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 2602.00743cs.ROcs.AI
收藏

SA-VLA:面向视觉-语言-动作强化学习的空间感知流匹配

SA-VLA: Spatially-Aware Flow-Matching for Vision-Language-Action Reinforcement Learning

Wuhan University(武汉大学);Agency for Science, Technology and Research (A*STAR)(新加坡科技研究局);National University of Singapore(新加坡国立大学);Chinese Academy of Sciences(中国科学院);Nanyang Technological University(南洋理工大学)

Xu Pan, Zhenglin Wan, Xingrui Yu, Xianwei Zheng, Youkai Ke, Ming Sun, Rui Wang, Ziwei Wang, Ivor Tsang

AI总结:

SA-VLA通过空间感知的RL适应框架,在强化学习微调中保持空间定位,提升视觉-语言-动作任务的鲁棒性和泛化能力。

摘要:

视觉-语言-动作(VLA)模型在机器人操作中表现出强大的泛化能力,但强化学习(RL)微调常常在空间分布偏移下降低鲁棒性。对于流匹配VLA策略,这种退化与RL适应过程中空间归纳偏置的削弱密切相关,因为稀疏奖励和空间无关的探索越来越倾向于短期视觉线索。为了解决这一问题,我们提出了SA-VLA,一种空间感知的RL适应框架,通过将表示学习、奖励设计和探索与任务几何对齐,从而在策略优化过程中保持空间定位。SA-VLA融合了隐式空间表示与视觉标记,提供反映几何进步的密集奖励,并采用SCAN,一种针对流匹配动态定制的空间条件退火探索策略。在具有挑战性的多物体和杂乱操作基准上,SA-VLA实现了稳定的RL微调,并提高了零样本空间泛化能力,产生了更鲁棒和可转移的行为。代码和项目页面可在https://xupan.top/Projects/savla获取。

英文摘要:

Vision-Language-Action (VLA) models exhibit strong generalization in robotic manipulation, yet reinforcement learning (RL) fine-tuning often degrades robustness under spatial distribution shifts. For flow-matching VLA policies, this degradation is closely associated with the erosion of spatial inductive bias during RL adaptation, as sparse rewards and spatially agnostic exploration increasingly favor short-horizon visual cues. To address this issue, we propose \textbf{SA-VLA}, a spatially-aware RL adaptation framework that preserves spatial grounding during policy optimization by aligning representation learning, reward design, and exploration with task geometry. SA-VLA fuses implicit spatial representations with visual tokens, provides dense rewards that reflect geometric progress, and employs \textbf{SCAN}, a spatially-conditioned annealed exploration strategy tailored to flow-matching dynamics. Across challenging multi-object and cluttered manipulation benchmarks, SA-VLA enables stable RL fine-tuning and improves zero-shot spatial generalization, yielding more robust and transferable behaviors. Code and project page are available at https://xupan.top/Projects/savla.

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。

↑