so100_brick / README.md
XNX25's picture
chore: sync metadata, drop original files
02b0561 verified
|
Raw
History Blame Contribute Delete
2.34 kB
---
license: apache-2.0
task_categories:
- robotics
tags:
- LeRobot
- so100
- tutorial
- tsfile
- format:tsfile
- time-series
- timeseries
modality:
- tabular
- timeseries
configs:
- config_name: default
data_files:
- split: train
path: data/so100_brick.tsfile
size_categories:
- 10K<n<100K
---
# so100_brick (TsFile)
Apache TsFile version of [`mldev19/so100_brick`](https://huggingface.co/datasets/mldev19/so100_brick), a LeRobot v2.1 demonstration dataset recorded with an SO-100 arm.
## Overview
This dataset was created using [LeRobot](https://github.com/huggingface/lerobot). It records one SO-100 manipulation task (brick) with per-frame joint state and action vectors, plus two camera streams (`laptop`, `phone`).
- **Robot:** SO-100 (`so100`)
- **Scale:** 55 episodes, 16,279 frames, 1 task, 30 fps
- **Split:** train
## Schema (TsFile structure)
The TsFile table is named `so100_brick`.
| Role | Columns |
| --- | --- |
| Time | `Time`, INT64 milliseconds |
| TAG | `episode_index`, `task_index` |
| FIELD | `frame_index`, `sample_index` |
| FIELD | `action_0``action_5`, FLOAT |
| FIELD | `observation_state_0``observation_state_5`, FLOAT |
The original 6-element `action` and `observation.state` vectors are flattened into scalar FLOAT measurements. The source `index` column is retained as `sample_index`.
## Conversion notes
- `Time = round(timestamp * 1000)` with millisecond precision; time restarts inside each episode, while `episode_index` and `task_index` identify the TsFile device.
- The original `timestamp` field is omitted because it is exactly represented by `Time / 1000`.
- The two camera streams (`observation.images.laptop`, `observation.images.phone`) are video and are NOT included in this repository; they remain in the [source videos tree](https://huggingface.co/datasets/mldev19/so100_brick/tree/main/videos).
- `meta/` is mirrored from the source. Aside from the redundant `timestamp` column and the excluded videos, no source rows or numeric fields are dropped.
## Read example
```python
from tsfile import TsFileReader
with TsFileReader("data/so100_brick.tsfile") as reader:
print(reader.get_all_table_schemas()["so100_brick"])
```
## Source & license
- Original dataset: https://huggingface.co/datasets/mldev19/so100_brick
- Author: mldev19
- License: apache-2.0