HuMI: Humanoid Whole-Body Manipulation from Humanoid-Free Demonstrations
HuMI: Humanoid Whole-Body Manipulation from Humanoid-Free Demonstrations
This paper introduces HuMI, a framework for learning whole-body manipulation tasks using data collected without humanoids. The method captures rich motion data via portable hardware to train hierarchical models that translate human motions into humanoid skills. Experiments on five tasks show improved data collection efficiency and performance in unseen environments.
The authors present a framework that claims to achieve a 3x increase in data collection efficiency compared to teleoperation and a 70% success rate in unseen environments, though these metrics are based on experiments across five specific tasks described in the abstract.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
Reported numbers
Reported success rate
70%
Reported trials
View original evidence
including kneeling, squatting, tossing, walking, and bimanual manipulation--demonstrate that HuMI achieves a 3x increase in data collection efficiency compared to teleoperation and attains a 70% success rate in unseen environments.
Open source S4data collection efficiency increase
3
View original evidence
including kneeling, squatting, tossing, walking, and bimanual manipulation--demonstrate that HuMI achieves a 3x increase in data collection efficiency compared to teleoperation and attains a 70% success rate in unseen environments.
Open source S4number of environments tested
5
View original evidence
ation Interface (HuMI), a portable and efficient framework for learning diverse whole-body manipulation tasks across various environments. HuMI enables humanoid-free data collection by capturing rich whole-body motion using portable hardware. This data drives a hierarchical learning pipeline that translates human motions into dexterous and feasible humanoid skills. Extensive experiments across five whole-body tasks--
Open source S3number of tasks
5
View original evidence
ation Interface (HuMI), a portable and efficient framework for learning diverse whole-body manipulation tasks across various environments. HuMI enables humanoid-free data collection by capturing rich whole-body motion using portable hardware. This data drives a hierarchical learning pipeline that translates human motions into dexterous and feasible humanoid skills. Extensive experiments across five whole-body tasks--
Open source S3
- paper: not_reported
- code: not_reported
- weights: not_reported
- dataset: not_reported
What remains unknown
Not established in the collected evidence: Environment, Control, Data origin.
Reported performance applies to the described task. It does not establish general autonomy or deployment readiness.
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
including kneeling, squatting, tossing, walking, and bimanual manipulation--demonstrate that HuMI achieves a 3x increase in data collection efficiency compared to teleoperation and attains a 70% success rate in unseen environments.
Open source S4
ation Interface (HuMI), a portable and efficient framework for learning diverse whole-body manipulation tasks across various environments. HuMI enables humanoid-free data collection by capturing rich whole-body motion using portable hardware. This data drives a hierarchical learning pipeline that translates human motions into dexterous and feasible humanoid skills. Extensive experiments across five whole-body tasks--
Open source S3
arXiv:2602.06643v3 Announce Type: replace Abstract: Current approaches for humanoid whole-body manipulation, primarily relying on teleoperation or visual sim-to-real reinforcement learning, are hindered by hardware logistics and complex reward engineering. Consequently, demonstrated autonomous skills remain limited and are typically restricted to controlled environments. In this paper, we present the Humanoid Manipul
Open source S2
HuMI: Humanoid Whole-Body Manipulation from Humanoid-Free Demonstrations
Open source S1
Source：arXiv Robotics — research abstracts · arxiv.org