Gallery · LIBERO-Pro

place white mug on plate and chocolate pudding to right

libero_10_lan, task 6 · seed 2. DeepSeek-V4.1-Flash, code mode: 7 decisions, 32 motions, solved.

2× · 95 s of motion
  1. 1 of 7 · grasp_mug.py
  2. 2 of 7 · grasp_mug2.py
  3. 3 of 7 · rim_grasp2.py
  4. 4 of 7 · place_mug2.py
  5. 5 of 7 · grasp_pudding.py
  6. 6 of 7 · pudding_grasp2.py
  7. 7 of 7 · place_pudding.py
  1. set_gripper
  2. move_pose
  3. move_pose
  4. set_gripper
  5. wait_ticks
  6. move_pose
  7. set_gripper
  8. move_pose
  9. move_pose
  10. set_gripper
  11. wait_ticks
  12. move_pose
  13. move_pose
  14. move_pose
  15. set_gripper
  16. wait_ticks
  17. move_pose
  18. set_gripper
  19. move_pose
  20. move_pose
  21. set_gripper
  22. wait_ticks
  23. move_pose
  24. move_pose
  25. set_gripper
  26. move_pose
  27. move_pose
  28. set_gripper
  29. wait_ticks
  30. move_pose
  31. move_pose
  32. move_pose
Decisions

The recording is the robot's own cameras at 2×; the time between motions, while the model thinks, is not filmed. The 3D replay is the camera's picture in depth after each decision, with the paths it planned, to turn and step through.