THULab/jetson1_060826_subtask_grab2

Jetson1 060826 Subtask Grab2 TsFile

This dataset is an Apache TsFile conversion of VibeCuisine/jetson1-060826-subtask-grab2, a LeRobot v3.0 robot dataset.

Modalities: Time-series. The converted repository contains numeric robot observations, actions, frame timing, task/episode tags, and source metadata. Camera videos remain in the original Hugging Face dataset.

Source Dataset

• Source dataset: VibeCuisine/jetson1-060826-subtask-grab2
• License: Apache-2.0
• Robot type: vibeboard_follower_tilt
• LeRobot codebase version: v3.0
• Split: train
• Scale: 55 episodes, 7,683 frames, 1 task, 20 fps
• Source frame layout: data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet
• Source video layout: videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4
• Task metadata: grab the cucumber close to one of the cucumber's end

Converted Files

• TsFile: data/jetson1_060826_subtask_grab2.tsfile
• Table: jetson1_060826_subtask_grab2
• Rows: 7,683
• Episodes: 55
• Time precision: milliseconds
• Metadata: meta/ is mirrored from the source, with meta/info.json updated to describe the converted TsFile artifact.

Schema

Time is synthesized as round(timestamp * 1000) milliseconds and restarts within each episode.

TAG columns:

• episode_index
• task_index

Scalar FIELD columns:

• frame_index
• sample_index, renamed from the source index

Flattened FLOAT FIELD groups:

• action[7] -> action_0 ... action_6
• observation.state[7] -> observation_state_0 ... observation_state_6

Conversion Notes

• The generic lerobot converter was used.
• All source frame Parquet shards are merged into one TsFile table. Use episode_index and task_index TAG filters to select an episode or task.
• Vector columns are flattened to scalar TsFile fields. Source column prefixes are preserved, with . replaced by _.
• The source timestamp column is dropped after Time synthesis because it is redundant with Time / 1000 seconds.
• The source index column is renamed to sample_index.
• Source camera video streams are not included in this repository: observation.images.top, observation.images.wrist, and observation.images.base. They remain available in the original dataset under videos/.
• No rows and no numeric time-series fields are intentionally dropped other than the redundant timestamp column noted above.

Validation

The converted TsFile was read back with the Apache TsFile Python SDK. The TsFile metadata row count and query readback both matched the staged Parquet: 7,683 rows.

Usage

from tsfile import TsFileReader

path = "data/jetson1_060826_subtask_grab2.tsfile"
reader = TsFileReader(path)

schemas = reader.get_all_table_schemas()
table_name = "jetson1_060826_subtask_grab2"
columns = [
    column.get_column_name()
    for column in schemas[table_name].get_columns()
    if column.get_column_name() not in {"Time"}
]

with reader.query_table(table_name, columns, batch_size=65536) as result:
    batch = result.read_arrow_batch()