Dataset · InternRobotics
Robot manipulators: 630K trajectories
InternRobotics/InternData-A1
Hybrid synthetic-real trajectory dataset for training robotic manipulators.
What's inside
Contains over 630k trajectories and 7,433 hours of data across 4 robot embodiments, 18 skills, 70 tasks, and 227 scenes. Covers handling rigid, articulated, deformable, and liquid objects. Data includes 3D, images, and text; description language is English.
What you can do with it
- Train a robot manipulator control policy
- Fine-tune a VLA model on 70 tasks
- Evaluate sim-to-real transfer
- Tasks
- Other, Robotics
- Languages
- English
- Size
- over 1T rows
- Downloads this month
- 71.9K
- Likes
- 118
- Access
- on request on Hugging Face
Models trained on it
- InternRobotics/InternVLA-A1.5-LiberoRobotics 141
- InternRobotics/InternVLA-A1.5-baseRobotics 138
- InternRobotics/InternVLA-A1-3BRobotics 97
- InternRobotics/InternVLA-A1.5-RoboTwinRobotics 84
- InternRobotics/InternVLA-A1.5-DOMINORobotics 32
- yuyangalin/ImageWAM-FLUX.2-4B-InternData-A1-EERobotics 30
- fan91/ImageWAM-FLUX.2-4BImage → image 12
How to load it
from datasets import load_dataset
ds = load_dataset("InternRobotics/InternData-A1")