paul mitchell, chairman and president, indy autonomous challenge
Listen on Apple Podcasts, Spotify, and almost all other outlets.
For a menu of other apps, click on the three dots on the upper right, select Subscribe, and pick your favorite play.er You won't have to subscribe to listen. Or just listen via the player above.
An unedited transcript is below.
Ali Tabibian (00:17):
That's great. Thank you. Paul, thank you so much for joining us here...
Paul Mitchell (00:17):
Thank you.
Ali Tabibian (00:20):
... at the Las Vegas Raceway.
Paul Mitchell (00:22):
Yeah.
Ali Tabibian (00:23):
What's the official name of this place?
Paul Mitchell (00:24):
Las Vegas Motor Speedway. This is our third time here for the Autonomous Challenge at CES.
Ali Tabibian (00:31):
Excellent. Paul, tell us a little bit, what are we about to see here?
Paul Mitchell (00:38):
So this is the Indy Autonomous Challenge. This is our third time coming to Las Vegas Motor Speedway. We were launched originally as a prize competition in 2020 with the goal of advancing technology and innovation in high-speed automation. And so we have 10 of the world's fastest autonomous race cars, fastest autonomous cars, top speed of 192 miles per hour. And the drivers of these cars are coded by top university teams from around the world. It's nine teams made up of 18 different universities in five countries, and the cars are identical, so the teams are really just competing on their AI driver software prowess. What you'll see today is a passing competition where it's two cars out on the track at the same time passing each other at increasingly higher speeds until one team either gives up or there's a crash.
Ali Tabibian (01:39):
Well, you know, people sometimes do watch the races with the crashes.
Paul Mitchell (01:42):
Yeah. And nobody gets hurt.
Ali Tabibian (01:43):
So even that'll add something.
Paul Mitchell (01:43):
We've got nobody in the car. So it's...
Ali Tabibian (01:47):
And so just to draw a visual image for our listeners, these are basically open-wheel Formula 1 style vehicles by the looks of it. Right?
Paul Mitchell (01:54):
So if people are familiar with IndyCar and we're based in Indianapolis Indiana, the home of IndyCar, the Indianapolis 500. This is the Indy Next Series race car. So it's an open-wheel race car, a little bit smaller than the full-size Indy car or a full-size Formula 1 car. But still it's a very fast open-wheel race car chassis built by Dallara with top speeds close to 200 miles an hour.
Ali Tabibian (02:20):
Wow. That's very impressive. Now, was this your idea to bring all this together and make it happen?
Paul Mitchell (02:24):
So I really credit Indy Autonomous Challenge to being just one of these successes of a public-private partnership. So it's set up as a nonprofit organization. And at the time we were talking to the Indianapolis Motor Speedway, the State of Indiana. Indiana is known for racing. It's also known for the automotive industry.
Ali Tabibian (02:24):
Right.
Paul Mitchell (02:43):
And so we said, okay, what better way to showcase the future of automotive and our history and legacy and racing than trying to do something around high-speed autonomous vehicle technology? And we talked to a lot of luminaries in the industry, people like Sebastian Thrun who had won the DARPA Grand Challenge and went on to found Waymo and Kitty Hawk. And we talked to investors like Reilly Brennan from Trucks and Chris Urmson who's with Aurora. And then we talked to a lot of academic experts to say like, you know, if we did this kind of grand challenge, sort of DARPA-esque activity, would it be worth it and would people actually want to participate? And we got really positive feedback. And so we launched it as a classic prize competition. We raised prize money from the Lilly Endowment, which is one of the largest private foundations in the world. It happens to be based in Indiana as well.
Ali Tabibian (03:40):
But the Eli Lilly Foundation?
Paul Mitchell (03:41):
It's Eli Lilly family money.
Ali Tabibian (03:43):
Family money.
Paul Mitchell (03:43):
It's not technically the same as the company.
Ali Tabibian (03:45):
Right.
Paul Mitchell (03:46):
And they put up some of the dollars that would be needed to give out a million dollar prize. And we put out a call to universities around the world to come and participate in this competition. And 2020 of February is when our registration closed and we had 41 universities, but within a month after that, the pandemic started. Right?
Ali Tabibian (04:08):
[inaudible 00:04:08].
Paul Mitchell (04:07):
And so we did see a pretty big drop-off and a number of universities just had to fall off completely. But those that stayed spent 2020 focusing on competitions that were simulation-based, so all virtual. And then during that time, we built these 10 race cars. We partnered with top industry collaborators, people like Luminar and Cisco and many others who donated hardware and or software to the challenge. And then we brought in partnership with Clemson University, which has a great prototyping program that works with big OEMs to do prototype projects. And so they took this on as a challenge to help us prototype the first car and then we replicated it. Team started running these cars in 2021. It was slow going for the first few months. You know, we had to really validate the cars could even work.
(05:02):
Getting them around the track at 30, 40 miles an hour was an accomplishment then. And what we saw was just an exponential improvement within a period of months. So by our first event of October 2021 when we gave out the million dollar prize, we had cars going 140 miles an hour at the Indianapolis Motor Speedway. And then a couple months later at CES 2022, we started introducing head-to-head racing, which had never been done in the world. Two separate AI systems competing with each other, not knowing what the other one's going to do. And the final round had the Polytechnic in Milano versus Technical University of Munich and Polytechnic in Milano won, but I think they were making a pass at 169 miles an hour. And that was really what woke us up, and I think in many ways woke the world up to this concept of extreme high-speed automation.
Ali Tabibian (05:52):
That's really quite impressive. Do you define the class in a way that limits the vehicle performance and turns it into a sort of software duel? Or how do you define the class?
Paul Mitchell (06:14):
Yeah. So I should say that the vehicles are spec vehicles. Every vehicle is completely identical.
Ali Tabibian (06:19):
Okay. Okay. I see.
Paul Mitchell (06:21):
Same engine, same tuning of the engine, same setup, same sensors, same hardware.
Ali Tabibian (06:25):
And is that something you offer to the universities...
Paul Mitchell (06:25):
Yes.
Ali Tabibian (06:28):
... and they were... Oh [inaudible 00:06:28].
Paul Mitchell (06:27):
Yes. Correct.
Ali Tabibian (06:27):
Gotcha. Gotcha.
Paul Mitchell (06:28):
So Indy Autonomous Challenge is responsible for sourcing the partnerships with industry and then for building and maintaining and repairing the vehicles, bringing them to the venues for testing and racing. Then the universities work on the code at their own campuses, do a lot of simulation and then come together for test events like the one we're doing here at CES. And, you know, this is really an applied research initiative to advance the state-of-the-art in autonomy to show that the hardware and the software systems can work at speeds greater than 100 miles an hour. And we feel like we felt at the time, and I think it's still there, there's a pretty big gap in the market right now. Most ADAS systems, maybe with the exception of Tesla, level three systems, they turn off at about 60 miles an hour.
(07:15):
Well, it turns out humans are pretty good at driving from zero to 60 miles an hour. Where we stop being really good at driving unless you're a Formula 1 race car driver is at speeds 60 miles an hour up or 100 miles an hour up. Right? So if we can get AI drivers to be able to operate safely at 120, 130 miles an hour, think of what that does to unlock the efficiency of mobility in terms of getting people and goods from point A to point B a lot faster. Now, there's a lot that has to go into that.
Ali Tabibian (07:15):
Right.
Paul Mitchell (07:43):
There's a lot of probably regulations and infrastructure support, but that's the kind of aspirations that I think a number of our universities and the companies that we're working with have to say, autonomy can be a lot more than just taking me from the airport to the hotel.
Ali Tabibian (08:00):
As an example, in California, we're spending tens of billions to connect to San Francisco to Los Angeles by rail.
Paul Mitchell (08:04):
Right.
Ali Tabibian (08:05):
People frequently say, why don't we just put high-speed buses on I-5, right, and the safety becomes an issue. And if you have something like your program as a testbed...
Paul Mitchell (08:14):
Yeah. Exactly.
Ali Tabibian (08:14):
... for these. Let's talk a little bit about that because, for example, you've mentioned as an example Luminar, well, they're definitely oriented toward commercial vehicles. Volvo, they just had a announcement with Mercedes.
Paul Mitchell (08:25):
Yeah. Yeah. But so Austin Russell, the CEO of Luminar was an early, early supporter of ours. We called him and asked him if he would supply LIDARs to us I think before his SPAC was done or if it was done, it was just happened. And he loved the DARPA Grand Challenge and he loved the ideas of getting academics to come together. He himself is not that old, so I think he kind of connected with the PhD students that are working on this challenge and they saw real value in it, I think, around testing the validity of their platform at extreme high speeds. You're not going to get that kind of testing from an OEM. You're going to need to push it that much further. So for a group like Luminar, they can say, look, our LIDARs, we're learning and gathering data of how they operate at these high speeds and how they operate and counting other vehicles using the same LIDARs at high speeds.
(09:18):
And there's a lot of value to that. And same thing's true for the radars we get from Conti or the computer systems that we get from dSPACE. There's a lot of testing and validation value. The other value that the industry partners get is access to the talent pipeline. We've got about 250 of the best and brightest minds from top universities around the world, you know Carnegie Mellon, MIT, Berkeley, Purdue University, Technical University of Munich, Korean Institute for Science and Technology that are all working on this challenge, you know, together. And they're working on a common platform and a common set of problems, but they're doing it in the context of Motorsport, which obviously creates a lot of, not just competitiveness, but sort of excitement and problem solving and working under pressure. And so that talent pipeline then can feed into the industry. And we're seeing that already with a number of our alumnis going into companies that are in the autonomous vehicle industry or going into academia or going into government roles.
Ali Tabibian (10:22):
That's pretty interesting. To what extent do your partners have access to the data and are the systems that are being developed essentially open systems, open source?
Paul Mitchell (10:32):
No. So the component level data is something that the sponsors have access to and their IP is their IP. And if we come up with something that maybe improves a component, then that becomes their IP.
Ali Tabibian (10:42):
Right.
Paul Mitchell (10:43):
The team's IP is around the software stack. And then the IP that we have at the nonprofit level of Indy Autonomous Challenge is more around the integration of all these autonomous hardware and base software systems into this very small form factor of a cockpit of a race car and making sure that it can handle a crash at 170 miles an hour and still be put back together and operate the next day.
Ali Tabibian (11:09):
Right. The aircraft black box [inaudible 00:11:12].
Paul Mitchell (11:11):
We do that all the time. Right?
Ali Tabibian (11:13):
Right. Okay. What's next in terms of the next year or two? Where would you like this all to go? What more?
Paul Mitchell (11:18):
So, I mean, I think what we've found is that, you know, the approach where you sort of have each team developing their own full stack AI driver, you know, leads to these gaps in performance where there's certain AI drivers that are just more advanced. And so you can either just kind of wait for those teams to try to catch up, or we can really embrace this kind of collaborative approach by having more cross-collaboration among the universities to work on different components that go into a full-stack AI driver. So an AI driver typically has four key components, perception, localization, vehicle dynamics, and path planning. Right? Now, each of those have to work perfectly for these cars to compete. Right?
Ali Tabibian (12:13):
Right.
Paul Mitchell (12:15):
But not every team is an expert in all four of those areas. Some are, and they're the ones that are really excelling, but you may have an institution that's really, really good at perception or like LIDAR localization.
Paul Mitchell (12:27):
So let's have them focus on that and they develop a module that can go into the full stack. And so that's something we're exploring with our teams. We're exploring with potential funders at the government level and corporate partners to see can we create more of an open innovation platform that would advance more rapidly the development of AI driver capabilities at these high speeds, recognizing that you probably don't need 10 or 12 of them. You probably need two or three of them that are kind of pushing one another to the point where you start to have some standards and some validation that would allow transferring this into commercial applications. And I'm not sure that it's necessarily competing with the traditional autonomous vehicle companies.
(13:11):
I mean, I think if you ask a lot of the executives from autonomous vehicle companies, like do they have any plans to have their AI drivers drive at 120 miles an hour, generally their answer is no. But it also is because it brings probably a lot of regulatory risk and a lot of infrastructure requirements that go beyond what an individual private company could do. But if you have a network of academic institutions, state governments, national governments, international governments working together, then you can find these niche use cases. Maybe it is a corridor down the center of California or connecting some airport in some country to an urban center. And, you know, that's the kind of thing we're looking to accomplish.
Ali Tabibian (13:56):
Excellent. Paul, I'd like to get one more question in, and I know you've got it stacked if you want, and...
Paul Mitchell (14:01):
Cars I'm hearing are starting to run, so I probably need to start paying attention to what's going on out there. [Note to Matt: somewhere around here, can we mix in an audio clip from the video snippet that I provided?]
Ali Tabibian (14:01):
I'll be quick.
Paul Mitchell (14:01):
Okay.
Ali Tabibian (14:05):
And it's a question I asked before, but I'm going to come back to it and I'd like to get a little bit more of your background because you strike me with the kind of intensity that I see from very accomplished startup founders and executives in my neighborhood, which is San Francisco. And I'd just like to hear a little bit more about how did you get involved, what's your background? What was your motivation?
Paul Mitchell (14:28):
So I'm born and bred in Indiana. My background is really, you know, I studied economics and public policy, but I really came out of government. I worked at a very young age. I was very fortunate to get a job as a policy director for then Governor Mitch Daniels, who I really believe is one of the most outstanding governors that we've had in the last century, did great work. And then he went on to become president of Purdue University and did some really amazing things there as well. And so I worked for Governor Daniels as a policy director on economic development, workforce, energy issues. And then basically from there on, starting in 2008 until now, and I plan to continue, I really focus on the intersection of public-private partnerships. So, you know, taking on challenges that are going to require government, university, and private sector to all be pushing in the same direction.
(15:22):
And I've always been fascinated with trying to get that sort of trifecta to work because when it does work, that's where you get the biggest impact and the biggest kind of stepwise gains. And, you know, I think this is a good example of that. Right? You're probably not going to have 150 mile an hour transport without government, academia, and industry working together. It's probably not something that just private capital is going to solve on its own. And there's other problems that I've worked on in that space around. The early deployment of electric vehicles was a big passion of mine, some of the early mobility as a service related activities. And then I'm also part of a group that put together and was successful in securing one of these billion dollar DOE Hydrogen Hub Awards...
Ali Tabibian (16:12):
Oh. Interesting.
Paul Mitchell (16:13):
... for the state of Indiana, Michigan...
Ali Tabibian (16:14):
The Midwest.
Paul Mitchell (16:15):
Midwest. [inaudible 00:16:16] too.
Ali Tabibian (16:17):
Right. We do a lot of work on hydrogen.
Paul Mitchell (16:18):
So I'm deep into that right now as well. So those are right now my two passion projects, the Hydrogen Hub effort, and then this Indy Autonomous challenge. Different but same concept. Right? You're still trying to get government industry and academia to align and... Yeah.
Ali Tabibian (16:37):
Well, I'm going to recommend our listeners keep, sort of, follow you carefully and keep an eye out. I've spent a lot of time covering entrepreneurs and people in the entrepreneurial set and definitely bookmark or whatever the right thing is Paul Mitchell...
Paul Mitchell (16:50):
Thank you.
Ali Tabibian (16:51):
[inaudible 00:16:51] his background. Let's get you out to the racetrack. What do you think?
Paul Mitchell (16:53):
Yes. Thank you very much. Great conversation.
Ali Tabibian (16:55):
Thank you so much.
Paul Mitchell (16:55):
Thanks.
Ali Tabibian (16:55):
Appreciate it.