arXiv cs.AI· Wen-Han Hsieh, Ken Jiankun Zheng, Chia-Chien Chen, David M. Chan, Trevor Darrell·· 1 小时前AI 评分25
HDH 框架为冻结 VLA 模型学习低延迟动作验证
Learning Low-Latency Action Verification for Frozen Vision-Language-Action Models
AI 导读
研究提出 HDH 框架，利用冻结 VLA 模型自身部署的后见数据训练轻量动作验证器，为每个动作块预测任务成功分数并选优执行，避免用高成本验证器带来的延迟。在 LIBERO-10 四个任务上平均成功率达 82.3%，三个真实操作任务较基线策略平均提升 22.6%。
正文 · 原文
Abstract:Vision-language-action (VLA) models have made significant progress in general-purpose robotic manipulation. However, sampled action sequences can be ineffective, while evaluating alternatives with computationally expensive verifiers adds latency at each policy query. In this work, we propose HDH, a lightweight framework that uses hindsight from a frozen VLA's own deployments to improve action verification. The framework has three stages. First, we deploy the policy and collect its autonomous rollouts. We then train a compact action verifier to assign each action chunk a predicted task-success score. Finally, the frozen VLA generates action candidates, and the verifier selects the candidate with the highest score for execution. Across four LIBERO-10 tasks, HDH achieves an average success rate of 82.3\% and on three real-world manipulation tasks, HDH improves average success over the baseline policy by 22.6\%.
| Subjects: | Robotics (cs.RO); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV) |
| Cite as: | arXiv:2611.14834 [cs.RO] |
| (or arXiv:2611.14834v1 [cs.RO] for this version) | |
| https://doi.org/10.48550/arXiv.2611.14834 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Ken Zheng [view email]
[v1]
Tue, 10 Nov 2026 23:32:12 UTC (3,185 KB)
来源：arXiv cs.AI · arxiv.org