Dataset Viewer
Auto-converted to Parquet Duplicate
Search is not available for this dataset
video
video
6.51
11
label
class label
2 classes
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
0observation.images.left_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head
1observation.images.right_head

YAML Metadata Warning:empty or missing yaml metadata in repo card

Check out the documentation for more information.

byang11259/stack_cubes_takes

43 episodes from 4 source recordings (10664 frames total), exported from raw portable-gripper captures.

Coordinate frame

Poses are in the marker frame, body frame tcp.

A world transform is baked into every pose, applied as Translate(world_offset) @ Rz(world_yaw) - rotate about the world origin first, then translate in the rotated frame:

  • world_yaw: -90.0 deg about +Z (rotates orientations too)
  • world_offset: [0.87, 0.175, 0.0107] m

A replay must therefore apply no additional yaw or workspace translation - the alignment is already in the data.

Episode windows

Each source recording holds several takes of the task; every kept window below was cut into its own episode (curation file stack_cubes_takes.json).

Windows are seconds on that episode's global synced clock (the shared two-camera time base), which is also what the source episode.yaml task_window/sync are expressed in. A window may be marginally narrower than the requested cut where real pose/video/gripper coverage ends sooner.

episode source window (s) frames
0 vid50 7.00-16.66 290
1 vid51 7.00-16.00 270
2 vid51 20.50-28.50 240
3 vid51 32.00-43.00 330
4 vid51 49.50-57.00 225
5 vid51 62.00-72.00 300
6 vid51 75.50-85.20 291
7 vid52 7.20-15.80 258
8 vid52 19.50-28.50 270
9 vid52 33.00-42.50 285
10 vid52 47.00-56.00 270
11 vid52 60.20-68.30 243
12 vid52 73.00-82.00 270
13 vid52 86.00-94.00 240
14 vid52 98.00-108.00 300
15 vid52 112.00-120.00 240
16 vid52 124.60-134.00 282
17 vid52 139.00-149.00 300
18 vid53 7.50-14.00 195
19 vid53 18.50-26.00 225
20 vid53 30.00-38.50 255
21 vid53 44.00-51.50 225
22 vid53 56.00-64.00 240
23 vid53 68.50-76.50 240
24 vid53 80.00-88.00 240
25 vid53 92.00-100.00 240
26 vid53 104.00-111.00 210
27 vid53 116.00-124.00 240
28 vid53 129.00-138.00 270
29 vid53 141.00-149.00 240
30 vid53 154.00-161.00 210
31 vid53 168.00-175.00 210
32 vid53 179.00-186.00 210
33 vid53 189.00-197.00 240
34 vid53 201.00-209.00 240
35 vid53 213.50-221.50 240
36 vid53 225.00-232.50 225
37 vid53 237.00-244.00 210
38 vid53 248.00-256.00 240
39 vid53 260.00-266.50 195
40 vid53 271.00-278.00 210
41 vid53 283.00-291.00 240
42 vid53 295.00-304.00 270

Requested cuts, by source

source takes requested windows (s)
vid50 1 7.00-16.66
vid51 6 7.00-16.00, 20.50-28.50, 32.00-43.00, 49.50-57.00, 62.00-72.00, 75.50-85.20
vid52 11 7.20-15.80, 19.50-28.50, 33.00-42.50, 47.00-56.00, 60.20-68.30, 73.00-82.00, 86.00-94.00, 98.00-108.00, 112.00-120.00, 124.60-134.00, 139.00-149.00
vid53 25 7.50-14.00, 18.50-26.00, 30.00-38.50, 44.00-51.50, 56.00-64.00, 68.50-76.50, 80.00-88.00, 92.00-100.00, 104.00-111.00, 116.00-124.00, 129.00-138.00, 141.00-149.00, 154.00-161.00, 168.00-175.00, 179.00-186.00, 189.00-197.00, 201.00-209.00, 213.50-221.50, 225.00-232.50, 237.00-244.00, 248.00-256.00, 260.00-266.50, 271.00-278.00, 283.00-291.00, 295.00-304.00

Provenance

  • generated: 2026-08-14T12:37:12
  • full per-episode source/window and the per-arm extrinsic chain: meta/export_provenance.json
Downloads last month
76