VIRAL: Visual Sim-to-Real at Scale for Humanoid Loco-Manipulation

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: He, Tairan, Wang, Zi, Xue, Haoru, Ben, Qingwei, Luo, Zhengyi, Xiao, Wenli, Yuan, Ye, Da, Xingye, Castañeda, Fernando, Sastry, Shankar, Liu, Changliu, Shi, Guanya, Fan, Linxi, Zhu, Yuke
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866908678753353728
author He, Tairan
Wang, Zi
Xue, Haoru
Ben, Qingwei
Luo, Zhengyi
Xiao, Wenli
Yuan, Ye
Da, Xingye
Castañeda, Fernando
Sastry, Shankar
Liu, Changliu
Shi, Guanya
Fan, Linxi
Zhu, Yuke
author_facet He, Tairan
Wang, Zi
Xue, Haoru
Ben, Qingwei
Luo, Zhengyi
Xiao, Wenli
Yuan, Ye
Da, Xingye
Castañeda, Fernando
Sastry, Shankar
Liu, Changliu
Shi, Guanya
Fan, Linxi
Zhu, Yuke
contents A key barrier to the real-world deployment of humanoid robots is the lack of autonomous loco-manipulation skills. We introduce VIRAL, a visual sim-to-real framework that learns humanoid loco-manipulation entirely in simulation and deploys it zero-shot to real hardware. VIRAL follows a teacher-student design: a privileged RL teacher, operating on full state, learns long-horizon loco-manipulation using a delta action space and reference state initialization. A vision-based student policy is then distilled from the teacher via large-scale simulation with tiled rendering, trained with a mixture of online DAgger and behavior cloning. We find that compute scale is critical: scaling simulation to tens of GPUs (up to 64) makes both teacher and student training reliable, while low-compute regimes often fail. To bridge the sim-to-real gap, VIRAL combines large-scale visual domain randomization over lighting, materials, camera parameters, image quality, and sensor delays--with real-to-sim alignment of the dexterous hands and cameras. Deployed on a Unitree G1 humanoid, the resulting RGB-based policy performs continuous loco-manipulation for up to 54 cycles, generalizing to diverse spatial and appearance variations without any real-world fine-tuning, and approaching expert-level teleoperation performance. Extensive ablations dissect the key design choices required to make RGB-based humanoid loco-manipulation work in practice.
format Preprint
id arxiv_https___arxiv_org_abs_2511_15200
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle VIRAL: Visual Sim-to-Real at Scale for Humanoid Loco-Manipulation
He, Tairan
Wang, Zi
Xue, Haoru
Ben, Qingwei
Luo, Zhengyi
Xiao, Wenli
Yuan, Ye
Da, Xingye
Castañeda, Fernando
Sastry, Shankar
Liu, Changliu
Shi, Guanya
Fan, Linxi
Zhu, Yuke
Robotics
A key barrier to the real-world deployment of humanoid robots is the lack of autonomous loco-manipulation skills. We introduce VIRAL, a visual sim-to-real framework that learns humanoid loco-manipulation entirely in simulation and deploys it zero-shot to real hardware. VIRAL follows a teacher-student design: a privileged RL teacher, operating on full state, learns long-horizon loco-manipulation using a delta action space and reference state initialization. A vision-based student policy is then distilled from the teacher via large-scale simulation with tiled rendering, trained with a mixture of online DAgger and behavior cloning. We find that compute scale is critical: scaling simulation to tens of GPUs (up to 64) makes both teacher and student training reliable, while low-compute regimes often fail. To bridge the sim-to-real gap, VIRAL combines large-scale visual domain randomization over lighting, materials, camera parameters, image quality, and sensor delays--with real-to-sim alignment of the dexterous hands and cameras. Deployed on a Unitree G1 humanoid, the resulting RGB-based policy performs continuous loco-manipulation for up to 54 cycles, generalizing to diverse spatial and appearance variations without any real-world fine-tuning, and approaching expert-level teleoperation performance. Extensive ablations dissect the key design choices required to make RGB-based humanoid loco-manipulation work in practice.
title VIRAL: Visual Sim-to-Real at Scale for Humanoid Loco-Manipulation
topic Robotics
url https://arxiv.org/abs/2511.15200