Skip to main content

Making AI for Robotics Accessible

LeRobot provides models, datasets, and tools for real-world robotics in PyTorch. Lower the barrier to entry and benefit from shared datasets and pretrained models.

Get Started in Minutes

Quick Installation

Install LeRobot via pip and start controlling robots in seconds

First Steps

Train your first policy and deploy it to a real robot

Core Concepts

Understand the unified Robot interface and LeRobotDataset format

Example Notebooks

Explore interactive Jupyter notebooks with real examples

Key Features

Hardware-Agnostic Control

Unified Python interface for diverse robots from low-cost arms to humanoids

Standardized Datasets

Parquet + MP4 format hosted on Hugging Face Hub for efficient storage and streaming

State-of-the-Art Policies

ACT, Diffusion, VQ-BeT, TDMPC, and VLAs ready for training and deployment

Real-World Ready

Policies proven to transfer from simulation to physical robots

Open Ecosystem

Comprehensive support for open-source robotics democratization

Simulation Integration

Built-in support for LIBERO, MetaWorld, and custom environments

Explore by Category

Train Policies

Learn imitation learning and reinforcement learning for robotics

Browse Hardware

SO-100, LeKiwi, Koch, Reachy2, Unitree G1, and more supported robots

Vision-Language-Action Models

Pi0, GR00T N1.5, SmolVLA, and X-VLA for multimodal robot learning

API Reference

Comprehensive API documentation for Robot, Dataset, and Policy classes

Community & Support

GitHub

Star the repo and contribute to the project

Discord

Join the LeRobot community server

X (Twitter)

Follow for the latest updates and announcements

Build docs developers (and LLMs) love