Skip to main content

Build self-driving robots with Python

A minimalist and modular framework for building autonomous vehicles. From hobbyist RC cars to research platforms, Donkeycar makes robotics accessible.

Get StartedLearn More

Quick Start

Get your autonomous car up and running in minutes

1

Install Donkeycar

Install the Donkeycar framework on your development machine or directly on your Raspberry Pi or Jetson Nano.
pip install donkeycar[pc]
Use donkeycar[pi] for Raspberry Pi or donkeycar[nano] for Jetson Nano
2

Create your car application

Use the CLI to create a new car project with your preferred template.
donkey createcar --path ~/mycar --template complete
This generates a complete application structure with configuration files, training scripts, and the vehicle loop.
3

Configure and calibrate

Edit myconfig.py to match your hardware setup, then calibrate your steering and throttle.
cd ~/mycar
python manage.py drive
Access the web interface at http://<your-car-ip>:8887 to drive manually and collect training data.
4

Train your autopilot

Train a deep learning model on the data you collected.
donkey train --tub ./data --model ./models/mypilot.h5
Deploy the trained model back to your car and watch it drive autonomously!

Explore by Topic

Everything you need to build, train, and deploy autonomous vehicles

Core Concepts

Understand the part-based architecture, vehicle loop, and memory system that powers Donkeycar.

Build Your Car

Step-by-step guides for assembling hardware, wiring components, and creating your car application.

Training & Autopilots

Learn about deep learning, computer vision, and GPS-based navigation for autonomous driving.

Parts Reference

Explore the extensive library of pre-built parts for cameras, controllers, actuators, and sensors.

CLI Reference

Complete command-line reference for creating cars, training models, and debugging.

API Reference

Detailed API documentation for the Vehicle class, Memory system, and all parts.

Key Features

Modular Architecture

Part-based design lets you mix and match components to build exactly the vehicle you need.

Multiple Autopilot Modes

Choose from deep learning, computer vision, or GPS-based navigation for your use case.

Built-in Simulator

Test and train your models in a virtual environment before deploying to real hardware.

Hardware Flexibility

Supports Raspberry Pi, Jetson Nano, and a wide range of cameras, sensors, and actuators.

TensorFlow & PyTorch

Train models using either framework with support for transfer learning and custom architectures.

Active Community

Join thousands of makers, students, and researchers building autonomous vehicles worldwide.

Ready to build your autonomous vehicle?

Join the Donkeycar community and start building self-driving robots with Python, deep learning, and computer vision.

Get Started NowView on GitHub