Merge branch 'main' into user/azouitine/2025-7-2-implement-pipeline

This commit is contained in:
Steven Palma
2025-08-06 14:15:01 +02:00
committed by Steven Palma
31 changed files with 1623 additions and 207 deletions

View File

@@ -19,6 +19,11 @@ on:
tags:
- 'v*.*.*' # Trigger on tags like v0.1.0, v1.0.0
# Sets up the environment variables
env:
UV_VERSION: "0.8.0"
PYTHON_VERSION: "3.10"
jobs:
# This job builds the Python package and publishes it to PyPI
build-and-publish:
@@ -50,6 +55,7 @@ jobs:
VERSION_NUMBER=${VERSION#v}
echo "tag_version=$VERSION_NUMBER" >> $GITHUB_OUTPUT
- name: Check if version matches pyproject.toml
if: startsWith(github.ref, 'refs/tags/v') && !contains(github.ref, '-')
# zizmor: ignore[template-injection]
run: |
TAG_VERSION=${{ steps.extract_info.outputs.tag_version }}
@@ -86,13 +92,29 @@ jobs:
env:
GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
# zizmor: ignore[template-injection]
run: gh release create ${{ github.ref_name }} --release-name "Release ${{ github.ref_name }}" --generate-notes ./dist/*
run: |
gh release create ${{ github.ref_name }} \
--title "Release ${{ github.ref_name }}" \
--generate-notes \
--draft=$([[ "${{ github.ref_name }}" == *-* ]] && echo true || echo false) \
--prerelease=$([[ "${{ github.ref_name }}" == *-* ]] && echo true || echo false) \
./dist/*
- name: Publish to PyPI
if: startsWith(github.ref, 'refs/tags/v')
- name: Publish to TestPyPI for pre-releases
# True for tags like 'v0.2.0-rc1'
if: startsWith(github.ref, 'refs/tags/v') && contains(github.ref, '-')
uses: pypa/gh-action-pypi-publish@v1.12.4 # zizmor: ignore[unpinned-uses, use-trusted-publishing]
with:
password: ${{ secrets.PYPI_API_TOKEN }}
repository-url: https://test.pypi.org/legacy/
verbose: true
print-hash: true
- name: Publish to PyPI
if: startsWith(github.ref, 'refs/tags/v') && !contains(github.ref, '-')
uses: pypa/gh-action-pypi-publish@v1.12.4 # zizmor: ignore[unpinned-uses, use-trusted-publishing]
with:
verbose: true
print-hash: true
# This job runs end-to-end tests on the release
test-release:
@@ -119,15 +141,31 @@ jobs:
enable-cache: true
version: ${{ env.UV_VERSION }}
python-version: ${{ env.PYTHON_VERSION }}
- name: Create uv virtual environment
run: uv venv
- name: Install lerobot release
run: uv run pip install lerobot==${{ needs.build-and-publish.outputs.version }} # zizmor: ignore[template-injection]
# zizmor: ignore[template-injection]
run: |
VERSION="${{ needs.build-and-publish.outputs.version }}"
if [[ "$VERSION" == *-* ]]; then
BASE_VERSION="${VERSION%%-*}"
echo "Installing pre-release version $BASE_VERSION from TestPyPI..."
uv pip install \
--index-url https://test.pypi.org/simple/ \
--extra-index-url https://pypi.org/simple \
--index-strategy unsafe-best-match \
"lerobot[all]==$BASE_VERSION"
else
echo "Installing release version $VERSION from PyPI..."
uv pip install "lerobot[all]==$VERSION"
fi
- name: Check lerobot version
run: uv run lerobot --version
run: uv run python -c "import lerobot; print(lerobot.__version__)"
- name: Run end-to-end tests
run: uv run make test-end-to-end
# TODO(Steven): Publish draft/pre-release and to test pypi
# TODO(Steven): Publish draft/pre-release and to test pypi weekly
# TODO(Steven): Separate build and publish job
# TODO(Steven): Tag documentation with the same version as the package

241
README.md
View File

@@ -1,25 +1,21 @@
<p align="center">
<picture>
<source media="(prefers-color-scheme: dark)" srcset="media/lerobot-logo-thumbnail.png">
<source media="(prefers-color-scheme: light)" srcset="media/lerobot-logo-thumbnail.png">
<img alt="LeRobot, Hugging Face Robotics Library" src="media/lerobot-logo-thumbnail.png" style="max-width: 100%;">
</picture>
<img alt="LeRobot, Hugging Face Robotics Library" src="https://raw.githubusercontent.com/huggingface/lerobot/main/media/lerobot-logo-thumbnail.png" width="100%">
<br/>
<br/>
</p>
<div align="center">
[![Tests](https://github.com/huggingface/lerobot/actions/workflows/nightly-tests.yml/badge.svg?branch=main)](https://github.com/huggingface/lerobot/actions/workflows/nightly-tests.yml?query=branch%3Amain)
[![Coverage](https://codecov.io/gh/huggingface/lerobot/branch/main/graph/badge.svg?token=TODO)](https://codecov.io/gh/huggingface/lerobot)
[![Tests](https://github.com/huggingface/lerobot/actions/workflows/nightly.yml/badge.svg?branch=main)](https://github.com/huggingface/lerobot/actions/workflows/nighty.yml?query=branch%3Amain)
[![Python versions](https://img.shields.io/pypi/pyversions/lerobot)](https://www.python.org/downloads/)
[![License](https://img.shields.io/badge/License-Apache%202.0-blue.svg)](https://github.com/huggingface/lerobot/blob/main/LICENSE)
[![Status](https://img.shields.io/pypi/status/lerobot)](https://pypi.org/project/lerobot/)
[![Version](https://img.shields.io/pypi/v/lerobot)](https://pypi.org/project/lerobot/)
[![Examples](https://img.shields.io/badge/Examples-green.svg)](https://github.com/huggingface/lerobot/tree/main/examples)
[![Contributor Covenant](https://img.shields.io/badge/Contributor%20Covenant-v2.1%20adopted-ff69b4.svg)](https://github.com/huggingface/lerobot/blob/main/CODE_OF_CONDUCT.md)
[![Contributor Covenant](https://img.shields.io/badge/Contributor%20Covenant-v2.1-ff69b4.svg)](https://github.com/huggingface/lerobot/blob/main/CODE_OF_CONDUCT.md)
[![Discord](https://dcbadge.vercel.app/api/server/C5P34WJ68S?style=flat)](https://discord.gg/s3KuuzsPFb)
<!-- [![Coverage](https://codecov.io/gh/huggingface/lerobot/branch/main/graph/badge.svg?token=TODO)](https://codecov.io/gh/huggingface/lerobot) -->
</div>
<h2 align="center">
@@ -29,10 +25,10 @@
<div align="center">
<img
src="media/hope_jr/hopejr.png?raw=true"
src="https://raw.githubusercontent.com/huggingface/lerobot/main/media/hope_jr/hopejr.png"
alt="HopeJR robot"
title="HopeJR robot"
style="width: 60%;"
width="60%"
/>
<p><strong>Meet HopeJR A humanoid robot arm and hand for dexterous manipulation!</strong></p>
@@ -51,20 +47,12 @@
</h2>
<div align="center">
<div style="display: flex; gap: 1rem; justify-content: center; align-items: center;" >
<img
src="media/so101/so101.webp?raw=true"
alt="SO-101 follower arm"
title="SO-101 follower arm"
style="width: 40%;"
/>
<img
src="media/so101/so101-leader.webp?raw=true"
alt="SO-101 leader arm"
title="SO-101 leader arm"
style="width: 40%;"
/>
</div>
<table>
<tr>
<td align="center"><img src="https://raw.githubusercontent.com/huggingface/lerobot/main/media/so101/so101.webp" alt="SO-101 follower arm" title="SO-101 follower arm" width="90%"/></td>
<td align="center"><img src="https://raw.githubusercontent.com/huggingface/lerobot/main/media/so101/so101-leader.webp" alt="SO-101 leader arm" title="SO-101 leader arm" width="90%"/></td>
</tr>
</table>
<p><strong>Meet the updated SO100, the SO-101 Just €114 per arm!</strong></p>
<p>Train it in minutes with a few simple moves on your laptop.</p>
@@ -76,7 +64,7 @@
<p>Want to take it to the next level? Make your SO-101 mobile by building LeKiwi!</p>
<p>Check out the <a href="https://huggingface.co/docs/lerobot/lekiwi">LeKiwi tutorial</a> and bring your robot to life on wheels.</p>
<img src="media/lekiwi/kiwi.webp?raw=true" alt="LeKiwi mobile robot" title="LeKiwi mobile robot" width="50%">
<img src="https://raw.githubusercontent.com/huggingface/lerobot/main/media/lekiwi/kiwi.webp" alt="LeKiwi mobile robot" title="LeKiwi mobile robot" width="50%">
</div>
<br/>
@@ -99,9 +87,9 @@
<table>
<tr>
<td><img src="media/gym/aloha_act.gif" width="100%" alt="ACT policy on ALOHA env"/></td>
<td><img src="media/gym/simxarm_tdmpc.gif" width="100%" alt="TDMPC policy on SimXArm env"/></td>
<td><img src="media/gym/pusht_diffusion.gif" width="100%" alt="Diffusion policy on PushT env"/></td>
<td><img src="https://raw.githubusercontent.com/huggingface/lerobot/main/media/gym/aloha_act.gif" width="100%" alt="ACT policy on ALOHA env"/></td>
<td><img src="https://raw.githubusercontent.com/huggingface/lerobot/main/media/gym/simxarm_tdmpc.gif" width="100%" alt="TDMPC policy on SimXArm env"/></td>
<td><img src="https://raw.githubusercontent.com/huggingface/lerobot/main/media/gym/pusht_diffusion.gif" width="100%" alt="Diffusion policy on PushT env"/></td>
</tr>
<tr>
<td align="center">ACT policy on ALOHA env</td>
@@ -110,23 +98,11 @@
</tr>
</table>
### Acknowledgment
- The LeRobot team 🤗 for building SmolVLA [Paper](https://arxiv.org/abs/2506.01844), [Blog](https://huggingface.co/blog/smolvla).
- Thanks to Tony Zhao, Zipeng Fu and colleagues for open sourcing ACT policy, ALOHA environments and datasets. Ours are adapted from [ALOHA](https://tonyzhaozh.github.io/aloha) and [Mobile ALOHA](https://mobile-aloha.github.io).
- Thanks to Cheng Chi, Zhenjia Xu and colleagues for open sourcing Diffusion policy, Pusht environment and datasets, as well as UMI datasets. Ours are adapted from [Diffusion Policy](https://diffusion-policy.cs.columbia.edu) and [UMI Gripper](https://umi-gripper.github.io).
- Thanks to Nicklas Hansen, Yunhai Feng and colleagues for open sourcing TDMPC policy, Simxarm environments and datasets. Ours are adapted from [TDMPC](https://github.com/nicklashansen/tdmpc) and [FOWM](https://www.yunhaifeng.com/FOWM).
- Thanks to Antonio Loquercio and Ashish Kumar for their early support.
- Thanks to [Seungjae (Jay) Lee](https://sjlee.cc/), [Mahi Shafiullah](https://mahis.life/) and colleagues for open sourcing [VQ-BeT](https://sjlee.cc/vq-bet/) policy and helping us adapt the codebase to our repository. The policy is adapted from [VQ-BeT repo](https://github.com/jayLEE0301/vq_bet_official).
## Installation
Download our source code:
LeRobot works with Python 3.10+ and PyTorch 2.2+.
```bash
git clone https://github.com/huggingface/lerobot.git
cd lerobot
```
### Environment Setup
Create a virtual environment with Python 3.10 and activate it, e.g. with [`miniconda`](https://docs.anaconda.com/free/miniconda/index.html):
@@ -151,7 +127,18 @@ conda install ffmpeg -c conda-forge
>
> - _[On Linux only]_ Install [ffmpeg build dependencies](https://trac.ffmpeg.org/wiki/CompilationGuide/Ubuntu#GettheDependencies) and [compile ffmpeg from source with libsvtav1](https://trac.ffmpeg.org/wiki/CompilationGuide/Ubuntu#libsvtav1), and make sure you use the corresponding ffmpeg binary to your install with `which ffmpeg`.
Install 🤗 LeRobot:
### Install LeRobot 🤗
#### From Source
First, clone the repository and navigate into the directory:
```bash
git clone https://github.com/huggingface/lerobot.git
cd lerobot
```
Then, install the library in editable mode. This is useful if you plan to contribute to the code.
```bash
pip install -e .
@@ -172,6 +159,34 @@ For instance, to install 🤗 LeRobot with aloha and pusht, use:
pip install -e ".[aloha, pusht]"
```
### Installation from PyPI
**Core Library:**
Install the base package with:
```bash
pip install lerobot
```
_This installs only the default dependencies._
**Extra Features:**
To install additional functionality, use one of the following:
```bash
pip install 'lerobot[all]' # All available features
pip install 'lerobot[aloha,pusht]' # Specific features (Aloha & Pusht)
pip install 'lerobot[feetech]' # Feetech motor support
```
_Replace `[...]` with your desired features._
**Available Tags:**
For a full list of optional dependencies, see:
https://pypi.org/project/lerobot/
### Weights & Biases
To use [Weights and Biases](https://docs.wandb.ai/quickstart) for experiment tracking, log in with
```bash
@@ -182,7 +197,7 @@ wandb login
### Visualize datasets
Check out [example 1](./examples/1_load_lerobot_dataset.py) that illustrates how to use our dataset class which automatically downloads data from the Hugging Face hub.
Check out [example 1](https://github.com/huggingface/lerobot/blob/main/examples/1_load_lerobot_dataset.py) that illustrates how to use our dataset class which automatically downloads data from the Hugging Face hub.
You can also locally visualize episodes from a dataset on the hub by executing our script from the command line:
@@ -212,7 +227,7 @@ Our script can also visualize datasets stored on a distant server. See `python -
A dataset in `LeRobotDataset` format is very simple to use. It can be loaded from a repository on the Hugging Face hub or a local folder simply with e.g. `dataset = LeRobotDataset("lerobot/aloha_static_coffee")` and can be indexed into like any Hugging Face and PyTorch dataset. For instance `dataset[0]` will retrieve a single temporal frame from the dataset containing observation(s) and an action as PyTorch tensors ready to be fed to a model.
A specificity of `LeRobotDataset` is that, rather than retrieving a single frame by its index, we can retrieve several frames based on their temporal relationship with the indexed frame, by setting `delta_timestamps` to a list of relative times with respect to the indexed frame. For example, with `delta_timestamps = {"observation.image": [-1, -0.5, -0.2, 0]}` one can retrieve, for a given index, 4 frames: 3 "previous" frames 1 second, 0.5 seconds, and 0.2 seconds before the indexed frame, and the indexed frame itself (corresponding to the 0 entry). See example [1_load_lerobot_dataset.py](examples/1_load_lerobot_dataset.py) for more details on `delta_timestamps`.
A specificity of `LeRobotDataset` is that, rather than retrieving a single frame by its index, we can retrieve several frames based on their temporal relationship with the indexed frame, by setting `delta_timestamps` to a list of relative times with respect to the indexed frame. For example, with `delta_timestamps = {"observation.image": [-1, -0.5, -0.2, 0]}` one can retrieve, for a given index, 4 frames: 3 "previous" frames 1 second, 0.5 seconds, and 0.2 seconds before the indexed frame, and the indexed frame itself (corresponding to the 0 entry). See example [1_load_lerobot_dataset.py](https://github.com/huggingface/lerobot/blob/main/examples/1_load_lerobot_dataset.py) for more details on `delta_timestamps`.
Under the hood, the `LeRobotDataset` format makes use of several ways to serialize data which can be useful to understand if you plan to work more closely with this format. We tried to make a flexible yet simple dataset format that would cover most type of features and specificities present in reinforcement learning and robotics, in simulation and in real-world, with a focus on cameras and robot states but easily extended to other types of sensory inputs as long as they can be represented by a tensor.
@@ -256,7 +271,7 @@ Dataset can be uploaded/downloaded from the HuggingFace hub seamlessly. To work
### Evaluate a pretrained policy
Check out [example 2](./examples/2_evaluate_pretrained_policy.py) that illustrates how to download a pretrained policy from Hugging Face hub, and run an evaluation on its corresponding environment.
Check out [example 2](https://github.com/huggingface/lerobot/blob/main/examples/2_evaluate_pretrained_policy.py) that illustrates how to download a pretrained policy from Hugging Face hub, and run an evaluation on its corresponding environment.
We also provide a more capable script to parallelize the evaluation over multiple environments during the same rollout. Here is an example with a pretrained model hosted on [lerobot/diffusion_pusht](https://huggingface.co/lerobot/diffusion_pusht):
@@ -280,13 +295,13 @@ See `python -m lerobot.scripts.eval --help` for more instructions.
### Train your own policy
Check out [example 3](./examples/3_train_policy.py) that illustrates how to train a model using our core library in python, and [example 4](./examples/4_train_policy_with_script.md) that shows how to use our training script from command line.
Check out [example 3](https://github.com/huggingface/lerobot/blob/main/examples/3_train_policy.py) that illustrates how to train a model using our core library in python, and [example 4](https://github.com/huggingface/lerobot/blob/main/examples/4_train_policy_with_script.md) that shows how to use our training script from command line.
To use wandb for logging training and evaluation curves, make sure you've run `wandb login` as a one-time setup step. Then, when running the training command above, enable WandB in the configuration by adding `--wandb.enable=true`.
A link to the wandb logs for the run will also show up in yellow in your terminal. Here is an example of what they look like in your browser. Please also check [here](./examples/4_train_policy_with_script.md#typical-logs-and-metrics) for the explanation of some commonly used metrics in logs.
A link to the wandb logs for the run will also show up in yellow in your terminal. Here is an example of what they look like in your browser. Please also check [here](https://github.com/huggingface/lerobot/blob/main/examples/4_train_policy_with_script.md#typical-logs-and-metrics) for the explanation of some commonly used metrics in logs.
![](media/wandb.png)
\<img src="https://raw.githubusercontent.com/huggingface/lerobot/main/media/wandb.png" alt="WandB logs example"\>
Note: For efficiency, during training every checkpoint is evaluated on a low number of episodes. You may use `--eval.n_episodes=500` to evaluate on more episodes than the default. Or, after training, you may want to re-evaluate your best checkpoints on more episodes or change the evaluation settings. See `python -m lerobot.scripts.eval --help` for more instructions.
@@ -305,26 +320,6 @@ reproduces SOTA results for Diffusion Policy on the PushT task.
If you would like to contribute to 🤗 LeRobot, please check out our [contribution guide](https://github.com/huggingface/lerobot/blob/main/CONTRIBUTING.md).
<!-- ### Add a new dataset
To add a dataset to the hub, you need to login using a write-access token, which can be generated from the [Hugging Face settings](https://huggingface.co/settings/tokens):
```bash
huggingface-cli login --token ${HUGGINGFACE_TOKEN} --add-to-git-credential
```
Then point to your raw dataset folder (e.g. `data/aloha_static_pingpong_test_raw`), and push your dataset to the hub with:
```bash
python lerobot/scripts/push_dataset_to_hub.py \
--raw-dir data/aloha_static_pingpong_test_raw \
--out-dir data \
--repo-id lerobot/aloha_static_pingpong_test \
--raw-format aloha_hdf5
```
See `python lerobot/scripts/push_dataset_to_hub.py --help` for more instructions.
If your dataset format is not supported, implement your own in `lerobot/datasets/push_dataset_to_hub/${raw_format}_format.py` by copying examples like [pusht_zarr](https://github.com/huggingface/lerobot/blob/main/lerobot/datasets/push_dataset_to_hub/pusht_zarr_format.py), [umi_zarr](https://github.com/huggingface/lerobot/blob/main/lerobot/datasets/push_dataset_to_hub/umi_zarr_format.py), [aloha_hdf5](https://github.com/huggingface/lerobot/blob/main/lerobot/datasets/push_dataset_to_hub/aloha_hdf5_format.py), or [xarm_pkl](https://github.com/huggingface/lerobot/blob/main/lerobot/datasets/push_dataset_to_hub/xarm_pkl_format.py). -->
### Add a pretrained policy
Once you have trained a policy you may upload it to the Hugging Face hub using a hub id that looks like `${hf_user}/${repo_name}` (e.g. [lerobot/diffusion_pusht](https://huggingface.co/lerobot/diffusion_pusht)).
@@ -341,34 +336,16 @@ To upload these to the hub, run the following:
huggingface-cli upload ${hf_user}/${repo_name} path/to/pretrained_model
```
See [eval.py](https://github.com/huggingface/lerobot/blob/main/lerobot/scripts/eval.py) for an example of how other people may use your policy.
See [eval.py](https://github.com/huggingface/lerobot/blob/main/src/lerobot/scripts/eval.py) for an example of how other people may use your policy.
### Improve your code with profiling
### Acknowledgment
An example of a code snippet to profile the evaluation of a policy:
<!-- prettier-ignore-start -->
```python
from torch.profiler import profile, record_function, ProfilerActivity
def trace_handler(prof):
prof.export_chrome_trace(f"tmp/trace_schedule_{prof.step_num}.json")
with profile(
activities=[ProfilerActivity.CPU, ProfilerActivity.CUDA],
schedule=torch.profiler.schedule(
wait=2,
warmup=2,
active=3,
),
on_trace_ready=trace_handler
) as prof:
with record_function("eval_policy"):
for i in range(num_episodes):
prof.step()
# insert code to profile, potentially whole body of eval_policy function
```
<!-- prettier-ignore-end -->
- The LeRobot team 🤗 for building SmolVLA [Paper](https://arxiv.org/abs/2506.01844), [Blog](https://huggingface.co/blog/smolvla).
- Thanks to Tony Zhao, Zipeng Fu and colleagues for open sourcing ACT policy, ALOHA environments and datasets. Ours are adapted from [ALOHA](https://tonyzhaozh.github.io/aloha) and [Mobile ALOHA](https://mobile-aloha.github.io).
- Thanks to Cheng Chi, Zhenjia Xu and colleagues for open sourcing Diffusion policy, Pusht environment and datasets, as well as UMI datasets. Ours are adapted from [Diffusion Policy](https://diffusion-policy.cs.columbia.edu) and [UMI Gripper](https://umi-gripper.github.io).
- Thanks to Nicklas Hansen, Yunhai Feng and colleagues for open sourcing TDMPC policy, Simxarm environments and datasets. Ours are adapted from [TDMPC](https://github.com/nicklashansen/tdmpc) and [FOWM](https://www.yunhaifeng.com/FOWM).
- Thanks to Antonio Loquercio and Ashish Kumar for their early support.
- Thanks to [Seungjae (Jay) Lee](https://sjlee.cc/), [Mahi Shafiullah](https://mahis.life/) and colleagues for open sourcing [VQ-BeT](https://sjlee.cc/vq-bet/) policy and helping us adapt the codebase to our repository. The policy is adapted from [VQ-BeT repo](https://github.com/jayLEE0301/vq_bet_official).
## Citation
@@ -376,83 +353,13 @@ If you want, you can cite this work with:
```bibtex
@misc{cadene2024lerobot,
author = {Cadene, Remi and Alibert, Simon and Soare, Alexander and Gallouedec, Quentin and Zouitine, Adil and Palma, Steven and Kooijmans, Pepijn and Aractingi, Michel and Shukor, Mustafa and Aubakirova, Dana and Russi, Martino and Capuano, Francesco and Pascale, Caroline and Choghari, Jade and Moss, Jess and Wolf, Thomas},
author = {Cadene, Remi and Alibert, Simon and Soare, Alexander and Gallouedec, Quentin and Zouitine, Adil and Palma, Steven and Kooijmans, Pepijn and Aractingi, Michel and Shukor, Mustafa and Aubakirova, Dana and Russi, Martino and Capuano, Francesco and Pascal, Caroline and Choghari, Jade and Moss, Jess and Wolf, Thomas},
title = {LeRobot: State-of-the-art Machine Learning for Real-World Robotics in Pytorch},
howpublished = "\url{https://github.com/huggingface/lerobot}",
year = {2024}
}
```
Additionally, if you are using any of the particular policy architecture, pretrained models, or datasets, it is recommended to cite the original authors of the work as they appear below:
- [SmolVLA](https://arxiv.org/abs/2506.01844)
```bibtex
@article{shukor2025smolvla,
title={SmolVLA: A Vision-Language-Action Model for Affordable and Efficient Robotics},
author={Shukor, Mustafa and Aubakirova, Dana and Capuano, Francesco and Kooijmans, Pepijn and Palma, Steven and Zouitine, Adil and Aractingi, Michel and Pascal, Caroline and Russi, Martino and Marafioti, Andres and Alibert, Simon and Cord, Matthieu and Wolf, Thomas and Cadene, Remi},
journal={arXiv preprint arXiv:2506.01844},
year={2025}
}
```
- [Diffusion Policy](https://diffusion-policy.cs.columbia.edu)
```bibtex
@article{chi2024diffusionpolicy,
author = {Cheng Chi and Zhenjia Xu and Siyuan Feng and Eric Cousineau and Yilun Du and Benjamin Burchfiel and Russ Tedrake and Shuran Song},
title ={Diffusion Policy: Visuomotor Policy Learning via Action Diffusion},
journal = {The International Journal of Robotics Research},
year = {2024},
}
```
- [ACT or ALOHA](https://tonyzhaozh.github.io/aloha)
```bibtex
@article{zhao2023learning,
title={Learning fine-grained bimanual manipulation with low-cost hardware},
author={Zhao, Tony Z and Kumar, Vikash and Levine, Sergey and Finn, Chelsea},
journal={arXiv preprint arXiv:2304.13705},
year={2023}
}
```
- [TDMPC](https://www.nicklashansen.com/td-mpc/)
```bibtex
@inproceedings{Hansen2022tdmpc,
title={Temporal Difference Learning for Model Predictive Control},
author={Nicklas Hansen and Xiaolong Wang and Hao Su},
booktitle={ICML},
year={2022}
}
```
- [VQ-BeT](https://sjlee.cc/vq-bet/)
```bibtex
@article{lee2024behavior,
title={Behavior generation with latent actions},
author={Lee, Seungjae and Wang, Yibin and Etukuru, Haritheja and Kim, H Jin and Shafiullah, Nur Muhammad Mahi and Pinto, Lerrel},
journal={arXiv preprint arXiv:2403.03181},
year={2024}
}
```
- [HIL-SERL](https://hil-serl.github.io/)
```bibtex
@Article{luo2024hilserl,
title={Precise and Dexterous Robotic Manipulation via Human-in-the-Loop Reinforcement Learning},
author={Jianlan Luo and Charles Xu and Jeffrey Wu and Sergey Levine},
year={2024},
eprint={2410.21845},
archivePrefix={arXiv},
primaryClass={cs.RO}
}
```
## Star History
[![Star History Chart](https://api.star-history.com/svg?repos=huggingface/lerobot&type=Timeline)](https://star-history.com/#huggingface/lerobot&Timeline)

3
docs-requirements.txt Normal file
View File

@@ -0,0 +1,3 @@
# docs-requirements.txt
hf-doc-builder @ git+https://github.com/huggingface/doc-builder.git@main
watchdog>=6.0.0

View File

@@ -20,7 +20,7 @@ To generate the documentation, you first have to build it. Several packages are
you can install them with the following command, at the root of the code repository:
```bash
pip install -e ".[docs]"
pip install -e . -r docs-requirements.txt
```
You will also need `nodejs`. Please refer to their [installation page](https://nodejs.org/en/download)

View File

@@ -294,7 +294,7 @@ dataset.push_to_hub()
#### Dataset upload
Locally, your dataset is stored in this folder: `~/.cache/huggingface/lerobot/{repo-id}`. At the end of data recording, your dataset will be uploaded on your Hugging Face page (e.g. https://huggingface.co/datasets/cadene/so101_test) that you can obtain by running:
Locally, your dataset is stored in this folder: `~/.cache/huggingface/lerobot/{repo-id}`. At the end of data recording, your dataset will be uploaded on your Hugging Face page (e.g. `https://huggingface.co/datasets/${HF_USER}/so101_test`) that you can obtain by running:
```bash
echo https://huggingface.co/datasets/${HF_USER}/so101_test
@@ -428,7 +428,7 @@ Your robot should replicate movements similar to those you recorded. For example
## Train a policy
To train a policy to control your robot, use the [`python -m lerobot.scripts.train`](../src/lerobot/scripts/train.py) script. A few arguments are required. Here is an example command:
To train a policy to control your robot, use the [`python -m lerobot.scripts.train`](https://github.com/huggingface/lerobot/blob/main/src/lerobot/scripts/train.py) script. A few arguments are required. Here is an example command:
```bash
python -m lerobot.scripts.train \
@@ -444,7 +444,7 @@ python -m lerobot.scripts.train \
Let's explain the command:
1. We provided the dataset as argument with `--dataset.repo_id=${HF_USER}/so101_test`.
2. We provided the policy with `policy.type=act`. This loads configurations from [`configuration_act.py`](../src/lerobot/policies/act/configuration_act.py). Importantly, this policy will automatically adapt to the number of motor states, motor actions and cameras of your robot (e.g. `laptop` and `phone`) which have been saved in your dataset.
2. We provided the policy with `policy.type=act`. This loads configurations from [`configuration_act.py`](https://github.com/huggingface/lerobot/blob/main/src/lerobot/policies/act/configuration_act.py). Importantly, this policy will automatically adapt to the number of motor states, motor actions and cameras of your robot (e.g. `laptop` and `phone`) which have been saved in your dataset.
3. We provided `policy.device=cuda` since we are training on a Nvidia GPU, but you could use `policy.device=mps` to train on Apple silicon.
4. We provided `wandb.enable=true` to use [Weights and Biases](https://docs.wandb.ai/quickstart) for visualizing training plots. This is optional but if you use it, make sure you are logged in by running `wandb login`.

View File

@@ -96,7 +96,7 @@ If you uploaded your dataset to the hub you can [visualize your dataset online](
## Train a policy
To train a policy to control your robot, use the [`python -m lerobot.scripts.train`](../src/lerobot/scripts/train.py) script. A few arguments are required. Here is an example command:
To train a policy to control your robot, use the [`python -m lerobot.scripts.train`](https://github.com/huggingface/lerobot/blob/main/src/lerobot/scripts/train.py) script. A few arguments are required. Here is an example command:
```bash
python -m lerobot.scripts.train \
@@ -111,7 +111,7 @@ python -m lerobot.scripts.train \
Let's explain the command:
1. We provided the dataset as argument with `--dataset.repo_id=${HF_USER}/il_gym`.
2. We provided the policy with `policy.type=act`. This loads configurations from [`configuration_act.py`](../src/lerobot/policies/act/configuration_act.py). Importantly, this policy will automatically adapt to the number of motor states, motor actions and cameras of your robot (e.g. `laptop` and `phone`) which have been saved in your dataset.
2. We provided the policy with `policy.type=act`. This loads configurations from [`configuration_act.py`](https://github.com/huggingface/lerobot/blob/main/src/lerobot/policies/act/configuration_act.py). Importantly, this policy will automatically adapt to the number of motor states, motor actions and cameras of your robot (e.g. `laptop` and `phone`) which have been saved in your dataset.
3. We provided `policy.device=cuda` since we are training on a Nvidia GPU, but you could use `policy.device=mps` to train on Apple silicon.
4. We provided `wandb.enable=true` to use [Weights and Biases](https://docs.wandb.ai/quickstart) for visualizing training plots. This is optional but if you use it, make sure you are logged in by running `wandb login`.

View File

@@ -1,15 +1,6 @@
# Installation
## Install LeRobot
Currently only available from source.
Download our source code:
```bash
git clone https://github.com/huggingface/lerobot.git
cd lerobot
```
## Environment Setup
Create a virtual environment with Python 3.10, using [`Miniconda`](https://docs.anaconda.com/miniconda/install/#quick-command-line-install)
@@ -40,12 +31,49 @@ conda install ffmpeg -c conda-forge
>
> - _[On Linux only]_ If you want to bring your own ffmpeg: Install [ffmpeg build dependencies](https://trac.ffmpeg.org/wiki/CompilationGuide/Ubuntu#GettheDependencies) and [compile ffmpeg from source with libsvtav1](https://trac.ffmpeg.org/wiki/CompilationGuide/Ubuntu#libsvtav1), and make sure you use the corresponding ffmpeg binary to your install with `which ffmpeg`.
Install 🤗 LeRobot:
## Install LeRobot 🤗
### From Source
First, clone the repository and navigate into the directory:
```bash
git clone https://github.com/huggingface/lerobot.git
cd lerobot
```
Then, install the library in editable mode. This is useful if you plan to contribute to the code.
```bash
pip install -e .
```
### Installation from PyPI
**Core Library:**
Install the base package with:
```bash
pip install lerobot
```
_This installs only the default dependencies._
**Extra Features:**
To install additional functionality, use one of the following:
```bash
pip install 'lerobot[all]' # All available features
pip install 'lerobot[aloha,pusht]' # Specific features (Aloha & Pusht)
pip install 'lerobot[feetech]' # Feetech motor support
```
_Replace `[...]` with your desired features._
**Available Tags:**
For a full list of optional dependencies, see:
https://pypi.org/project/lerobot/
### Troubleshooting
If you encounter build errors, you may need to install additional dependencies: `cmake`, `build-essential`, and `ffmpeg libs`.

View File

@@ -0,0 +1,14 @@
## Paper
https://tonyzhaozh.github.io/aloha
## Citation
```bibtex
@article{zhao2023learning,
title={Learning fine-grained bimanual manipulation with low-cost hardware},
author={Zhao, Tony Z and Kumar, Vikash and Levine, Sergey and Finn, Chelsea},
journal={arXiv preprint arXiv:2304.13705},
year={2023}
}
```

View File

@@ -0,0 +1,14 @@
## Paper
https://diffusion-policy.cs.columbia.edu
## Citation
```bibtex
@article{chi2024diffusionpolicy,
author = {Cheng Chi and Zhenjia Xu and Siyuan Feng and Eric Cousineau and Yilun Du and Benjamin Burchfiel and Russ Tedrake and Shuran Song},
title ={Diffusion Policy: Visuomotor Policy Learning via Action Diffusion},
journal = {The International Journal of Robotics Research},
year = {2024},
}
```

View File

@@ -0,0 +1,14 @@
## Paper
https://arxiv.org/abs/2506.01844
## Citation
```bibtex
@article{shukor2025smolvla,
title={SmolVLA: A Vision-Language-Action Model for Affordable and Efficient Robotics},
author={Shukor, Mustafa and Aubakirova, Dana and Capuano, Francesco and Kooijmans, Pepijn and Palma, Steven and Zouitine, Adil and Aractingi, Michel and Pascal, Caroline and Russi, Martino and Marafioti, Andres and Alibert, Simon and Cord, Matthieu and Wolf, Thomas and Cadene, Remi},
journal={arXiv preprint arXiv:2506.01844},
year={2025}
}
```

View File

@@ -0,0 +1,14 @@
## Paper
https://www.nicklashansen.com/td-mpc/
## Citation
```bibtex
@inproceedings{Hansen2022tdmpc,
title={Temporal Difference Learning for Model Predictive Control},
author={Nicklas Hansen and Xiaolong Wang and Hao Su},
booktitle={ICML},
year={2022}
}
```

View File

@@ -0,0 +1,14 @@
## Paper
https://sjlee.cc/vq-bet/
## Citation
```bibtex
@article{lee2024behavior,
title={Behavior generation with latent actions},
author={Lee, Seungjae and Wang, Yibin and Etukuru, Haritheja and Kim, H Jin and Shafiullah, Nur Muhammad Mahi and Pinto, Lerrel},
journal={arXiv preprint arXiv:2403.03181},
year={2024}
}
```

View File

@@ -25,7 +25,7 @@ discord = "https://discord.gg/s3KuuzsPFb"
[project]
name = "lerobot"
version = "0.2.0"
version = "0.4.0"
description = "🤗 LeRobot: State-of-the-art Machine Learning for Real-World Robotics in Pytorch"
readme = "README.md"
license = { text = "Apache-2.0" }
@@ -125,7 +125,6 @@ hilserl = ["lerobot[transformers-dep]", "gym-hil>=0.1.9", "lerobot[grpcio-dep]",
async = ["lerobot[grpcio-dep]", "matplotlib>=3.10.3"]
# Development
docs = ["hf-doc-builder @ git+https://github.com/huggingface/doc-builder.git@main", "watchdog >= 6.0.0"]
dev = ["pre-commit>=3.7.0", "debugpy>=1.8.1", "lerobot[grpcio-dep]", "grpcio-tools==1.73.1"]
test = ["pytest>=8.1.0", "pytest-timeout>=2.4.0", "pytest-cov>=5.0.0", "mock-serial>=0.0.1 ; sys_platform != 'win32'"]
video_benchmark = ["scikit-image>=0.23.2", "pandas>=2.2.2"]
@@ -147,7 +146,6 @@ all = [
"lerobot[smolvla]",
"lerobot[hilserl]",
"lerobot[async]",
"lerobot[docs]",
"lerobot[dev]",
"lerobot[test]",
"lerobot[video_benchmark]",

625
requirements-macos.txt Normal file
View File

@@ -0,0 +1,625 @@
# This file is autogenerated by pip-compile with Python 3.10
# by the following command:
#
# pip-compile --output-file=requirements-macos.txt requirements.in
#
-e .[all]
# via -[all]
absl-py==2.3.1
# via
# dm-control
# dm-env
# dm-tree
# labmaze
# mujoco
accelerate==1.9.0
# via lerobot
aiohappyeyeballs==2.6.1
# via aiohttp
aiohttp==3.12.15
# via fsspec
aiosignal==1.4.0
# via aiohttp
annotated-types==0.7.0
# via pydantic
asttokens==3.0.0
# via stack-data
async-timeout==5.0.1
# via aiohttp
attrs==25.3.0
# via
# aiohttp
# dm-tree
# jsonlines
# rerun-sdk
av==15.0.0
# via lerobot
blinker==1.9.0
# via flask
certifi==2025.7.14
# via
# requests
# sentry-sdk
cffi==1.17.1
# via pymunk
cfgv==3.4.0
# via pre-commit
charset-normalizer==3.4.2
# via requests
click==8.2.1
# via
# flask
# wandb
cloudpickle==3.1.1
# via gymnasium
cmake==4.0.3
# via lerobot
cmeel==0.57.3
# via
# cmeel-assimp
# cmeel-boost
# cmeel-console-bridge
# cmeel-octomap
# cmeel-qhull
# cmeel-tinyxml2
# cmeel-urdfdom
# cmeel-zlib
# coal-library
# eigenpy
# eiquadprog
# pin
# placo
# rhoban-cmeel-jsoncpp
cmeel-assimp==5.4.3.1
# via coal-library
cmeel-boost==1.87.0.1
# via
# coal-library
# eigenpy
# eiquadprog
# pin
cmeel-console-bridge==1.0.2.3
# via cmeel-urdfdom
cmeel-octomap==1.10.0
# via coal-library
cmeel-qhull==8.0.2.1
# via coal-library
cmeel-tinyxml2==10.0.0
# via cmeel-urdfdom
cmeel-urdfdom==4.0.1
# via pin
cmeel-zlib==1.3.1
# via cmeel-assimp
coal-library==3.0.1
# via pin
contourpy==1.3.2
# via matplotlib
coverage[toml]==7.10.1
# via pytest-cov
cycler==0.12.1
# via matplotlib
datasets==3.6.0
# via lerobot
debugpy==1.8.15
# via lerobot
decorator==5.2.1
# via ipython
deepdiff==8.5.0
# via lerobot
diffusers==0.34.0
# via lerobot
dill==0.3.8
# via
# datasets
# multiprocess
distlib==0.4.0
# via virtualenv
dm-control==1.0.14
# via gym-aloha
dm-env==1.6
# via dm-control
dm-tree==0.1.9
# via
# dm-control
# dm-env
docopt==0.6.2
# via num2words
draccus==0.10.0
# via lerobot
dynamixel-sdk==3.7.31
# via lerobot
eigenpy==3.10.3
# via coal-library
einops==0.8.1
# via lerobot
eiquadprog==1.2.9
# via placo
exceptiongroup==1.3.0
# via
# ipython
# pytest
executing==2.2.0
# via stack-data
farama-notifications==0.0.4
# via gymnasium
feetech-servo-sdk==1.0.0
# via lerobot
filelock==3.18.0
# via
# datasets
# diffusers
# huggingface-hub
# torch
# transformers
# virtualenv
flask==3.1.1
# via lerobot
fonttools==4.59.0
# via matplotlib
frozenlist==1.7.0
# via
# aiohttp
# aiosignal
fsspec[http]==2025.3.0
# via
# datasets
# huggingface-hub
# torch
gitdb==4.0.12
# via gitpython
gitpython==3.1.45
# via wandb
glfw==2.9.0
# via
# dm-control
# mujoco
grpcio==1.73.1
# via
# grpcio-tools
# lerobot
grpcio-tools==1.73.1
# via lerobot
gym-aloha==0.1.1
# via lerobot
gym-hil==0.1.10
# via lerobot
gym-pusht==0.1.5
# via lerobot
gym-xarm==0.1.1
# via lerobot
gymnasium==0.29.1
# via
# gym-aloha
# gym-hil
# gym-pusht
# gym-xarm
# gymnasium-robotics
# lerobot
# pettingzoo
gymnasium-robotics==1.2.4
# via gym-xarm
hf-transfer==0.1.9
# via huggingface-hub
hf-xet==1.1.5
# via huggingface-hub
hidapi==0.14.0.post4
# via
# gym-hil
# lerobot
huggingface-hub[cli,hf-transfer]==0.34.3
# via
# accelerate
# datasets
# diffusers
# lerobot
# tokenizers
# transformers
identify==2.6.12
# via pre-commit
idna==3.10
# via
# requests
# yarl
imageio[ffmpeg]==2.37.0
# via
# gym-aloha
# gym-hil
# gymnasium-robotics
# lerobot
# scikit-image
imageio-ffmpeg==0.6.0
# via imageio
importlib-metadata==8.7.0
# via diffusers
iniconfig==2.1.0
# via pytest
inquirerpy==0.3.4
# via huggingface-hub
ipython==8.37.0
# via meshcat
ischedule==1.2.7
# via placo
itsdangerous==2.2.0
# via flask
jedi==0.19.2
# via ipython
jinja2==3.1.6
# via
# flask
# gymnasium-robotics
# torch
jsonlines==4.0.0
# via lerobot
kiwisolver==1.4.8
# via matplotlib
labmaze==1.0.6
# via dm-control
lazy-loader==0.4
# via scikit-image
lxml==6.0.0
# via dm-control
markupsafe==3.0.2
# via
# flask
# jinja2
# werkzeug
matplotlib==3.10.5
# via lerobot
matplotlib-inline==0.1.7
# via ipython
mergedeep==1.3.4
# via draccus
meshcat==0.3.2
# via placo
mock-serial==0.0.1
# via lerobot
mpmath==1.3.0
# via sympy
mujoco==2.3.7
# via
# dm-control
# gym-aloha
# gym-hil
# gym-xarm
# gymnasium-robotics
multidict==6.6.3
# via
# aiohttp
# yarl
multiprocess==0.70.16
# via datasets
mypy-extensions==1.1.0
# via typing-inspect
networkx==3.4.2
# via
# scikit-image
# torch
nodeenv==1.9.1
# via pre-commit
num2words==0.5.14
# via lerobot
numpy==2.2.6
# via
# accelerate
# cmeel-boost
# contourpy
# datasets
# diffusers
# dm-control
# dm-env
# dm-tree
# gymnasium
# gymnasium-robotics
# imageio
# labmaze
# matplotlib
# meshcat
# mujoco
# opencv-python
# opencv-python-headless
# pandas
# pettingzoo
# rerun-sdk
# scikit-image
# scipy
# shapely
# tifffile
# torchvision
# transformers
opencv-python==4.12.0.88
# via gym-pusht
opencv-python-headless==4.12.0.88
# via lerobot
orderly-set==5.5.0
# via deepdiff
packaging==25.0
# via
# accelerate
# datasets
# huggingface-hub
# lazy-loader
# lerobot
# matplotlib
# pytest
# scikit-image
# transformers
# wandb
pandas==2.3.1
# via
# datasets
# lerobot
parso==0.8.4
# via jedi
pettingzoo==1.24.3
# via gymnasium-robotics
pexpect==4.9.0
# via ipython
pfzy==0.3.4
# via inquirerpy
pillow==11.3.0
# via
# diffusers
# imageio
# matplotlib
# meshcat
# rerun-sdk
# scikit-image
# torchvision
pin==3.4.0
# via placo
placo==0.9.14
# via lerobot
platformdirs==4.3.8
# via
# virtualenv
# wandb
pluggy==1.6.0
# via
# pytest
# pytest-cov
pre-commit==4.2.0
# via lerobot
prompt-toolkit==3.0.51
# via
# inquirerpy
# ipython
propcache==0.3.2
# via
# aiohttp
# yarl
protobuf==6.31.0
# via
# dm-control
# grpcio-tools
# lerobot
# wandb
psutil==7.0.0
# via
# accelerate
# imageio
ptyprocess==0.7.0
# via pexpect
pure-eval==0.2.3
# via stack-data
pyarrow==21.0.0
# via
# datasets
# rerun-sdk
pycparser==2.22
# via cffi
pydantic==2.11.7
# via wandb
pydantic-core==2.33.2
# via pydantic
pygame==2.6.1
# via
# gym-hil
# gym-pusht
# lerobot
pygments==2.19.2
# via
# ipython
# pytest
pymunk==6.11.1
# via
# gym-pusht
# lerobot
pyngrok==7.2.12
# via meshcat
pynput==1.8.1
# via
# gym-hil
# lerobot
pyobjc-core==11.1
# via
# pyobjc-framework-applicationservices
# pyobjc-framework-cocoa
# pyobjc-framework-coretext
# pyobjc-framework-quartz
pyobjc-framework-applicationservices==11.1
# via pynput
pyobjc-framework-cocoa==11.1
# via
# pyobjc-framework-applicationservices
# pyobjc-framework-coretext
# pyobjc-framework-quartz
pyobjc-framework-coretext==11.1
# via pyobjc-framework-applicationservices
pyobjc-framework-quartz==11.1
# via
# pynput
# pyobjc-framework-applicationservices
# pyobjc-framework-coretext
pyopengl==3.1.9
# via
# dm-control
# mujoco
pyparsing==3.2.3
# via
# dm-control
# matplotlib
pyrealsense2-macosx==2.54.2
# via lerobot
pyserial==3.5
# via
# dynamixel-sdk
# feetech-servo-sdk
# lerobot
pytest==8.4.1
# via
# lerobot
# pytest-cov
# pytest-timeout
pytest-cov==6.2.1
# via lerobot
pytest-timeout==2.4.0
# via lerobot
python-dateutil==2.9.0.post0
# via
# matplotlib
# pandas
pytz==2025.2
# via pandas
pyyaml==6.0.2
# via
# accelerate
# datasets
# draccus
# huggingface-hub
# pre-commit
# pyngrok
# pyyaml-include
# transformers
# wandb
pyyaml-include==1.4.1
# via draccus
pyzmq==27.0.0
# via
# lerobot
# meshcat
regex==2025.7.34
# via
# diffusers
# transformers
requests==2.32.4
# via
# datasets
# diffusers
# dm-control
# huggingface-hub
# transformers
# wandb
rerun-sdk==0.22.1
# via lerobot
rhoban-cmeel-jsoncpp==1.9.4.9
# via placo
safetensors==0.5.3
# via
# accelerate
# diffusers
# lerobot
# transformers
scikit-image==0.25.2
# via
# gym-pusht
# lerobot
scipy==1.15.3
# via
# dm-control
# scikit-image
sentry-sdk==2.34.1
# via wandb
shapely==2.1.1
# via gym-pusht
six==1.17.0
# via
# pynput
# python-dateutil
smmap==5.0.2
# via gitdb
stack-data==0.6.3
# via ipython
sympy==1.14.0
# via torch
termcolor==3.1.0
# via lerobot
tifffile==2025.5.10
# via scikit-image
tokenizers==0.21.4
# via transformers
toml==0.10.2
# via draccus
tomli==2.2.1
# via
# cmeel
# coverage
# pytest
torch==2.7.1
# via
# accelerate
# lerobot
# torchvision
torchcodec==0.5
# via lerobot
torchvision==0.22.1
# via lerobot
tornado==6.5.1
# via meshcat
tqdm==4.67.1
# via
# datasets
# dm-control
# huggingface-hub
# transformers
traitlets==5.14.3
# via
# ipython
# matplotlib-inline
transformers==4.51.3
# via lerobot
typing-extensions==4.14.1
# via
# aiosignal
# exceptiongroup
# gymnasium
# huggingface-hub
# ipython
# multidict
# pydantic
# pydantic-core
# rerun-sdk
# torch
# typing-inspect
# typing-inspection
# wandb
typing-inspect==0.9.0
# via draccus
typing-inspection==0.4.1
# via pydantic
tzdata==2025.2
# via pandas
u-msgpack-python==2.8.0
# via meshcat
urllib3==2.5.0
# via
# requests
# sentry-sdk
virtualenv==20.32.0
# via pre-commit
wandb==0.21.0
# via lerobot
wcwidth==0.2.13
# via prompt-toolkit
werkzeug==3.1.3
# via flask
wrapt==1.17.2
# via dm-tree
xxhash==3.5.0
# via datasets
yarl==1.20.1
# via aiohttp
zipp==3.23.0
# via importlib-metadata
# The following packages are considered to be unsafe in a requirements file:
# setuptools

650
requirements-ubuntu.txt Normal file
View File

@@ -0,0 +1,650 @@
#
# This file is autogenerated by pip-compile with Python 3.10
# by the following command:
#
# pip-compile --output-file=requirements-ubuntu.txt requirements.in
#
-e .[all]
# via -[all]
absl-py==2.3.1
# via
# dm-control
# dm-env
# dm-tree
# labmaze
# mujoco
accelerate==1.9.0
# via lerobot
aiohappyeyeballs==2.6.1
# via aiohttp
aiohttp==3.12.15
# via fsspec
aiosignal==1.4.0
# via aiohttp
annotated-types==0.7.0
# via pydantic
asttokens==3.0.0
# via stack-data
async-timeout==5.0.1
# via aiohttp
attrs==25.3.0
# via
# aiohttp
# dm-tree
# jsonlines
# rerun-sdk
av==15.0.0
# via lerobot
blinker==1.9.0
# via flask
certifi==2025.7.14
# via
# requests
# sentry-sdk
cffi==1.17.1
# via pymunk
cfgv==3.4.0
# via pre-commit
charset-normalizer==3.4.2
# via requests
click==8.2.1
# via
# flask
# wandb
cloudpickle==3.1.1
# via gymnasium
cmake==4.0.3
# via lerobot
cmeel==0.57.3
# via
# cmeel-assimp
# cmeel-boost
# cmeel-console-bridge
# cmeel-octomap
# cmeel-qhull
# cmeel-tinyxml2
# cmeel-urdfdom
# cmeel-zlib
# coal-library
# eigenpy
# eiquadprog
# pin
# placo
# rhoban-cmeel-jsoncpp
cmeel-assimp==5.4.3.1
# via coal-library
cmeel-boost==1.87.0.1
# via
# coal-library
# eigenpy
# eiquadprog
# pin
cmeel-console-bridge==1.0.2.3
# via cmeel-urdfdom
cmeel-octomap==1.10.0
# via coal-library
cmeel-qhull==8.0.2.1
# via coal-library
cmeel-tinyxml2==10.0.0
# via cmeel-urdfdom
cmeel-urdfdom==4.0.1
# via pin
cmeel-zlib==1.3.1
# via cmeel-assimp
coal-library==3.0.1
# via pin
contourpy==1.3.2
# via matplotlib
coverage[toml]==7.10.1
# via pytest-cov
cycler==0.12.1
# via matplotlib
datasets==3.6.0
# via lerobot
debugpy==1.8.15
# via lerobot
decorator==5.2.1
# via ipython
deepdiff==8.5.0
# via lerobot
diffusers==0.34.0
# via lerobot
dill==0.3.8
# via
# datasets
# multiprocess
distlib==0.4.0
# via virtualenv
dm-control==1.0.14
# via gym-aloha
dm-env==1.6
# via dm-control
dm-tree==0.1.9
# via
# dm-control
# dm-env
docopt==0.6.2
# via num2words
draccus==0.10.0
# via lerobot
dynamixel-sdk==3.7.31
# via lerobot
eigenpy==3.10.3
# via coal-library
einops==0.8.1
# via lerobot
eiquadprog==1.2.9
# via placo
evdev==1.9.2
# via pynput
exceptiongroup==1.3.0
# via
# ipython
# pytest
executing==2.2.0
# via stack-data
farama-notifications==0.0.4
# via gymnasium
feetech-servo-sdk==1.0.0
# via lerobot
filelock==3.18.0
# via
# datasets
# diffusers
# huggingface-hub
# torch
# transformers
# virtualenv
flask==3.1.1
# via lerobot
fonttools==4.59.0
# via matplotlib
frozenlist==1.7.0
# via
# aiohttp
# aiosignal
fsspec[http]==2025.3.0
# via
# datasets
# huggingface-hub
# torch
gitdb==4.0.12
# via gitpython
gitpython==3.1.45
# via wandb
glfw==2.9.0
# via
# dm-control
# mujoco
grpcio==1.73.1
# via
# grpcio-tools
# lerobot
grpcio-tools==1.73.1
# via lerobot
gym-aloha==0.1.1
# via lerobot
gym-hil==0.1.10
# via lerobot
gym-pusht==0.1.5
# via lerobot
gym-xarm==0.1.1
# via lerobot
gymnasium==0.29.1
# via
# gym-aloha
# gym-hil
# gym-pusht
# gym-xarm
# gymnasium-robotics
# lerobot
# pettingzoo
gymnasium-robotics==1.2.4
# via gym-xarm
hf-transfer==0.1.9
# via huggingface-hub
hf-xet==1.1.5
# via huggingface-hub
hidapi==0.14.0.post4
# via
# gym-hil
# lerobot
huggingface-hub[cli,hf-transfer]==0.34.3
# via
# accelerate
# datasets
# diffusers
# lerobot
# tokenizers
# transformers
identify==2.6.12
# via pre-commit
idna==3.10
# via
# requests
# yarl
imageio[ffmpeg]==2.37.0
# via
# gym-aloha
# gym-hil
# gymnasium-robotics
# lerobot
# scikit-image
imageio-ffmpeg==0.6.0
# via imageio
importlib-metadata==8.7.0
# via diffusers
iniconfig==2.1.0
# via pytest
inquirerpy==0.3.4
# via huggingface-hub
ipython==8.37.0
# via meshcat
ischedule==1.2.7
# via placo
itsdangerous==2.2.0
# via flask
jedi==0.19.2
# via ipython
jinja2==3.1.6
# via
# flask
# gymnasium-robotics
# torch
jsonlines==4.0.0
# via lerobot
kiwisolver==1.4.8
# via matplotlib
labmaze==1.0.6
# via dm-control
lazy-loader==0.4
# via scikit-image
lxml==6.0.0
# via dm-control
markupsafe==3.0.2
# via
# flask
# jinja2
# werkzeug
matplotlib==3.10.5
# via lerobot
matplotlib-inline==0.1.7
# via ipython
mergedeep==1.3.4
# via draccus
meshcat==0.3.2
# via placo
mock-serial==0.0.1
# via lerobot
mpmath==1.3.0
# via sympy
mujoco==2.3.7
# via
# dm-control
# gym-aloha
# gym-hil
# gym-xarm
# gymnasium-robotics
multidict==6.6.3
# via
# aiohttp
# yarl
multiprocess==0.70.16
# via datasets
mypy-extensions==1.1.0
# via typing-inspect
networkx==3.4.2
# via
# scikit-image
# torch
nodeenv==1.9.1
# via pre-commit
num2words==0.5.14
# via lerobot
numpy==2.2.6
# via
# accelerate
# cmeel-boost
# contourpy
# datasets
# diffusers
# dm-control
# dm-env
# dm-tree
# gymnasium
# gymnasium-robotics
# imageio
# labmaze
# matplotlib
# meshcat
# mujoco
# opencv-python
# opencv-python-headless
# pandas
# pettingzoo
# rerun-sdk
# scikit-image
# scipy
# shapely
# tifffile
# torchvision
# transformers
nvidia-cublas-cu12==12.6.4.1
# via
# nvidia-cudnn-cu12
# nvidia-cusolver-cu12
# torch
nvidia-cuda-cupti-cu12==12.6.80
# via torch
nvidia-cuda-nvrtc-cu12==12.6.77
# via torch
nvidia-cuda-runtime-cu12==12.6.77
# via torch
nvidia-cudnn-cu12==9.5.1.17
# via torch
nvidia-cufft-cu12==11.3.0.4
# via torch
nvidia-cufile-cu12==1.11.1.6
# via torch
nvidia-curand-cu12==10.3.7.77
# via torch
nvidia-cusolver-cu12==11.7.1.2
# via torch
nvidia-cusparse-cu12==12.5.4.2
# via
# nvidia-cusolver-cu12
# torch
nvidia-cusparselt-cu12==0.6.3
# via torch
nvidia-nccl-cu12==2.26.2
# via torch
nvidia-nvjitlink-cu12==12.6.85
# via
# nvidia-cufft-cu12
# nvidia-cusolver-cu12
# nvidia-cusparse-cu12
# torch
nvidia-nvtx-cu12==12.6.77
# via torch
opencv-python==4.12.0.88
# via gym-pusht
opencv-python-headless==4.12.0.88
# via lerobot
orderly-set==5.5.0
# via deepdiff
packaging==25.0
# via
# accelerate
# datasets
# huggingface-hub
# lazy-loader
# lerobot
# matplotlib
# pytest
# scikit-image
# transformers
# wandb
pandas==2.3.1
# via
# datasets
# lerobot
parso==0.8.4
# via jedi
pettingzoo==1.24.3
# via gymnasium-robotics
pexpect==4.9.0
# via ipython
pfzy==0.3.4
# via inquirerpy
pillow==11.3.0
# via
# diffusers
# imageio
# matplotlib
# meshcat
# rerun-sdk
# scikit-image
# torchvision
pin==3.4.0
# via placo
placo==0.9.14
# via lerobot
platformdirs==4.3.8
# via
# virtualenv
# wandb
pluggy==1.6.0
# via
# pytest
# pytest-cov
pre-commit==4.2.0
# via lerobot
prompt-toolkit==3.0.51
# via
# inquirerpy
# ipython
propcache==0.3.2
# via
# aiohttp
# yarl
protobuf==6.31.0
# via
# dm-control
# grpcio-tools
# lerobot
# wandb
psutil==7.0.0
# via
# accelerate
# imageio
ptyprocess==0.7.0
# via pexpect
pure-eval==0.2.3
# via stack-data
pyarrow==21.0.0
# via
# datasets
# rerun-sdk
pycparser==2.22
# via cffi
pydantic==2.11.7
# via wandb
pydantic-core==2.33.2
# via pydantic
pygame==2.6.1
# via
# gym-hil
# gym-pusht
# lerobot
pygments==2.19.2
# via
# ipython
# pytest
pymunk==6.11.1
# via
# gym-pusht
# lerobot
pyngrok==7.2.12
# via meshcat
pynput==1.8.1
# via
# gym-hil
# lerobot
pyopengl==3.1.9
# via
# dm-control
# mujoco
pyparsing==3.2.3
# via
# dm-control
# matplotlib
pyrealsense2==2.56.5.9235
# via lerobot
pyserial==3.5
# via
# dynamixel-sdk
# feetech-servo-sdk
# lerobot
pytest==8.4.1
# via
# lerobot
# pytest-cov
# pytest-timeout
pytest-cov==6.2.1
# via lerobot
pytest-timeout==2.4.0
# via lerobot
python-dateutil==2.9.0.post0
# via
# matplotlib
# pandas
python-xlib==0.33
# via pynput
pytz==2025.2
# via pandas
pyyaml==6.0.2
# via
# accelerate
# datasets
# draccus
# huggingface-hub
# pre-commit
# pyngrok
# pyyaml-include
# transformers
# wandb
pyyaml-include==1.4.1
# via draccus
pyzmq==27.0.0
# via
# lerobot
# meshcat
regex==2025.7.34
# via
# diffusers
# transformers
requests==2.32.4
# via
# datasets
# diffusers
# dm-control
# huggingface-hub
# transformers
# wandb
rerun-sdk==0.22.1
# via lerobot
rhoban-cmeel-jsoncpp==1.9.4.9
# via placo
safetensors==0.5.3
# via
# accelerate
# diffusers
# lerobot
# transformers
scikit-image==0.25.2
# via
# gym-pusht
# lerobot
scipy==1.15.3
# via
# dm-control
# scikit-image
sentry-sdk==2.34.1
# via wandb
shapely==2.1.1
# via gym-pusht
six==1.17.0
# via
# pynput
# python-dateutil
# python-xlib
smmap==5.0.2
# via gitdb
stack-data==0.6.3
# via ipython
sympy==1.14.0
# via torch
termcolor==3.1.0
# via lerobot
tifffile==2025.5.10
# via scikit-image
tokenizers==0.21.4
# via transformers
toml==0.10.2
# via draccus
tomli==2.2.1
# via
# cmeel
# coverage
# pytest
torch==2.7.1
# via
# accelerate
# lerobot
# torchvision
torchcodec==0.5
# via lerobot
torchvision==0.22.1
# via lerobot
tornado==6.5.1
# via meshcat
tqdm==4.67.1
# via
# datasets
# dm-control
# huggingface-hub
# transformers
traitlets==5.14.3
# via
# ipython
# matplotlib-inline
transformers==4.51.3
# via lerobot
triton==3.3.1
# via torch
typing-extensions==4.14.1
# via
# aiosignal
# exceptiongroup
# gymnasium
# huggingface-hub
# ipython
# multidict
# pydantic
# pydantic-core
# rerun-sdk
# torch
# typing-inspect
# typing-inspection
# wandb
typing-inspect==0.9.0
# via draccus
typing-inspection==0.4.1
# via pydantic
tzdata==2025.2
# via pandas
u-msgpack-python==2.8.0
# via meshcat
urllib3==2.5.0
# via
# requests
# sentry-sdk
virtualenv==20.32.0
# via pre-commit
wandb==0.21.0
# via lerobot
wcwidth==0.2.13
# via prompt-toolkit
werkzeug==3.1.3
# via flask
wrapt==1.17.2
# via dm-tree
xxhash==3.5.0
# via datasets
yarl==1.20.1
# via aiohttp
zipp==3.23.0
# via importlib-metadata
# The following packages are considered to be unsafe in a requirements file:
# setuptools

9
requirements.in Normal file
View File

@@ -0,0 +1,9 @@
# requirements.in
# requirements-macos.txt was generated on macOS and is platform-specific (macOS 15.5 24F74 arm64).
# Darwin MacBook-Pro.local 24.5.0 Darwin Kernel Version 24.5.0: Tue Apr 22 19:54:43 PDT 2025; root:xnu-11417.121.6~2/RELEASE_ARM64_T8132 arm64
# requirements-ubuntu.txt was generated on Linux and is platform-specific (Ubuntu 24.04.2 LTS x86_64).
# Linux mlerobot-linux 6.14.0-27-generic #27~24.04.1-Ubuntu SMP PREEMPT_DYNAMIC Tue Jul 22 17:38:49 UTC 2 x86_64 x86_64 x86_64 GNU/Linux
-e .[all]

View File

@@ -82,5 +82,9 @@ def calibrate(cfg: CalibrateConfig):
device.disconnect()
if __name__ == "__main__":
def main():
calibrate()
if __name__ == "__main__":
main()

View File

@@ -27,6 +27,7 @@ from huggingface_hub.constants import CONFIG_NAME
from huggingface_hub.errors import HfHubHTTPError
from lerobot.configs.types import FeatureType, NormalizationMode, PolicyFeature
from lerobot.constants import ACTION, OBS_STATE
from lerobot.optim.optimizers import OptimizerConfig
from lerobot.optim.schedulers import LRSchedulerConfig
from lerobot.utils.hub import HubMixin
@@ -119,8 +120,8 @@ class PreTrainedConfig(draccus.ChoiceRegistry, HubMixin, abc.ABC):
@property
def robot_state_feature(self) -> PolicyFeature | None:
for _, ft in self.input_features.items():
if ft.type is FeatureType.STATE:
for ft_name, ft in self.input_features.items():
if ft.type is FeatureType.STATE and ft_name == OBS_STATE:
return ft
return None
@@ -137,8 +138,8 @@ class PreTrainedConfig(draccus.ChoiceRegistry, HubMixin, abc.ABC):
@property
def action_feature(self) -> PolicyFeature | None:
for _, ft in self.output_features.items():
if ft.type is FeatureType.ACTION:
for ft_name, ft in self.output_features.items():
if ft.type is FeatureType.ACTION and ft_name == ACTION:
return ft
return None

View File

@@ -286,7 +286,7 @@ def save_images_from_all_cameras(
print(f"Image capture finished. Images saved to {output_dir}")
if __name__ == "__main__":
def main():
parser = argparse.ArgumentParser(
description="Unified camera utility script for listing cameras and capturing images."
)
@@ -313,3 +313,7 @@ if __name__ == "__main__":
)
args = parser.parse_args()
save_images_from_all_cameras(**vars(args))
if __name__ == "__main__":
main()

View File

@@ -61,5 +61,9 @@ def find_port():
raise OSError(f"Could not detect the port. More than one port was found ({ports_diff}).")
if __name__ == "__main__":
def main():
find_port()
if __name__ == "__main__":
main()

View File

@@ -0,0 +1 @@
../../../../docs/source/policy_act_README.md

View File

@@ -0,0 +1 @@
../../../../docs/source/policy_diffusion_README.md

View File

@@ -0,0 +1 @@
../../../../docs/source/policy_smolvla_README.md

View File

@@ -0,0 +1 @@
../../../../docs/source/policy_tdmpc_README.md

View File

@@ -0,0 +1 @@
../../../../docs/source/policy_vqbet_README.md

View File

@@ -393,5 +393,9 @@ def record(cfg: RecordConfig) -> LeRobotDataset:
return dataset
if __name__ == "__main__":
def main():
record()
if __name__ == "__main__":
main()

View File

@@ -112,5 +112,9 @@ def replay(cfg: ReplayConfig):
robot.disconnect()
if __name__ == "__main__":
def main():
replay()
if __name__ == "__main__":
main()

View File

@@ -286,6 +286,10 @@ def train(cfg: TrainPipelineConfig):
policy.push_model_to_hub(cfg)
if __name__ == "__main__":
def main():
init_logging()
train()
if __name__ == "__main__":
main()

View File

@@ -80,5 +80,9 @@ def setup_motors(cfg: SetupConfig):
device.setup_motors()
if __name__ == "__main__":
def main():
setup_motors()
if __name__ == "__main__":
main()

View File

@@ -153,5 +153,9 @@ def teleoperate(cfg: TeleoperateConfig):
robot.disconnect()
if __name__ == "__main__":
def main():
teleoperate()
if __name__ == "__main__":
main()

View File

@@ -27,11 +27,13 @@ from lerobot import available_policies
from lerobot.configs.default import DatasetConfig
from lerobot.configs.train import TrainPipelineConfig
from lerobot.configs.types import FeatureType, NormalizationMode, PolicyFeature
from lerobot.constants import ACTION, OBS_STATE
from lerobot.datasets.factory import make_dataset
from lerobot.datasets.utils import cycle, dataset_to_policy_features
from lerobot.envs.factory import make_env, make_env_config
from lerobot.envs.utils import preprocess_observation
from lerobot.optim.factory import make_optimizer_and_scheduler
from lerobot.policies.act.configuration_act import ACTConfig
from lerobot.policies.act.modeling_act import ACTTemporalEnsembler
from lerobot.policies.factory import (
get_policy_class,
@@ -363,6 +365,54 @@ def test_normalize(insert_temporal_dim):
unnormalize(output_batch)
@pytest.mark.parametrize("multikey", [True, False])
def test_multikey_construction(multikey: bool):
"""
Asserts that multiple keys with type State/Action are correctly processed by the policy constructor,
preventing erroneous creation of the policy object.
"""
input_features = {
"observation.state": PolicyFeature(
type=FeatureType.STATE,
shape=(10,),
),
}
output_features = {
"action": PolicyFeature(
type=FeatureType.ACTION,
shape=(5,),
),
}
if multikey:
"""Simulates the complete state/action is constructed from more granular multiple
keys, of the same type as the overall state/action"""
input_features = {}
input_features["observation.state.subset1"] = PolicyFeature(type=FeatureType.STATE, shape=(5,))
input_features["observation.state.subset2"] = PolicyFeature(type=FeatureType.STATE, shape=(5,))
input_features["observation.state"] = PolicyFeature(type=FeatureType.STATE, shape=(10,))
output_features = {}
output_features["action.first_three_motors"] = PolicyFeature(type=FeatureType.ACTION, shape=(3,))
output_features["action.last_two_motors"] = PolicyFeature(type=FeatureType.ACTION, shape=(2,))
output_features["action"] = PolicyFeature(
type=FeatureType.ACTION,
shape=(5,),
)
config = ACTConfig(input_features=input_features, output_features=output_features)
state_condition = config.robot_state_feature == input_features[OBS_STATE]
action_condition = config.action_feature == output_features[ACTION]
assert state_condition, (
f"Discrepancy detected. Robot state feature is {config.robot_state_feature} but policy expects {input_features[OBS_STATE]}"
)
assert action_condition, (
f"Discrepancy detected. Action feature is {config.action_feature} but policy expects {output_features[ACTION]}"
)
@pytest.mark.parametrize(
"ds_repo_id, policy_name, policy_kwargs, file_name_extra",
[