INDIANAPOLIS—At a block party down the road from Indianapolis Motor Speedway (IMS) the day before the Indy 500, engineering students from across the world gathered for a very different kind of car reveal. The AV-21, with its sharp, green-blue Dallara chassis, looks like a typical Indy Lights racecar. However, there’s one notable exception: no driver. Its cockpit is crammed with wires, processors, sensors, and a motherboard—essentially, a supercomputer.

“It’s shaped like a car, but I don’t view it as one,” said Chanyoung Jung, an engineering student from the Korea Advanced Institute of Science and Technology. “I view it as a robot.”

Automobile, robot, or something in between, the AV-21’s unveiling was a significant step for the Indy Autonomous Challenge (IAC), a competition set for October 23 in which driverless cars will race wheel-to-wheel around IMS at potential speeds of more than 180 mph. The winning team will earn a $1.5 million university prize. Organized by IMS and Energy Systems Network (ESN), an Indianapolis-based nonprofit, the race seeks to significantly advance the safety and speed of autonomous vehicles.

And—if you ask many of the participants—it might even shape the future of motorsports.

A proving ground for new tech

The Indy 500 has long been the home of automotive innovation: rear-view mirrors (1911), seatbelts (1922), and four-wheel drive (1932) made their debuts at the iconic race before becoming standard features in consumer cars. That’s why, according to ESN’s President and CEO, Paul Mitchell, the Speedway is the perfect track to showcase the safety of fully autonomous vehicles.

“It could wake up the average motorsports fan and the average citizen,” Mitchell said during his remarks at the block party. “They’ll say, ‘Gee, if these cars can go 180 or 200 miles-per-hour without a driver, then maybe [an autonomous] car on the highway is something I can feel comfortable with and safe using.”

The upcoming competition is, in many ways, the spiritual successor of the DARPA Grand Challenge, a robotics race from the early 2000s. Back then, the winning team from Stanford—led by famed computer scientist Sebastian Thrun, who is now an advisor to the IAC—completed the 132-mile race in nearly seven hours. The average speed was about 19 mph, a monumental achievement at the time that seems mundane by today’s standards.

“If you could bring back the excitement of the DARPA Grand Challenge,” Mitchell continued, “and apply it to a really challenging edge use case, like high-speed racing, then that can leap the industry from where it is to where it needs to be to help us realize our autonomous future.”

The engineering students competing in the IAC are an eclectic group, representing universities on four continents. Many of them are massive motorsports fans, like Italian engineering student Andrea Finazzi. (At the car launch, Finazzi was lamenting Ferrari star Charles Leclerc’s misfortune at the Monaco Grand Prix the week prior.) Other students are more interested in robotics than cars, and a few—like Nayana Suvarna from the University of Pittsburgh—don’t even have a driver’s license of their own.

“It’s kind of a big irony, I know,” she said with a laugh.

Many of the students are hopeful that this race could prove the value and intrigue of fully autonomous racing, with the concept eventually working its way into various aspects of motorsports, if not spawning a dedicated series of its own.

“I can see this going down like Formula E,” said Abhishek Bhagwat, a student from Clemson University who helped design the AV-21. “Autonomous racing is something completely different from Formula 1 or IndyCar, with its own set of unique challenges… and this is how you start it, by building something cool and inspiring the next generation to push it forward.”

Finazzi, meanwhile, suggested that wheel-to-wheel races between humans and AI seem like the inevitable next step. “Maybe within 10 years, a fully autonomous F1 car can challenge a champion like Lewis Hamilton, similar to how AI challenged chess masters,” he said, alluding to the infamous Kasparov vs. Deep Blue chess battles that dominated the media in the late ’90s.

Overtaking isn’t easy

That being said, the question remains: How exactly will these engineering students get their cars to zoom around IMS at breakneck speeds without a driver and without crashing? According to Jung, it comes down to three components: perception, planning, and control. The car first utilizes its cameras and sensors to perceive its surroundings, using that data to plan its next maneuver. Once a path is devised, the optimal steering controls are calculated and enacted.

“But our goal, really, is to overtake,” Jung explained. “So we have to determine the correct way to pass, collision-free.”

For Alexander Wischnewski, team leader from the Technical University of Munich, this is where the fun begins. The IAC’s unique challenge is that the cars will be racing wheel to wheel and will have to account for the unpredictable movements of rival cars.

“Here’s an easy and interesting example,” said Wischnewski. “Let’s say you put your vehicle on a safety margin of two meters all around, meaning you must stay away from all other objects around you by two meters. What do you do if someone else’s car moves into your safety module?”

The true challenge, then, is to develop an algorithm that allows the racecar to react and adapt to unpredictable scenarios, including other cars that “break the rules.” Wischnewski noted that this could have an enormous impact on the safety of autonomous cars as they begin to share the road with human drivers in the future.

“Because on the road, no one adheres to the rules!” he joked.

IAC competitors are also grappling with another unique problem: How can they ensure the car can process all of this data while moving at 180 mph (290 km/h), approximately a football field per second?

“You set up something called a ‘shared memory pool,'” said Wischnewski, adding that doing so could shave valuable milliseconds off of the car’s processing time. “And, of course, you try to design the algorithm as efficiently as possible.”

The IAC hosted a computer simulation race on June 30, providing an opportunity for the teams to test their algorithms against one another in anticipation of the on-track competition four months later. One of the main question marks hanging over the race is whether a team will push its car to the limit and successfully break the current autonomous land-speed record of 175 mph.

Our autonomous future

Less than a decade ago, car companies were investing more in autonomy than electrification, according to Steve Patton, Americas Mobility Sector Leader at Ernst & Young. Patton consulted Clemson students during the creation of the AV-21. But negative press surrounding driverless cars, as well as advancement of electric batteries, flipped that. Electric vehicles are now in the midst of a massive surge in adoption—and motorsports have helped to fuel that.

“Formula E was a little fringe thing at the beginning, with people sitting back and saying, ‘OK, this is kind of fun,'” Patton said. “Once it becomes a bit more mainstream… all of a sudden there’s acceptance of the technology, and that was a bit of a tailwind for electric vehicles, I believe.”

The same could be true for autonomy. Patton points to the IAC—as well as Roborace, a hybrid human/machine time-trial contest championed by racecar driver Lucas Di Grassi—as important methods of raising public awareness.

“We’re continuing to try to prove out that the new technologies are as good as the old,” Patton noted. “So we’ll see the same thing with automation versus human drivers.”

And once there is more acceptance of automation, and once technology has advanced just a bit further, Patton said it will be as if a flip has been switched in our daily lives, especially in certain sectors like delivery vehicles.

“It won’t be too long until, maybe on a ubiquitous basis, we see an autonomous vehicle,” he said. “Although it’s not likely you’ll go out and buy an autonomous car any time soon.”

As for the IAC, competitors will continue to tweak their algorithms and test their cars in the months ahead, hoping to outclass the competition and take home the prize money. While a dedicated, fully autonomous racing series could be an interesting future project, most competitors, like Jung, are focused right now on the race at hand—and one goal in particular.

“No crashing,” Jung said with a laugh. “Our main goal is to cross the finish line in one piece.”

Listing image by Gregory Leporati

https://arstechnica.com/?p=1778606