Niki: I’m Niki Christoff, and welcome to Tech’ed Up. Today we’re talking self-driving technology with Lia Theodosiou-Pisanelli, who runs partnerships, programs, and operations at Aurora. We discuss the difference between driver assist technology - like Tesla - and fully autonomous vehicles, which don’t need a human at all. I’m from the midwest, and I love to drive. But I also love efficiency, so I had to try hard not to be too much of a hype girl on this one.
Niki: Good morning, Lia. Thank you for coming on Tech’ed Up. Thank you for coming into the studio first thing.
Lia: Yeah, happy to; it's nice to do something in person every once in a while.
Niki: It is! It's like Earth won! [both chuckle] It's almost like old times. Okay. I have to let our listeners know before we get started, that I am a total super fan of self-driving trucks. To the point where everyone who knows me has to listen to me talk about it.
So, I'm going to try to calibrate my enthusiasm for what you do for a living. But you work for Aurora, which is a self-driving autonomous vehicle tech company. You're starting out with trucks, but let's start with you. What do you do for a living? What's your role?
Lia: So, I am Vice President of Partner Programs and Operations at Aurora. So what that means, I run a few different departments, but a huge part of what we do at Aurora is we partner with others for the things that we aren't best suited to do. So, we know that we can't launch this technology all by ourselves. So it was really important from the founding of the company to build deep partnerships across industries.
So automotive companies, logistics companies, ride-sharing companies, fleet management. So with each of those partnerships, it's not just, kind of, a business development deal, and then you're good to go. It's a deep engagement, most of the time technical. So my team runs all of those engagements. And then, I also lead our operations teams.
Niki: And I want to talk about Aurora specifically. What is Aurora?
Lia: [chuckles] So, Aurora- we're a self-driving technology company, which, we're; basically, we're building something called the Aurora Driver. And so, that's a combination of hardware, software, and data services that is integrated into multiple different vehicle platforms. The way we've built it is such that you can integrate it into passenger vehicles, into trucks, you can use it to move people, you can use it to move goods. And that is possible because of the way we architected it from the very beginning, knowing that we would want to be able to use this across a number of different industries and a big, um, big component of that is we leaned really heavily into simulation. And so you don't, just, the Aurora Driver doesn't just benefit from what it's learned on the roads, in a car, or on a track, but it's actually learned an enormous amount from the millions and millions and millions, possibly even billions, at this point, I don't keep track of, of miles in simulation and experiences in simulation that we train this system that lives on both different types of vehicle platforms.
Niki: So you and I…I don't want to say we were bitter enemies, but you were at Lyft for a while, working on autonomous vehicle technology. I was at Uber when we also had, a, an autonomous vehicle team, and something really cool about Aurora in the past year is you acquired the Uber self-driving engineering team [Lia: Yep!] and brought on that talent. [Lia: Yep!] And, I think that part of the reason you guys went public this year was to get more capital so you could build out that engineering talent.
Lia: This technology is really expensive to develop, and we've had the benefit of investors and partners who have been with us all along the way, who, who understand that this is a long-term play and really understand that it takes time to build this right, do it safely before you launch it. And so, as we've been, kind of, preparing ourselves for not just developing this product, but bringing it to market, it's really, uh, it's really a battle for talent and technology.
And when it came down to it, we knew that Uber ATG was, was, in trouble. And that Uber, just, and I knew this from being at Lyft and knowing where things were kind of going on the Lyft AV side, we knew that Uber was looking for, needed to focus on its core business. And so there's this opportunity, a fantastic talent, great technology, bring them under our wing at Aurora.
Niki: Aurora was founded by Chris Urmson, who ran some of the self-driving work at Google. In fact, when I was there, he was a legend at Google, partly because he’s a great robotics engineer and very technical, but also incredibly nice. [chuckles] Which in the tech industry is somewhat rare. And he attended Carnegie Mellon University. So, you guys are based in Pittsburgh?
Lia: That's right.
Niki: And Uber, when I was there, you say ATG, that's the acronym for Uber’s self-driving project [Lia: Thank you], which had these amazing engineers, but who are incredibly expensive because everyone's fighting over the talent of people who understand autonomous vehicle technology. And it was costing Uber $300 million a year to keep this technology going. And, no shade, but Uber's head of self-driving is a criminal. [chuckles] And I'm not just saying that! Was, was actually guilty of committing crimes and stealing trade secrets. He was pardoned the last day of Donald Trump's presidency, but then you have this great team of engineers sort of demoralized in this program. And so what they've done is you guys have acquired the talent, and you're partnering with Uber, meaning that you can license the technology back to Uber in a partnership, and they can focus on what they do best, which is ride-sharing. Am I getting that right?
Lia: [hesitating] Close. [Niki: Okay] Yeah. So, I would say, I won't talk about what happened at Uber before, before, Aurora was involved. So, one, yes, that's true about Chris, and even before that, he led the DARPA grand challenge. So, he has been in this industry for a long time, has figured out how, has tried a number of different ways to develop the technology and deploy it.
Whether y’ know, when he was doing it from a research perspective, when he was doing it at Google. And then you have the other two co-founders, Drew Bagnell, who came from Uber ATG also was a Carnegie Mellon, not just student, but professor, and one of the foremost experts, in, in machine learning but as it supplied to robotics. I won't nerd out too, too, much farther, deeper, but he's brilliant. And then you have Sterling Anderson, who came from Tesla, who knew how to launch, a, an actual vehicle into the market. And so, Chris kind of brought together these, these three brains that came from different backgrounds, and, and kind of brought them together to form this company.
Now, when you then turn and look at, okay, what was the opportunity for acquiring Uber's team? It was, here's this brilliant team who has had all the resources in the world to build whatever it was that Uber wanted. And one of the things about Aurora is that one of our key values is focus. And so, we've been very intentional in everything that we've built, why we've built it, and in what sort of order and what time to market, what time to development.
And so, there's this opportunity to bring this talent in, and kind of apply that sort of broad strategic focus. And once you take those resources and you harness them in the right direction, that's, that's where the magic happens.
Niki: So let's talk, focus. [Lia: Yep!] So, you, there is a long-term vision for Aurora. And for your autonomous vehicle technology, but I want to talk for a minute about the difference between, I think, in pop culture, when people hear “self-driving vehicle,” they think of Tesla, but Tesla is not actually a self-driving vehicle. It's, it's essentially driver assistance technology. [Lia: That’s right] And what Aurora is building is completely autonomous vehicle technology, meaning the vehicle can just, no human has to even be in it.
Lia: That's right. So we're building, definitely a big difference between driver assist, which is something that assists a human driver and then self-driving technology, which means you don't have a driver in the vehicle. We are essentially offering Driver as a service. So, our technology is built such that there will not need to be and will not be a human in the vehicle. Now, today, we have human operators, but they're test operators. They're there to monitor the system to make sure that everything is working well, but there's never a time where we're going to say, here's our vehicle with the Aurora Driver technology in it-
Niki: [interrupts] And so, let's talk about focus for a second. [Lia: Yep!] So, you are starting with, your first marketed product is going to focus on trucks [Lia: Yep!] and freight [Lia: Yep!]. And I find this incredibly interesting because a major economic issue that we have, which is there's a huge shortage of long-haul truck drivers in the United States.
I think the estimates are 80,000 openings for truck drivers. It's a, it's a, profession that is declining, that people are aging out of, that young people are less interested in. And if you guys can figure out how to create the Aurora Driver to do some of that long haul work, as completely autonomous, that helps our supply chain, and it helps our logistics problems that we're facing right now.
Is that why you guys are focused on it? I mean, you did it pre-pandemic. You started focusing on this…
Lia: Certainly, there was a driver shortage before the pandemic. Right? And so, we saw that there was a huge business opportunity to, and, and not just business opportunity, but need, right?
You can build the best technology in the world, but if the market isn't ready for it or doesn't want it, then, you know, what's the point? Right? So in the sort of logistics category, we have all of these customers, and we have everybody who's kind of impacted in the logistics supply chain saying this would be hugely meaningful, not just from a safety perspective, but from an efficiency perspective. If you kind of rethink, okay, if trucks weren't limited by the hours of service, which is the limited amount of time that a human driver can operate the truck before they need to take a break or sleep. Right? So, if there, if you don't have that limit and you can run any time of day, what does that mean for the future of our supply chain?
So, we plan to deploy first by using two, by using terminals. So, what that means is that we'll have sort of a controlled endpoint. It's not right off the highway it's, it's off the highway, but you have terminals, let's say Dallas to Houston. FedEx will drop off a trailer, and then we will take that trailer in our autonomous truck, and we will operate it autonomously all the way down to Houston, drop it off at our terminal. FedEx comes to that terminal, picks up that trailer, and we go back with another load. And the beauty of that is we, we, know the highway very well. We know the terminals, and we're able to take that long haul that oftentimes most truck drivers don't want to be driving. Dallas to Houston is actually pretty short, but you think about how this expands Dallas to El Paso, Dallas to Phoenix. You go on and on and on. But then what you're doing is, you're saying, okay, the human operators, they'll do those shuttles. And so they'll take the, they'll pick up those trailers, and then they'll bring them to another hub, y’know, a certain number of miles away, and they'll get to sleep at home at night and, and, they'll continue to be able to operate close to where they live, which is something we've heard is, is, a really big desire.
When you think about launching technology, particularly self-driving technology, you're looking for scenarios that are similar to each other. So, let's say if you can safely operate on a highway with certain features, then you can safely operate on any highway with those same features.
There are a lot of highways in America that look a lot like each other, rather than city centers that look a lot like each other. So that's not to say that the technology isn’t transferable to different use cases, but if you just think about focus and let's get this technology out there as quickly as possible, going the sort of trucking route is beneficial.
Niki: Yes! Okay. Just for everyone- there's a little background motorcade. [Lia: Yeah] Welcome to Washington. This is a regular weekly occurrence [Lia: Yeah]. We keep it in. [chuckles] So, I just want to reiterate what you said. So, when you think about the idea of truly autonomous vehicles operating in cities for passenger cars. [Lia: Yep] You're talking crosswalks. Potholes. If you're in Washington, DC, sinkholes. [Lia: Oh, yeah] Roads collapsing, [chuckles], bad street signs, scooters in the crosswalk. It's much more complicated, but for the route you're talking about, if you have good lane lines and you have a repeated set of information where you know that route it's much, it's much more realistic that this is going to happen faster.
So that's what you guys are doing first. I think that these truly self-driving passenger vehicles are just much farther out because it's just going to be more complicated to, y’know, figure out all of those scenarios of city driving.
Lia: I have confidence that it's not a matter of if; it's just a matter of when. Right? And so, it's this idea of, let's just, we know the benefits of this technology, and we want to get it out, y’know, of course, first, safely, but then second quickly and, and, if we know that we can do it a certain way, let's focus on that. Let's get it out there and then continue to benefit from the learnings to then be able to launch subsequent products like our ride-sharing product. And you'll actually see the first ride-sharing types of trips that we'll do are going to be more similar. So, they're going to be highway trips, right? Because why not leverage sort of the vast experience that we have from the work that we've already been doing on trucks?
Niki: So you, basically, one thing I just want to think about is you're kind of an enterprise company, meaning you're licensing your technology to, like, a FedEx?
Lia: No. [Niki: okay] So we're not, we're not, licensing our technology we're offering, Driver as a service. So, um, similar, so from a logistics perspective, FedEx says I have, y’know, X number of loads that I need to move between A and B at this time. And we say, great. We'll do it, right? And we move the goods just like they would get that service from their service providers today. Similarly, on the ride-sharing side, we actually won't be licensing any of our technology. Instead, it’s, Uber has, y’know, a certain amount of demand, and we'll say here's the supply that we can support so long as it lives within this, kind of, operational design domain, which is a fancy way of saying space where we can operate. So, it's really, it's, it's the service rather than the technology.
Niki: So, it is actually your trucks!
Lia: Well, so, that's a good point. So, right now, it is our trucks, but over time through the partnership that we have with OEMs, with automotive companies, whether that's Packer trucks or Volvo trucks, or Toyota. We're working with them to figure out, possibly it's sort of a three-way partnership where we offer the Driver as the service, the automotive company offers the truck, maybe in a leasing capacity, and then we combine to offer that sort of joint, that joint, product together. So right now, we own them, but over time you can guess we probably don't want to have too many vehicles, too many assets that we control ourselves purely from the fact that that's not what we're best at, right? We're best at the technology. So, we'll leverage our partners for this sort of maintenance and, and, servicing and fleet ownership.
Niki: So, what is the timeline for, say, this Texas route for drivers where they pick up a load, they drop it off? What's the timeline of that being fully automated for you guys that you're expecting?
Lia: Yeah, our goal is end of 2023. [Niki: Okay!] Yeah! That's the, that's it, that's what we're all staring at every day when we, when we talk about, kind of, what we need to do, have the schedules going and, and look, that's the goal, and we're marching towards it. But again, y’know, the great things about, about, Chris Urmson, our CEO, but another one is just his focus on safety. So, we're not doing anything until we have complete confidence that to remove the driver is, is, the safest thing to do and that we're ready to do so. So, that's our goal, end of 2023. And, and, and we're on track to get there.
Niki: And I think it's an interesting point to make about timelines because, I remember in 2016, I was at a conference, and people were talking about the fear that self-driving Ubers were going to replace Uber drivers. And I kept thinking, these cars can't even work in the rain [Lia: chuckles] because they're too highly sensitive. Precipitation freaks them out, on bridges, they would just stop because they couldn't figure out from the sides of the vehicles; they didn't have buildings to figure out what, where they were. So they'd just sort of like stop on bridges. And I kept thinking, this is not happening soon. Like, when you're talking about autonomous vehicle technology, it really is a long game to get it right partly because of the safety so important. And partly because the technology is really hard [Lia: yeah] to get right. But if we do get it right, which I know that you guys will, specifically for freight, it's going to supplement a supply chain that is bleeding drivers, that needs more efficiency, that can create these back and forth in a much faster way.
And hopefully, that can happen in, what I think, is a very short amount of time. [Lia: Oh, yeah] It sounds like the end of 2023 doesn't sound short to other people, but when it comes to this technology, that’s soon!
Lia: It’s evolved so much. And, and that's the thing, it's this benefit of, kind of, focusing on the right use cases and prioritizing resources towards, towards, those to make sure that you can tackle it. But, I mean, we're, we're going pretty long stretches today not needing to intervene. It's incredible. And, and, and it's not just Aurora. I mean, there are other companies that have, have really achieved a lot in this space. So the technology is coming. It's, it will be here, y’know, very soon. And in some cases, there are probably some companies that would say it's here today. I think, um, you have both the technology and the, the, need to be able to educate people on it and prove its safety, and, and just get folks comfortable with, with, what it is.
Niki: When you're talking about humans driving their cars, there's a really wide range of how much of an assist people want. [Lia: Yep!] I'm a driver. I drive a stick shift. I really don't want the car driving for me. [Lia: Yeah] I'm not interested! [Lia: Yeah] I find it really disconcerting. And when I have driven friends’ Teslas, I find it incredibly disconcerting because it's also, it's like, why am I in the left lane? [Lia: chuckles] Y’know, why am I hanging out in the left lane? [Lia: Yeah] But there are some people I look at, y’know, teenagers don't even want to learn to drive. A lot of them are refusing to get their license. So, maybe it's generational, maybe it's regional, maybe it's just certain people. But when you talk about the market, I don't know how many people are gonna want driver-assist.
Lia: Absolutely! Really important distinction. I'm the same as you. I don't feel comfortable with any sort of driver-assist technology. I need to be a hundred percent alert and I don't want anything letting me think that, that, I don't need to be in full control the entire time. Right? And so, that's something that's, that's, really important to us as we talk about this technology. What we're building is, is, not to be handed over to a driver. It's, it's truly, a fully, fully autonomous system that nobody should be sitting in the front seat of.
Niki: I think it's so exciting. What's something I'm missing? What's something that we haven't covered that you think is really cool?
Lia: We've wanted to do trucking for a long time. And one of the things that was holding us back in the early days is, in order, in order to be able to operate safely at high speeds, you need to be able to see really, really far away because that's how you can predict what's happening on the road. And so, only in a coup…only a couple years ago, were we able to find the sensing technology that would allow us to do that. So we bought a company named Blackmore; Bozeman, Montana. So we now have a Bozeman outfit.
Um, and this technology is really mind-blowing. It's, it's, called frequency modulated continuous wave LIDAR, which essentially just means it can see really far and it can see if an object is moving towards you or away from you. And so, that's really useful. And it was this moment that we realized, okay, now that we've secured this technology, now we can do high-speed operation.
So, this is a super nerdy, y’know, piece of hardware. It's inspiring to me for two reasons. One, because it's enabled us to do
this high-speed. Two, because it's, it's, designed and made in the United States in Bozeman, Montana, which is crazy. You don't hear about that. And, and so, I always liked to talk about that, cause it's kind of a little bit, sort of, behind the curtain of the crazy technology that's, that's, making this work.
Niki: I love it! I love it! I love an American-first technology [Lia: Yeah] that we're using. I think this company, the fact that it's based in Pittsburgh, is really cool and different for a tech company. I think the fact that it's going to help our logistics and supply chain in, in a relatively short amount of time, which is a couple of years from now, as we continue to lose long-haul drivers and make it a better living for the people who do want to stay in the trucking industry, but they don't have to do those long trips. And then the trips themselves are going to be safer and more efficient because you don't have humans who are tired and time-limited and, and having this, you know, it's a, it's a grind that, that lifestyle. And if you can get people to do the shorter parts with a fully autonomous middle part, I'm for it!
Lia: We're excited. It’s, and we have this, this benefit of, just having these incredible partners too, which, which, just makes all the difference in the world. Right? I think you see in a lot of cases with, with, tech companies, they want to build everything. And we knew, from the beginning, that makes no sense, right? So, we have incredible automotive partners. We have FedEx; we have these great investors. And so, there's just this sort of industry camaraderie and feeling like we're all in this together because we all stand to benefit. And so, that's pretty powerful and gives me confidence that we'll be able to pull this off.
Niki: We will end on that high note [Lia: Yeah] because I think that's exactly right. Sometimes people worry about consolidation. I mean, obviously, we're in Washington [Lia: chuckles], but in this case consolidating in a way where you're actually working together, you're getting so much of this talent, the Uber talent brought in to Aurora, still finding ways to partner with these other companies. I'm really excited about this. I'm excited about what you're doing! Thank you so much for coming on today.
Lia: Thanks for the opportunity. It was nice to chat.
Niki: Next week, our guest is Kat Mauler, and we’re talking about parenting in the age of iPads and smart cribs. Even for the child-free among us, it’s still a kicky conversation with some fascinating insights. Be sure to follow Tech’ed Up wherever you get your podcasts.