无动作推理助力策略泛化
Action-Free Reasoning for Policy Generalization
AI总结：
针对端到端模仿学习在新场景泛化难的问题，提出RAD方法，结合机器人演示数据和无动作人类视频数据的推理标签训练策略，发布新数据集，有效弥合具身差距并提升泛化能力。
摘要：
端到端模仿学习为训练机器人策略提供了一种有前景的方法，但在新场景中的泛化仍然是一个重大挑战。尽管大规模机器人演示数据集已显示出诱导泛化的潜力，但扩展这些数据集需要大量资源。相比之下，人类视频数据丰富多样，是一种有吸引力的替代方案。然而，这些人类视频数据集缺乏动作标签，使其在模仿学习中的应用变得复杂。现有方法试图提取基础动作表示（例如手部姿势），但由此产生的策略难以弥合人类和机器人动作之间的具身差距。我们提出了一种替代方法：利用人类视频中基于语言的推理（这对指导机器人动作至关重要）来训练可泛化的机器人策略。基于最近在推理驱动策略架构方面的进展，我们引入了通过无动作数据推理（RAD）。RAD 从机器人演示数据（带有推理和动作标签）和无动作人类视频数据（仅带有推理标签）中学习。机器人数据教会模型将推理映射到低级动作，而无动作数据增强推理能力。此外，我们将发布一个包含 3,377 个人手演示的新数据集，该数据集带有与 Bridge V2 基准兼容的推理注释，旨在促进未来推理驱动机器人学习的研究。我们的实验表明，RAD 能够有效跨具身差距迁移，使机器人能够执行仅在无动作数据中见过的任务。此外，扩大无动作推理数据显著提高了策略性能和对新任务的泛化能力。这些结果凸显了从无动作数据集进行推理驱动学习在推进可泛化机器人控制方面的潜力。项目页面：https://rad-generalization.github.io
英文摘要：
End-to-end imitation learning offers a promising approach for training robot policies. However, generalizing to new settings remains a significant challenge. Although large-scale robot demonstration datasets have shown potential for inducing generalization, they are resource-intensive to scale. In contrast, human video data is abundant and diverse, presenting an attractive alternative. Yet, these human-video datasets lack action labels, complicating their use in imitation learning. Existing methods attempt to extract grounded action representations (e.g., hand poses), but resulting policies struggle to bridge the embodiment gap between human and robot actions. We propose an alternative approach: leveraging language-based reasoning from human videos-essential for guiding robot actions-to train generalizable robot policies. Building on recent advances in reasoning-based policy architectures, we introduce Reasoning through Action-free Data (RAD). RAD learns from both robot demonstration data (with reasoning and action labels) and action-free human video data (with only reasoning labels). The robot data teaches the model to map reasoning to low-level actions, while the action-free data enhances reasoning capabilities. Additionally, we will release a new dataset of 3,377 human-hand demonstrations with reasoning annotations compatible with the Bridge V2 benchmark and aimed at facilitating future research on reasoning-driven robot learning. Our experiments show that RAD enables effective transfer across the embodiment gap, allowing robots to perform tasks seen only in action-free data. Furthermore, scaling up action-free reasoning data significantly improves policy performance and generalization to novel tasks. These results highlight the promise of reasoning-driven learning from action-free datasets for advancing generalizable robot control. Project page: https://rad-generalization.github.io
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。
