Open source software for autonomous cars
- Donkey: Keras and TensorFlow-based CNN code for Raspbery Pi controlled RC car. Instructions here.
- OpenMV: Computer vision car using the OpenMV board and the standard Donkey chassis.
- Duckietown: ROS-based small cars that can navigate a model city, part of a large MIT and ETH education initiative
- Udacity self-driving car code
- Simulator for Udacity code; uses Unity game engine
- PX4:Rover
- APM:Rover
- OSCC/Polysync autonomous car OS
- Autonomia: (Cloud connected car and autopilot in keras/tensorflow. “We make use of our video and data streams management cloud API, and have the car streaming video at any time, including for training (with embedded telemetry) and while in autopilot for live view. We have also implemented a remote control API. Will be happy to give all DIY Robocars participants free use of our Cloud API.)
- Comma.ai OpenPilot code
- OpenAI Universe training agents
- Audi 1/10th scale code
- CNN-based Carputer
- Lane-tracking CV code (Python). Works with APM:Rover, ROS, Raspberry Pi.
- Automotive Grade Linux
- Autoware (urban autonomous driving)
- Dataspeed (DBW, Speed Control, CAN interface)
- CompoundEye: OpenCV based code for RaspberryPi. Won first DIYRobocars time trial
- Suiron: TensorFlow machine learning on a RC car