ZETA：面向桌面操作的零样本跨 embodiment VLA 迁移的受控研究

中文摘要 AI 辅助

对未见 embodiment 的零样本泛化对于可泛化的视觉-语言-动作（VLA）模型至关重要，因为机器人硬件不断发展，且特定任务的数据采集成本高昂。然而，对该问题的系统理解仍然有限，部分原因是文献缺乏统一的零样本迁移定义，以及能将 embodiment 变化与任务、场景或协议差异隔离开的受控评估设置。为解决这一差距，我们首先区分严格零样本迁移（目标 embodiment 未出现在所有训练数据中）与预训练暴露型零样本迁移（目标 embodiment 仅在预训练期间出现）。随后，我们引入了一个受控基准，涵盖模拟和真实世界验证中的14个保留目标 embodiment。在该框架内，我们对四个因素进行了受控分析：状态-动作表示、预训练 embodiment 多样性、辅助联合训练目标以及目标 embodiment 暴露。实验结果表明，局部末端执行器（EEF）状态-动作表示、源 embodiment 多样性以及辅助联合训练分别将跨 embodiment 迁移性能提升了约15、18和7个百分点。我们进一步发现，在预训练期间仅添加5%的目标 embodiment 数据，可使平均目标 embodiment 进展提升13.4个百分点，这表明严格零样本迁移与预训练暴露型零样本迁移是不同的，应分别报告。这些发现共同为评估和改进两指夹爪桌面操作场景中的跨 embodiment VLA 迁移提供了实用指导，同时推动未来对更广泛设置的研究，包括移动基座控制、灵巧手和长 horizon 任务。

英文摘要

Zero-shot generalization to unseen embodiments is important for generalizable vision-language-action (VLA) models as robot hardware evolves and task-specific data collection remains costly. However, a systematic understanding of this problem remains limited, in part because the literature lacks a unified zero-shot transfer definition and controlled evaluation settings that isolate embodiment changes from differences in tasks, scenes, or protocols. To address this gap, we first distinguish strict zero-shot transfer, where the target embodiment is absent from all training data, from pretrain-exposed zero-shot transfer, where it appears only during pretraining. We then introduce a controlled benchmark spanning 14 held-out target embodiments across simulation and real-world validation. Within this framework, we conduct a controlled analysis of four factors: state-action representations, pretraining embodiment diversity, auxiliary co-training objectives, and target-embodiment exposure. Experimental results show that local end-effector (EEF) state-action representations, the source embodiment diversity, and auxiliary co-training improve cross-embodiment transfer by around 15, 18, and 7 percentage points, respectively. We further find that adding only 5% target-embodiment data during pretraining improves average target-embodiment progress by 13.4 percentage points, showing that strict and pretrain-exposed zero-shot transfer are distinct and should be reported separately. Together, these findings provide practical guidance for evaluating and improving cross-embodiment VLA transfer in stationary tabletop manipulation with two-finger grippers, while motivating future investigation of broader settings including mobile-base control, dexterous hands, and long-horizon tasks.

发表机构

• Galbot（智元机器人）
• CFCS, School of CS, Peking University（北京大学计算机学院CFCS）
• Peking University（北京大学）
• Renmin University of China（中国人民大学）
• Xiamen University Malaysia（马来西亚厦门大学）
• The University of Hong Kong（香港大学）
• Beihang University（北京航空航天大学）

机构由 AI 辅助整理，请以论文原文为准。

相关深度报道

• 智元机器人等测14种机型：只加5%目标数据，迁移进展升13.4点