VLA 模型知道自己会失败吗？解读物理 AI 系统中的成功与失败模式
Do VLAs Know They Will Fail? Interpretable Success and Failure Patterns in Physical AI Systems
AI 导读
一项发表于 CoRL 2026 物理AI安全研讨会的研究显示，OpenVLA 激活的稀疏自编码器（SAE）表示包含结果预测信息，可在机器人进入不可恢复状态前预测任务失败。研究者识别出与成败阶段相关的单个 SAE 特征，并通过 VLM 解释流程生成行为描述加以验证。
正文 · 原文
Abstract:Vision-Language-Action (VLA) models have emerged as powerful general-purpose robot policies, but their successes and failures remain difficult to diagnose and understand. In this work, we show that sparse autoencoder (SAE) representations of OpenVLA activations contain outcome-predictive information and can be used to predict task failure before the robot reaches an unrecoverable state. Moreover, to provide insights into what common behaviors may contribute to these outcomes, we identify individual SAE features whose strong activations are associated with specific stages of success or failure, and uncover recurring behavioral patterns by localizing temporal intervals where these features exhibit significant activation changes. We further develop a VLM-based interpretation pipeline to generate behavioral descriptions of these patterns and validate their correspondence to feature activations on held-out intervals.
| Comments: | 12 pages (4 main pages + 8 pages appendix), 4 figures. Accepted at the Science of Physical AI Safety Workshop at the 2026 Conference on Robotic Learning (CoRL 2026, non-archival) |
| Subjects: | Robotics (cs.RO); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) |
| Cite as: | arXiv:2610.32208 [cs.RO] |
| (or arXiv:2610.32208v1 [cs.RO] for this version) | |
| https://doi.org/10.48550/arXiv.2610.32208 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Christian Rogers [view email]
[v1]
Fri, 23 Oct 2026 06:02:12 UTC (901 KB)
来源：arXiv cs.AI · arxiv.org