ember-lab-berkeley/robocasa365-target-atomic
Robotics Dataset
RoboCasa365 target-atomic is a large-scale human teleoperation dataset of 18 atomic kitchen tasks (500 demos each) in 10 held-out kitchens, with 9,126 episodes and over 2.2 million frames, providing RGB camera views, state/action data, and natural language task descriptions.
Catalog metadata
- Repository owner
- ember-lab-berkeley
- Source
- Hugging Face / LeRobot
- Robot
- PandaOmron
- Episodes
- 9126
- Frames
- 2231347
- Cameras
- 3
- Frames per second
- 20
- License
- mit
Browse the catalog to compare robots, tasks, sensors, and licenses. Sign in to open protected download and model links.