Unveiling the Impact of Data and Model Scaling on High-Level Control for Humanoid Robots

TL;DR

This paper introduces a large-scale humanoid robot motion dataset and a scalable learning framework, demonstrating significant improvements in high-level control and motion quality through data and model scaling, validated both in simulation and real robots.

Contribution

The paper presents Humanoid-Union, a new extensive dataset, and SCHUR, a scalable learning framework, advancing high-level control for humanoid robots with improved motion quality and alignment.

Findings

37% reconstruction improvement under MPJPE

25% alignment improvement under FID

Effective deployment on real humanoid robots

Abstract

Data scaling has long remained a critical bottleneck in robot learning. For humanoid robots, human videos and motion data are abundant and widely available, offering a free and large-scale data source. Besides, the semantics related to the motions enable modality alignment and high-level robot control learning. However, how to effectively mine raw video, extract robot-learnable representations, and leverage them for scalable learning remains an open problem. To address this, we introduce Humanoid-Union, a large-scale dataset generated through an autonomous pipeline, comprising over 260 hours of diverse, high-quality humanoid robot motion data with semantic annotations derived from human motion videos. The dataset can be further expanded via the same pipeline. Building on this data resource, we propose SCHUR, a scalable learning framework designed to explore the impact of large-scale data on high-level control in humanoid robots. Experimental results demonstrate that SCHUR achieves high robot motion generation quality and strong text-motion alignment under data and model scaling, with 37\% reconstruction improvement under MPJPE and 25\% alignment improvement under FID comparing with previous methods. Its effectiveness is further validated through deployment in real-world humanoid robot.