The team Unicamp E-Racing, formed by more than 70 students from the University, it became the first team in the Americas to have a car without a driver for Formula SAE, a student competition organized by Society of Automotive Engineers. The prototype of the car was ready in 2019 and the developers' expectation is that it can be used for competitions in Europe in 2021.
Also known as an autonomous car or, in English, driverless, the model developed by the E-Racing team, uses artificial intelligence technology. Students Nikollas Javier, Gabriel Almeida and Lucas Costa e Lima, who are part of E-Racing, highlight that this type of car for Formula SAE competitions does not yet exist in Latin America or the United States.
“The part of driverless It only exists in Europe. We started researching in 2018 and in 2019 we made the first functional prototype and it was already running and working. We set the goal that the driverless It will be the team’s focus from now on. As we are recognized as the first formula in Latin America to be driverless, we have this prerogative. The goal is to compete in Europe in 2021. We are now rebuilding the prototype in 2020, adapting and improving it”, says Nikollas, who studies Mechanical Engineering at Unicamp and leads the E-Racing team.
Lucas, who studies Medical Physics and worked directly on developing the software part, explains how the car works. “What every autonomous system tries to do is imitate what a human would do. As our competition is marked by cones, the idea is to identify where these cones are. We use a camera that simulates human vision, with two objectives, which would be the eyes, and it can identify the cones using an artificial intelligence algorithm, a neural network”, he points out.
To do this, it was first necessary to train the system with images of the cones. “It's basically taking several images of cones, saying where the cones are. Afterwards, the algorithms can identify it alone. After that, the process is to try to locate the car in relation to the cones. For us it is intuitive, to look and have an idea of where they are, but for the car it is not intuitive and the process involves cons and complex mappings”.
Furthermore, it is necessary to train the car for the trajectories to be developed in competitions. Lucas points out that this step is less complex, and the middle of the track is used as a reference to make the car follow this parameter.
The most difficult thing about developing an autonomous car, according to the student, is not implementing a way to do this, but designing this approach, thinking about how to solve problems without copying teams that are already a reference in the subject. “There is a team from Zurich, which published a paper in 2018 about their model and what many teams do today is look at their paper and copy it. We didn't copy, we went for another way of making autonomous cars, which in our opinion is much better. It is a light, fast, very reactive system that can make decisions very quickly. The results are very good”, he assesses.
Goals
With the prototype ready, which is autonomous but can also be driven, E-Racing is now working on improving it. Although the pandemic does not allow group meetings and workshops, students say that even with the restrictions, more people joined the team, through an online selection process, and the goals continue to be strengthened.
Unicamp E-Racing, which works with electric cars, is already renowned for holding the record for the fastest car in the Americas, reaching acceleration from zero to 100km/h in three seconds. Furthermore, the team came first in all Brazilian competitions in the electric car category since 2011, with the exception of 2015, and is two-time world champion. Now, as the students point out, the goal is to focus on the autonomous car and the 2021 European competitions.
“The idea is that we can participate in as many competitions as our autonomous car can compete in,” says Nikollas. Even with less financial investment, he notes, the idea is that this disparity can be alleviated with performance. “We can overcome this with the performance of algorithmic logic. We see a lot of potential in getting good results there even if we have a big financial difference.”
Boosting technology in the country
All codes used by the team in developing the autonomous car are freely available to be consulted on the internet, according to Gabriel, a Chemical Engineering student. For him, this is important to encourage more Brazilian teams to be motivated to create driverless cars, helping Brazil to become a reference on the subject. Thus, there is no greater selfishness or competition than this goal. “Our code is open and free in open source. Other teams that have no idea how to do it can look and get an idea”, says Gabriel.
Check out the car's promotional video made by E-Racing: