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 2503.06960cs.CVcs.RO
收藏

面向机器人学习的预训练视觉模型的以数据为中心的重审

A Data-Centric Revisit of Pre-Trained Vision Models for Robot Learning

The University of Hong Kong(香港大学);University of Edinburgh(爱丁堡大学);Shanghai AI Laboratory(上海人工智能实验室)

Xin Wen, Bingchen Zhao, Yilun Chen, Jiangmiao Pang, Xiaojuan Qi

AI总结:

该研究针对预训练视觉模型在机器人学习中的最优配置问题,发现物体中心表征能力是关键,提出SlotMIM方法并通过多数据集实验验证其在多项任务上的显著提升与良好扩展性。

摘要:

预训练视觉模型(PVMs)是现代机器人技术的基础,但其最优配置仍不明确。通过系统评估,我们发现尽管DINO和iBOT在视觉运动控制与感知任务上的表现优于MAE,但它们在非(单)物体中心(NOC)数据上训练时表现不佳——这一局限性与其学习物体中心表征的能力下降密切相关。该研究表明,从非物体中心的机器人数据集中形成物体中心表征的能力,是PVMs取得成功的关键。受这一发现启发,我们设计了SlotMIM,该方法通过引入语义瓶颈减少原型数量以促进“物体性”的涌现,同时加入跨视图一致性正则化来鼓励多视图不变性,从而诱导生成物体中心表征。我们的实验涵盖在物体中心、场景中心、网络爬取和自我中心数据上的预训练。在所有设置下,我们的方法都能学习到可迁移的表征,并且在图像识别、场景理解和机器人学习评估中相比先前工作取得了显著提升。当扩展到百万级数据集时,我们的方法还展现出更优的数据效率和可扩展性。我们的代码和模型已在https://github.com/CVMI-Lab/SlotMIM 公开发布。

英文摘要:

Pre-trained vision models (PVMs) are fundamental to modern robotics, yet their optimal configuration remains unclear. Through systematic evaluation, we find that while DINO and iBOT outperform MAE across visuomotor control and perception tasks, they struggle when trained on non-(single-)object-centric (NOC) data--a limitation strongly correlated with their diminished ability to learn object-centric representations. This investigation indicates that the ability to form object-centric representations from the non-object-centric robotics dataset is the key to success for PVMs. Motivated by this discovery, we designed SlotMIM, a method that induces object-centric representations by introducing a semantic bottleneck to reduce the number of prototypes to encourage the emergence of objectness as well as cross-view consistency regularization for encouraging multiview invariance. Our experiments encompass pre-training on object-centric, scene-centric, web-crawled, and ego-centric data. Across all settings, our approach learns transferrable representations and achieves significant improvements over prior work in image recognition, scene understanding, and robot learning evaluations. When scaled up with million-scale datasets, our method also demonstrates superior data efficiency and scalability. Our code and models are publicly available at https://github.com/CVMI-Lab/SlotMIM.

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。

↑