Team-RAMEN2026cc-by-4.0
RAMEN G1 IKEA Assembly — RAMEN-Ori insert_table_leg
Training dataset for the RAMEN-Ori policy consisting of 1,664 episodes of a Unitree G1 humanoid robot with Dex1-1 gripper performing the insert_table_leg assembly task on a children's table. The dataset includes multi-camera RGB and infrared imagery, proprioceptive state, and action commands with pre-rendered image caches featuring YOLO-OBB object detections and augmentations.
Downloads218
Episodes1,664
Hours4.5
Why This Matters for Physical AI
This dataset provides high-quality demonstrations of complex assembly manipulation by a humanoid robot with dexterous hands, directly supporting research in physical AI for long-horizon task learning and object interaction understanding.
Technical Profile
- Modalities
- rgbinfraredproprioception
- Robot Embodiments
- Unitree G1Dex1-1
- Action Space
- joint_positions
- Environment
- lab
- Task Types
- assemblyinsertionmanipulation
- Episodes
- 1,664
- Total Hours
- 4.5
- Data Format
- LeRobot
- Annotation Types
- bounding_boxescuration_labels
- License
- cc-by-4.0
Access
Need custom rgb data?
Claru builds purpose-built datasets for lab applications with dense human annotations and quality assurance.
Request a Sample Pack