PHUMA: Physically-Grounded Humanoid Locomotion Dataset

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Lee, Kyungmin, Kim, Sibeen, Park, Minho, Kim, Hyunseung, Hwang, Dongyoon, Lee, Hojoon, Choo, Jaegul
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866914124140642304
author Lee, Kyungmin
Kim, Sibeen
Park, Minho
Kim, Hyunseung
Hwang, Dongyoon
Lee, Hojoon
Choo, Jaegul
author_facet Lee, Kyungmin
Kim, Sibeen
Park, Minho
Kim, Hyunseung
Hwang, Dongyoon
Lee, Hojoon
Choo, Jaegul
contents Motion imitation is a promising approach for humanoid locomotion, enabling agents to acquire humanlike behaviors. Existing methods typically rely on high-quality motion capture datasets such as AMASS, but these are scarce and expensive, limiting scalability and diversity. Recent studies attempt to scale data collection by converting large-scale internet videos, exemplified by Humanoid-X. However, they often introduce physical artifacts such as floating, penetration, and foot skating, which hinder stable imitation. In response, we introduce PHUMA, a Physically-grounded HUMAnoid locomotion dataset that leverages human video at scale, while addressing physical artifacts through careful data curation and physics-constrained retargeting. PHUMA enforces joint limits, ensures ground contact, and eliminates foot skating, producing motions that are both large-scale and physically reliable. We evaluated PHUMA in two sets of conditions: (i) imitation of unseen motion from self-recorded test videos and (ii) path following with pelvis-only guidance. In both cases, PHUMA-trained policies outperform Humanoid-X and AMASS, achieving significant gains in imitating diverse motions. The code is available at https://davian-robotics.github.io/PHUMA.
format Preprint
id arxiv_https___arxiv_org_abs_2510_26236
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle PHUMA: Physically-Grounded Humanoid Locomotion Dataset
Lee, Kyungmin
Kim, Sibeen
Park, Minho
Kim, Hyunseung
Hwang, Dongyoon
Lee, Hojoon
Choo, Jaegul
Robotics
Motion imitation is a promising approach for humanoid locomotion, enabling agents to acquire humanlike behaviors. Existing methods typically rely on high-quality motion capture datasets such as AMASS, but these are scarce and expensive, limiting scalability and diversity. Recent studies attempt to scale data collection by converting large-scale internet videos, exemplified by Humanoid-X. However, they often introduce physical artifacts such as floating, penetration, and foot skating, which hinder stable imitation. In response, we introduce PHUMA, a Physically-grounded HUMAnoid locomotion dataset that leverages human video at scale, while addressing physical artifacts through careful data curation and physics-constrained retargeting. PHUMA enforces joint limits, ensures ground contact, and eliminates foot skating, producing motions that are both large-scale and physically reliable. We evaluated PHUMA in two sets of conditions: (i) imitation of unseen motion from self-recorded test videos and (ii) path following with pelvis-only guidance. In both cases, PHUMA-trained policies outperform Humanoid-X and AMASS, achieving significant gains in imitating diverse motions. The code is available at https://davian-robotics.github.io/PHUMA.
title PHUMA: Physically-Grounded Humanoid Locomotion Dataset
topic Robotics
url https://arxiv.org/abs/2510.26236