arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2604.13645cs.ROcs.AIcs.LG
收藏

生成机器人策略中模拟与现实联合训练的机制分析

A Mechanistic Analysis of Sim-and-Real Co-Training in Generative Robot Policies

Department of Computer Science, The University of Texas at Austin(得克萨斯大学计算机科学系);Amazon FAR(亚马逊FAR);NVIDIA(英伟达)

Yu Lei, Minghuan Liu, Abhiram Maddukuri, Zhenyu Jiang, Yuke Zhu

AI总结:

本文通过理论分析和实验研究,揭示了联合训练中结构表示对齐和重要性再加权效应,为生成机器人策略的联合训练提供了统一解释和改进方法。

摘要:

联合训练通过结合有限的领域真实数据与丰富的替代数据,如模拟或跨身体机器人数据,被广泛用于训练生成机器人策略。尽管其经验成功,决定何时以及为何联合训练有效的机制仍不清晰。我们通过理论分析和实证研究探讨联合训练的机制,并识别出两个决定性能的内在效应。第一个是“结构表示对齐”,反映了跨领域表示对齐与领域辨别能力之间的平衡,并在下游性能中起主要作用。第二个是“重要性再加权效应”,源于领域依赖的行动加权调节,并在次级层面发挥作用。我们通过受控实验在玩具模型上以及广泛的模拟与模拟和模拟与现实机器人操作实验中验证了这些效应。我们的分析为最近的联合训练技术提供了统一的解释,并激励了一种简单的方法,该方法一致地改进了先前的方法。更广泛地说,我们的目标是研究联合训练的内部运作,并促进该领域的研究。

英文摘要:

Co-training, which combines limited in-domain real-world data with abundant surrogate data such as simulation or cross-embodiment robot data, is widely used for training generative robot policies. Despite its empirical success, the mechanisms that determine when and why co-training is effective remain poorly understood. We investigate the mechanism of sim-and-real co-training through theoretical analysis and empirical study, and identify two intrinsic effects governing performance. The first, \textbf{``structured representation alignment"}, reflects a balance between cross-domain representation alignment and domain discernibility, and plays a primary role in downstream performance. The second, the \textbf{``importance reweighting effect"}, arises from domain-dependent modulation of action weighting and operates at a secondary level. We validate these effects with controlled experiments on a toy model and extensive sim-and-sim and sim-and-real robot manipulation experiments. Our analysis offers a unified interpretation of recent co-training techniques and motivates a simple method that consistently improves upon prior approaches. More broadly, our aim is to examine the inner workings of co-training and to facilitate research in this direction.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑