arXiv 2607.064422026-07-08cs.RO

SIEVE：用于基于VLA模型的模仿学习的结构感知数据选择

SIEVE: Structure-Aware Data Selection for Imitation Learning with VLA Models

East China Normal University（东华师范大学）；Zhongguancun Academy（中关村学院）；Harbin Institute of Technology（哈尔滨工业大学）；Beihang University（北航大学）；Huazhong University of Science and Technology（华中科技大学）；The Hong Kong University of Science and Technology (Guangzhou)（香港科技大学（广州））；Zhongguancun Institute of Artificial Intelligence（中关村人工智能研究院）；DeepCybo Training Demonstrations Policy Execution（深科博培训演示政策执行）

Changti Wu, Bin Yu, Zhaolong Shen, Shijie Lian, Xiaopeng Lin, Cong Huang, Zhirui Zhang, Lei Zhang, Kai Chen

备注：: The code is available at \href{https://github.com/ChangtiWu/SIEVE}{SIEVE}

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

研究针对VLA模型模仿学习中数据冗余等问题，提出结构感知数据选择方法SIEVE。该方法将演示视为原语和接口组合，通过发现原语、分配预算和选择轨迹来选数据，实验证明其优于基线，能在少数据少步骤下超越全数据训练。

摘要：

视觉语言动作（VLA）模型通常通过对大规模机器人演示数据集进行模仿学习来训练，但由于冗余、噪声和覆盖不均，更多数据不一定能产生更好的策略。现有数据选择方法常从轨迹或状态动作层面评估演示，忽略了构成长期行为的可复用结构。本文提出SIEVE，一种用于VLA模仿学习的结构感知数据选择方法。SIEVE将演示视为可复用原语和过渡接口的组合，先从分段轨迹中发现视觉运动原语，再通过在收益递减下最大化复用感知结构曝光为组合模式分配选择预算，最后在每个组合模式桶内选择类中值轨迹以保留核心、稳定且利于模仿的演示。跨多个数据集、基准和VLA模型的实验表明，SIEVE始终优于有竞争力的数据选择基线。值得注意的是，SIEVE仅使用50%的演示和50%的训练步骤就能超越全数据训练，这表明通过原语和过渡捕获的可复用结构是高效VLA模仿学习的重要信号。

英文摘要：

Vision-Language-Action (VLA) models are typically trained by imitation learning on large-scale robot demonstration datasets, but more data does not necessarily yield better policies due to redundancy, noise, and uneven coverage. Existing data selection methods often assess demonstrations at either the trajectory or state-action level, missing the reusable structures that compose long-horizon behaviors. In this paper, we propose SIEVE, a structure-aware data selection method for VLA imitation learning. SIEVE views demonstrations as compositions of reusable primitives and transition interfaces. It first discovers visuo-motor primitives from segmented trajectories, then allocates selection budgets to composition patterns by maximizing reuse-aware structural exposure under diminishing returns. Finally, it selects medoid trajectories within each composition-pattern bucket to retain central, stable, and imitation-friendly demonstrations. Experiments across multiple datasets, benchmarks, and VLA models show that SIEVE consistently outperforms competitive data selection baselines. Notably, SIEVE can surpass full-data training while using only 50% of demonstrations and 50% of training steps, suggesting that reusable structure, captured through primitives and transitions, is an important signal for efficient VLA imitation 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。