面向人形全身控制的端到端语言-动作模型
End-to-End Language-Action Model for Humanoid Whole Body Control
把语言指令与本体感知直接映射到低层动作的端到端流匹配策略，配合残差动作头做真机精修，可作为视觉-语言-动作（VLA）类具身控制中绕开中间表示与模块化流水线的可复用范式。
原文依据 The model directly maps language commands and proprioceptive inputs to low-level actions without any intermediate representation.
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论现有人形控制系统往往依赖遥操作，或依赖将语言理解与物理执行分离的模块化生成流水线。然而，前者完全由人驱动，后者则缺乏语言指令与物理行为之间的紧密对齐。本文提出 SENTINEL，一个用于人形全身控制的完全端到端语言-动作框架。我们利用预训练的全身控制器在仿真中跟踪人体运动，并结合其文本标注，构建了一个大规模数据集。该模型直接将语言指令与本体感知输入映射为低层动作，不依赖任何中间表示。模型使用流匹配生成动作分块，随后可由残差动作头进行精修以用于真实世界部署。我们的方法在仿真与真实世界部署中均展现出强语义理解能力与人形机器人上的稳定执行能力，并且通过将输入转换为文本还支持多模态扩展。
英文原文摘要 1039 字符
Existing humanoid control systems often rely on teleoperation or modular generation pipelines that separate language understanding from physical execution. However, the former is entirely human-driven, and the latter lacks tight alignment between language commands and physical behaviors. In this paper, we present SENTINEL, a fully end-to-end language-action framework for humanoid whole-body control. We construct a large-scale dataset by tracking human motions in simulation using a pretrained whole body controller, combined with their text annotations. The model directly maps language commands and proprioceptive inputs to low-level actions without any intermediate representation. The model generates action chunks using flow matching, which can be subsequently refined by a residual action head for real-world deployment. Our method exhibits strong semantic understanding and stable execution on humanoid robots in both simulation and real-world deployment, and also supports multi-modal extensions by converting inputs into texts.