One of the biggest problems of the 21st century in modern metropolitan cities is vehicle traffic. With the rise of robotics in the last decade the solution to that problem is obvious – make vehicles completely autonomous. This will make traffic both safer and faster and will allow us to focus on more meaningful tasks, rather than spend an hour each day on commuting. Additionally, if the same practice is applied to racing vehicles this will allow car manufacturers to rapidly test new technology and progress innovation further.
We are looking to form a team of passionate students to research and design the necessary software for a fully autonomous f1-style racing vehicle and compete with it at the Formula Student UK (FSUK) competition in July 2018. The aim will be to develop a low-cost software package which will be both safer and faster than a human driver, whilst also being modular and implementable on various vehicles. The AI driving it will be capable of making decisions by itself, acknowledging the results and improving itself on the spot with the use of deep machine learning algorithms. All of this will be implemented only in simulations for now and on an actual Formula Student car at the competition but we also want to make the first steps towards transforming our full-size formula student car into an autonomous vehicle.
The car will be able to create a map of the track it is driving on and decide how to go fast around it. External data will be fed through a set of cameras, lidars, and various other sensors. The design flow will revolve around processing vision and sensor data, artificial intelligence, and machine learning, meaning that the vehicle has to adapt to the circumstances of the race track at any given moment and improve itself on the spot with only one goal in mind – speed. In addition, it must also be able to drive a normal vehicle in everyday traffic.
Here is a slight teaser for what we might be doing. Courtesy of our friends at AMZ Racing Driverless.