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 2611.05368cs.ROcs.LG
收藏

GUILD:用于模仿学习的引导式迭代数据收集

GUILD: Guided Iterative Data Collection for Imitation Learning

University of Texas at Austin(德克萨斯大学奥斯汀分校)

Shivin Dass, Kevin Nguyen, Rohan Yelandur, Junhong Xu, Roberto Martín-Martín

AI总结:

提出GUILD引导式迭代数据收集方法,通过迭代任务提议与数据评估闭环,在仿真和真实机器人中验证其相比被动策略等能提升样本效率与部署性能。

摘要:

机器人领域的模仿学习策略通常基于通过成本高昂的现实世界遥操作收集的数据集进行训练。数据收集过程往往缺乏引导,依赖遥操作者的直觉,导致数据对整体策略性能的贡献很小。近期研究表明,整理此类无引导数据集可显著提升策略性能,但这些方法均在收集后对现有数据集进行过滤,因此无法阻止无信息数据的获取。在本研究中,我们提出GUILD,一种新颖的引导式数据收集(GDC)方法,该方法主动推理应收集哪些数据以最大化下游策略性能。我们引入一种迭代框架,在以下两个步骤之间交替进行:1)任务提议与数据收集;2)数据评估,以形成闭环。对于任务提议,我们利用视觉语言模型的语义先验来识别用于数据收集的相关环境、对象配置和目标。然而,由于语义相关性并不意味着策略改进,我们进行数据评估。通过利用预测数据归因,我们近似收集的轨迹如何影响目标部署场景下的策略性能。该信号随后被用作反馈以指导后续数据收集,从而关闭数据收集循环。我们在仿真和真实机器人设置中评估了我们的方法,结果表明,与被动数据收集策略、多个基线方法以及经验丰富的机器人专家相比,引导式数据收集显著提高了样本效率和部署性能。更多信息可访问此https URL。

英文摘要:

Imitation learning policies in robotics are trained on datasets collected through costly real-world teleoperation. Often, the data-collection process is unguided, relying on the teleoperator's intuition, resulting in data that contributes little to overall policy performance. Recent work has shown that curating such unguided datasets can substantially improve policy performance, but these approaches operate by filtering existing datasets after collection and therefore do not prevent the acquisition of uninformative data. In this work, we propose GUILD, a novel Guided Data Collection (GDC) method that actively reasons about which data should be collected to maximize downstream policy performance. We introduce an iterative framework that alternates between 1) task proposal and data collection and 2) data evaluation to close the loop. For task proposal, we leverage semantic priors from vision-language models to identify relevant environments, object configurations, and goals for data collection. However, since semantic relevance does not imply policy improvement, we perform data evaluation. By leveraging predictive data attribution we approximate how the collected trajectories influence policy performance on a target deployment setting. This signal is then used as feedback to guide subsequent data collection, closing the data collection loop. We evaluate our approach in simulation and real robot settings, demonstrating that guided data collection substantially improves sample efficiency and deployment performance compared to passive data collection strategies, multiple baselines and experienced roboticists. More information at https://robin-lab.cs.utexas.edu/guild2026/

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。

↑