With the rise of autonomous cars into the mainstream in recent years, our team responded in the only way we know – with curiosity, enthusiasm, and dedication. Thus we founded the AI branch of the team which seeks to give students an opportunity to get into the exciting field of driverless cars, apply their classroom knowledge to a real project, obtain experience in collaborating in a group and of course race the car at Formula Student events around the globe!
In October 2018, EUFS expanded with a brand new project, namely AI. This project is to run alongside the original goal of EUFS to develop and build a formula-style race car within a single year. This year, the goal of the AI team is to develop both the AI for a completely driverless car and to modify one of our existing cars into a self-driving one.
The hardest part of the challenge is that prior to the event, the car cannot have any data about the track itself which means that it must process its surroundings in real time. To perceive the environment, our cars are equipped with state-of-the-art Lidar and multiple cameras which when combined allow us to extract features from the environment using modern deep learning approaches. These features are then used in a SLAM algorithm to create a map of the racecourse and once it has a full map, the car completely release the brake pedal and use machine learning to improve itself on the fly and go faster!
With this, we managed to win 1st place in the DDT class in the first ever Formula Student UK Driverless event!
The Formula Student driverless competition is split into Static and Dynamic events. Static events include Engineering Design, Real World Autonomous and Business presentation which challenges the student’s knowledge and comprehension, engineering design, understanding of challenges faced in the industry and how prepared are they to transform a concept into a business.
The dynamic events test the performance of the car and its autonomous algorithms. Acceleration sees the car accelerate from a full stop to 75m as fast as it can. Skidpad sees the car go around a figure of 8. Trackdrive tests the robustness of the car by challenging it complete 10 laps with no prior data about the track.
The UK competition will be split into two categories and our team is intending on participating in both!
In the DDT Class, teams will be provided with a car onto which they can upload their AI software in order to make it fully driverless. As the car is the same between teams, the only variable will be the algorithms. This is a wonderful opporunity for new teams to evaluate their software before they enter the ADS class.
In the ADS Class, the concept is still the same but teams are now required now make their own autonomous racecar and bring it to the competition. This allows for greater flexibility in terms of hardware but it also significantly more challenging.
Here is our software running on the mule vehicle at the 2018 FSUK competition.
Applications for the team are now open, go to our Join Us page to apply!
The team is split into two subteams which work closely together to create the autonomous car.
The software team develops the brains of the self-driving car. Covering perception, mapping, planning and control this subteam uses artificial intelligence, machine learning, SLAM and computer vision we are delving into the most exciting and high-tech computer science projects to date. To achieve fast development times we work with simulations and datasets from previous runs before deploying to real hardware.
Note: We are using ROS for software development. If you are planning on applying for this team, please make sure you are comfortable with installing ROS on your personal machine.
If you are not a student but are interested in the project please contact us here.
If you are interested in sponsoring the project please look at our sponsorship page.