Breaking new ground: The Abu Dhabi Autonomous Racing League
On April 27, the Abu Dhabi Autonomous Racing League, or A2RL, will attempt what has never been done before: put four autonomously driven racecars on track to compete together for the first time. Cue the cries of the robots taking over.
Well, maybe not.
“The gladiatorial element of humans competing on a field of play is not something we ever intend to replace, and never should be,” says Dr. Thomas McCarthy, Executive Director of Aspire, who have created this $2.25 million Grand Challenge that has brought the best and the brightest in coding, robotics and race engineering together to push the performance boundaries of autonomy in the context of driving to the furthest point currently imaginable.
The race will take place at the Yas Marina Circuit using identical Dallara SF23 cars currently used in the Super Formula Series in Japan. Competing are eight teams from around the world, including Code 19 Racing from the United States.
The cars themselves will be looked after by a team of independent mechanics and technicians in terms of car preparation, set-up and maintenance. At present the teams are only competing on the basis of which one can develop the most effective code derived from what the AI learns from a multitude of sensors and cameras on the car. Indeed, the only connection the teams have with the car is when it returns from the track and they can gather and process the data accrued. Once the car leaves the pit box again, it does so completely autonomously without any human interaction except for a kill switch as a safety measure.
Teams have already been testing since early March of this year at the Yas Marina Circuit. Come race week, teams will first compete in single-car style qualifying. Successful teams will then square-off in a one-on-one “attacker-defender” style race with the four most successful teams finally competing wheel-to-wheel in a multi-lap final.
Adding to the drama will be an ‘AI versus Human’ race that will see ex-F1 driver Daniil Kvyat go head-to-head with an autonomous car. In that race, Kvyat will get a head start and the challenge will be to see if the autonomous car can close the gap.
So, if the goal here isn’t to eliminate human drivers, what is it?
“There has been a lot of conversation about a future centered around ‘driverless cars,’ but we believe that such a thing would be massive mistake,” argues McCarthy. “There has been a long history of technology created that has either not been accepted by people within the market, or it has been used in ways other than how it was intended. Consumer acceptance of technology is critical.
“Moreover, the automotive industry has always been focused on what we would call ‘secondary safety,’ meaning what happens in the event of and immediately in the aftermath of a crash,” McCarthy continues. “We are trying to promote technology that focuses on ‘primary safety’ to prevent the crash from happening at all. But for that to happen, drivers need to not just accept AI as a co-pilot but demand it. So, to develop that technology and prove its capability to the point that people will accept it, we created this race series.”
Code 19 Racing, the first professional team, is led by Lawrence Walter. He has assembled a diverse team of male and female engineers from a variety of backgrounds with experience in motorsports, automotive, robotics and coding. Along with the pro team running the A2RL car, there’s a junior team of STEM students brought together by the Boys & Girls Club of Southeastern Michigan fielding a scaled down racecar with similar autonomous capabilities.
“What a racecar driver does is among the great feats of human athletic ability,” says Walter. “We are using this as a platform for our start-up to recruit and develop talented minds in this field by giving them a monumental challenge of racing a car completely through the AI code they write and develop.”
While aviation has utilized advanced auto-pilot systems for decades, Walter points out the challenge here is much higher.
“Planes have a lot of airspace to work with and large runways to land on. Being off by a few feet isn’t likely to negatively affect them. But in a racecar going for the best performance, being off by a few centimeters could put the car in the wall, and that would be a bad day,” he says. “The individuals on these teams are all highly competitive and they are pushing the limits to develop the incredibly complex coding needed to get the maximum performance out of the racecar.”
Walter adds that racecar drivers succeed by sensing what the car is doing and analyzing the conditions and then filtering that through their experience developed over many seasons of racing to determine what their inputs are. A2RL coders are having to take the data from an array of sensors along with what the AI program has gathered and learned to develop new or additional code to refine the autonomous process. Just as race engineers help a human driver, they are striving to help the AI improve its performance.
A big part of that is developing the code that allows the AI to learn and understand what effects things like tire degradation and fuel load have on performance to make the correct driving adaptations.
Interested to see how it all pans out? Follow all the action live from Abu Dhabi, April 27, on A2RL’s YouTube or Twitch channels.
Anyone with virtual reality goggles can get an even more immersive experience. Visit a2rl.io to learn more.
