liorbenhorin-nv2026
SO-101 Vials-to-Rack — GR00T N1.7 Sim250 + Real10 — 50 Real Rollouts
Autonomous policy rollouts recorded on a physical SO-101 follower robot performing vial-to-rack placement tasks. The dataset contains 50 episodes (22 successes, 28 failures) with RGB observations from wrist and external cameras.
Downloads36
Episodes50
Hours0.123
Why This Matters for Physical AI
This dataset provides real-world policy rollout data for training and evaluating vision-language robot models (GR00T) on manipulation tasks, demonstrating the practical performance of learned policies on physical hardware with naturalistic success/failure distributions.
Technical Profile
- Modalities
- rgb
- Robot Embodiments
- SO-101 follower
- Action Space
- relative joint positions (6-DOF: shoulder_pan, shoulder_lift, elbow_flex, wrist_flex, wrist_roll, gripper)
- Environment
- lab
- Task Types
- manipulationpick_and_place
- Episodes
- 50
- Total Hours
- 0.123
- Data Format
- LeRobot v3.0
- Annotation Types
- action_labelsreward_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