---
license: apache-2.0
task_categories:
- robotics
tags:
- LeRobot
- so101
- vision-language-action
- robotics
- manipulation
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
dataset_info:
  features:
  - name: observation.state
    sequence: float32
  - name: action
    sequence: float32
  - name: episode_index
    dtype: int64
  - name: frame_index
    dtype: int64
  - name: timestamp
    dtype: float32
  - name: next.done
    dtype: bool
  - name: index
    dtype: int64
  - name: task_index
    dtype: int64
  splits:
  - name: train
    num_bytes: 547591
    num_examples: 5944
  download_size: 237673
  dataset_size: 547591
---

# SO101 VLA Dataset v2.1

This dataset contains 20 episodes of robot demonstrations for the "pick up the black tape" task using the SO101 follower robot.

## Dataset Details

- **Format**: LeRobot v2.1
- **Episodes**: 20
- **Total Frames**: 5,944
- **FPS**: 30
- **Robot Type**: so101_follower
- **Task**: pick up the black tape

## Features

- **Actions**: 6D joint positions (shoulder_pan.pos, shoulder_lift.pos, elbow_flex.pos, wrist_flex.pos, wrist_roll.pos, gripper.pos)
- **States**: 6D joint positions (same as actions)
- **Images**: 
  - Top camera: 720x1280 RGB
  - Wrist camera: 240x320 RGB
- **Actions are UNNORMALIZED** (in degrees)

## Dataset Structure

```
data/chunk-000/episode_XXXXXX.parquet  # Per-episode data files
videos/chunk-000/observation.images.top/episode_XXXXXX.mp4
videos/chunk-000/observation.images.wrist/episode_XXXXXX.mp4
meta/info.json
meta/episodes.jsonl
meta/tasks.jsonl
meta/episodes_stats.jsonl
```

## Usage

```python
from lerobot.datasets.lerobot_dataset import LeRobotDataset

dataset = LeRobotDataset("abuod0/so101_vla_dataset_v21")
```

## Citation

```bibtex
@dataset{so101_vla_dataset_v21,
  title={SO101 VLA Dataset v2.1},
  author={Your Name},
  year={2024},
  url={https://huggingface.co/datasets/abuod0/so101_vla_dataset_v21}
}
```