ToolsBuild with AI

LeRobot: a Python toolkit for robotics and VLA learning

A Python robotics library for recording demonstrations, working with datasets and training learned policies.

Hugging Face LeRobot

What LeRobot brings together

LeRobot is Hugging Face’s Python library for robotics. It connects robot control with recording, training and deployment. Its dataset format combines synchronized video with action and state data; supported policies include ACT and VLA models such as SmolVLA. Overview

Choose a starting path

Hardware ownership is optional for exploring existing datasets and simulation. Use the VLA learning exercise to frame a small experiment.

Read the installation page for the extras your workflow needs: dataset, training and hardware dependencies are separated. Check platform compatibility and current instructions before copying older commands. Installation

Keep exploring

Explore the collection

ArticlesLearn & research

Robotics after the VLA headline: design a useful first learning experiment

Understand VLA inputs and limits, plan a small robotics lesson, and make evaluations comparable with LeRobot documentation.

CurioFuse editorial · Hugging Face and OpenVLA documentation

VideosLearn & research

Robotics training in Colab: Trossen’s LeRobot video walkthrough

A six-minute robotics training introduction with an exact-video creator companion; the 2024 commands need version checks.

Trossen Robotics11 Oct 2024

NewsLearn & research

REKI gives robot actions a short visual memory

Kyutai and its research partners introduced REKI, a robot-control model with 2.4 billion parameters.

MarkTechPost17 Oct 2026