In the following interview, I sat down with Matt Peak, one of the co-founders and architects of the upcoming Indy Autonomous Challenge. Check our recent story about the final ten teams competing in the Indy Autonomous Challenge. The event culminates on October 23, 2021 at the Indianapolis Motor Speedway.
Mike Oitzman – Mobile Robot Guide (MRG): Hi Matt, can you give us an update of what the next month is going to look like in the Indy Autonomous Challenge (IAC)?
Matt Peak – Indy Autonomous Challenge:
“It’s a phenomenal time in the world of the IAC right now. Race cars are on the track every single day, this month, at our sister track, Lucas Oil Raceway. We’re just coming off a pretty big day at the Indianapolis Motor Speedway (IMS) with a practice day at IMS, where every single one of our remaining teams drove multiple laps around the famous oval and really, it’s inspiring watching them.
Starting off with the most basic of exercises as their cars are going around and collecting information and their sensors are firing and all of our great sponsor-provided-equipment is collecting data. Then they use that to advance the next time they go out on the track.
Can you give me a two minute review of how the concept came to reality? How it got started and who some of those early folks were that helped make this happen?
“The IAC is a competition to attract and stimulate the next generation of robotics and automotive leaders, and to facilitate an expedited pathway to the arrival of these advanced technologies that can not only drive cars by themselves, but even save lives on the road today. We started this because we were very much inspired by the DARPA Grand Challenge.
What the DARPA GC did was basically it created our commercial autonomous vehicle industry today. The winning team was Stanford University, and was led by a gentleman named Sebastian Thrun, who after leading that effort, went on to cofound Google with Larry Page and then went on to found Udacity. We hosted him at the Indy 500 in 2018. He was reflecting on his experiences being part of DARPA, and he basically said; “There’s nothing today that equals the challenge that we and our team experienced back then, and it’s desperately missing. We need a modern day successor to DARPA, in order to get these technologies advanced to overcome the software issues”. We think that should take place at the world’s leading racetrack and he pointed right at us. So that’s what started the juices flowing. We thought, well, there’s an intriguing idea anded, we started developing the challenge event.
Do you think this is going to be a one time event or do you think you’re going to start the cycle over again at the end of this first challenge?
“Nothing’s been announced. But I will say that we now have teams from universities all around the world that own cars, that are learning day by day how to better operate them and who are looking to put those into use well beyond October. So we’ve already started talking with them and with some others about what things might look like after this October and hopefully in the not too distant future, we’ll have something to announce.
How similar is the Dallara AV-21 race car to the Indy Lights vehicle, raced by human drivers?
“The chassis and the components are virtually identical. The suspension and chassis are all supplied by the same suppliers as in the Indy Lights. We have improvements to our vehicle that we’re very proud of, including our engine, which is built by Four Piston Racing (a local Indianapolis based company).
The cars run on Bridgestone tires that are custom engineered specifically for this race and these vehicles. There are some things that come with the Indy Light vehicles, that for our purposes are not useful. Like, for instance, additional crash protection for human race car drivers. For the most part they are very similar.
So what’s unique about the tires for this use case for AV’s as opposed to human drivers?
“Bridgestone has a whole innovation unit that’s focused entirely on developing tires for automated vehicles. An incredible amount of R&D goes into the tire design and compounds. In fact, the tires are controlled when no one is around so that there are no trade secrets getting away. This is the Bridgestone secret sauce.
All ten of the race vehicles are identical, right? It’s a single class, so the advantage isn’t in the hardware platform, it’s in the software in this race, correct?
“Yes, and that’s by design. Early on, we looked at the different ways that we could design this competition. In the DARPA competition it was open to everything from the vehicle design, to sensors, to software.
When we looked at this, two thoughts come to mind. One is that’s an enormous scope, if the team had to design their own hardware. That dramatically increases the resources that teams would need to participate.
More importantly, we realized that sensors and computers and equipment have reached a sophisticated maturity, and it’s not the limiting factor in autonomous vehicles.
All of the hardware is state of the art and there didn’t need to be improvement in that area. What we wanted to challenge the teams on was software algorithm design. We know that the team that crosses the finish line first on October 23, is the most innovative in the area of (control) software.
I agree. One of the things that I found innovative for the IAC is the use of simulation for the early qualifying rounds. Also, the fact that the same software stack functions identically in both simulation as well as in live racing. How has that worked out?
“It’s pretty exciting. This was by design. We knew that it would be very resource intensive to develop these algorithms. The first cost effective way to advance development was simulation.
We were fortunate to to have a relationship with the ANSYS who became our exclusive simulation sponsor, to basically create an entire virtual world. The IMS track and the characteristics of our vehicle are embedded in the simulation. Teams can develop their software and run a ton of laps safely at a simulated Indy Speedway, even before they hit the road in their physical vehicles.
And make all the control mistakes without ever without breaking or bending any metal?
“Yeah. And that led to a pretty exciting simulation finale in June. There, we had PoliMOVE, our Italian team from Politecnico di Milano, come in first to earn the prize of $100,000. The second place prize of $50,000 went the German team of TUM Autonomous Motorsport, representing the Technische Universität München.
What’s the total cost of each of these race vehicles? Is it over a million dollars each?
“The full costs for somebody building a car like this would be north of a million dollars. We have very generous sponsors, and gathering from our components, suppliers and whatnot to the state of Indiana, to the Lilly endowment, who have dramatically reduced the cost of these vehicles.
They’re still a substantive cost, but they’re not seven figures to us. But if we just simply add up, you know, the chassis, wheels, tires, all the sensor equipment – you’re north of a million dollars.
We’ve been very fortunate in the relationships that we’ve had throughout this journey, one of which is with Clemson University’s International Center for Automotive Research. They did all of the engineering for the autonomous system that integrates into the car, and they did all of the engineering of the power plants that we’re using.
I would estimate that with that level of engineering, and with costs in the open market, it’s likely that each vehicle costs $5 million.
It is amazing to see what you’ve pulled together for this challenge. Have there been any accidents in qualifying and testing? Has anybody rubbed the wall?
“Fortunately, nothing has happened. The teams are really careful. They start with the most basic of operations. It’s probably the least exciting laps you’ve ever seen at IMS. They are starting at 30 miles an hour. During the first laps, they’re collecting information and creating redundant systems such that they can then increase their speed without risk of the unknown.
We admire the approach that they’re taking in the fight and we think that’s the the core component as to why these cars are operating strong at the moment.
What’s been the fastest speed in the qualifying session so far?
“We’ve limited speeds for two reasons: First is the team’s experience. We don’t push them to undertake any risks that they feel are premature. We made very clear to them that it’s up to their development cycles and the finish line is October 23.
Second, we’re still using a chase car for some basic communicative and safety functions. That chase car is following the race car around the track, not to control it, but rather to collect diagnostic data.
So teams generally don’t want to go above around 70 miles per hour with the chase car. Now, what’s exciting is that we’re in a phase right now where they’re ditching that chase car. Over the next couple of weeks, we’re going to be operating at Lucas Oil Raceway without a chase car.
By the time we get back to IMS, on our next practice dates, Oct 20 and 21, there isn’t going to be a chase car there. And so we’ll probably see speeds escalate at that point.
I originally thought that this would be an elimination type of event where teams would drop out of the competition, but it looks like you consolidated the remaining teams down and put them onto ten final teams. So everybody that was participating up to this point, gets to continue to participate through to the end?
“When the competition was launched, and our registration period initially closed, we had 42 universities from around the world that had signed. As of this morning, we’re down to 21 universities. So there’s definitely been that weeding out.
But at the same time, we’ve made an explicit attempt to recognize the talent and commitment of various teams and individuals. Some of the talented individuals may not have made it all the way to the top with their initial teams. But they definitely have a role in a broader team.
So we’ve done our best to matchmake. Some of this happened on its own across several of the teams. Some universities have come together with others.
One of the things that we’re most proud of, is the four way collaboration between MIT University of Pittsburgh, Rochester Institute of Technology and University of Waterloo. There are really talented components of all of those different universities. And due to a number of different reasons, the original teams weren’t likely to advance. That was an effort of our matchmaking, but we’ve also seen a very collaborative competition so far.
I think that I like that outcome better than what I originally anticipated would happen. Especially for the students on the team that may have been on a team that didn’t have all of the pieces to be a competitive team in the end. So what’s going on now with the race teams on a daily basis? I assume they’re all camped out in Indianapolis now until October 23rd, and they’re down on the track on a daily basis, working to improve their vehicles?
“We’ve got every team in town, and each of the teams has a large number of their team members in town. They’re currently going to our sister track, Lucas Oil Raceway, testing their cars on a daily basis. All of them are working towards the October 23 race date.
Sounds like fun. So the final race will be a head-to-head race for 20 laps?
“Yes, all of the race rules are posted on the IAC website. The final race will be a head-to-head format.