Datasets:
pick_tblock_mp_pdscale_20_new2 TsFile
This dataset is a TsFile conversion of the LeRobot dataset
younghyopark/pick_tblock_mp_pdscale_20_new2.
The source dataset was created with LeRobot
and is licensed under apache-2.0.
Dataset Details
- Source dataset:
younghyopark/pick_tblock_mp_pdscale_20_new2 - Source format: LeRobot
v2.1 - Robot type:
DualPanda - Modalities: Time-series robot state and action data
- Split:
train - Episodes:
1000 - Frames / rows:
153000 - Tasks:
1 - Sampling rate:
50 fps - Source videos:
0(video_path: null) - Converted data file:
data/pick_tblock_mp_pdscale_20_new2_train.tsfile
TsFile Layout
The conversion writes one TsFile table for the train split:
pick_tblock_mp_pdscale_20_new2_train.
Timeis the TsFile time column in milliseconds.episode_indexandtask_indexare TAG columns.frame_indexandsample_indexare scalar FIELD columns.- Vector features are flattened into scalar FIELD columns.
Vector column flattening preserves the source column name, replaces . with
_, and appends the element index. For example, observation.state becomes
observation_state_0 through observation_state_29.
Flattened source vectors:
observation.state: 30 float32 fieldsobservation.environment_state: 7 float32 fieldsaction.relative_ee_pose: 16 float32 fieldsobservation.qvel: 30 float32 fieldsobservation.env_qvel: 6 float32 fieldsobservation.ee_pose: 16 float32 fieldsaction.ctrl: 16 float32 fieldsaction.qpos: 16 float32 fieldsaction.relative_qpos: 16 float32 fieldsaction.ee_pose: 16 float32 fields
Conversion Notes
Time = round(timestamp * 1000)and restarts per episode.- The source
timestampcolumn is dropped because it is represented byTime / 1000seconds. - The source
indexcolumn is renamed tosample_index. - No numeric time-series rows are dropped; the converted TsFile has
153000rows. - Source
meta/files are mirrored, withmeta/info.jsonupdated to describe the converted TsFile artifact. - No videos are uploaded here. The source metadata reports
total_videos: 0andvideo_path: null.
Read Example
Use the Apache TsFile SDK or TsFile tools to read:
data/pick_tblock_mp_pdscale_20_new2_train.tsfile
For episode-level filtering, query the table with the TAG column, for example
WHERE episode_index = 0.
- Downloads last month
- 98