Event
Fourier releases ActionNet dexterous bimanual humanoid manipulation dataset
Key points
- Fourier introduced ActionNet as a dexterous bimanual manipulation dataset for humanoid robots with more than 30,000 teleoperated trajectories and around 140 hours of interaction data.
- The dataset includes Fourier GR1-T1, GR1-T2, and GR2 humanoid recordings plus Fourier dexterous hands, with language annotations reviewed for imitation learning, VLA, and world-model training.
- It matters because this is a first-party dataset release from Fourier rather than a product demo, and it gives the company research tab a concrete research artifact.
Company context
Develops robotics systems across rehabilitation, exoskeletons, and humanoid platforms, with a focus on human-centered applications and physical assistance technologies.
Context
- Company
- Fourier Intelligence
- Segment
- Humanoid
- Event type
- Research Publication
- Geography
- Shanghai · China