Self-driving cars are almost certainly inevitable. The only question is when they’ll become commonplace. The two primary challenges we need to overcome to get there are the development of robust and reliable software, and gaining the general public’s trust. By any objective measure, self-driving cars are already far safer than human drivers in most conditions. But to really understand how self-driving cars work, you can follow Piotr Sokólski’s guide to build the smartphone-controlled DeepRC robot car.
Sokólski is a software engineer, and started this project in order to develop a platform for safely testing self-driving car software and hardware. Aside from the high cost of converting an actual car to autonomous control, it’s also very difficult to experiment with them in a safe way. Nobody should be comfortable with a DIY autonomous car driving around on city streets. DeepRC provides an affordable way to test and modify various autonomous driving hardware and software.
To achieve that, Sokólski cut costs dramatically by using a smartphone as the brain of the robot car. Just about everyone has a powerful smartphone, so you may as well take advantage of it. In this case, the smartphone’s camera can even be used thanks to a clever mirror periscope that allows it see forward. The smartphone handles some of the tasks, such as communicating with and controlling the drivetrain, and a separate computer handles the heavy lifting.
DeepRC is built on a 3D-printed frame, which houses the smartphone and other hardware. That hardware includes a control board built on an nRF52 SoC that has an integrated Bluetooth LE radio to connect to the smartphone. That, in turn, controls the two drive motors through a pair of ESC modules. It also operates the steering mechanism via a standard hobby servo motor. If you’re looking for a way to experiment with autonomous vehicle software, DeepRC is a great platform to do it.