Bionic Human-Motion Style Transfer for Physically Executable Whole-Body Control of Humanoid Robots

Tianchen Huang, Mingkuan Zhao, Yang Gao, Feiyang Yuan, Junchi Gu, Xiaohu Zhang, Dongdong Zhao, Shi Yan, Yu Wang, Wei Gao, Shiwu Zhang · 2026

Read Bionic Human-Motion Style Transfer for Physically Executable Whole-Body Control of Humanoid Robots by Tianchen Huang, Mingkuan Zhao, Yang Gao, Feiyang Yuan, Junchi Gu, Xiaohu Zhang, Dongdong Zhao, Shi Yan, Yu Wang, Wei Gao, Shiwu Zhang (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2606-03536.txt.

Description

Huang and co-authors transfer human motion style onto humanoid robots that must stay physical. Expressive whole-body motion helps robots that work near people. Style cues ride in gait rhythm, posture, arm swing, and body sway. Most robot motion still comes from fixed demos or hand-built scripts. A shown style is thus hard to reuse on new motion content.

A generation-to-control framework solves the reuse problem. It takes a short human style example plus a target motion. A physics-aware diffusion model fuses style, content, and path conditions. Guidance strength sets how strong the style feels, with no retraining. Contact and smoothness rules during training cut artifacts. The output is converted for the Unitree G1 robot and run by a tracking policy. Real-robot trials reach a 96 percent success rate over 125 runs. Contact and jitter faults fall versus animation-based rivals.

Catalog record

CREATORTianchen Huang, Mingkuan Zhao, Yang Gao, Feiyang Yuan, Junchi Gu, Xiaohu Zhang, Dongdong Zhao, Shi Yan, Yu Wang, Wei Gao, Shiwu Zhang
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE52 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2606.03536

Browse the full library · About the Tower of Babel