OpenHLM: An Empirical Recipe for Whole-Body Humanoid Loco-Manipulation
Overview
This paper introduces OpenHLM, an open-source framework for whole-body humanoid loco-manipulation. The authors conduct a systematic empirical study across three phases—whole-body teleoperation, VLA model design, and heterogeneous co-training—to build a native vision-language-action (VLA) model that maps language and pixels directly to all degrees of freedom. Key findings include the superiority of a joint-based teleoperation interface, effective transfer of VLA models pretrained on static and wheeled dual-arm platforms to humanoids, and policy extension via co-training with HuMI. In long-horizon tasks spanning a wide vertical range, OpenHLM outperforms two state-of-the-art baselines (GR00T N1.6 and Ψ₀) using less than half the total demonstration time. Code, training data, and model checkpoints are available at https://openhlm-project.github.io/.
Generated from attributed reports · Updated 2 hours ago
Event evidence and corrections
0 attributed source owners. Ownership does not establish independent confirmation. Quantities are reported separately and are never added together.
Artifact availability · code: available
“ipe for whole-body humanoid loco-manipulation. In a challenging long-horizon task that spans a wide vertical range of the humanoid, OpenHLM outperforms two state-of-the-art humanoid VLA baselines (GR00T N1.6 and $\Psi_0$) using less than half the total demonstration time. Our code, training data, and model checkpoints are available at [https://openhlm-project.github.io/].”
Exact source · revision 1Source owner not reported
Report timeline
Follow attributed reports and material updates.
- arXiv Robotics — research abstractsSignalOpenHLM: An Empirical Recipe for Whole-Body Humanoid Loco-Manipulation
This paper introduces OpenHLM, an open-source framework for whole-body humanoid loco-manipulation. The authors conduct a systematic empirical study across three phases—whole-body teleoperation, VLA model design, and heterogeneous co-training—to build a native vision-language-action (VLA) model that maps language and pixels directly to all degrees of freedom. Key findings include the superiority of a joint-based teleoperation interface, effective transfer of VLA models pretrained on static and wheeled dual-arm platforms to humanoids, and policy extension via co-training with HuMI. In long-horizon tasks spanning a wide vertical range, OpenHLM outperforms two state-of-the-art baselines (GR00T N1.6 and Ψ₀) using less than half the total demonstration time. Code, training data, and model checkpoints are available at https://openhlm-project.github.io/.
Event coverage history
There is not enough continuous observation data to show a trend.
Timezone · UTC
Article dates follow your selected timezone. Briefing editions use Hong Kong time (UTC+8).