which requires no policy rollouts. Visual shortcuts consistently enter action representations in early layers, with models differing in the extent to which later layers correct them by incorporating language information. To boost models' attention to language, we introduce task scrubbing, a new domain-adversarial training method that decreases models' likelihood of using visual shortcuts and improves VLAs' generalization. Experiments in both simulation and the real world across multiple VLAs and visual cues show that task scrubbing improves out-of-distribution robustness and often eliminates visual shortcut learning.

作者：Jasper Gerigk, Kenzo Aspuru-Takata, Chin-Hsuan Wu, Mohammad Mohammadi, Shuhong Zheng, Igor Gilitschenski
:::

## 原始参考来源 (Sources)

- [arXiv 2610.10912：When Listening Becomes Easier: Scrubbing Visual Cues for Shortcut-Free VLAs](https://arxiv.org/abs/2610.10912)