A physically-grounded humanoid locomotion dataset created by Luckyt1, leveraging large-scale human motion data. The dataset was curated to overcome physical artifacts through physics-constrained retargeting, resulting in a high-quality resource for robotics research.
Use Cases
- Training reinforcement learning agents for bipedal locomotion based on physics-constrained motion data.
- Benchmarking humanoid motion imitation algorithms based on curated, large-scale human motion data.
- Developing physics-based simulation models for humanoid robots based on the retargeted locomotion sequences.
Strengths
- Dataset leverages large-scale human motion data, suggesting a substantial volume of source material.
- Data curation and physics-constrained retargeting were applied to overcome physical artifacts, indicating a focus on quality.
Limitations
- Column-level documentation is absent; field semantics must be inferred after download.
- Row count is unknown, which may limit suitability assessment.
- Last updated 2026-05-07 03:04:40; freshness should be verified.
Provenance
- Source
- huggingface
- Collection Method
- Data curation and physics-constrained retargeting applied to large-scale human motion data.