daixianjie
Controlled Anchor — Task Suite V2
A collection of three task suites for imitation learning, with Task Suite 2 focusing on controlled candy pouring with 1920 training and 132 held-out support/oracle episodes totaling 734887 frames at 20 Hz.
Downloads26
Episodes2052
Why This Matters for Physical AI
This dataset provides high-quality imitation learning demonstrations for manipulation tasks with preserved state and action values, supporting research in learning robot control policies from expert examples.
Technical Profile
- Modalities
- rgbproprioceptionstateaction
- Environment
- lab
- Task Types
- pouringmanipulation
- Episodes
- 2052
- Data Format
- LeRobot
- Annotation Types
- action_labels
Access
Need custom rgb data?
Claru builds purpose-built datasets for lab applications with dense human annotations and quality assurance.
Request a Sample Pack