域算术：环境偏移下的单次VLA自适应

摘要

视觉-语言-行动（VLA）模型在环境变化（如相机视角改变、迁移至不同但相似的机器人平台（例如从Panda机器人切换至UR5e机器人））时，常无法正确执行已习得的任务。为使这些模型适应变化后的环境（即目标域），通常需要针对每个任务收集多组演示数据进行训练，这导致采集成本高昂。为减轻数据整理与训练负担，我们提出一种基于类比的方法，通过权重向量算术运算与领域特定信息加法实现VLA模型的环境适应，称为域算术（Domain ARiThmetic, DART）。与现有方法不同，DART仅需收集单一演示数据即可实现高效适应。为准确提取用于加法的领域特定信息，DART对权重向量中的奇异分量进行子空间对齐，以滤除噪声分量。在仿真与真实世界实验中，DART在跨多种视觉与本体迁移的单次学习场景下，均优于现有VLA适应方法。代码已开源至 https://github.com/snumprlab/dart。

English

Vision-Language-Action (VLA) models often fail to perform the same learned tasks under environmental shifts, such as changes in camera pose and shifts to a different but similar robot (e.g., from Panda to UR5e). Adapting these models to the shifted environment (i.e., target domain) often requires training on multiple demonstrations for each task, which are costly to collect. To reduce the burden of data curation and training, we propose an analogy-based method that adapts VLA models under environmental shifts through weight vector arithmetic with domain-specific information addition, named Domain ARiThmetic (DART). Unlike prior approaches, DART requires collecting only a single demonstration, enabling efficient adaptation. To accurately isolate domain-specific information for addition, DART performs subspace alignment between singular components in weight vectors to filter out noisy components. In both simulated and real-world experiments, DART outperforms existing VLA adaptation methods in one-shot scenarios across diverse visual and embodiment shifts. Code is available at https://github.com/snumprlab/dart.