Sam Abuelsamid 0:03 Welcome to Episode 128 wheel bearings. We have a very special guest with us today. First of all, I'm Sam Abuelsamid from Navigant research. Dan Roth 0:12 I'm Dan Roth from Forbes. Rebecca Lindland 0:13 I'm Rebecca Lindland from Rebecca drives, Dan Roth 0:15 know if you were going to go or not. Sam Abuelsamid 0:19 And joining us today from across the table from me, here is Bryan Salesky 0:24 Bryan Salesky. Nice to see all of you again. Dan Roth 0:27 Welcome, Brian. Right. Sam Abuelsamid 0:28 Great to be here with you, Brian. We're we're sitting here in in Dearborn. Brian's in town for for a day or two. And we managed to wrangle them for a little conversation about autonomous vehicles. So I get to talk to Brian on a pretty regular basis. So I don't want to dominate the conversation today. So why don't I let you to go first with your questions? Dan Roth 0:52 All right. You know what? I'll let Rebecca go first semester. She has better questions. Bryan Salesky 0:58 You're also polite with one another? Sam Abuelsamid 1:01 Door deteriorates very quick. It goes off into tangents that you wouldn't believe. Rebecca Lindland 1:10 So, Brian, it's so nice to talk to you. We met a couple years ago in San Francisco at I coincidentally at a Ford event as well, when I moderated a panel, I think that we were on and speaking. So when in that conversation, you know, really Argo was fairly new and and the investment that Ford had made was fairly new as well. And there's been a lot of a lot of changes since then. But one of the things that I'll never forget is that we talked about the geographical specificity of driving behaviors. And you had mentioned like the, I think it was like the Pittsburgh left, and I talked about the Michigan left, and then the Boston jug handles and all these crazy ways that we manage left hand turns in our world. And so I wondered, is that still part of the conversation? If so, how's that coming along? Just kind of where did that Stand. Bryan Salesky 2:01 Yeah, no, it's, it's one of my favorite topics is all the possible left turns styles in, in the US and abroad. So, you know, an hour ago we're we're testing in five different cities now all have different styles and geographies. So we've got, you've got Austin and Miami and DC and, you know, in in just those three alone, there's a lot of differences. And so, you know, by taking that sort of testing approach, it's been very helpful because we get to encounter all the different, you know, myriad of types of that just left turns, but more broadly speaking, how the different actors in the world behave differently, right, so we get a we get a sensor a measure of where the dangerous intersections might be, where do pedestrians likely jaywalk, we get a sense for how things change over time and over seasons. And all of that helps us just Make the system better, and help us to better anticipate and predict what those you know what those road users are going to do. Miami, there's a huge scooter population, same thing in Austin becoming more prevalent in DC, and you know, getting experience and how they tend to navigate and the types of routes that they take and how they behave. All of that information helps us just make our system better. So it's definitely in the conversation. It's not just left turns. It's all variety of things that we that we learned when we go to a city. Rebecca Lindland 3:31 It's fascinating. Sam Abuelsamid 3:32 Brian, you know, one of the things that is particularly challenging about this is, you know, how, when you think about going to a production system with this, you know, you've got all these different types of driving styles. Is that something that your system is going to be able to just automatically adapt to those when you go into a new city? Or are you going to have you know, maybe different calibration sets or different rule sets, you know, that you deploy in different Cities? Bryan Salesky 4:00 Yeah, I think a good way to put it as different calibration sets. So it's one system. It's one platform. It's not like we're engineering a whole new system for each city. But we're also not going so far as to allow it to do what would be called online learning. So the ability for an algorithm to sort of Teach itself on the fly. And the reason we don't do that is because what if it learns the wrong thing by accident, that would be bad, right? And so what we want to do is, we want to take these systems are vehicles that we map with and that we we get when we first go into a city, we sort of a launch plan of how we not only map the city, but also get experience all the experience that I just that I just talked about. And so there's sort of an add to adaptation period. And we get to see how well the system we've built works in the city where some of the areas for improvement, and then we make sort of equivalent of parameter changes that that tune it to that environment. And that has to do with sort of how confident it drives. It has to do with how it behaves at certain intersections or maybe some intersections where there's a social protocol where you don't ever block we call it block the box or don't block the intersection. There's other intersections where if you don't do that, you're never going to make your way through. And so we learn those types of things and and teach the autonomous driving system, what the best way is to, to approach to approach those types of intersections and other types of scenarios. Dan Roth 5:24 So, back in 2017, you told Kristen karasek of the the she was at the verge at the time that you wanted to give cars, the eyes, the ears, the brains, they need to operate without humans. And your timeline was 2021. So that's now a year and a half away. So how are you doing with that? Is that goals still? 2021 are we we learning as we go that we're going to need more time? Bryan Salesky 5:52 Well, you know, the goal and what I said at the time is the goal was 21. And and you know that still is an important year, and a important milestone for us where, you know, we expect to have initial deployments. But the scale of those deployments is going to vary over time, and it's going to be a slower ramp than I think what most people would suggest in the industry. You know, our view is that these vehicles are going to get deployed and, and operate in small sections of the city, limited number of vehicles, and then it will grow over time. It's going to grow where we learned the demand is it's going to grow based on the types of partnerships we have with businesses in these cities. And and it's going to, you know, it's going to be a gradual, gradual deployment. In terms of and that's that's just sort of the natural cadence of building any business, right? Yes, it's an it's an it's an evolution. It's not a revolutionary thing. Sure. Yeah. It's not a gotcha Dan Roth 6:51 question. It's just checking it Yeah, no, no, no, I've just Bryan Salesky 6:54 given I'm given the full answer, right. Because I think this is important topic to you know, clarify for people. Now Now I separate everything I just said from when is the driver removed from the car, you can start to build these businesses and learn in market with the driver in the vehicle, and and still monitoring the self driving system. And I think the question of when the driver gets pulled is a whole separate question and is unparalleled, everything I just described. And for me, it's it's not just, you know, when does the data tell us that it's safe, and it's ready, but it's also it? When is the community ready to accept that and I think that may vary city by city. I think I think, you know, policymakers will obviously have a say in this and and, you know, to me, that's that's, that's even you know, another question in addition to what you just asked, Sam Abuelsamid 7:46 When waymo launched WeMo, one in Chandler, Arizona last December. And one of the surprises to a lot of people was that they still kept the safety drivers in the vehicles. And and they in fact, they're still there today. Do you anticipate following a similar pattern, you know, in 2021, you know, watching commercial deployments, but perhaps, you know, either in all or some of those, at least three cities that you that have been identified now, Miami, DC and Austin anticipate potentially retaining the safety drivers for some period of time after that commercial launch. Bryan Salesky 8:20 Yeah. So, you know, for me, I it's hard for me to know exactly what commercial launches. You know, as I said earlier, I think it's there's these initial deployments where we're going to be serving people and businesses, potentially, and and you know, when it actually launches, or what launch is, to me is sort of a fuzzy thing, right? And as you start to learn, and as you build these businesses, to me, that takes a separate evolutionary path from when you pull the driver. I don't tie pulling the driver to any of those particular milestones. To me, that's a separate question. And, you know, it's so it's hard, it's hard to link it to a date, but I think what you saw with the way deployment, you know, makes sense to me in terms of the net, again, the natural evolution of things, which is they're simultaneously trying to stand up a business, as well as prove out their self driving technology. And and I think, you know, for them, that's, that's obviously two threads of development. Dan Roth 9:16 I mean, and when it actually does deploy as well, when you talk about commercial deployment, what does that actually look like? I mean, is it still individual cars? Or is it some other offering? That's more efficient? I mean, you're thinking about efficiency to the you get back to a bus with a safety driver? Well, that's that's a train. Now, like, what how does it How does commercial deployment actually sort of get out of the gate? And does that look like people think it looks now where it's everybody's own individual car just drives itself? Or there's some other way that's it's more of a walled garden in some ways, you know, it starts off in ports, which, you know, they're working on that solution, or, you know, farms already have automated vehicles to it gets, where does it actually start? Bryan Salesky 10:00 Right now it's good questions, we probably have to unpack it a little bit. So I think for the first question is, is it personal car ownership? Or is it some sort of shared fleet or a service. And in our view, it's the latter, it's, it's some sort of, it's some sort of service that's provided. And the reason for that is, because of the cost of the technology is relatively high, most people wouldn't be able to afford that to equip their own personal vehicle with with all the self driving tech. So the better way to offer it is to actually provide a service, whether it be to deliver something or to move people and to spread the cost of the technology over the lifetime of the asset. So, you know, imagine if the vehicle is able to operate for roughly 300,000 miles with some routine maintenance associated with it. You can imagine spreading the cost over those 300,000 miles and all of a sudden on a per mile basis is not that expensive. So that's the idea on how I think this will be deployed initially, in terms of the types of business models you know, there's so many and that's what makes this field so Exciting in my view, even if you just you know, a lot of people talk about, well, is it ride hailing? Or is it goods movement? Or is it both? Well, to me, there's, there's a whole bunch of slices within each of those. There's tons of different use cases in the goods delivery side loan, whether it be last mile delivery door to door fits, going from one sort of, you know, pick up drop off point to another, you know, there's all sorts of ways you can sort of slice this problem. And and, you know, to me, that's the challenge that I think everybody in the industry has is to sort their way through that and figure out what's the best way to marry what the tech can do today with the opportunity and how does that change as the system matures? And, and as, you know, as logistics changes over time, so that, that that whole thing is kind of a giant Rubik's cube that I think everybody's wrestling with. Yeah, well, sure, and Dan Roth 11:53 especially, go ahead, Rebecca. Rebecca Lindland 11:55 I just I I just wanted to to Because we talk a lot about the sort of long term plans of all this, but I want to get into a little bit more of like today. What? In your mind, Brian, what's been it? Was there one thing that was easier to accomplish? I, that you were surprised, like pleased about or is there something that has just proven so much more difficult than you expected? And kind of more? I don't think a day thing. Yeah, I don't think anything's easy. Bryan Salesky 12:31 To be fair, I don't know George hawks seem to make it sound super easy. Just consumer stuff and use it off label and it drives your car. Yeah, I mean, it is. I mean, it's, it isn't too difficult to, you know, put together a smart team and, and, and to build a prototype right. But that's sort of follows the 8020 rule, right. I think to get 80% of the way there is not too hard, but the last 20% is super hard, and, you know may take even longer than it took the time to get to the 80% So that, you know, there's that definitely applies here in terms of what's difficult. I mean, look, I think, I think the question that's before all of us is, how do we best assemble a company? That is, this is what I spent a lot of my time on, how do I build? How do we build a company that's going to be around and last for a long period of time, and that's gonna, that's going to actually solve this in a meaningful way. Right? I want to assemble a team and a team that's committed and mission driven, that wants to, that wants to see the evolution of all these businesses develop and see the technology mature and move from one platform to the next. And from one partner to the next. How do we build a lasting brand and a lasting product that's going to really solve this problem in a meaningful way? And and that's and that's what well, and that's what we set out to do at Argo and I'm, that's what I'm really proud about is that it all starts with the people. It starts with getting mission driven people Whoo, you know, there's no drama, they just want to come to work and they want to solve this problem they want to make, you know, they all have their different stories, but generally speaking, they want to see transportation be safer and more accessible. And I think I think that, that while we're one piece in the overall puzzle, and that we're focused on the self driving system, there's a ton of challenging problems to work on there. And, and, you know, we're driven by, by the sense that we can build this and deploy it in a truly meaningful product, not just a prototype, but something that's actually going to be embedded as sort of Argo inside, in all of these vehicles and become a platform that that's going to, you know, enable more efficient, more efficient ways to get from A to B in the future. I mean, that's what we're, that's what we're after. That's what we're that's what we're driven by. And sure, there's a lot of challenges along the way. And and there's a ton of hard problems, but but I think, you know, that's that's why we exist, that's what gets us up on a bed in the morning Sam Abuelsamid 14:59 and you Got to expand on that a little bit. You've got partners here at Ford, you know, and even within Ford, there's multiple teams that are they're part of this. You've got a team within autonomous vehicles LLC business unit that is working on the business models of, you know, how, what's the what's the consumer facing business, how they're going to deploy these things. There's Ford earlier this summer acquired quantum signal, which is going to help a lot of the Ford teams with things like, you know, working around on the user experience around the vehicle, how people interact with the vehicle, right? You know, it's it's not nearly as simple as just building it. Not that it building an AV stack is simple, but I mean, that is only one piece of a very large puzzle. Bryan Salesky 15:43 Oh, you're absolutely right. And this you have to partner in order to be successful here. There's so many different aspects to the problem, whether it be user experience or the functional safety aspect. You know, how to build in all the redundant control systems, which you and I've talked about before, Sam right into the vehicle platform. You know, conventional vehicles today that are meant to be human driven are not at all ready to accept an autonomous vehicle stack, there's a lot of changes that need to be made there. And so there are partnerships with Ford. And now VW enable us to have the right close relationship with these automakers to be able to make the changes necessary to adapt the vehicle to the AV stack, and vice versa. And you know, it takes one team one mission, one, it takes it takes, it takes a coalition of people across the supply base and our partners to actually be able to bring this to the world in a meaningful way. Dan Roth 16:34 If you're going to own the cars to it sounds like this is something that you've thought about in terms of as it comes out, you know, that it's less individual ownership than it is some sort of service within your own your own machine and you're trying to maximize its use. So it's got to also handle all kinds of conditions, you know, and that's, that's right. Bryan Salesky 16:54 Yeah. And no tell you one of the benefits there is you can actually, you can you can do the routine maintenance. In a predictable way, right? Because you own the you own the vehicles you can have you can have great you know, maintenance history and track every aspect of the lifetime of a vehicle you can do proactive maintenance you know, the other thing is that with our with our partners right you know, automakers one piece of the business that maybe doesn't get talked about a lot is their parts delivery and service network, you know, forward in the US is is enormous. The their footprint. Oh, yeah, that's actually dealer. That was one of the things that in in the police fleet side of things. Yes. Ford was really really, really good at that. And that's why a lot of fleets stuck with Ford even when Daimler Chrysler jumped back in with the charger this day we can't get the parts we can't get the support for has that gets right up so they know what it's right. And so when that's right, so when it comes to keeping vehicles on the road and being able to maintain them and to be able to get, you know, parts and service quickly and done properly. You know, there's that They have a huge advantage there in terms of fleet management. Dan Roth 18:03 To follow up on that sort of eight car handle all conditions. Question, you know, there's chatter about LIDAR and about, you know, stuff like FLIR, like infrared and stuff. So what, what is your stack look like? What sensors are going to enable you to, to actually drive fully self drive in adverse weather, without any kind of, you know, sort of special measures, you know, like a person coming up and scraping the sensors off. It's like, how are you gonna manage that? Is that something that you've, you've cracked yet? Bryan Salesky 18:41 Well, so we've aligned around, LIDAR, radar and camera. Each sensing modality has its strengths and weaknesses. You know, we're a big believer that LIDAR is an important part of the equation. It's a, you know it it operates in three dimensions. and is able to get fair amount of range if you're using the right instrument. And it's, it's an important part of the overall equation. You know, strengths in one complement, the weaknesses in the other. All three modes are super important. All three modes have issues in in really bad weather. So, you know, as an example, fog is, is probably the hardest type of sort of bad weather scenario to solve, mainly because both LIDAR and camera are heavily affected by it ultimately changes completely what it can see. And it becomes a very different problem. Of course, when you think as a human, how does a human drive through fog? They make a heck of a lot of assumptions, right? Unknown Speaker 19:50 Totally. And so it's all clear. Sam Abuelsamid 19:52 That's right. Bryan Salesky 19:54 And so it's right. And so this sort of requires what we would call optimistic planning. And, and so is it impossible to solve? No, but it's but it's, it's definitely among the more challenging weather conditions to deal with. And it's something that is on the roadmap that's kind of further out. Now. Now, light falling rain, we can handle that relatively well today, the challenge there is to just have a really good sensor cleaning system to build to build either blow or wipe the droplets off. You know, falling snow a little bit harder, because it sticks to stuff, it can freeze. Anyway, so you know, each each type of precipitation comes with its own set of problems. And yeah, there's a long tail associated with solving all of those things. You know, it's hard to put a timeline on any of it. Because Because, and, and really, we can build a great business, really just going to cities where the weather is decent. And that's what I think you're going to see a lot of companies do and including us. And then over time, we'll start as we solve and tackle those hard, hard weather problems will start to scale elsewhere. Sam Abuelsamid 20:57 Yeah, recently early The summer and you guys start getting your latest generation of test vehicles from Ford that the the sensor stack on the roof now has cleaning system it does integrated into that how's Bryan Salesky 21:10 that working out for it's actually works remarkably well and we've been able to operate in falling rain that's relative that sort of light to moderate let's say pretty well. It isn't perfect, you know, we're still iterating It was the first generation that that came out but yeah, I mean that was it's a really great system that that Ford came up with and how they part of how they integrated the sensors onto the car. You know, and will continue to refine it. I think one Sam Abuelsamid 21:35 of the one of the funniest things about it, when they when they announced it was you know some of the testing they did they developed the bug blocker Yes, they want the cameras and they're using the the air curtains to help deflect the bugs. Bryan Salesky 21:48 Absolutely. And and the the people who get to test these vehicles, sometimes I'm envious because of some of the stuff that they get to do. We we absolutely torture the vehicle on the test track and they're coming up with all kinds of crazy things. Remote controlled skateboards, you know, you know, fake pedestrians that jump out in the middle of nowhere drop from the sky. Rebecca Lindland 22:14 I remember being on actually on the Ford test track where they had, like, big giant things come out to test the emergency braking. It's terrifying. Unknown Speaker 22:21 It is terrifying. Rebecca Lindland 22:22 It's absolutely terrifying. Unknown Speaker 22:24 I'm somebody who's very satisfying when you see the system work well, though. That's the yes, Rebecca Lindland 22:27 no, you're you're absolutely right. It is. One of our Twitter followers wanted to know, I thought this was an interesting question about infrastructure. And do you see LIDAR potentially being used in infrastructure to help with self driving cars like busy intersections and stuff? Bryan Salesky 22:46 Well, we absolutely see that as part of the the equation. We're not building the system in such a way that we have to have it. But the way we look at it is that as cities begin to invest in smart infrastructure, whether that be you know, can cameras in the intersections that are monitoring throughput, whether it be vehicle to vehicle or vehicle to infrastructure communications, whatever the source, it may be, we just view that data as another sensor input, right, we view it as another input that helps complete our picture of what's around the vehicle and the state of the world. And we'll use that to to help us navigate it. You know, one of the places where it's it's it's really useful is where there's what we call occlusions at an intersection. So imagine, we're getting ready to take a right turn and we'd like to take a right on red, where it's allowed. And but there's a concrete pole or there's a barrier or something that prevents us from able to see around the corner. With the infrastructure, we can have better assurances of what's around the corner and what the safe action is to take. Without it. We're sort of creeping very delicately, much like a human would to sort of peer around the corner and see if it's safe to go. So you know, that that sort of infrastructure, I think is a key piece. Is it certainly a key piece and where it's available, we want to use it and Sam Abuelsamid 23:59 yeah Add to that, you know, another one of the other companies is developing this stuff may mobility, you know the work they're doing here in Detroit, for example, one of the things that they do is they have roadside units mounted. They're operating on fixed routes around downtown Detroit. And they have roadside units mounted on poles, with cameras and radio transmitters, so that they're up where they can always see the traffic signals. So because they're operating these little, tiny gem electric vehicles, yeah. And, you know, a lot of times when you're driving around the city, you might be stuck behind a truck or a bus, you can't see the traffic signals. So that, you know, they put those up there where they can always have a clear view of the traffic signal, and they can transmit that information back to the vehicles and you know, information like that can can be useful to help make the system a little more robust to add to that situational awareness. Bryan Salesky 24:51 Absolutely. We talk about safety in a lot of different terms. But, you know, one of the ways to add to safety is to just have additional inputs about the state of the world, whether it be traffic lights of the positions of various objects, whatever it might be absolutely. Sam Abuelsamid 25:06 Sorry to interrupt, Rebecca. You know, one more thing now that you're, you've got a relationship with Volkswagen. And I don't know if you've actually started working with them yet, or you're waiting until the deal is closed. But, you know, Audi, for example, for the last couple of years is they've had some vehicle to infrastructure communications in various cities where they're, they're using the LTE connection that's in the vehicle to get traffic signal information from the city's central traffic control computer. Is that something that you're looking at integrating into into your system, that sort of capability? Bryan Salesky 25:41 Yeah, it's something we would love to integrate in. We haven't had those discussions yet. Just because as you said, we're sort of in this quiet period where we're waiting for regulatory approval for the deal to officially close. Okay. Rebecca Lindland 25:54 I was wondering, you know, one of the areas that I think is really exciting is the commercial space. And you touched on this a little bit earlier, but, you know, the long haul truckers that we see and and such, and are you working in that space as well to get some of this technology into the sort of semis and, you know, as they travel across the country? Bryan Salesky 26:14 Yeah, we're definitely looking at it. It is. It's absolutely another interesting business case. It's part of the, you know, overall goods delivery stack, right. And like I said, you know, goods delivery is, there's so many different use cases, different vehicle types and platforms and trips. And you know, definitely long haul trucking is is one piece of is one piece of that overall pie. Rebecca Lindland 26:38 Yeah. Do you see most of these vehicles, whatever they eventually come to look like? Do you see them as primarily electric or do you see them as having internal combustion engines as well? A combination. Just talk about that a little bit if you would? Bryan Salesky 26:55 Well, you know, to in order to support the self driving system, the vehicle needs to be electrified in some way, shape or form. And whether that's a hybrid or a full electric or, you know, a giant alternator, whatever it might be. You know, we're flexible in that regard as long as we can, as long as the power system can meet our requirements. I think the I think the broader question, though, of of what is sort of the, you know, what, what is the way of the futures? I think electrification is inevitable, that's where everybody's going. And and the key is, though, to work with the vehicle manufacturers to to sort of understand what we want to do near term and what we want to do in the long term and plug into their overall strategy, which, let's face it's not being influenced so much by autonomous vehicles or autonomy. It's been influenced by fuel economy and regulation and standard tracks, right. So I want to be able to just sort of plug into their roadmaps. Rebecca Lindland 27:55 Yeah. Where does regulation fit into this? I mean, are they Is this something that you're constantly talking to them about? Or do you feel like it's, it's Is it a barrier? Bryan Salesky 28:06 Oh, no, it's not a barrier at all. We welcome discussions with regulators, we've had great discussions at the federal, state and local levels where we, where we operate. You know, we welcome those discussions. We love to educate, and show folks where we are and sort of help help make, make what we're doing, you know, transparent, so that they can understand how the technology works to some degree, right, understand how we're thinking about the how deployment might work. And, you know, we see the policymakers as a sort of an important constituency and folks that we need to be absolutely working very closely with. When you talk to regulators Do you, do you feel like they have a sufficient understanding of the complexities of technology and that they grasp the issues around you know, what, what needs to be done or perhaps what should be done? In terms of regulation, Well, it certainly varies. I mean, you know, these folks are are, you know, whether elected officials or staff, they come from all different backgrounds, right. And so a lot of times, what we find is, is we just need to understand what their starting point is. And then, and then sort of, you know, begin the educational process. I would say, though, that every, every rate every person in sort of a regulator or policymaking context that we've met, they've all been extremely eager to understand and get that education and want to understand, you know, in basic terms, look, what is what are the good things this is going to do? And what are the things that we have to watch out for and how do we deploy it in the in the right way? And so as long as you have that sort of common starting point, I think everything goes really well. Dan Roth 29:47 Well, I think that's the the thinking that we have to we have to train us in the media and you folks in the industry that it's building these things is we should have to train everybody who's paying attention. Just got to be really frustrating when you talk about this stuff. And a lot of the coverage is teaching the public to expect that these are just going to pop out fully formed perfect work for everyone everywhere in all conditions. And, you know, regular cars don't do that now. Bryan Salesky 30:16 So yeah, and we we aren't doing that, right. And that at least at our goal, we're not doing that. Neither is neither as for, you know, the way I look at it is the education part of it is not what frustrates me and we enjoy that that aspect to it. And that's an important part of what we should be doing. But you're right, the the, the expectation that somehow has been set in the press or in the mass media, I think is that, like you said, these are right around the corner, they're going to be operating everywhere. They're going to be in your lives, you know, day in and day out. That just simply isn't the case. Right it going back to what we discussed earlier. It's going to be relatively modest deployments that will grow over time. And it will be done in concert with the with the city and in the community that we're serving. And it may not even be all that exciting, right? There's a lot of low hanging fruit. They can let me tell you this stuff is not exciting when it works it's just a safe driver. That's the Sam Abuelsamid 31:06 that's what we're that's what we're shooting for. Yeah, no Rebecca Lindland 31:09 drama applies there as well. Sam Abuelsamid 31:11 I travel a lot and you know, when when I arrived somewhere and people inevitably ask, How was your trip? uneventful, which is just the way I like Exactly. Dan Roth 31:22 I don't know, I kind of like in their pockets and stuff. Rebecca Lindland 31:25 You're so weird. Bryan Salesky 31:27 I'm not. I'm not flying with you. Rebecca Lindland 31:30 So Brian, what are what does it look like in other parts of the world? In terms of development usage just stopped me globally? Bryan Salesky 31:40 Yeah, I mean, I can speak to what I know, which is obviously China's leaning in really hard to, you know, ai in general is is a set as a priority for that country and they want to, they want to sort of gain ground as quickly as possible. And, you know, it looks like they're entertaining ways to, you know, potentially even build cities from scratch, right which I can do with all the infrastructure necessary is to accelerate the Yeah, it's crazy the deployment is it's crazy what they're able to do. You know, you look at Singapore, which is also leaning, weigh in on this, and I think is is, is also taking a, you know, a certainly am measured but as aggressive approach as possible to see this sort of technology get deployed, and they also are in the, you know, they have the right sort of system government system in place to be able to also be an accelerant to the deployment of this sort of, Rebecca Lindland 32:35 yeah, there's and there's this right, culturally, it's appropriate, as well. Exactly. Bryan Salesky 32:38 Yeah. Perfect. Yeah, exactly. And then, you know, she turned, he turned into Europe. And I think Europe's a very complex area. It's, it's, you know, it's an area that I think we're going to see. We're going to see a lot more unfolding in the coming years, but each area sort of has its own story to tell and about who's bullish on it and who isn't. It almost really comes down even potentially to sort of a city by city level. In in Europe. Sure. Unknown Speaker 33:09 That's fascinating. Sam Abuelsamid 33:11 But one of the questions we had from Twitter again was, was Asia. You know, you mentioned Asia, you know, China has the, because of the nature of the government there, they have the ability to do a lot of things from scratch if they choose to to sell. Yeah. But you're not operating in China. And I don't think you have any, do you have any current plans to do that? Or, or it because there are some restrictions on China Bryan Salesky 33:37 apps? Absolutely. Yeah. No, our, our focus right now is is us first. And then and then with VW coming on will be the Europe you know, particular parts of Europe. We have nothing to announce in the way of China or elsewhere at the moment. Okay. Dan Roth 33:53 When you do want to break that news, please let us know. we'll have you back on. Bryan Salesky 33:58 We still be looking The The answer is we, you know, we're always studying the market all the markets, right and just being open with you about where our plans are. Dan Roth 34:07 And and certainly it's going to be taking a system that's sort of designed and optimized for the for the US, you might have some things to learn. If you go to a place like India or China, for sure, Bryan Salesky 34:17 there will be plenty there, there'll be plenty to learn. And in China in particular, you know, the government has come out with, you know, a fairly, fairly rigid but open set of requirements that says, look, we, you know, they want to be heavily involved in the cyber security, they want to own the data. You know, there's a number they want it, they want to, you must you must be working with the Chinese company in order to do the mapping work. There's a number of requirements in China that makes it for makes for it to be a very different system. It's not just plug and play with, you know, what we would be doing in the US. Sam Abuelsamid 34:50 And, and speaking of mapping, you know, one of the questions on Twitter was about, you know, gating factors to move into more cities. You know, you talked about You know, understanding the the realities, you know, the behavior of those cities, but what about, you know, once you choose to go into a new city? Is that an expensive proposition to go into a city? You know, you do HD maps, you know, what can talk a little bit about, you know, once you've chosen silicon opted to go to a new city, what's what's the what's involved in that process? Bryan Salesky 35:23 Right. So when we first go to a city, we start to understand as much as we can about the where the demand is for the different types of businesses that we're that we're looking at. And then we we send a small fleet is sometimes it's not even more than two cars, and we start mapping the city. We start driving the streets, we get information about sort of how the lanes are connected. You know, how wide each of the lanes are, and you know, where the traffic lights are positioned and various aspects of rules of the road that gives us sort of innovation. formation, you get the basics, we take that back, process it and then create a map, that map then gets loaded onto the vehicles and is used to test out the autonomy system. And then again, just with a few, you can do with a pretty small number of vehicles, or sometimes we'll put, we'll put more depending on the size and you know, how we're prioritizing things. And, and then we'll start to test the autonomous system in validate that the map is accurate, and also start to learn pretty quickly, you know, where the system works well, and where it needs some refinement. And then we have a whole team that, you know, piles through the information that comes back from those tests and then starts to create, you know, work for the development team. And that becomes a very much a process that happens day in and day out. And in general, the process can work pretty quick, pretty quickly. We can, we can map a small downtown area in a day and be working, be operational within a week and then sort of expand gradually from there. Sam Abuelsamid 36:57 And I visited your offices last summer or summer 2018. And saw, you know the product the process that you've you guys have built up internally for building maps. And, you know, today you do all your own maps, is that something going forward that you foresee continuing to do is you expand into new areas, doing your own maps or working with outside vendors? Bryan Salesky 37:21 Well, I'll tell you, I would love to buy it if I could. The issue is that the way in which the vehicle knows where it is in the city to a kind of a centimeter level of resolution, so we're talking very, very accurate. It's better than even what basic unaided GPS could give you. The way it does that is it actually uses these maps uses this map information and compares it to what its sensors see in real time and does sort of a matching process, okay, without getting too much detail that's essential to how we do the the, the what we call localization or how does the vehicle know where it is. It turns out that because of that process, Because we're using the sensors on the vehicle, the visual sensors to be able to do that localization process, that the maps are very coupled to the sensing configuration on the cars. And so if someone wanted to sell me an HD map, right, if if the data was presented in a, in a very different format, than what we're expecting in terms of the types of sensors that are in our car, then becomes very challenging to be able to use it, right? I can take any map in in order to get street information addresses, street names, that kind of thing. But the types of maps we're talking about have is not so much that type of data, it's more so information about sort of the broad 3d structure of the city all around me so that I can do enough so that I can apply it to this, this so called localization problem. Sam Abuelsamid 38:48 You know, there's there's another guy that runs a company that's doing some of the stuff that claims he don't actually need those HD maps you foresee a time when that might be true. Bryan Salesky 38:58 Not probably in my lifetime. It's just, it's it's too important of a piece of information that helps the vehicle understand everything about the world around it. And, and, you know, if nothing else, let's say algorithms advanced over the next couple of decades where it becomes less and less reliant on the map, you're still going to want a map to sort of gut check that those algorithms are doing the right thing. It's a great any sort of prior information that can go through a QA process is is is really helpful and part is essential part of the safety case at the end of the day can be used to make sure the system is operating properly. Dan Roth 39:36 Where does it go? And let's look out 10 years is this on you know, Argo, Ai, chips you guys are making and providing to for sale or to Ford, and are they using it in you know, now the technology has been downsized Moore's laws sort of applies right where it just it gets smaller and cheaper with every generation exponentially? What does it look like? Where do you guys think it goes? Just Just to Sam Abuelsamid 40:07 add to that these do you foresee using custom customized silicon? For your systems? That's simple. Bryan Salesky 40:13 I can see us doing that at some point. I think that the algorithms are maturing at such a rapid rate that there's a risk that if you invest in custom silicon too soon, that you've actually created a chip that no longer is relevant to the software that's running on it. And so there's a chicken I bought a whole year like that. Yeah, yeah. It's that's not good point. It's not a not uncommon, even in the consumer world, right? Or, you know, you get the latest video game and it no longer runs, you got to get new hardware, right. So it there's a there's a, there's a process that that we need to follow here, where as when we think that the algorithms have stabilized to a point where it's worthwhile to invest. We'll do it we're not at that point yet. Sam Abuelsamid 40:56 So so right now it's better to stay with a more gentle purpose, compute platform, Bryan Salesky 41:02 then then a customer right now we're preserving for flexibility. And then over time that will start to change. Absolutely, yeah. Rebecca Lindland 41:09 Yeah. Yeah. Are there other industries that you're working with as well like shipping or error? aviation? Bryan Salesky 41:17 Well, you know, our partners are talking to all sorts of different businesses to see what sort of services that that we can provide and to come up with as many use cases as possible for for automated vehicles. So you know, that they're talking to people across all different types of all types of industries. And I think the challenge from what we were talking about earlier is how do you prioritize and sort of slice that up so that you can marry up you know, the, the, the problem or the opportunity with kind of what the technology is able to do now near into the future? Sam Abuelsamid 41:53 What about, you know, one of the challenges when you start looking at, you know, the business side of this is not just the vehicle, you know, transporting goods around the city or transporting people, but also, you know, the last hundred foot problem of how do you get whatever it is from the vehicle to a doorstep or to a lock or whatever it might be. Are you looking also at, you know, building the systems to power, maybe, you know, small, smaller robots that handle that that last hundred foot problem. Bryan Salesky 42:25 So we're not actively working in that area. As you know, there's a lot of startups actually doing the kind of small delivery but in that small delivery bot space, you know, the system we're building is very much focused at the moment on the urban core and dealing with these, you know, kind of crazy city driving scenarios. It can certainly be adapted, but that's not a focus at the current time. Sam Abuelsamid 42:52 One more here from from Twitter. You've got Ford, and now VW is as OEM partners. Are you looking At additional partners or you know, making your platform available as a product to additional companies if they maybe if they aren't necessarily Equity Partners. Bryan Salesky 43:10 It's something that we're open to in the future we're not doing it just yet in terms of making it open to, you know, non Equity Partners. It's, it's, it's, we have the option to do that, if the if it's the right business case, but you know, we're after already a huge opportunity here and just getting the Ford and VW fleets sort of powered with our technology and that's very much our focus you know, again on another OEM or another investor you know, the there's there's still fundraising to go right. This is as I've always said, this is a time people and capital intensive endeavor. So we'll see where those roads take us. But you know, nothing to announce at this time. Something that something that we're, you know, always looking for someone, someone rings our phone, I'll answer it. How's that? Rebecca Lindland 43:58 Brian, since I always send us So look at the consumer side of this. I'm interested in understanding how to I mean, right now people don't even like electric vehicles because they don't understand them. And you mentioned the educational part of it. And but, you know, how, how do we work on getting people to trust these vehicles and to understand them and to see the benefits of them? Bryan Salesky 44:23 Right, yeah, this is a really important question around trust. And I'm not going to pretend that I have the answer. I don't think anyone does. I will say that our approach is absolutely to be as transparent as we can and to serve an educational role as much as possible so that so that we can help make what we're doing less kind of science fiction and more tangible and real. I find that when we give people rides in the car, and we and we show them the care that we take in developing it and all the tests that we've gone through many people their comment is, wow, this is way better than myself. You're 17 year old, right? And, and, Dan Roth 45:05 you know, I think to me, we can we can, Bryan Salesky 45:08 we can climb that, that mountain, a fairway with just those two simple ingredients of transparency and education. And those are things that we're, we're looking to do to do our part. You know, we have a, some of the cities were particularly interested in just, well, how do you test these vehicles while the technology is not ready. And so we built out a portion on our website called how we test which I think is probably the most exhaustive description that any companies put out there in terms of, you know, just how we how much care we take in training our test drivers and, and and how we, how we are developing the technology. And we're going to continue to add on to that and also through other forums to try to, you know, bring people along for the journey who are interested in knowing more. Rebecca Lindland 45:53 Yeah, a big Sam Abuelsamid 45:54 part of that, I think is the the no parking podcast that you're doing. You're going to be launching soon. Which is, you know, part of that educational effort. I'm really looking forward to listening to that. And we actually just before the show, you and I and Alex Roy, I recorded an episode of that that'll be coming up in the coming weeks. Yep. Bryan Salesky 46:14 Alex are always entertaining hosts is sort of leading the charge, we've got a good mix of, you know, folks coming out of the cities that we're testing in that are, you know, interested in what we're doing, as well as folks who are kind of more technical. And so it's a great mix of dialogue around all things. automated vehicles, so should be pretty cool. Rebecca Lindland 46:33 Well, I think one of the things that I really love about this technology, even though I think we can all agree we all have to drive is just the accessibility that it gives to people that can't drive that are disabled in some way, the blind, you know, people that I now have to depend on others. I just love the idea of them having that type of independence, especially aging populations and such. It's just it's really, really Really exciting. I'm so happy you've been able to join us Bryan Salesky 47:04 what we do as well, that's it's a huge, hugely important part of, you know, the constituency that I think is obviously has an interest in seeing this technology get deployed, and we want to work with those with those folks as well. Sam Abuelsamid 47:19 All right. Well, Brian, I really want to thank you again for for joining us today. This as always, great conversation. It's always great to talk to you about about what's going on in the space. And thanks and Rebecca, Dan, any last thoughts? Dan Roth 47:35 I you know, I just wonder if you weren't doing if you're doing autonomous vehicles what would you be doing? Bryan Salesky 47:46 Oh, wow. You know, I love woodworking if I was independently wealthy. I probably would still be doing this actually. Unless I love the self driving car things. I Sam Abuelsamid 47:56 want to be Ron Swanson. But yeah, Bryan Salesky 47:57 since to some degree Yeah. Who's guy from Parks and Recreation Francois Is that right? Yeah, exactly. Yeah, exactly. I would that that would be great. He seems to kind of work when he wants to otherwise he's in his barn like building something that's that's me Dan Roth 48:15 being able to afford anything. That's right. Unknown Speaker 48:18 That's right. Well, thanks for having me. All right, Brian, Sam Abuelsamid 48:22 wrap it up, and we'll talk to you all next time. Transcribed by https://otter.ai