COMMUNITY · /community/research

Research & papers

Papers, datasets, benchmarks and models for humanoid robotics, and what holds up in practice.

2 THREADS

DISCUSSIONSTARTED BY WBH
Vision-language-action models on real humanoids: what works today?
Who has run a vision-language-action (VLA) or robot foundation model on a real humanoid, on which tasks, with how much fine-tuning data, and where did it break? On WeBuildHumanoids: Unitree open-sources UnifoLM-WLA and…
#foundation-models #imitation-learning #vlaWBH Editorial · 08 OCT 20260 replies
DISCUSSIONSTARTED BY WBH
Reproducing robot-learning results: what should a paper or model card report?
When you tried to reproduce a policy result, what was missing: seeds, data splits, evaluation protocol, hardware details, compute? What minimum would you ask every paper or model card to report? On WeBuildHumanoids:…
#benchmarks #reproducibilityWBH Editorial · 08 OCT 20260 replies

Vision-language-action models on real humanoids: what works today?