Dataset Viewer
Auto-converted to Parquet Duplicate
Search is not available for this dataset
video
video
30.1
36
label
class label
2 classes
020260925_182240_88fe4bd3
020260925_182240_88fe4bd3
020260925_182240_88fe4bd3
120260925_182910_413ba05c
120260925_182910_413ba05c
120260925_182910_413ba05c

Bottle Straw

Two YAM robot teleoperation recordings in the original yam-abc-reproduce raw episode format (schema version 1). The original folder names, metadata, arrays, and videos are preserved.

Folder Episode Control frames Control rate
bottle_straw_01 20260925_182240_88fe4bd3 902 30 Hz
bottle_straw_02 20260925_182910_413ba05c 1080 30 Hz

Total: 2 episodes and 1,982 control frames.

Each episode contains:

  • metadata.json: recording configuration and frame count.
  • left-images-rgb.mp4, right-images-rgb.mp4, and top-images-rgb.mp4: RGB camera recordings, configured at 640 x 480 and 30 FPS.
  • *-joint_pos.npy and *-gripper_pos.npy: recorded arm observations.
  • action-*-joint.npy and action-*-gripper.npy: recorded arm actions.
  • *-timestamp.npy: camera timestamps.
  • write_complete.flag: recording completion marker.

Both arms have six joints and a gripper. NumPy arrays can be loaded with numpy.load(path, allow_pickle=False) and videos with an MP4-compatible decoder. These files retain the raw recording format; they have not been converted to LeRobot or another training dataset format.

Downloads last month
67