DecentVLAapache-2.0
moplan_Cube_Red_on_Green_franka_left
A dataset of 100 successful cube-stacking episodes where a Franka Panda robot in Isaac Sim places a red cube on a green cube, with actions generated by a motion planner discovered through GEPA (LLM-driven program search).
Downloads15
Episodes100
Why This Matters for Physical AI
This dataset demonstrates learning from motion-planner-generated trajectories rather than teleoperation, providing a controlled alternative for training vision-language-action models with guaranteed task success and consistent policy behavior across environmental variations.
Technical Profile
- Modalities
- rgbproprioception
- Robot Embodiments
- Franka Panda
- Action Space
- joint_positions
- Environment
- simulation
- Task Types
- manipulationpick_and_place
- Episodes
- 100
- Data Format
- LeRobot
- License
- apache-2.0
Access
Need custom rgb data?
Claru builds purpose-built datasets for simulation applications with dense human annotations and quality assurance.
Request a Sample Pack