---
license: apache-2.0
task_categories:
- robotics
pretty_name: ARX Left Cube YuHai HDF5 Dataset
size_categories:
- n<1K
tags:
- robotics
- arx
- teleoperation
- hdf5
- visuomotor
---

# ARX Left Cube YuHai HDF5 Dataset

This dataset contains ARX-X5 left-arm single-arm teleoperation episodes for a cube manipulation task.
Each episode is stored as one HDF5 file.


https://huggingface.co/datasets/Xia-2004/arx-left-cube


## Files

- `episode_000000.hdf5` ... `episode_000200.hdf5`
- 201 episodes
- 63,619 total frames
- RGB images are stored as `uint8`
- Actions are stored as `float32`

## HDF5 Format

Each `episode_*.hdf5` contains:

| Key | Shape | Dtype | Meaning |
| --- | --- | --- | --- |
| `action` | `(T, 5)` | `float32` | Delta end-effector action for each frame |
| `timestamp` | `(T,)` | `float64` | Collection timestamp for each frame |
| `pixels/cam_high` | `(T, 224, 224, 3)` | `uint8` | Third-person/top camera RGB frames |
| `pixels/cam_left_wrist` | `(T, 224, 224, 3)` | `uint8` | Left wrist RealSense RGB frames |
| `goal_pixels/cam_high` | `(224, 224, 3)` | `uint8` | Final goal image from `cam_high` |
| `goal_pixels/cam_left_wrist` | `(224, 224, 3)` | `uint8` | Final goal image from `cam_left_wrist` |
| `observations/pixels/cam_high` | soft link | - | Link to `pixels/cam_high` |
| `observations/pixels/cam_left_wrist` | soft link | - | Link to `pixels/cam_left_wrist` |

The action columns are:

```text
[dx, dy, dz, dyaw, d_gripper]
```

where:

- `dx`, `dy`, `dz`: delta end-effector translation
- `dyaw`: delta end-effector yaw
- `d_gripper`: delta gripper command

The HDF5 attribute `action_order` also records this column order.

## Python Loading Example

```python
import h5py

path = "episode_000000.hdf5"

with h5py.File(path, "r") as f:
    actions = f["action"][:]
    cam_high = f["pixels/cam_high"][:]
    cam_left_wrist = f["pixels/cam_left_wrist"][:]
    goal_high = f["goal_pixels/cam_high"][:]
    goal_left_wrist = f["goal_pixels/cam_left_wrist"][:]

print(actions.shape)
print(cam_high.shape)
print(goal_high.shape)
```

## Download

Install the Hugging Face Hub CLI:

```bash
pip install -U huggingface_hub
```

Download the full dataset:

```bash
huggingface-cli download arx-left-cube \
huggingface-cli download Xia-2004/arx-left-cube \
  --repo-type dataset \
  --local-dir ./arx-left-cube
```

Or with Python:

```python
from huggingface_hub import snapshot_download

snapshot_download(
    repo_id="Xia-2004/arx-left-cube",
    repo_type="dataset",
    local_dir="./arx-left-cube",
)
```