Build Your Own Self-Driving Car
Donkeycar is an open-source DIY self-driving platform for small scale cars. Built with Python and powered by machine learning, it lets you train your own autopilot using your own track and data.
Why Donkeycar?
Everything you need to build, train, and race your own autonomous vehicle.
Autonomous Driving
Train your car to drive itself using behavioral cloning and end-to-end deep learning. Collect data, train a model, and watch your car navigate the track autonomously.
Machine Learning
Leverage TensorFlow and PyTorch to build neural networks that learn from your driving data. Experiment with different architectures and training strategies.
Hardware Flexibility
Run on Raspberry Pi, Jetson Nano, or any Linux computer. Support for a wide range of RC car platforms, cameras, and sensors gives you complete freedom.
Open Source Community
Join thousands of builders worldwide. Contribute to the codebase, share your builds, participate in virtual and real-world races, and learn together.
Get Up and Running
From zero to autonomous driving in four simple steps.
Install Donkeycar
Set up your development environment and install the Donkeycar Python package with a single command.
pip install donkeycar[pc]Build Your Car
Assemble your RC car with a Raspberry Pi or Jetson Nano, camera module, and motor controller.
donkey createcar --path ~/mycarTrain Your Model
Drive your car manually to collect training data, then train a neural network to learn your driving behavior.
donkey train --tub ./data --model ./models/mypilot.h5Drive Autonomously
Load your trained model and let your car drive itself around the track. Iterate and improve!
python manage.py drive --model ~/mycar/models/mypilot.h5Join the Community
Be part of a global network of makers, engineers, and AI enthusiasts building the future of autonomous vehicles.