Your submission was sent successfully! Close

You have successfully unsubscribed! Close

Thank you for signing up for our newsletter!
In these regular emails you will find the latest updates about Ubuntu and upcoming events where you can meet our team.Close

Taking Ubuntu for a spin (literally)

This article was last updated 1 year ago.

The designers of the Indianapolis Motor Speedway never could have predicted that unmanned autonomous vehicles would someday race on their track – much less robots that can see the checkered flag while their ‘drivers’ kiss the bricks. But after more than a century, what started as a gravel-and-tar track hosted the most advanced driving competition to this date. And in the process, it made history. Let us tell you how.

On Saturday 23 of October, nine teams raced in Indy to see who was the fastest. A total of 21 universities from 9 countries competed, programming Dallara AV-21 racecars to win and take home $1.5M in prizes. The year-long challenge for innovating the field of autonomous vehicles started with more than 25 teams, and finished with nine finalists.    

While many skilled teams took part, open-source won the day. It powered the cars and teams, helping them shape the future of autonomous vehicles. It was also seen in the collaboration between the teams, and between competitors during the weeks before the race. ROS was there, and Ubuntu as well. Focal Fossa donned his racing suit and drove in the AV-21.

For more than ten years together, we’ve been powering robotics developers, and we couldn’t be more proud of our involvement. Ubuntu 20.04 provided the foundation for this exciting competition, powering every phase from simulations to execution. You can find everything that was used in the winning car here.

Autonomous vehicles challenges

The first comparable racing challenge was DARPA in 2004, where competitors developed unmanned vehicles that drove through the Mojave Desert region of the United States along a 150-mile/240 km route. None of the autonomous vehicles finished the race. Luckily for the fans, this was not the case in Indianapolis.

While the first DARPA challenge spurred the development of the technologies needed to create the first fully autonomous ground vehicles, the Indy Autonomous Challenge expanded our knowledge about solving this edge case scenario. Autonomous cars going at more than 150 mph experience challenges that occur only under extreme operating parameters, such as processing massive IO of logging data in real time, avoiding unanticipated obstacles at high speeds while maintaining vehicular control, air turbulence, warming up the wheels and weaving, general safety, and more. 

Racing autonomous vehicles on

The competition didn’t start smoothly. The weather was colder than expected and the rain washed away the rubber on the track. A cold track lacking traction poses a challenge for any racing car. Therefore, the teams needed to reconsider their strategies. Abstracting to a high level, teams needed to decide on the speed on the front-straights and during the turns. The times each competitor posted during the trials and the race heavily informed their thinking.

During the first round of competition, teams needed to leave the pitlane and complete a warm up lap, followed by two timed laps and a cool-down lap. In this final lap, teams had to navigate around several inflatable barriers on the front stretch to test the cars’ ability to avoid obstacles. Only the 3 fastest teams would move on to the final round. The final round also involved cars leaving the pitlane, four warm up laps and two timed laps. The team that registered the highest average speed on the final two laps would win it all. 

PoliMOVE, made up of University of Alabama and Politecnico di Milano, opened the first round and gracefully completed the challenge with an average speed of 124.450 mph. Technische Universitat Munchen (TUM) Autonomous Motorsport momentary jumped to first place at 129.237 mph. Finally, EuroRacing, made up of schools from Italy, Switzerland and Poland, averaged a speed of 131.148 mph in this first round, becoming the round’s fastest team and surprising all the competitors after a rough week. These were the three teams that moved to the final round. 

From the remaining teams, only two were able to finish the race. The South Korea team, KAIST, completed a two-lap run of just 84.355 mph. The Cavalier Autonomous Racing (CAR), from University of Virginia clocked at 119.883 mph, but unfortunately ran an unexplained extra lap, forcing the judges to disqualify it. This showed that the cars were indeed fully autonomous – once they were racing, nothing could be done to change their operation. And this behaviour returned in the final round as well. 

From the remaining teams, Black & Gold Autonomous Racing failed to make it past the pit exit when their car crashed with a system malfunction. Our friends at AI Racing Tech from Hawaii unfortunately couldn’t complete the first round either. Their car took off into the infield before Turn 1 and spun dramatically on the grass. A technical problem with the GPS cost them the race. We are proud of their work, and as our sponsored team, we would like to thank them for the exciting ride. 

The cars from MIT, University of Pittsburgh, Rochester Institute of Technology and the University of Waterloo closed the first round, and took all the media attention. Unfortunately, the car crashed. A GPS failure triggered the car’s safety stop that made it drive down the front-straight and made a hard-left turn into the inside wall.

Yes, another wreck. But it was not only the MIT-PITT-RW team and the AI Racing Tech team that experienced these issues, but also the PoliMOVE and Black & Gold Autonomous Racing teams. The EuroRacing team experienced this same problem during the week before the race, forcing the team to change the entire system.

At these speeds, navigation is nothing without GPS. Lidar sensors can’t keep up. GPS is a key technology in autonomous mobile robots, as it allows them to calculate their positioning to address their navigation and timing. Accordingly, if any of the competing cars had their GPS jammed, they executed their safety stop. But, as seen with the MIT-PITT-RW car, sometimes this was not enough. After losing its two GPSs, the car activated its emergency mode, but it crashed anyway.

Which was the fastest autonomous vehicle?

Despite the crashes and setbacks for many of the competitors, the final round went off without a hitch. Our three finalists rushed to simulate new speeds after the first round and assigned these parameters to the cars. With so much code being changed, there was always a chance that mistakes could happen. But the final round was not waiting for anyone. The automotive industry was watching. 

PoliMOVE opened again, warming up its tires, getting a good pace and following the lines. They stayed away from the wall, and remained stable as they were going at 108 mph. But after the first lap, they lost the first GPS, and after the second they lost the second. The team that won the virtual challenge and became the benchmark for the final race was out before long. 

TUM went second. Their car started slowly, warming up the tires so it could go faster. But that changed as soon as the tires were ready to race. The team sped at 151 mph on straight sections and slowed down during the turns. When their car completed the last lap, it had registered a massive ~136 mph. 

This put all the pressure on the EuroRacing team. A strong competitor throughout the competition, they nonetheless raised some eyebrows due to hardware issues during the last week. The team needed to change its entire GPS system. And despite never testing it before the Saturday race, they posted the fastest speed on the first lap. The car started slow, below 30 mph, weaving back and forth trying to gain speed. It was beautiful to watch. With the green flag, we started our last two laps of this year’s challenge. And EuroRacing delivered one of the most graceful laps we’ve ever seen. With an average speed of 139 mph, the car was consistent throughout both the turns and straights. For some turns, it was even running at 144 mph, which was never before seen. It was looking promising – but it proved too good to be true. 

The second lap started, and the car was slowing down. It looked like there was an issue since the car had stopped its amazing behaviour. But there was nothing the team could do. The car was simply going too slowly, and the average speed was dropping and dropping. As it turned out, they didn’t program the car to complete the two laps – only one.

TUM took the glory after this logistical error. Or should we say ‘human error?’ 

Open source robotics won, so did you

You could be disappointed with that outcome. I was. But it helps to remember all the projects that you worked on throughout the years, all the robots and code you put together trying to reach various deadlines, and how you often had to test and deploy everything at the last minute. We were all students at one point, making these or bigger mistakes. But how many of us were working on a Dallara AV-21, speeding at more than 150 mph? Kudos to all the teams for achieving the unthinkable!

Covid also impacted this challenge, making it nearly impossible for teams to work together for some time. Much of the work was done online, sharing screens during video calls while building this sophisticated software stack. This added barriers to an already challenging activity. But all in all, it was a success.  

I like talking about technology perception; that is to say, the impact of deployed technologies in society. Here we can see a remarkable case. We see open source software, such as ROS and Ubuntu, running autonomous cars at more than 150 mph – cars that are not taking any input from the teams once they start racing. This shows how much progress the community has made, but more importantly, it proves that open source is all you need to bring this future to life today. 

The most advanced software available today for autonomous racing cars was droved in Indy. Teams are making these codes open, due to the global benefits they represent. These algorithms will end up in research centres, industries, and startups, all of whom will build upon them. The impact of this competition will live on for many years. And we at Canonical are proud to keep supporting innovators.

Talk to us today

Interested in running Ubuntu in your organisation?

Newsletter signup

Get the latest Ubuntu news and updates in your inbox.

By submitting this form, I confirm that I have read and agree to Canonical's Privacy Policy.

Are you building a robot on top of Ubuntu and looking for a partner? Talk to us!

Contact Us

Related posts

Discover the future of Autonomous Mobile Robots (AMR) with Canonical and Husarion at ROSCon 2023

At a time when Autonomous Mobile Robots (AMR) keep rapidly reshaping industries like logistics and manufacturing, we’re thrilled to introduce you to our...

Do more with Vulcanexus snaps; your all-in-one ROS 2 toolset

eProsima’s advanced toolset for ROS 2 is now available in the Snap Store A surprise is awaiting all ROS developers in the Snap Store with the release of the...

ROS architectures with snaps

Choosing the right architecture can be hard, but we want to help. In this blog, we’ll look at different architectures and their pros and cons. We’ll also show...