SENTINEL: A Fully End-to-End Language-Action Model for Humanoid Whole Body Control

TL;DR

SENTINEL is an end-to-end model that directly translates language commands into humanoid robot actions, achieving strong understanding and stable execution in simulation and real-world scenarios.

Contribution

The paper introduces SENTINEL, a novel fully end-to-end language-action model for humanoid control that bypasses traditional modular pipelines and demonstrates effective real-world deployment.

Findings

Strong semantic understanding demonstrated

Stable execution in simulation and real-world

Supports multi-modal input extensions

Abstract

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 model 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.