arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2603.01474cs.RO
收藏

ROSER：用于可扩展机器人学习的少样本机器人序列检索

ROSER: Few-Shot Robotic Sequence Retrieval for Scalable Robot Learning

University of Nevada, Las Vegas（内华达大学拉斯维加斯分校）；University of Toronto（多伦多大学）；LatentWorlds AI

Zillur Rahman, Eddison Pham, Alejandro Daniel Noel, Cristian Meo

备注：
2026 ICLR DATA-FM Workshop
阅读 PDF
arXiv 页面 ↗HTML 原文 ↗

AI总结：

ROSER通过少样本检索框架有效解决机器人序列数据稀缺问题，提升机器人学习的数据可用性。

摘要：

机器人学习中的关键瓶颈是任务标记、分段训练数据的稀缺性，尽管存在大量长连续交互日志记录的大规模机器人数据集。现有数据集包含大量多样化的行为，但结构上与需要干净分段、任务特定轨迹的现代学习框架不兼容。我们通过正式化机器人序列检索来解决这一数据利用危机：即使用仅几个参考示例从未标记的日志中提取可重用的、以任务为中心的片段。我们介绍了ROSER，一个轻量级的少样本检索框架，它在时间窗口上学习任务无关的度量空间，使得在仅需3-5个演示的情况下就能实现准确的检索，而无需任何任务特定的训练。为了验证我们的方法，我们建立了全面的评估协议，并在三个大规模数据集（如LIBERO、DROID和nuScenes）上将ROSER与经典对齐方法、学习嵌入和语言模型基线进行比较。我们的实验表明，ROSER在准确性和效率上均优于所有先前方法，实现每匹配亚毫秒的推理时间，同时保持优越的分布对齐。通过将数据整理重新框架化为少样本检索，ROSER为解锁未充分利用的机器人数据集提供了实用路径，从根本上提高了机器人学习的数据可用性。

英文摘要：

A critical bottleneck in robot learning is the scarcity of task-labeled, segmented training data, despite the abundance of large-scale robotic datasets recorded as long, continuous interaction logs. Existing datasets contain vast amounts of diverse behaviors, yet remain structurally incompatible with modern learning frameworks that require cleanly segmented, task-specific trajectories. We address this data utilization crisis by formalizing robotic sequence retrieval: the task of extracting reusable, task-centric segments from unlabeled logs using only a few reference examples. We introduce ROSER, a lightweight few-shot retrieval framework that learns task-agnostic metric spaces over temporal windows, enabling accurate retrieval with as few as 3-5 demonstrations, without any task-specific training required. To validate our approach, we establish comprehensive evaluation protocols and benchmark ROSER against classical alignment methods, learned embeddings, and language model baselines across three large-scale datasets (e.g., LIBERO, DROID, and nuScenes). Our experiments demonstrate that ROSER consistently outperforms all prior methods in both accuracy and efficiency, achieving sub-millisecond per-match inference while maintaining superior distributional alignment. By reframing data curation as few-shot retrieval, ROSER provides a practical pathway to unlock underutilized robotic datasets, fundamentally improving data availability for robot learning.

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。

↑