Gallery · LIBERO-Pro

Open the middle layer of the drawer

libero_goal_object, task 0 · seed 3. DeepSeek-V4.1-Flash, code mode: 3 decisions, 4 motions, solved.

2× · 24 s of motion
  1. 1 of 3 · step_pregrasp.py
  2. 2 of 3 · step_grasp.py
  3. 3 of 3 · step_pull.py
  1. move_pose
  2. move_pose
  3. set_gripper
  4. 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.