Accordion

• Transcript

  Paul Mitchell (00:07):
  No one's really tested autonomous vehicle technology in any consistent manner at a 150 miles per hour and above.
  
  James Kotecki (00:19):
  This is CES Tech Talk. I'm James Kotecki. The most influential tech event in the world is back in Las Vegas, January 5th through 8th and we are here to preview CES 2022. Today, a look at the future of mobility or is it the future of sports? With the Indy Autonomous Challenge, you get both. Here's the pitch. University teams program self-driving cars to compete at the Indianapolis Motor Speedway. Paul Mitchell is the president and CEO of the organization behind this challenge, Energy Systems Network. Paul, welcome to the show. Imagine it's the day of the challenge, we're sitting in the stands. Describe the experience, what are we seeing?
  
  Paul Mitchell (01:05):
  You're seeing history being made. You're seeing for the first time, high speed race cars that are already exciting to watch and to hear and almost to feel as they go by. And then at some point, it clicks that there's nobody driving the cars and that they're completely autonomous and it just kind of sends you to this next level of awe. And I think that was the feeling I had and I think certainly the feeling of most that were in the audience. It's just kind of a, I'm witnessing something that has just never happened before and there's something we're really exciting about that.
  
  James Kotecki (01:44):
  And this challenge took place at the Indianapolis Motor Speedway on October 23rd, 2021. That was the first time that this challenge had happened. Is that right?
  
  Paul Mitchell (01:53):
  That's right. We had been running these autonomous vehicles all summer getting ready for the competition but the first time that the people from the public and sponsors and kind of the broader world got a look at these vehicles running autonomously was on the most famous racetrack in the world, often called the racing capital of the world, the Indianapolis Motor Speedway. That was a fitting venue, given the track's history and legacy as a proving ground for early automotive technology at the turn of the century. And now, we're in a new century proving out the next generation of automotive innovation.
  
  James Kotecki (02:30):
  And if you didn't tell me that it was an autonomous challenge and I just showed up at the track that day, would I be able to figure out pretty quickly that something's different? Or would it look pretty similar to a human driven race?
  
  Paul Mitchell (02:42):
  I think as the cars go by, they're open wheel race cars, they're moving a 130, 150, topped out around a 160 almost miles per hour. And the sound of the car is going to be the same. As they go by, they're going by pretty quick and so kind of figuring out there's no driver in there might take a couple of times if no one told you. But the reality is where there's a cockpit and you usually can see a person's helmet and the head that's maybe exposed a little bit, there's a really sleek aerodynamic, 3D printed cockpit cover that covers up our suite of autonomous sensors and advanced supercomputers. And so, I think if you look at the car, you can kind of tell there's probably no room for a person in it in the way it's designed, which is kind of neat.
  
  James Kotecki (03:34):
  When these cars are being designed and programmed, what exactly are the individual teams changing and tinkering with? What is allowed to be different about each vehicle and what has to be the same?
  
  Paul Mitchell (03:46):
  The Indy Autonomous Challenge, the prize competition that has been going on for two years and with the final prize given out on October 23rd was always about university teams competing on software. It was a software competition to design the algorithms that could pilot these fully autonomous race cars. The car, which was frankly its own challenge to design, engineer, build and test and validate is meant to be exactly the same for each team. Same components, same hardware, same base software that helps all the components talk to one another and exact same setup in terms of the down force, the tires, the air pressure, everything is set up exactly the same. And the engine tuning is meant to be the same so that the only variable that's different is the algorithms that are loaded into that robot race car that tells it what to do and how to operate.
  
  Paul Mitchell (04:47):
  You ask what can teams do leading up to competition? They can't really do anything with respect to the hardware on the vehicle. They can't make any modifications to the engines or the drive train or the tires or wheels. We actually have a racing crew that's a mixture of traditional motorsports backgrounds and autonomous vehicle technology backgrounds, including some people we actually hired from Clemson University that helped design the car, some graduate students that are the ones that are really taking care of these vehicles and making sure that we keep the teams supported with what they need but also avoiding teams having to make modifications that could somehow give them an advantage or change out the configurations.
  
  James Kotecki (05:34):
  If this is about software and all the cars are basically the same in terms of hardware, what was the difference between the winning team on the day of the original challenge and the team that came in last place? How much of a difference, I guess I'm asking, does software really make?
  
  Paul Mitchell (05:49):
  Well it makes a huge difference. I think one of the things that was really clear with our event on the 23rd at the Indianapolis Motor Speedway was that while all of our teams have come a very long way, every single one of the nine teams that competed had completed fully autonomous laps with their cars at the Indianapolis Motor Speedway, that was a pre-requirement for participating in the event that day. But the difference was very visible. We had some teams that were unable to make it out of the pits without an accident or couldn't get out of the pits at all. A couple of teams that were in that unfortunate situation. And then we had teams that were making very clean runs, doing everything that the competition required, which wasn't just high speed operations. It also required constant following of race control commands that are sent virtually to the car and also an obstacle detection and avoidance at high speed.
  
  Paul Mitchell (06:46):
  There was a gap between top handful of teams and those that were still in development. But there's a lot of reasons for that. It's not a knock on the university or anything of that nature. I think it's really just time and background coming into the competition. Some of our teams have been working on high speed automation for a few years now and also brought more resources to the table. Other teams, they hadn't really touched one of these types of autonomous high speed vehicles until maybe July and just had really a short few months to get ready for a competition that had never been achieved by anybody. While there is a variety of capabilities among our teams, everyone walks away a winner in the sense that everybody advanced the state of the art in autonomous technology, some just maybe advanced it a little further.
  
  James Kotecki (07:37):
  I really want to talk about what kind of advancements were made over the course of this competition but I do not want to skip over the big news, the exciting news for people who are looking forward to CES 2022 and what they're going to be able to experience from the Indianapolis Autonomous Challenge in Las Vegas, being brought to Las Vegas. Can you tell us what's what you're going to be doing at CES 2022?
  
  Paul Mitchell (07:57):
  Yeah. We began getting asked before October 23rd, "Well, what's next? What are you going to do next?" And we thought long and hard about that and our friends at the Consumer Technology Association and CES, who we had such a great partnership with when we debuted our car there at virtual CES 2021. We got together and had this idea of what if we ran a racing event at the Las Vegas Motor Speedway during CES and made it an official part of the CES agenda and programming. And to me, it's a great opportunity to amplify what we've done, take it to the next level and showcase it with an audience that really truly understands the importance of the technology that's being developed. We're going to be running our cars at the Las Vegas Motor Speedway on the afternoon of January 7th as part of CES. And I couldn't be more excited to showcase it to the entire audience that people that are both going to be there at the Speedway and others that are going to be watching on the virtual CES.
  
  James Kotecki (09:08):
  Well, we really look forward to that. CES is of course, always about innovations, advancements, what's next? You mentioned that this guy competition had advanced the state of the art in the technology. Is this challenge more about major breakthroughs? Is it more about kind of slow and steady improvements? How do you see that?
  
  Paul Mitchell (09:26):
  I think there's kind of both. On the hardware side, I think we are making some, some slow and steady advancement. What we learned is that the components that we used for these race cars, we call it the Dallara AV-21, autonomous vehicle for 21, the year it was engineered, are off the shelf but they're still very cutting edge hardware and components, LIDAR, radar, optical sensors, supercomputers, GPS systems, all from leading suppliers that are supplying the automotive industry. But those components often have not been put into an application where they're required to go a 130, a 150 miles an hour in a race car. And so we're seeing things that can be designed differently or updates that can be made to make these components work even better at high speeds. And we saw that come out of the competition. And a lot of our sponsors maybe didn't know that going into this challenge, that they would get applied research benefits from it. But they clearly got that.
  
  Paul Mitchell (10:29):
  The big kind of stepwise improvements or innovations that are more breakthrough, I think is in the algorithm, the software algorithm that can determine how a vehicle can operate fully autonomously in high speeds, close counter interactions with other autonomous vehicles. And so I think the code and the algorithms developed by our teams, that's where there's some breakthrough being done. I do not believe, we certainly could be you wrong and we'd have to sort of test it out to see, but I don't think you could take an off the shelf commercial algorithm being used in some of the low speed autonomous applications that are being tested around the world and just drop it into our car and have it work effectively. I think some real advancements were made to handle high speed, sensor fusion and split second decision making of an autonomous driver stack.
  
  James Kotecki (11:22):
  I imagine there are a lot of unique things about programming self-driving cars for a high speed racing environment. What are the applications that you can bring from this racing environment back to the consumer technology environment, especially on the saw software side, as we see such breakthroughs happen in the racing environment?
  
  Paul Mitchell (11:41):
  Yeah. I think what you're really proving out and validating or what we talk about in the autonomous industry is edge cases. How do you test out an algorithm's reaction to make a decision in a very short millisecond timeframe when something unexpected is happening? And I think the other element of this is just do you start to see any breakdown in the software's ability to perceive its surroundings based on the hardware that is supporting that algorithm at extreme high speeds? And I think we're going to keep going faster. I will say that one of the things that hurt us on the 23rd was the track was very cold that day and the tires were really cold. We had teams that were very much ready to go a 150, a 160 miles an hour and I think we're prepared to get there. And we saw some of those speeds in the straightaways. I'm excited about Vegas because it's a little warmer, track should be a little warmer and we should see higher speeds.
  
  Paul Mitchell (12:42):
  No one's really tested autonomous vehicle technology in any consistent manner at a 150 miles per hour and above. Just that test environment is in and of itself a great incubation tool. I think it's testing edge cases, it's pushing the hardware and the software to the absolute limit. I think also the vehicle to vehicle communication elements is an important area that we're working on. Cisco was the presenting sponsor of the Indy Autonomous Challenge event that was held on October 23rd at the Indianapolis Motor Speedway. And they're the provider of some of our vehicle to vehicle communication technology, where you're talking about a very low latency network, wireless network that's built at the track so that these cars can know where each other are and can essentially be able to communicate back with race control. And that's also a new technology area that needs to be proven out.
  
  James Kotecki (13:35):
  I'm curious to know a little bit more about that vehicle to vehicle connectivity. If you're talking about driving on the road, I obviously want to know where every other vehicle is but if you're talking about a strategic racing environment, were there kind of overarching safety parameters or connectivity guidelines that you had over and above the kind of raw cutthroat strategic considerations you might have in a race?
  
  Paul Mitchell (13:58):
  There were. And keep in mind on the 23rd, the competition was limited to time trials. It was one vehicle at time, although they did have to do obstacle avoidance and detection and avoidance. I will tell you our plans for Las Vegas Motor Speedway when we run at CES is that we will have multi vehicles.
  
  James Kotecki (14:18):
  Wow.
  
  Paul Mitchell (14:18):
  At the same time with passing. We're going to introduce a passing competition, a format that I'm really excited about. Two cars at a time but where they're passing each other at increasingly higher speeds to see kind of who, it's almost like a game of chicken to see who can pass at higher and higher speeds and eventually somebody can't do it. But we do have kind of protocols in place for primarily how race control can see the cars and see where the cars are on the track and have that communication. It's not so much the individual teams that get to see that information or even the other cars to make decisions about it. It's more of a ability of race control to monitor what's going on, to black flag a car if it needs to slow down and go into the pits. Frankly, so that the competition can progress.
  
  Paul Mitchell (15:02):
  The reality is, if you've got multiple cars on the track at the same time and there's no communication with it, that's not how racing works. Crews are talking to the drivers constantly on headsets and they're telling the driver, "So and so's coming on your left. So and so's coming on your right. Speed up, slow down." We're trying to figure out how do you create some of that information that you need in order for passing and things to occur while at the same time, not overstepping this line of ensuring that everything is done fully autonomously.
  
  Paul Mitchell (15:35):
  It's an evolving rule set. The reality is the autonomous challenge at Indy and when we run it at Las Vegas, we work on the rule set in a way that we think reflects a fairness component, a technology innovation component and allows us to demonstrate where we are in the journey toward eventual full head to head racing with all the cars on the track at the same time. And so that rule set it can't be too rigid otherwise you don't have a product to share with the world.
  
  James Kotecki (16:06):
  Yeah. And how is the wider world of motorsports perceived what you're up to? I imagine that they're not ready to ditch human drivers anytime soon, that's still very exciting but are they learning things from your competition?
  
  Paul Mitchell (16:19):
  Yeah. The interest level, particularly in the Indy Autonomous Challenge and coming out of the 23rd has been very broad globally and it's been a mix of traditional news, broadcast news from the US, from Europe and Asia. The tech community, CES and others have obviously loved this and are all over it and want to see more and more. I think the racing and kind of traditional automotive and motorsports community is interested in it and we've got great coverage in some of the leading magazines and industry outlets around motorsports and they always ask that question of, "Is this about a new motorsport series? Or is this going to replace drivers?" And I think our answer is, no, this is not about replacing drivers in motorsport. I think the whole concept of man versus machine and man taking on the risk of racing is always going to be an important motorsport for many years to come.
  
  Paul Mitchell (17:15):
  But where we're seeing the real interest is in the technology development that's occurring on these autonomous cars, how could those migrate to human driven race cars in ways that would allow motorsports to advance? I would like to see human driven race cars at 300 miles an hour someday and we're not there. Indy car is probably one of the fastest motorsport series and maybe the fastest in the world and they touch 240. But that leap from say 240 to 250 to 300, it increasingly risks human life. And if there's a way to add in 360 degree perception, ability to add some level of crash avoidance, to navigate when there's smoke and the driver can't see anything, then that's a pathway to faster motorsports that's also safer. And I think that's what everyone in motorsports has always wanted. How do you go faster and do so safely?
  
  James Kotecki (18:12):
  As we're talking about that future, the future that you're building at the Indy Autonomous Challenge is largely being driven by university students who are the ones who are actually involved in the challenge. Why did you have the competition specifically for university students? Why did you do it globally? Tell me about some of the thought process there and the learnings that came out of it.
  
  Paul Mitchell (18:29):
  Well, we've given a lot of credit to the concept of the Indy Autonomous Challenge prize competition to the predecessor prize competition of the DARPA Grand Challenge that took place in 2004 or five and the DARPA Urban Challenge later. And that really ushered in the modern autonomous vehicle industry. And you see so many companies that came out of that and so many luminaries and innovators that came out of those university led teams from Stanford and Carnegie Mellon and MIT and University of Michigan and others. And so we wanted to bring that back and we knew that prize competitions are a great way to inspire the best and brightest minds, that are largely inside these elite engineering powerhouses and research universities around the world. And so I think that was always obvious that they were the only kind of teams that could drop everything and dedicate themselves to this for two years without a clear return on investment, that's financial. It's not financially motivated. It's motivated by advancing the state of the art and it aligns with their research goals.
  
  Paul Mitchell (19:40):
  The winning team from Technical University of Munich, they had 14 PhD students that made up this team. They were supported by a total of, I think there was a total of 40 people on the team. 14 of them were PhD students but then you had master's and undergraduate students back in Germany working on this, led by an amazing team leader, a Professor Markus from TUM. But it's very hard to find a startup company or even a big for profit company that would say, "Hey, I'm going to allocate 14 of my best engineers and let them work on this challenge for two years." That's why we went the university route.
  
  Paul Mitchell (20:16):
  The global nature of it kind of happened unexpectedly. The DARPA Grand Challenge really didn't attract international teams. It was run by US Department of Defense, DARPA. It's unlikely to get a lot of international participation. What was unique about our competition is we have a lot of our top competitors coming from Europe, from South Korea. At one point we had, I think, 11 different countries that had universities that signed up. And I think we had nine countries or eight countries represented in it on the 23rd. This was very much global in nature. And the benefit we had was that a lot of the top talent is actually coming from Europe and the European teams placed first, second and third in the final competition. I think it also is an opportunity for us to take stock with where we're at and realize that our US institutions are going to have to continue to invest and progress in order to keep up with some of our neighbors across the pond, they say.
  
  James Kotecki (21:09):
  And as you look at the research that came out of this, is this all going to be open and available to the next group of students who do this challenge or a similar challenge next time? The teams will all have the advantage of everything that the winning team did and everything everyone else did.
  
  Paul Mitchell (21:21):
  The short answer is yes. It is up to individual teams whether or not they take their final software algorithms and make them available. But TUM, Technical University of Munich that won the competition, made the announcement that they are open sourcing their algorithm. And so I don't know if it's already been posted. It probably has but it is available to the entire world. And so you asked earlier about what were some of the big breakthroughs? I believe that algorithm is a major breakthrough and it proved itself and now it's available to everybody. And so that's exciting. I think some of the other teams will do the same. Others may have a different strategy. Obviously university's value, IP development and need to find ways to monetize that and attract applied research.
  
  Paul Mitchell (22:06):
  But overall, yes, our goal is to have this continue to progress, to provide a pathway for perhaps even new universities who want to get involved to start with simulation tools that can get them up and going and get exposed to what's required but then eventually get their own vehicle and begin to compete. But open source approach is always been our preference.
  
  James Kotecki (22:32):
  As we look to other industries that might be thinking, hey, doing a competition sounds like a great idea to advance the state of the art in my corner of the tech world, what advice would you have to other tech leaders who want to do a competition like this?
  
  Paul Mitchell (22:46):
  Well, I heard about and we've certainly read about the power of prize competitions, whether it's the XPRIZE or the DARPA Grande Challenge and others but I've witnessed over the last two years that there's really nothing like it to get broad buy in and focus on a grand challenge. I think my advice though, is that you have to listen to the voice of the customer. Before we ever issued a rule set related to the Indy Autonomous Challenge, we brought together not all, but many of the universities that ended up competing and put them together in a room at the Indianapolis Motor Speedway for a workshop and basically bounced a lot of ideas off of them and got their input and got their buy in before anything was announced.
  
  Paul Mitchell (23:33):
  I think prize competitions are a great way to motivate but you've got to also make sure that you're setting the prize competition up in a way that is pushing the state of the art but also achievable to the audience that you're trying to attract. That's even as simple as aligning it with academic calendars, if you're using universities, making sure that there's resources available to those universities to be successful. Because prize competitions where nobody can succeed or really compete, that doesn't do anybody any good. It's a great tool but you got to do it the right way.
  
  James Kotecki (24:09):
  The first DARPA challenge, as you're mentioning, I think famously nobody completed it, right? Nobody actually.
  
  Paul Mitchell (24:15):
  Yeah, that one. And I think they learned a lot and they went back and made adjustments and luckily.
  
  James Kotecki (24:22):
  Look how far we've come.
  
  Paul Mitchell (24:23):
  Yeah. Luckily a lot of the best teams stayed involved. I'm glad that didn't happen on the 23rd. I'm glad we were able to progress things. Although keep in mind, we did adjust our rule set and our rule set moved to closer to where teams were. Had we stuck with our original rule set of, it has to be a head to head, fully autonomous competition, I don't think it would've gone as well as it did on the 23rd. We might have still had some teams make it past that finish line but there would've been a lot of carnage along the way and I'm not sure that was the best way to preserve the assets of these multimillion dollar race cars. I think even DARPA, the lesson was learned in the first year, they didn't make any adjustments. In that second DARPA Grand Challenge, if you talk to Sebastian Thrun or Chris Urmson or people that participated in that, they did actually adjust the rule set all the way up until the end. And so I think a lesson was learned that you can't be too rigid on those rule sets.
  
  James Kotecki (25:20):
  At CES, we are going to see head to head racing.
  
  Paul Mitchell (25:24):
  We're going to see at least two cars on the track at the same time.
  
  James Kotecki (25:27):
  Passing.
  
  Paul Mitchell (25:27):
  Doing a passing and overtaking.
  
  James Kotecki (25:30):
  And do you foresee a near term future of all the cars on the track at the same time, just like a regular Indy 500?
  
  Paul Mitchell (25:35):
  Yeah. That's the progression that we're headed toward. First we went fast, now we will pass and then eventually we'll be onto racing.
  
  James Kotecki (25:45):
  Put that on a t-shirt. That's great.
  
  Paul Mitchell (25:47):
  Yeah. That's right.
  
  James Kotecki (25:49):
  Paul, do you have any final thoughts before we close out?
  
  Paul Mitchell (25:51):
  Yeah. I think it's just important to understand that this competition is not just about automotive or motorsport and while we are excited about all of the people that'll be in the west hall and the automotive pavilion, this is really a competition in data analytics, sensor fusion, machine learning and that applies to the entire CES universe. We want folks to be interested and engaged in what's happening. The fact that we're doing this in the form factor of a race car is really just about the use case but the core technologies can be applied to virtually every industry from life sciences to automotive and aerospace and beyond.
  
  James Kotecki (26:37):
  Well, thanks so much for joining us, Paul and we will see you at CES 2022.
  
  Paul Mitchell (26:42):
  See you there.
  
  James Kotecki (26:43):
  Well, that's our show for now but there's always more tech to talk about. Subscribe to this podcast so you don't miss a moment and get more CES at ces.tech. That's C-E-S DOT T-E-C-H. Our show today is produced by Tina Anthony and Kiersten Hizak, recorded by Andrew Lynn and edited by Third Spoon. Special thanks to CTA's John Lindsay for the studio help. I'm James Kotecki, talking tech on CES Tech Talk.