外观
论文arXiv 2610.0796122 天前
IronMan：用于机器人操作的信息约束视频-动作学习
原文链接：IronMan: Information-Constrained Video-Action Learning for Robot Manipulation（arXiv 2610.07961 · 发布日期：2026-10-06）
IronMan: Information-Constrained Video-Action Learning for Robot Manipulation
针对VAM中过多视觉细节削弱泛化的问题，IronMan用信息瓶颈约束视频特征，并在仿真和真实实验中验证。
核心要点
- VAM将视觉动态建模与动作生成耦合，但过量视觉细节会损害策略泛化。
- IronMan基于信息瓶颈，抑制无关视觉信息并保留动作相关动态线索。
- 动态感知瓶颈将噪声、纠缠的一步视频特征蒸馏为紧凑世界表征。
- 在LIBERO达99.0%、RoboTwin clean2clean达79.4%，OOD下LIBERO-Plus达79.1%。
验证范围与局限：验证范围包括仿真和真实世界实验，涉及LIBERO、RoboTwin clean2clean和LIBERO-Plus。
英文摘要 (Abstract)
Video Action Models (VAMs) couple visual dynamics modeling with action generation for robot manipulation. However, video representations are not naturally suited to action generation, as exposing the action policy to excessive visual detail can impair its generalization ability. Therefore, we introduce IronMan (Information-constRained videO-actioN learning for robot MANipulation), a robust video-action learning framework built on the information bottleneck principle. The core principle of this framework is to impose information constraints that suppress irrelevant visual information while preserving action-relevant dynamics cues. IronMan employs a dynamics-aware bottleneck that distills noisy, entangled one-step video features into compact world representations. Extensive simulation and real-world experiments demonstrate strong in-distribution (ID) performance and out-of-distribution (OOD) robustness while maintaining efficient inference. IronMan achieves success rates of 99.0% on LIBERO and 79.4% on RoboTwin clean2clean, outperforming all the evaluated baselines. Under OOD shifts, IronMan achieves a success rate of 79.1% on LIBERO-Plus, exceeding the strongest baseline by 10.4 percentage points. Project page: https://youngsoul0731.github.io/ironman-project-page/
作者：Yuanshuo Zhang, Wenzhe Zhao, Zixing Lei, Bin Chen, Siheng Chen
来源
本页事实对应的公开页面，以原文为准。