Pushing the Boundaries of Automated Driving
The Indy Autonomous Challenge stages events in which high speed defines the operational design domain.
In late April 2022, SAE Media editor Bruce Bennett spoke with Paul Mitchell, CEO of Energy Systems Network, a 13-year-old nonprofit organization that organizes and manages various public-private partnerships in the energy and transportation sectors. Energy Systems Network created the Indy Autonomous Challenge (IAC), a technical competition involving nine teams from 21 universities competing in a series of driving challenges to develop software and systems capable of controlling fully automated racecars at speed. An edited version of their exchange follows.
Explain the Indy Autonomous Challenge and how it went from concept to reality.
The Indy Autonomous Challenge started out three years ago as a prize competition. It was hoping to bring back the excitement of, and the achievements that occurred, when the DARPA Grand Challenge ran in the 2004-2005 timeframe. It really jumpstarted the autonomous-vehicle [development] industry, and we thought, you know, what could we do that would get the level of commitment and interest from top universities, not only across the U.S., but actually make it global and get International Universities involved.
We set it up as a as a prize competition with a million-dollar prize and we brought university experts together and helped design the rule set with their input, which was important. And where we wound up was that the teams would compete by writing software that would control a fully autonomous race car running at the Indianapolis Motor Speedway. Each of the teams would be given the same car.
That was different from DARPA. DARPA involved the teams having to design the car and the software. We debated that, but came to the conclusion when you're talking about racecars, it's hard to design a racecar. So the competition is around the software to control the vehicles.
On October 23, 2021, we ran the first event at the Indianapolis Motor Speedway and gave out that million-dollar prize to the German team from Technical University of Munich. At that point, that was potentially the bookend of the Indy Autonomous Challenge. That's what we set out to do. We wanted to get the best and brightest minds from around the world to advance the state-of-the-art of autonomous technology and maybe win some hearts and minds along the way by getting people exposed to this technology. And we decided to go beyond that.
Tell us about some of the technology behind these racecars.
The technology is going to be common to what you see on autonomous vehicles that are being tested around the world today. We've got a number of different sensors: lidar, radar, optical. We've got an advanced computer onboard, essentially a supercomputer. We've got drive-by-wire systems, GPS systems for localization. Then we've got wireless-communication technology for the vehicles to communicate to race control and to each other through our partnership with Cisco.
All of those technologies are on vehicles or are used in testing and development of autonomous vehicles today. The real difference is, we're pushing them to the absolute limit of their performance by running them at 150 mph (241 km/h), 175 mph (282 km/h).
Any plans to go electric at some point in the future?
Right now, we have our hands full just trying to get the autonomous systems to operate and to work, and our goal is extreme high speed. And so it's really important that we keep going faster, because our objective is to prove out the hardware and the software at these extreme speeds so that when human [-occupied] cars encounter each other on the highway at 100 mph (161 km/h), they can use ADAS systems or autonomous systems and do so safely.
I believe that at some point, we will have autonomous vehicles that can safely travel on highways at 150 mph or more. Think about the time that we’d save. Think about if they can do that and do so more safely, then we have a much better supply chain. We have a much better quality of life and so on and so forth.
Is this just a technical and engineering exercise, or are there plans to turn this into a real racing series someday?
It’s more than a technical exercise. That's its main objective, but it also seeks to be a way to attract talent into the industry. Not just engineering talent. Vehicle dynamics, business- development talent. When they see autonomous race car passing another autonomous race car at 170 miles an hour, that wakes them up to a future that they can actually get excited about and think differently about.
I don't see us running a proper series where you're running, you know, 10, 12 events, even eight events a year, and they're back-to-back, and you're going from city to city. I would say it this way: We will continue to run events, but our events will be based on the timeline of technology advancements that we want to showcase.
Do you think racing fans will pay to watch driverless computer-controlled cars compete?
I don't know. They might, I have no idea. But that's not something that we're going to test. We don't view this as something that is a spectator sport where we're selling tickets. Our events will have fans. I think we'll eventually have lots of fans. But those fans may be people that are attending conferences like this at AUVSI [Association for Unmanned Vehicle Systems International] or at CES, where we ran in Las Vegas, people that understand the technology or are interested in it.
The first challenge event was held at IMS last year. And then you held the second event at CES. Did they go according to your expectations?
October  was not as grandiose as we had hoped, at Indianapolis Motor Speedway. And then Las Vegas was more than I ever expected. So, the difference was that that we really wanted to have cars on the track at the same time passing each other at the Indianapolis Motor Speedway. And we just frankly ran out of time. Weather didn't help us either. It was cold. The cars were spinning out because of tire temperatures.
But the ability to have the cars racing head-to-head was always what we were seeking, and that's what we did in Las Vegas. And the format of that competition, two cars at a time passing each other at increasingly higher speeds until one either gives up – or there's a wreck – is really exciting.
You just completed two days of testing at the Kennedy Space Center and you set a new automated-vehicle speed record. What were you doing and what did you learn?
It was a test run to figure out how the new engine package was going to work. We wanted to boost the speeds of the cars because we kind of hit the limit in Las Vegas, and we want to make sure the next time these cars run they can go faster. It just happened to be that the new package went over 300 km/h (186 mph), so that resulted in a new land speed record for autonomous vehicles. We’re honored and excited to have that record. I'm confident we'll probably continue to break it over the years.
Can those monitoring the cars take control if they see a problem?
No. Race control that is monitoring all the cars has the kind of red button, and that kill switch is electronic and it's sent to the car. So we can tell the car to turn off. We can also tell the car to come back to the pits. So we can give it a black flag. All the flags are electronic. There's a green flag, a yellow flag, a white flag, a black flag. We have our own flags, like a purple flag, which is our version of kill all the cars on the track at the same time.
Are there any plans to hold an actual race with a full field of cars, to see how intelligently they interact?
Not anytime soon. We did originally. Our original hypothesis was that the Indy Autonomous Challenge would be a full field of vehicles running against each other. And I think we realized that while we could do it, that the reality is that it wouldn't be a very productive use of the assets in the sense that, inevitably, some of the cars are going to mis-perform, just like you have in a race. That's going to cause wrecks, cause accidents. And with our vehicles, one of the challenges is, if it wrecks or causes an accident, there's not a driver [to guide the racecar out of danger].
To spend the time programming the algorithms to [handle] vehicle recoveries and safety spins and all that kind of stuff, it's not really the best use of the coder’s time. What we're trying to do is validate vehicle interactions and dynamics at high speeds, which is replicable to the commercial industry.