10Kh-RealOmin-OpenData
VerifiedLargest open-source dual-hand robotics dataset with 13,000+ hours of trajectory data.
What is 10Kh-RealOmin-OpenData?
This is a large-scale collection of embodied motion data focused on dual-hand manipulation, recorded in everyday environments with full trajectory details.
It is intended for use in robotics research and reinforcement-learning model training that requires real-world manipulation examples.
What you can build with 10Kh-RealOmin-OpenData
Train robotic manipulation models
Use the synchronized dual-hand trajectories to train imitation learning policies for household object interactions in simulation or on real robots.
Develop precise hand tracking systems
Leverage frame-level clips and sub-millisecond alignment to benchmark and improve multi-camera or wearable-based hand pose estimation algorithms.
Analyze human activity patterns
Study motion data across 10,000+ scenarios to build statistical models of everyday bimanual actions for ergonomics or AR/VR applications.
Load 10Kh-RealOmin-OpenData
from datasets import load_dataset
ds = load_dataset("genrobot2025/10Kh-RealOmin-OpenData")- 1pip install datasets
- 2from datasets import load_dataset
- 3dataset = load_dataset('genrobot2025/10Kh-RealOmin-OpenData')
- 4Access splits and view motion clip metadata via dataset['train']
- 5Load individual trajectory arrays for model training
10Kh-RealOmin-OpenData: pros & cons
Pros
- +Over 13,000 hours of real-world recordings
- +High-precision reconstruction with sub-millisecond sync
- +Diverse data from 3,000+ participants
- +5+ million individual clips ready for ML
Cons
- –Extremely large dataset size may require significant storage
- –No text or language annotations provided
- –Category listed as NLP but content is motion capture only
Frequently asked questions
A large collection of dual-hand motion recordings from real household scenarios with precise trajectory data.
User reviews
Verified reviews from the community shape this listing's rating.
Loading reviews…
Similar datasets
Other text & nlp options worth comparing.