We found this self-driving RC (remote-controlled) car on YouTube. It uses Raspberry Pi to detect the road ahead so it can figure out how to steer itself around a track.
How does it work?
The Raspberry Pi camera sends images to a Raspberry Pi Zero W, which predicts the steering angle required using what its maker describes as “deep [machine] learning magic”. Then Raspberry Pi Zero sets a servo motor to steer the car at the correct angle.
A custom 3D-printed bodykit fits on top of the car’s chassis. It has a perfectly sized slot on the back for a power bank, plus mounts for the camera and Raspberry Pi.
Where we’re going, we only need toilet roll roads
This project’s maker set out a beautiful high-spec track on their lounge floor for the car to display its skills. They created this high-spec track using toilet roll. It’s actually kind of genius, because it takes only seconds to lay, you can tease it into curves, and it provides a clear, thick, white border for the car to recognise against the brown wooden floor.
It is unclear whether the machine learning model used to train this car to drive itself can recognise anything other than toilet roll as the edge of a track. Perhaps it thinks the world is made of toilet-paper pavements. Who knows.