No driver required: Indy Autonomous Challenge is AI racing at its best

“Drivers, start your engines!” is the traditional call at Indianapolis Motor Speedway. But what if the drivers were replaced by computers? Then you’d have the Indy Autonomous Challenge, a new competition featuring the world’s fastest autonomous race cars. Purdue AI Racing recently showcased its car maneuvering autonomously around the track at speeds up to 140 miles an hour.

 

“The Indy Autonomous Challenge is really a software competition,” said Dan Williams, professor of engineering practice in mechanical engineering, and team lead for Purdue AI Racing. “It’s a spec series, so all the teams get the same car with the same sensors and the same actuators. Our students develop software that connects the sensors to the actuators to make the car go as fast as possible, as safely as possible.”

The cars are built by IndyCar manufacturer Dallara, roughly equivalent to their lower-level Indy NXT cars. But the driver’s cockpit has been replaced with a suite of cameras and sensors, as well as a powerful computer to process the data.

“Each car has redundant GPS antennas, four lidars, two radars, and six cameras,” said Sashank Modali, a Ph.D. student in aeronautics and astronautics. “We write software that interprets that data and transmits commands to the steering, the throttle, and the brakes. It also has an LTE antenna, so we receive live telemetry back at our base station.”

“It’s crazy the massive amount of data these cars can collect,” Williams said. “We can download it after the fact, but during the race it all has to be processed and acted upon live, by the car itself.”

There are no drivers in the Indy Autonomous Challenge, so it’s the engineers that become the superstars. (Purdue University photo/Jared Pike)

Start your computers

The Indy Autonomous Challenge (IAC) began its early roadmapping in 2019. After being briefly interrupted by the COVID pandemic, they began running races in earnest in 2021. Purdue University participated from the very start, collaborating on an entry with the United States Military Academy.

Since then, IAC’s profile has risen tremendously. Races have taken place in Indianapolis, Las Vegas, Texas, and the Monza circuit in Italy. This summer, an IAC car set the autonomous record for the famous hillclimb at the Goodwood Festival of Speed in England.

Purdue rebooted its IAC team last year, renaming it Purdue AI Racing (PAIR). They also recruited Dan Williams, who had previously spent 37 years at ZF working on vehicle dynamics and control systems. “This is kind of full circle for me,” Williams explained. “In 1989 I wrote a master’s thesis on near time-optimal control of racing vehicles. Then about five years ago, I noticed that paper started receiving a ton of citations again. Autonomous vehicles are now a hugely growing field, and this racing series is evidence of that.”

During summer practices at Kentucky Speedway, Williams found that their car’s current setup wouldn’t allow it to go faster than 110 miles an hour. So his team of students got to work.

“They did a great job figuring out the root causes of the issues,” he said, “working through it in a systematic and methodical way, the way engineers should. They analyzed the data and decided that at Indy, we could open up our risk tolerance just a bit.”

Members of Purdue AI Racing wheel their autonomous race car past the famous pagoda at Indianapolis Motor Speedway. (Purdue University photo/Jared Pike)

Pushing the envelope

In September 2024, Purdue AI Racing headed to Indianapolis Motor Speedway for their “hometown” race. Ten cars filled the field, coming from across the United States, Germany, Italy, and South Korea. Given its relatively new team and practice session results, Williams decided that they would only compete in the time trials, skipping the head-to-head competition.

Their target? Anything faster than 110 miles an hour. And they had just five minutes to do it.

Once the green flag dropped, the car accelerated out of pit lane (on its own, of course). Students nervously watched the telemetry from pit lane, hoping that their programming had done its job. The car zoomed around the 2.5-mile oval, headed toward the start/finish line.

The speed for lap one: 131 miles an hour.

The students allowed themselves a few tentative fist pumps and high-fives, but there were still several minutes left in the run, where anything could happen. The car continued dead straight down the middle of the backstretch. The second lap speed popped up on the scoreboard: 135 miles an hour. Another lap went by: 137 miles an hour. And then, just before the five minutes expired, they registered a speed of 139.530 miles an hour. The students’ jaws dropped, celebrating at their laptops by high-fiving their faculty advisors and family supporters.

“This was a thrill for the entire team,” Williams said. “We had confidence that we would run our fastest lap ever at this competition, but I think we were all pleasantly surprised by the results.”

“Watching this multi-disciplinary team of students put their software and controls learning into action was thrilling,” said Milind Kulkarni, the Michael and Katherine Birck Head and Professor of Electrical and Computer Engineering. “Projects like this exemplify Purdue Engineering’s drive to put theory into practice — this is literally when the rubber hits the road! I love that we are able to give students this kind of experience in the racing capital of the world.”

“This is an amazing achievement and I am extremely proud of the team’s effort,” said Eckhard Groll, the William E. and Florence E. Perry Head of Mechanical Engineering, and Reilly Professor of Mechanical Engineering. “Motorsports is all about pushing the envelope, and these students and their faculty advisors did just that. We can’t wait to see what they do in the future.”

Purdue AI Racing team members watch remote telemetry of their autonomous race car. Their car reached nearly 140 miles an hour, faster than it had ever gone. (Purdue University photo/Jared Pike)

The future of autonomy

Williams sees this as just the start of Purdue’s presence in the IAC, whose next big race is planned for Las Vegas in January 2025. “We’ve shown that we can challenge the fastest cars in this series, and we’re excited to keep progressing in the future.”

Purdue Motorsports also enjoys a hometown advantage. The Purdue Motorsports Engineering program, the only such accredited program in the country, recently relocated to Dallara’s headquarters in Speedway, Indiana — just steps from the famous Brickyard at Indianapolis Motor Speedway.

“We’re so excited to welcome Purdue students from both West Lafayette and Indianapolis,” said Williams. “Our team needs computer programmers, electrical engineers, mechanical engineers, vehicle dynamicists, and controls experts. We’re also looking for sponsorships, which is an important part of motorsports. The more folks we have on our team, the further we all can advance. This is the future, and we are always looking for ways to go faster!”

The Indy Autonomous Challenge showcases Purdues expertise in motorsports, engineering, and artificial intelligence. (Purdue University photo/Jared Pike)

 

Purdue AI Racing is part of a College of Engineering initiative for Autonomous and Connected Systems (ACS) incorporating faculty from many different schools, researching everything from agricultural robots to cyber security to human-computer interaction. It’s also a part of the University’s larger Purdue Computes initiative, whose research thrusts include physical implementations of artificial intelligence.

 

Source: Dan Williams, dewillia@purdue.edu

Writer: Jared Pike, jaredpike@purdue.edu, 765-496-0374