Hugging Face's logo

Datasets:
jzpeng
/
team13_two_balls_stacking

Tasks:
Robotics
Modalities:
Time-series
Tags:
LeRobot
so100
hackathon
tsfile
timeseries
License:
Dataset card Files
xet
 Community
team13_two_balls_stacking / README.md
jzpeng's picture
jzpeng
Add TsFile (converted from roboticshack/team13-two-balls-stacking)
4b3d378 verified
preview code
|
Raw
History Blame Contribute Delete
3.84 kB
metadata 
license: apache-2.0
task_categories:
  - robotics
tags:
  - LeRobot
  - so100
  - hackathon
  - tsfile
  - timeseries
modality:
  - timeseries
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/team13_two_balls_stacking_train.tsfile

Team 13 Two-Balls Stacking (TsFile)

Converted from roboticshack/team13-two-balls-stacking at pinned revision 7df39c37cad86c865ae9194461160824afd46d4a. Modalities: Time-series.

Dataset description

This LeRobot v2.1 SO100 dataset records demonstrations of stacking two balls on top of each other.

The pinned task instruction is: “Stack the balls on top of each other.”

  • Source repository owner/publisher: roboticshack
  • License: apache-2.0 (declared by the source card).
  • Paper/homepage/citation: no completed paper, homepage, or citation is documented in the pinned source card unless linked above.

Dataset Scale

Split Episodes Tasks Source trajectory Parquets TsFile rows Sampling rate TsFile files/shards
train 51 1 51 13,476 30 Hz 1

The staged Parquet has 13,476 rows and 17 columns including Time; TsFile chunk metadata independently reports the same 13,476 rows.

TsFile schema

Column Role TsFile type Observed/source range
Time TIME INT64 0–11,767 ms; restarts per episode
episode_index TAG STRING source integer 0–50
task_index TAG STRING source integer 0–0
frame_index FIELD INT64 0–353 within an episode
sample_index FIELD INT64 0–13,475 globally

Exact remaining FIELD names/ranges and imported types:

  • action_0–action_5 (FLOAT): main shoulder pan/lift, elbow flex, wrist flex/roll, and gripper action
  • observation_state_0–observation_state_5 (FLOAT): the corresponding SO100 state

Conversion

  • All source episodes in the train split are merged into data/team13_two_balls_stacking_train.tsfile; episode_index and task_index are TAG dimensions.
  • Time = round(timestamp * 1000) in milliseconds. The source timestamp column is omitted because it is redundant with Time / 1000 seconds.
  • frame_index is retained. Source index is retained as sample_index.
  • Every vector is fully flattened: the complete source name is kept, . becomes _, and element indices are appended. Float vectors are imported as single-precision FLOAT fields.
  • Other scalar source columns shown above are retained; no trajectory rows are intentionally dropped.
  • Source metadata is mirrored for publication, with copied meta/info.json rewritten to describe the converted data path and conversion semantics.

Video policy

The pinned metadata declares 102 source videos across side and RealSense camera streams. MP4 files were not downloaded or uploaded; they remain in the pinned source videos/ tree.

Minimal read example 

from tsfile import TsFileReader

reader = TsFileReader("data/team13_two_balls_stacking_train.tsfile")
print(reader.get_all_table_schemas().keys())
reader.close()

Source and provenance