Datasets:
SO100 PnP Block TsFile
This Apache TsFile dataset is derived from
samsitol/so100_PnPblock,
a LeRobot v2.1 SO100 robot-manipulation dataset.
Source Dataset
- Original dataset:
samsitol/so100_PnPblock - Author, repository owner, and sole contributor: Sam (
samsitol) - License: Apache-2.0
- Robot type:
so100 - LeRobot codebase version:
v2.1 - Task: Grasp the yellow block and put it in the fixed green bin.
- Split:
train - Scale: 50 episodes, 32,709 frames, 1 task, 30 fps
- Source frame files: 50 Parquet files
- Source frame layout:
data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet - The source card does not provide a paper or completed citation.
TsFile Data
- Path:
data/samsitol_so100_pnpblock.tsfile - Table:
samsitol_so100_pnpblock - Rows: 32,709
- Time precision: milliseconds
- Episodes are represented by TsFile TAG values rather than separate files.
Schema
Time is round(timestamp * 1000) milliseconds and restarts within each episode.
| Column | TsFile role | Type | Source mapping |
|---|---|---|---|
Time |
TIME | INT64 |
round(timestamp * 1000) |
episode_index |
TAG | STRING device/tag value |
Source episode_index |
task_index |
TAG | STRING device/tag value |
Source task_index |
frame_index |
FIELD | INT64 |
Preserved |
sample_index |
FIELD | INT64 |
Renamed from source index |
action_0 ... action_5 |
FIELD | FLOAT |
Flattened from action[6] |
observation_state_0 ... observation_state_5 |
FIELD | FLOAT |
Flattened from observation.state[6] |
The six action and state dimensions are main_shoulder_pan,
main_shoulder_lift, main_elbow_flex, main_wrist_flex,
main_wrist_roll, and main_gripper.
Conversion Notes
- All 50 episode Parquet files are merged into one table-model TsFile. Filter
by
episode_indexandtask_indexto select a device/episode. - Vector columns are flattened to scalar fields. Full source prefixes are
retained, with
.replaced by_. - The source
timestampcolumn is omitted after creatingTimebecause the same value is recoverable asTime / 1000seconds. - No numeric rows, episodes, action dimensions, or state dimensions are omitted.
- Numeric columns use the requested compact TsFile codecs with LZ4, while TAG values use the TsFile table/device mechanism. This source has no BOOLEAN field.
Videos
Videos are not included here. They remain in the original repository under
videos/chunk-000:
Each stream has one MP4 per episode, for 150 source videos total. Numeric rows
remain frame-aligned through episode_index and frame_index.
Usage
from tsfile import TsFileReader
reader = TsFileReader("data/samsitol_so100_pnpblock.tsfile")
table_name = "samsitol_so100_pnpblock"
columns = [
"episode_index",
"task_index",
"frame_index",
"sample_index",
"action_0",
"observation_state_0",
]
with reader.query_table(table_name, columns, batch_size=65536) as result:
batch = result.read_arrow_batch()
print(batch.to_pandas().head())
- Downloads last month
- 46