The Luddy School of Informatics, Computing and Engineering’s Autonomous Racing team doesn’t just keep pushing the boundaries of what’s possible in autonomous racing, it does it faster and more efficiently than anyone else in the renowned Indy Autonomous Challenge.
If it leads to world-changing breakthroughs in building safer cars for everyone, all the better.
“If something works in a race car,” said Indiana University Luddy Team lead Durgakant Pushp, a fourth-year Intelligent Systems Engineering Ph.D. student, “it’s close to being transferred to society.”
Hassan Jardali, a third-year ISE Ph.D. student who is the planning leader focused on developing features for the car’s controller added: “If we can control our race car at high speeds, the same principles can be applied to normal roads, enabling safer and faster navigation among people.”
The IU Luddy Team once again surpassed the Indy Autonomous Challenge norm, this time during January’s Las Vegas Motor Speedway (LVMS) time trial. The team overcame limited resources and minimal artificial intelligence driver on-track developmental time to do what few expected and fewer could match.
Despite windy conditions, the IU Luddy team’s 1,400-pound Dallara AV24 race car reached a top Las Vegas speed of 163 mph during the 12-lap time trial. That was 35 mph faster than what it achieved during last September’s time trial at the Indianapolis Motor Speedway.
“We are among the top teams even though we have the least practice time,” said Lantao Liu, Luddy associate professor of Intelligent Systems Engineering and Computer Science, and team director through his Vehicle Autonomy and Intelligence Lab. “That is amazing.”
Even more impressive -- the car has not yet reached its top-speed limit.
“To push it to 163 was a dream outcome,” Pushp said. “We hadn’t planned for it, but the car was stable, and our preparation was good. Everyone is happy with the result.”
Lack of track test time that could have been a deal breaker was overcome by the team’s ambitious agenda and impressive dedication. Countless hours were put in through simulation, planning, research and sheer hard work.
“I’m very proud of our students,” Liu said. “They are really dedicated and efficient. They are amazing. They might be more ambitious than I am.”
Ambition comes from opportunity IU and the Luddy School provide through Dean Joanna Millunchick.
“I appreciate the school’s support,” Liu said. “Dean Millunchick has supported us. Our students are very encouraged that they have done this.”
Jardali praised the Luddy School and IU for “providing a special platform to test our research.” He said team members are fortunate to have an autonomous race car to work on that delivers stable performance at high speeds.
“We do applied research in robotics and engineering,” he said. “With the racing car, we don’t bind ourselves with just simulation.
“The combination of research and engineering application is my motivation. I can apply what I’m researching on with this car.”
Since getting its $1 million Indy Autonomous Challenge Dallara AV-24 car, powered by a 488 horsepower Honda racing engine, last July and August, the IU Luddy team has had 11 hours of track testing time last summer and 6.5 hours this winter. That is far less than the other nine participating teams from around the world, many of which had at least five days of track time.
“We are among the elite of global autonomous racing teams,” Liu said. “Considering our lack of track time, that is amazing.”
How fast can the IU Luddy Team’s car go?
“We’re not yet full throttle,” Liu said. “We have not touched our limit. At 163 mph, the car was stable, so how fast we can push? We don’t know.
“We have to push step by step. The reason we have not reached the limit is we haven’t had enough track time to do it.”
The IU Luddy team has rapidly closed the speed gap with the top teams. In the fall, the difference was around 50 mph. Now, it’s less than 10 mph.
“From where we started and how far we’ve come,” Pushp said, “I’d like to maintain that pace. If we put forth a little more effort and work the same way we have been, we can become among the top teams in terms of speed.”
Added speed has come even as the IU Luddy team reconfigured its software and tested a new controller this winter. It had on-track sessions in Las Vegas in December and in January.
“We went from zero and achieved 163 mph in only 6.5 hours of winter on-track testing,” Liu said. “That’s hard to imagine. We used every minute as efficiently as possible.”
Other teams noticed.
“Our team is very good and very respected,” Liu said. “I heard a lot of positive discussions from other teams talking about Indiana. Our students are like soldiers, favored and respected by all performers from all teams based on what I heard.”
The Indy Autonomous Challenge, with a total of 18 teams from around the world, is part of a collaboration with the U.S. Defense Advanced Research Projects Agency. The goal is to improve AI training for autonomous systems and correct the simulation-to-real-world gap in that training.
Competition began three years ago, although the IU Luddy team is only in its first year. It has run 305 miles in its 17.5 hours of on-track testing with only one crash. No other team can match that.
Other IU Luddy team members who participated in this LVMS competition are ISE Ph.D. students Mahmoud Ali, Youwei Yu, Ihab Mohamed, Md. Al-Masrur Khan and Alejandro Murillo Gonzalez, along with ISE undergrad Paul Quigley.
In addition to those present at the competition, several other dedicated teams members worked behind the scenes from Bloomington, providing support for data processing and analysis.
The Indy Autonomous Challenge features a progressive three-tiered format. Tier 1 is a time trial with one car on a track trying to attain its fastest speed. Tier 2 is a passing competition between two cars. Tier 3 is a more advanced passing competition between three to four race cars in an open racing format designed to test the limits of AI driver multi-agent interaction and high speeds. Only a few teams have advanced to the Tier 3 stage.
The IU Luddy team hopes to participate in Tier 2 and Tier 3 at the next race, set for this summer at the world-renowned Monza Circuit in Milan, Italy.
“We want to be Tier 3 as soon as possible,” Jardali said.
From a technical standpoint, Pushp said the IU Luddy team has developed all the required Tier 2 codes, and validated the car’s detection, competence and adapted cruise control. It has successfully over-taken vehicles in simulation, but not yet on a track.
“We need more practice and detailed work,” Pushp said.
The work continues. After the Las Vegas event, Liu attended an autonomous racing symposium at the University of California-Berkeley to help develop new strategies that will not only produce more impressive results, but also deliver career-lasting benefits for Luddy students. Pushp hopes more resources will be provided so that more IU and Luddy students will participate moving forward.
“This helps our students a lot,” Liu said. “They are working on real-world advanced cars and real-world things. We are pushing the boundaries of real racing to see how far we can go.”