---
license: apache-2.0
datasets:
- lerobot/pusht
pipeline_tag: robotics
tags:
- robotics
---

# PushT Diffusion Policy - Robot Control Model

This model is an implementation of Diffusion Policy for the PushT environment, which simulates robotic pushing tasks.

## Model

This model uses a conditional diffusion architecture to predict robotic actions based on visual observations.

## Performance

The model achieves a success rate of 100.0% in the PushT environment with different initial configurations.

## Demonstration Videos

The repository includes demonstration videos in the `videos/` folder.

## Usage

```python
from lerobot.common.policies.diffusion.modeling_diffusion import DiffusionPolicy

policy = DiffusionPolicy.from_pretrained("RafaelJaime/pusht-diffusion")
```

Published on 2025-04-28