zjt24/jetson1_062626_grab_doris

Jetson1 062626 Grab Doris (TsFile)

This dataset is a TsFile conversion of the Hugging Face dataset VibeCuisine/jetson1-062626-grab-doris. It contains the numeric LeRobot frame data for a vibeboard_follower_tilt robot dataset.

Modalities: Time-series. The original dataset also contains videos; those video files are not mirrored in this converted repository.

Source dataset

• Original dataset: VibeCuisine/jetson1-062626-grab-doris
• Source commit used for conversion: d66b4da477651afb45bb5b112836f7945eb526f4
• Original organization: VibeCuisine
• License: apache-2.0
• LeRobot codebase version: v3.0
• Robot type: vibeboard_follower_tilt
• Task: Grab the cucumber close to one of the cucumbers ends

Dataset scale

• Episodes: 55
• Frames / converted TsFile rows: 3,828
• Tasks: 1
• Split: train (0:55)
• Sampling rate: 20 FPS
• Converted files: 1 TsFile file
• Original video streams: observation.images.base, observation.images.top, and observation.images.wrist

Converted layout

data/jetson1_062626_grab_doris.tsfile
meta/
  episodes/chunk-000/file-000.parquet
  info.json
  stats.json
  tasks.parquet
  vibedata/episode_sources.jsonl

The uploaded meta/info.json mirrors the source metadata and adds a tsfile_conversion object describing the TsFile conversion.

TsFile schema

Table name: jetson1_062626_grab_doris

• Time: millisecond timestamp, computed as round(timestamp * 1000).
• TAG columns: episode_index, task_index.
• FIELD columns: frame_index, sample_index, aux_elevator_mm, aux_spinner_active, aux_limit_home, aux_limit_gripper, action_0 to action_6, and observation_state_0 to observation_state_6.

The seven action_* and seven observation_state_* fields preserve the source joint order:

shoulder_pan.pos
shoulder_lift.pos
elbow_flex.pos
wrist_flex.pos
wrist_roll.pos
gripper.pos
tilt.pos

Conversion notes

• Source vector column action is flattened to action_0 through action_6.
• Source vector column observation.state is flattened to observation_state_0 through observation_state_6.
• Source scalar columns with dots are renamed for TsFile compatibility: aux.elevator_mm to aux_elevator_mm, aux.spinner_active to aux_spinner_active, aux.limit_home to aux_limit_home, and aux.limit_gripper to aux_limit_gripper.
• Source column timestamp is mapped to the TsFile Time column in milliseconds and is not retained as a separate field because it is equivalent to Time / 1000 seconds.
• Source column index is renamed to sample_index.
• Source video features observation.images.base, observation.images.top, and observation.images.wrist are omitted from this TsFile conversion. The original videos remain available in the source dataset under videos/.

Minimal read example

import pyarrow as pa
from tsfile import ColumnCategory, TsFileReader

path = "data/jetson1_062626_grab_doris.tsfile"
table_name = "jetson1_062626_grab_doris"

reader = TsFileReader(path)
schema = reader.get_all_table_schemas()[table_name]
columns = [
    c.get_column_name()
    for c in schema.get_columns()
    if c.get_category() in (ColumnCategory.FIELD, ColumnCategory.TAG)
]

batches = []
with reader.query_table(table_name, columns, batch_size=65536) as result_set:
    while True:
        batch = result_set.read_arrow_batch()
        if batch is None:
            break
        if batch.num_rows:
            batches.append(batch)

table = pa.concat_tables(batches)
df = table.to_pandas()
print(df.head())