VisualMimic: Visual Humanoid Loco-Manipulation via Motion Tracking and Generation

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Yin, Shaofeng, Ze, Yanjie, Yu, Hong-Xing, Liu, C. Karen, Wu, Jiajun
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866909900193398784
author Yin, Shaofeng
Ze, Yanjie
Yu, Hong-Xing
Liu, C. Karen
Wu, Jiajun
author_facet Yin, Shaofeng
Ze, Yanjie
Yu, Hong-Xing
Liu, C. Karen
Wu, Jiajun
contents Humanoid loco-manipulation in unstructured environments demands tight integration of egocentric perception and whole-body control. However, existing approaches either depend on external motion capture systems or fail to generalize across diverse tasks. We introduce VisualMimic, a visual sim-to-real framework that unifies egocentric vision with hierarchical whole-body control for humanoid robots. VisualMimic combines a task-agnostic low-level keypoint tracker -- trained from human motion data via a teacher-student scheme -- with a task-specific high-level policy that generates keypoint commands from visual and proprioceptive input. To ensure stable training, we inject noise into the low-level policy and clip high-level actions using human motion statistics. VisualMimic enables zero-shot transfer of visuomotor policies trained in simulation to real humanoid robots, accomplishing a wide range of loco-manipulation tasks such as box lifting, pushing, football dribbling, and kicking. Beyond controlled laboratory settings, our policies also generalize robustly to outdoor environments. Videos are available at: https://visualmimic.github.io .
format Preprint
id arxiv_https___arxiv_org_abs_2509_20322
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle VisualMimic: Visual Humanoid Loco-Manipulation via Motion Tracking and Generation
Yin, Shaofeng
Ze, Yanjie
Yu, Hong-Xing
Liu, C. Karen
Wu, Jiajun
Robotics
Computer Vision and Pattern Recognition
Machine Learning
Humanoid loco-manipulation in unstructured environments demands tight integration of egocentric perception and whole-body control. However, existing approaches either depend on external motion capture systems or fail to generalize across diverse tasks. We introduce VisualMimic, a visual sim-to-real framework that unifies egocentric vision with hierarchical whole-body control for humanoid robots. VisualMimic combines a task-agnostic low-level keypoint tracker -- trained from human motion data via a teacher-student scheme -- with a task-specific high-level policy that generates keypoint commands from visual and proprioceptive input. To ensure stable training, we inject noise into the low-level policy and clip high-level actions using human motion statistics. VisualMimic enables zero-shot transfer of visuomotor policies trained in simulation to real humanoid robots, accomplishing a wide range of loco-manipulation tasks such as box lifting, pushing, football dribbling, and kicking. Beyond controlled laboratory settings, our policies also generalize robustly to outdoor environments. Videos are available at: https://visualmimic.github.io .
title VisualMimic: Visual Humanoid Loco-Manipulation via Motion Tracking and Generation
topic Robotics
Computer Vision and Pattern Recognition
Machine Learning
url https://arxiv.org/abs/2509.20322