Sebastian Thrun was one of the early pioneers of the self-driving car, and spent years working at Google and elsewhere to make autonomous vehicles a reality. Then he ditched the industry entirely and went for something even bigger: flying cars.
Except, wait, don't call them flying cars. Thrun, now the CEO of Kitty Hawk, calls them "electric vertical take-off and landing aircrafts," or eVTOLs for short. (It's not quite as catchy.) But whatever the name, Thrun is betting that they'll be transformative. No more dealing with existing infrastructure and outdated systems, no more worrying about the human driver next to you. He imagines a fully autonomous, fully safe, much more environmentally-friendly skyway system that doesn't have to worry about terrestrial matters at all. And he's convinced that's all coming much faster than you might think.
Thrun joined the Source Code podcast to talk about the state of flying cars — sorry, eVTOLs — along with his vision for the future, what it'll take to get there, why batteries are the bane of everyone's existence and whether he's nervous to be the first human passenger inside Kitty Hawk's latest vehicle, Heaviside.
You can hear our full conversation on the latest episode of the Source Code podcast, or by clicking on the player above. Below are excerpts from our conversation, edited for length and clarity.
When you're starting out on a project like Kitty Hawk's, you have this big goal and futuristic vision in mind. And obviously, helicopters and planes exist, so we know how to make things fly. But where do you start? Literally, what's step one?
Step one is a spreadsheet. Larry Page and I sat there and said, "How much energy is there in a battery? And how much energy do we need to lift a person? And how much energy do we need to fly a person 100 miles at this and that speed?" And with very simple math equations — equations that every undergraduate could learn when they study aviation — you can already make a case that battery technology has now matured to a level that's feasible to do electric flight.
Why is electric so important? Electric, at the surface, looks like a step backwards. Because the amount of energy you can put in a battery per weight or per volume is a small fraction of fossil fuels, like gasoline or jet fuel. But it's a big advantage, because electric allows for a very simple architecture with very simple electric motors. It allows for distributed architecture: We have many of those, so if one fails it's not a big deal. And in principle, it makes these vehicles very, very quiet, because you have no combustion engine that burns fuel that makes a lot of noise. And all these things have become true as we build these prototypes.
That was one thing I was going to ask you, is the question or assumption you build out of. And it sounds like that question was: How do we get, and use, and harness energy?
If you ask me what I love the most and what I hate the most in what I do, the answer is always exactly the same: It's batteries. When you want to use batteries for anything, you tie your hands behind your back. It was really hard for automotive companies to get ranges of 300 miles out of a single battery. It's an amazing accomplishment that is not obvious, and those batteries are much, much bigger than your gas tank.
For flight, even more so. We are in a world where every ounce of weight punishes you. We really want to get rid of any amount of weight, because guess what? Flying is a gravity shield. You have to combat gravity for a while in the air. And the more gravity pulls you down, the more energy you have to spend to stay up there. So for us the fundamental question is if our battery is good enough. And it turns out they are. It turns out, modern batteries are plenty good to build electric vehicles that can take off and land like a helicopter and still have a range of 100 miles or more.
How long has that been the case?
I'd say for about eight years. And it's getting better by the month. The big push in automotive has certainly helped us in this niche field of VTOL. And we're using effectively automotive batteries today.
There are all these questions about the job these vehicles should do. Should it be a taxi? Should people be able to own them? Should it be something that can be summoned like an Uber? Should there be airports? How has your sense of the job that these vehicles should do for people, both at Kitty Hawk and in general, changed over time?
My objective has been unwavering. We want everybody to be able to use these vehicles, every day, whenever they go from one place to another. And the benefit would be that there would never be traffic again, because the sky is very, very large. It would be much, much faster: We fly at speeds of up to 180 miles per hour, on a straight line, without ever stopping for a red light. That's really nice. It would be greener: We have a small energy consumption compared to even the most efficient electric cars. And it would be safer, certainly more safe than the ground.
So how do you get there? I would love to be able to take over land anywhere. We really have a strong focus at Kitty Hawk to really get the costs down, to make it comparable to automotive costs. And that's a differentiating factor for us that we're really pushing hard. We like to make it super easy to use: You get on board, and you don't have to have a pilot's license or anything, you just maybe have a day of training or maybe zero days of training, and you can still use it. Maybe your kids can use it, your grandparents can use it. And we'd like to be pervasive in cities, we'd like to be everywhere, we'd like to be able to go to every place.
Now, if you start selling those, and you're going to buy them from us, then you have a problem. There's things sitting around all day that you don't need, it might not fit in your garage. And it's just not a pleasant experience. What the world has taught us in the last year is that instead, we're entering this new economy where people can share things. So we love the Uber model, we love the Lyft model, where you go to our app and say, "Hey, I need a ride," and the thing comes and lands in front of you. You hop in, and then punch in your coordinates, it goes there and it disappears. And it makes it much better economics. I can prove to you that the cost to you as an end consumer, if you just use us when needed, is a small fraction of what you would pay if you try to own a vehicle.
Is the goal to take a version of the transportation system we have now, and just put it in the air, where there's less traffic and more ability to move around? Or is it to make it an additive alternative to what we have now? Like, if you fast forward 40 years, are cars part of the future that you're imagining?
In 40 years, I want every major OEM in the car space to have stopped making cars, and only making what we sometimes call flying cars.
When we look back 100 years, that transition from horses to cars obviously took a while, but in historical proportion it was very, very sudden. And it turned out, for daily transportation needs, cars are the better solution than horses. By and large society has switched over.
When I look at what we're doing today, we fly roughly up to 10 times as fast as a person stuck in average Bay Area traffic, which is, last I checked, about 18 miles per hour on the ground. We go much faster, we go in a straight line, we fly on a small fraction of energy. So even if the entire Bay Area turned into Teslas — and I admire Tesla — we'd still just be a third of the energy that's being used for it. So much, much greener. So if the Bay Area wants to really make a significant contribution to global warming, they would look for technologies like ours. And I have this very firm belief that when everything is said and done, you're going to be safer.
What's interesting about that is, the folks working on autonomous cars say very similar things. Not quite to the same extent that you do, but the idea of it being safer and more efficient and better for the environment is very much the same thing that folks at those companies — companies you used to work for! — say. Having been in both of those spaces, what feels different about this, as opposed to that?
Many of those companies have this vision of building a taxi service for self-driving cars. They're still subject to the same limitations of being on the ground. So if there's a big traffic jam, then they'll be stuck in it, and they will contribute to it. If you really want to be transformational, being able to make your car fly is actually really transformational.
I left self-driving cars as a field, and entered this new field, when I realized that no matter how hard you work as a self-driving car person, all you're going to get is you're going to be cheaper than cars and safer than cars. That's really what you're after. But you're not going to fundamentally transform how traffic works. If a self-driving car company succeeds and we succeed at the same time, then we're going to offer to a customer transportation at the same cost, equally safe. So safety and costs will be the same, but we can still go 10 times as fast, or five times as fast.
How do you think about sequencing that? Because I think the other big learning of the self-driving world has been that you can see the big future, but it's very hard and very expensive to get there. And it takes a long time. And I think the mistake a lot of those companies have made is not figuring out steps along the way to make it work. And I wonder how you think about that too. Is it enough to build toward the big future? Or do you have to find victories along the way?
Anybody who's ever had the chance to work with Larry Page, the Google co-founder, will know that there is no amount of impatience that would describe him accurately. I go to him and say, "We can do something in a year that would have taken a lifetime," and he says, "Why not in a day?" And it's refreshing to think about, because that's what happened in the tech industry in the last decades: For anything that wise people would have told you would take 10 years, now takes a month.
We are right now in a sprint: By Q3 2022, a year from now, we hope to have 100 hours flying pretty much every day. That's our objective right now: Fly people around as passengers and really get this kind of working a mini-city type situation, where we really get the prototype to a stage where it's safe. They'll be operated daily, and there will be dozens of them, and we really have the basic technology under control. We're almost there. We now have about five or so flying prototypes, and we fly them three times a week. But we haven't yet been able to fly 10 of them at the same time. So we are inching our way up.
And then, when we're really confident that the technology is really under control, and we have a safety case made and documented, and all that work in progress, we want to find a partner. We're going to ask the world, "Hey, you 5 million mayors who have a traffic problem that your people want you to solve, you 5 million cities in the world, who of you would be interested in trying out a new, safe and scalable solution?" And we're going to see, is there anybody in the world who wants us? And our conditions will be, when we come to you, we won't be for you the city politicians, you won't be for you the rich people, we will be for everybody.
You mentioned proving and documenting safety, which is going to be a really interesting challenge both from a regulatory perspective and from a public opinion perspective. People get afraid of flights, but they don't get afraid of getting in the car, even though it's statistically massively more dangerous to get in the car and drive to the grocery store than to get on a plane to Europe. So what does that look like?
That's certainly one of the questions that will keep us busy. We've been working very closely with the FAA; one of our advisers is Mike Huerta, who used to run the FAA as administrator. We're really investigating how we take safety practices into this new field. What you'll find is that the regulatory framework that exists today isn't quite ready yet. And the biggest difference is, today, over any busier space in the Bay Area, you might have at any point in time maybe 30 or 40 aircraft in the air. And I want to have 30,000 aircraft in the air. There's nothing that can manage this today. Having said this, we have a very active safety team inside our company that really engineers our safety situation to the same level as a typical small aircraft would have.
The FAA has a regulatory framework for these little Cessnas and Cirruses that fly around. Those typically have a safety metric of one accident in 106 hours. And we are working to fly at exactly the same safety level. From there, I want to go much, much higher. I want to get to the point that these guys never crash. And there's one reason to assume we can accomplish this: Our vehicles are just incredibly simple. They have almost no moving parts, except for a few rotors. It's much simpler than a combustion engine in a car that is basically a chemical factory.
Our Heaviside vehicle has eight independent motors, and if you lose one, they can keep flying as if nothing happened. It won't do this, it'll typically land, but we can keep flying as if nothing happened. This fundamental redundancy, I think, is one of the big differences. Helicopters often have just a single engine, and they have a single bolt that holds two blades together. And that bolt is called Jesus Bolt. Because if that thing breaks, praying is your last option. We don't have that. We have this complete distributed system where any component can fail. And we typically have at least two or three replacement components that can just act on its behalf.
How do you balance that wanting to be safe and redundant with things like design and weight and wanting something to be inexpensive? Those things seem so incredibly at odds all the time.
I would say when it comes to safety, it's safety first. We're not willing to give up redundancy just to save a buck or two. And the reason is very simple. I think if this technology proves to be unsafe, I would not believe that the public would love it.
Frankly, I think we love cars, and we've gotten used to the fact that they're unsafe. But if I told people today, "I have a new technology that's going to cost 30,000 lives a year in this country," I think people would ask the question, "Is this something you want?" If you ever lost a loved one in a traffic accident, it's devastating. It's something that has influenced my life. I lost a loved one when I was 18. And it has transformed what I want to work on, because I think safety is so paramount.
But it turns out in electric aviation, we have a leg up, because it is so easy to build an airplane with many, many, many, many motors. It is so hard to build a jet with many, many, many jet engines. In fact, modern jets have only two engines. If you lose one engine, they can still safely fly. But God forbid if you lose both, you're in for a hard landing.
This is a totally random aside, but you just made me think of this. Is GPS good enough as a technology to do all of the stuff we are going to need it to do in the coming decades?
Yes, with one caveat. GPS is unbelievably amazing. The fact that I can take my phone and figure out not just what street I'm on, but what lane I'm in, blows my mind. It's something we should really all appreciate. It's plenty sufficient for air navigation and for every aircraft. In fact, when you look at how the FAA has adapted, it has made GPS a prime method of navigation over other methods that preceded it.
The only downside is, we can't quite guarantee that GPS always works. When GPS came out, many trucking companies used GPS to monitor truck drivers, to see if they're breaking the speed limit. And truck drivers, being smart, bought things called GPS jammers. And in places like Newark airport, these highways are so close to landing strips that on planes, all of a sudden, just before landing, the GPS didn't work because some truck driver underneath was driving with a GPS jammer. So there are some safety cases that we all have to address, because it can't be that there's a GPS jamming event going on and all of a sudden all airplanes crash. The public would not accept this, and as a result, the FAA will not accept this. But there are some very, very good solutions underway: There are ways to navigate that get you even through GPS outages. The Google self-driving car, it can drive through GPS outages very nicely for hours at this point, and we can do the same for aircraft.
There's a lot of interest in this space right now. You have Joby and Archer going public, and there's a lot of money being flown around and a lot of demos being done. Is that because we're at some really exciting inflection point with all this? Why is it so frothy right now?
I love it. I mean, I used to work on self-driving cars, and for the longest time when I told people what I'm doing, they smiled at me and said, "I'm so happy, Sebastian, that someone pays you for your work." And then I still remember, around 2017, when Uber got really involved and so on, all of a sudden it was everyone's news. And people realized, "Oh my god, that's coming," and took it seriously.
And I think we're going through the same transformation right now in this eVTOLs space. Before it was kind of more like, "Oh, you're working on flying cars. That's funny." And now it gets to the point where serious people say that might actually become reality. It takes more than one company to do it, so I'm really happy that we have so much bandwidth right now in the public mindshare for what we're doing.
I wish I could give people something tomorrow morning that they could try. But I think that we are close. I think it's going to happen. Definitely this decade, definitely by the middle of this decade, and hopefully much, much faster.
Speaking of, if I'm reading the news correctly, you're currently slated to be the first person to do that in a Heaviside at some point in the near future, right?
Yeah! I wish it wouldn't come across as like a media stunt, because we're very, very serious engineers. But yeah, we had an all-hands, and we talked about the safety case and so on. And some person raised their hand and said, "Sebastian, you're so confident, can you please fly first?" and I spontaneously said, "Absolutely!" But yeah, I'm very, very confident. I'm super excited.
You're a licensed pilot, right?
I am. A licensed commercial pilot, even.
Is it going to be a little sad that you can't fly this thing?
I tell you what: I became a pilot, and I love being a pilot today, because it gets me to places. But I became a pilot because I really wanted to understand how airspace is being managed. And almost everything I do as a pilot is somewhat ridiculous.
Imagine you are in a safety-critical situation, but you have the lives of your passengers in your hands, and you're getting the directions over a cell phone, but you're on the same cell phone with 50 other people at the same time. And the person yells out not your name, but your codename: "November Four Four Zulu Echo." And then you have to repeat what you heard. If you make a mistake, you will put safety in jeopardy. And then halfway through, that person yells a new frequency at you, because now your cell phone channel changed, and you dial this frequency into a machine by hand, and if you make a mistake you're off the ether and you're becoming a safety hazard to everybody else. That's what flying means today. We don't even have a real cell phone, where I can talk to somebody uninterrupted without repeating everything and having 20 other people listening and stepping on me.
There is a lot of work to be done. When you look at the future, I don't understand why I as a pilot am a link in the safety chain. At the end of the day, all I do is operate the autopilot. If my computer could talk to my airplane directly, and stop dialing in frequencies, and tell it exactly the heading and so on, I, Sebastian, would be a safer pilot today. 99% of aviation is already computerized and automated; we're all going to recognize the computer is safer than people. And not just safer, but more affordable.