Italy’s Formula 1 city of Imola has hosted Europe’s first multi-car race for fully autonomous cars.
Five teams from the Abu Dhabi Autonomous Racing League (A2RL) competed at the historic Imola circuit, with cars racing without steering wheels, pedals or even drivers.
The teams represented Germany, Italy and the United Arab Emirates (UAE).
Even if the cars drive themselves, making them competitive still requires a team of humans capable of developing the best AI-driven racer.
“The teams that come here are all about winning,” said Stephane Timpano, CEO of ASPIRE, the organisation behind the event.
“They are scientists. They look like scientists, and they behave like scientists, but once the competition starts, it’s about showing that they have the best software, the best code in the car, and what they want is to win”.
All teams used the same SF23 single-seaters developed by Dallara for Japan’s Super Formula series, regarded as one of the fastest racing championships outside Formula 1.
But the human controls have been taken out. Instead, the cars use cameras, radar and lidar sensors to perceive their surroundings, while an onboard computer controls the car.
Just like in any other motorsport competition, mechanics work on the cars, engines roar and engineers closely monitor streams of telemetry data.
Organisers also hope putting autonomous cars on a familiar racing stage can help build public trust in AI.
“Create trust. I think this is the biggest issue we will have with AI going forward,” Timpano said.
“A lot of people are reluctant to engage with AI, are worried about what is going to happen with their jobs, their life and what AI is able to do,” he added.
Pushing autonomous driving to its limits
For autonomous driving experts, however, the competition is fought largely through programming, simulator sessions and the constant search for a balance between speed and risk.
“The league addresses one of the biggest challenges in developing autonomous vehicles, which is testing risky scenarios in a controlled environment, at high speeds,” said Alexander Winkler, A2RL’s head of sporting.
He said racing against other vehicles travelling at similarly high or different speeds creates situations and data that would be difficult to reproduce in ordinary road testing.
“To me, as a professor, this is probably one of the most difficult engineering problems you have today. It is not just autonomous driving, but autonomous driving at the very, very limit of handling,” explained Marko Bertogna, a professor at the University of Modena and Reggio Emilia.
And once the race starts, the engineers have to let their code do the driving.
“When the race starts, it’s hands off,” said Sascha Buettner, team lead of TUM, the participating team from the Technical University of Munich in Germany.
“And so this is basically just letting a child do its thing and just looking at them and, you know, wishing them all the best, essentially hoping to be proud. But, of course, that can go wrong. Our AI driver can stumble, it is all possible.”
Despite the cars now approaching professional drivers in outright performance, organisers say AI still cannot reproduce everything a human brings to racing.
“Still today, this car is not able to reproduce the fantasy of a driver. You don’t see that in the car,” Timpano continued.
“With time, it may get there. But I see it the other way around. Our human pilot can use AI to have a better performance in the safest way.”
The winner of Saturday’s 12-lap autonomous race was Kinetiz from the UAE.
Founded in 2023, A2RL is based in Abu Dhabi in the United Arab Emirates.