G1 Carrying Box
100 annotated carrying box demonstrations from robot Charlie, recorded on
2026-09-29. LeRobot v3.0, 30 FPS, 82,033 frames
(0.760 hours).
The feature names, dimensions, camera layout and nested language schema follow BitRobot/HIW-500-LeRobot. These are separately converted source recordings, not copied reference episodes.
Features
observation.images.head: 480 x 1280 RGB stereo image, kept side-by-side.observation.images.left_wrist,observation.images.right_wrist: 480 x 640 RGB.observation.state: 29 measured joint positions, ordered as listed inmeta/info.json.observation.state.wbc: pivot (7), measured end-effector poses (12), gripper controls (4).action: pivot (7), commanded end-effector poses (12), gripper controls (4).language_persistent: main-task and subtask timeline with times relative to the episode start.language_events: empty lists, matching the reference dataset's representation.
The same pivot and gripper-control values accompany measured and commanded end-effector poses, following the reference's 23-dimensional layout. Head images are not split, and no camera flips or IR streams are added. Videos use AV1.
Segmentation and synchronization
Each episode runs from its exact carrying box annotation through the matching
end, sampled on a 30 Hz grid. Every feature uses the most recent MCAP sample
at or before each tick, without interpolation. Source sample holds are bounded
by 2 seconds. Original UTC times and annotation IDs are retained in
meta/source_segments.json.
This release contains exactly 100 episodes as requested, selected from
147 validated episodes. Source recording proportions are preserved
using largest-remainder quotas; within each source, episodes with smaller
maximum sample age are preferred, then longer duration and annotation ID.
See meta/selection.json for the complete deterministic selection record.
47 otherwise valid episodes were omitted only for the requested count;
2 other original intervals had previously failed data-quality
checks. Both categories are recorded in meta/excluded_segments.json.
Healthy video packets are copied without re-encoding. Selected episode boundaries, numeric values, and language timelines are unchanged. Export episode indices are reassigned contiguously; previous indices remain in source metadata.
Language timelines are repeated on every frame of their episode, as in the
reference. The full source annotation timeline is retained, including future
subtasks; applications should filter by timestamp if they need only the current
instruction. Calibration files are preserved under calibration/.
Validation and compatibility
All numeric frames were checked for finite values, correct dimensions,
contiguous episode/frame indices and 30 Hz timestamps. First, middle and last
frames of every episode were read back through the native LeRobot v3 reader and
all three camera streams were decoded. See meta/validation.json.
The container uses LeRobot 0.4.4. Its native reader needs a feature/tensor adapter
for the reference dataset's additive language type; the converter includes
enable_hiw_language_reader() for that purpose. Numeric data, indexing and video
decoding use the original LeRobot implementation. The exported Arrow language
columns and feature declarations match the reference schema.
- Downloads last month
- 1,489