Latest / Autonomy Insiders / Germany's Only Full-Stack Level 4 AV Startup: Inside MOTOR Ai

Transcript

Germany's Only Full-Stack Level 4 AV Startup: Inside MOTOR Ai · Autonomy Insiders

• Daniel Abreu Marques: Welcome to Autonomy Insiders. I'm your host Daniel and today I have a guest who's building something that almost nobody outside the German deep tech scene is talking about yet. Roy Uhlmann is the CEO and co-founder of Motor.ai, a Berlin based startup that claims to be the only German company developing a full level four autonomous driving software stack from scratch. So not an ADAS feature, not a level two highway assist. It's a full autonomous driver built in Berlin. Today with Roy Uhlmann Roy Uhlmann: Hmm. Daniel Abreu Marques: Welcome to Autonomy Insiders, the show where global industry leaders in autonomous driving unpack their real-world insights. funded with the recent 20 million seed round and heading for EU type approvals. So Roy, welcome to the show. Happy to have you on. Roy Uhlmann: Thanks a lot for inviting me. Daniel Abreu Marques: Sure. So Roy, our listeners who haven't heard of MOTOR Ai yet, are you, what are you building and what's vision? Roy Uhlmann: Yeah, so are MOTOR Ai ⁓ 100 people, actually founded the company 2017, so already like nine years old, ⁓ for quite a while and then attracted investors. You know, it's pretty difficult to attract investors in Germany. You have revenues, you have a final product, the risk is gone, but autonomous driving is a bit of the opposite. So it took a while to really attract investors, but then we actually hit the ground, no, left the ground, sorry, not hit the ground, left the ground. And yeah, we built a full stack autonomous driving system, meaning open source or something like that. Everything is developed in-house, ⁓ but just the software, we also developed some key components, ⁓ which we are crucial for the ⁓ value chain you want to really launch autonomous vehicles. which is the HD maps. So we built an engine which creates HD maps out of satellite data. So we can scale pretty fast when it comes to new driving or service areas. Second, also a drive-by-wire. So meaning we're not depending on specific vehicles already have a steer or drive-by-wire. We can actually retrofit them. And we also do the technical surveillance, which is being ⁓ necessary law that you actually monitor the tag of a vehicle, so like the AD ⁓ and built that as well to actually monitor ⁓ and a lot of over Germany and Europe. Daniel Abreu Marques: Sounds good and maybe let's make it concrete. So if I'm a transit authority in German district, for example, and I call motor. So what exactly am I buying from you? So maybe walk us through the product and service. Roy Uhlmann: Mm-hmm. So well, so what we do is like, first all, we think the market is splitting up. ⁓ There be a market for autonomous driving systems being deployed in OEMs, like vehicles of OEMs. ⁓ So like private passenger vehicle ⁓ the maybe like in the near future. However, ⁓ there totally makes to use systems like end-to-end systems, et cetera, which have a crucial problem. You can't really like ⁓ mitigate them. You can't really like ensure the safety argumentation about that since these are black box systems. So totally makes sense that you still have some sort of like a fallback solution inside the vehicle, which is the driver. And therefore it totally makes sense to deploy these sort of technologies within passenger vehicles as a level two plus plus. Then on the other side, there are systems which you can really like mitigate and you can like follow a safety argumentation on that. Meaning when you think about the European Union and its strict safety argumentation and safety standards actually, and actually it's one of the one areas which has a safety standard. The US it's actually a bit lower. And thinking about that, we think the market is splitting up ⁓ into for OEMs like as a feature, autonomous driving as a feature. and the other one is autonomous driving as a mobility service. And that's what we offer. We offer mobility service a bit like the same like Waymo does in the US, meaning we're putting a fleet of autonomous vehicles onto public streets and we fleet them into ride-hailing and on-demand services. So it's not only public transport, for example, it's also the private operated fleets. Daniel Abreu Marques: Okay, so basically you have then customer segments more or less public transport and also Correct? Roy Uhlmann: At the end, it's a fleet owner or actually it's like one has the final contact to the end customer. And for us, it doesn't make a difference. Yeah, because the autonomous driving system, what we think is crucial that you actually control in the early beginning, the service area, because autonomous driving obviously is not that you just put a car on the street and you don't see it for the next 30,000 kilometers. You actually need to have control like, what is it doing, where is it doing what. So every millisecond, you're like technical supervising it. And basically that's our differentiation between different sort of models, to be honest. that's why we choose ride-hailing on demand since their actually kind of like the same, just only a different ⁓ customer and not being the one who's providing a system for OEMs. Daniel Abreu Marques: in ⁓ I call it now, autonomy as a service model. ⁓ owns what? So does the customer own the vehicle or you? Roy Uhlmann: Mm-hmm. To be fair, we started with a model ⁓ the customer owns the vehicle ⁓ now we realize that it totally makes sense ⁓ maybe we own it or to be frank Waymo doesn't in ⁓ the future won't own all vehicles anymore. Someone else will do it like some sort of like financing vehicle or whatever. And I think this is a proper and like really doable framework we can apply here in the European Union and Europe as well and we will go for that as well. So meaning we will possess the vehicle but we might be not ⁓ owner of the vehicle ⁓ the future then. Daniel Abreu Marques: OK, makes sense. And maybe let's get back to the topic that you already talked about a bit. So let's talk about the architecture. So the AV industry is more or less in a fundamental debate right now. So you have modular stacks with separate perception, prediction planning models versus the end-to-end neural networks that you mentioned. So basically, you need to go from sensor inputs straight to the driving commands. So where does MOTOR Ai sit in this Roy Uhlmann: Mm-hmm. Mm-hmm. Daniel Abreu Marques: spectrum and also why. ⁓ Roy Uhlmann: Yeah, mean, maybe I'm going to explain a bit about the details and the technical details. So build a modular system. ⁓ So it's an end-to-end system. ⁓ Some people think this outdated actually when it comes to safety, it's not ⁓ because we actually prove safety layer by layer. ⁓ And safety not sort of like... that you just have some data about like driven kilometers or disengagement or something like that. Safety is actually the mitigation process about every risk and threats your system can have. And we have to ensure that we have like one failure per 100 million operating hours. And you obviously can't prove that by just driving around. So therefore it's really crucial to build a system in different sort of layers. And then you can prove layer by layer that this actually, this failure rate. ⁓ exists and that you hit it actually, like, mean, you save it and it's fairly all right. And when it comes to the decision-making process in the vehicle, meaning like the driving maneuver, they are like sort of different sort of strategies or technical approaches. And one of them is a classical generative ⁓ AI approach. You're data, you're creating new sort of... situations and scenarios out of these data via simulation this is not what we do. So what we do is actually in the we encounter a specific traffic scenario, we create a out of that, based on some sort like a world model ⁓ and the ⁓ let's determinism that world model. Meaning this helps us to really solve long tail at the end. And it's some sort of a system which is capable to understand when its own knowledge about a situation is not enough, meaning a car slowly approaching, for example, an intersection or something to gather more data to make a decision that it now can drive or it cannot drive. there's a huge advantage when it comes to the driving itself, but the much bigger advantage is actually in the costs ⁓ because don't need actually these huge computing powers of like what we know of generative AI models. ⁓ some sort of autonomous chat GPT anymore. It's much more than that. ⁓ ⁓ also means we can leave this world, which ⁓ we know right now in autonomous driving and the obviously known chip providers, et cetera, ⁓ and a different sort of approach in encountering that. And ⁓ at end, this means we can drive autonomously with much, much less money to spend. on data, on simulations, and so on. We still need data. We still need to simulate things, this is obvious but not in the amount of what you before from autonomous driving. And this actually helped us to build such a system. So this is the big difference, to be honest. Daniel Abreu Marques: walk through your sensor stack. So ⁓ what's on the and why did you choose this configuration? Roy Uhlmann: Mm-hmm. Mm-hmm. So I mean, we follow a classical ⁓ sensor stack, radar, lidar, camera. And this is specifically because that's what you need when you want to drive in the European Union. Sensors have to be redundant. So because you have to redundantly identify objects, for example, ⁓ and a in the dark doesn't see anything, that's given. And you need to ensure that at least two sensors can an object for example and so therefore you need at least two and since like one of them could always fail you have three and this is like just a simple explanation of it. Overall it's sensor stack to be frank everyone else in the industry has when it comes real autonomous driving meaning like taking out the driver and not having like some supervised system or so on and this is radar, camera, a classical GPU, IMU, etc. Daniel Abreu Marques: and you also mentioned that you're not dependent on specific vehicles. So on to what platform are you integrating your stack currently? Roy Uhlmann: Well, mean, there are some, let's say, limitations right now. The limitation is rather like a legal one. ⁓ you're like a truck or let's say a 12 meter bus and you want to turn right, you have to really like drive into the opposing, ⁓ traffic. basically this creates a problem your autonomous vehicle is not supposed to ⁓ violate the traffic regulations. So therefore, we are going to vehicles up to let's say about six meters. This makes ⁓ And it's a at this point right now, which are testing is a classical V-Class. a van and you can see it on Berlin Street, driving around autonomously and doing its tests. ⁓ But there be another platform then coming soon as well, which is then rather for the scaling of a fleet. Daniel Abreu Marques: So what does it cost to equip a single vehicle with the full motor AI package more or less, or sensors, compute, integration, everything? Roy Uhlmann: Was there anyone in this podcast ever answering that question? don't know. I think I can't answer that. Yeah, that's totally true. answer is really, really interesting also for me. ⁓ now to be frank, I mean, we're still in the process ⁓ like create ⁓ scaling economics, let's say it this way. ⁓ every number I would say right now wouldn't be like really reflect the reality. ⁓ Daniel Abreu Marques: No, but I still try to ask it. Someone needs to be the first one. Roy Uhlmann: or like the near future of like what does it really cost to have like, let's say for example, 1,000 vehicles on the streets. I mean, I can tell you like the first test vehicle was a couple of hundred thousands, but we're ways ways far away from that for the ones who are then finally coming on public streets in the scaling mechanism. Daniel Abreu Marques: So we on the product ⁓ and the tech. I think also super interesting to talk about is what makes building a level four vehicle in Europe fundamentally different from doing it in the US or China. You already teased it a bit. So Germany passed the AFGBV in 2022, the first legal framework for four vehicles on public roads. And yet Roy Uhlmann: Mm-hmm. Mm-hmm. Daniel Abreu Marques: Four years later, are still zero type approved level four vehicles operating in Germany. So in your opinion, why is it taking so long? Roy Uhlmann: Yeah. There are a couple of reasons for that, I think. mean, to be frank, Germany and even European Union isn't that slow. I mean, the situation is the following. In the US, ⁓ you don't let's say, the same sort of regulation for ⁓ autonomous vehicles we have here. Actually, in a couple of states, there doesn't even exist one. ⁓ And ⁓ let's say this way, has a rather more opportunistic approach. And at the end, meaning a legislation is what I said at beginning, is some sort of like a safety framework. ⁓ And have a process here in the European Union which describes that. ⁓ So meaning you a testing permission ⁓ driving doing ⁓ the of like every sort of like scenario and you need prove and validate for the regulator. which is the, in Germany, the KBA. It's some sort of like the National Traffic Authority, something like that. And then you have like the technical assessment companies ⁓ like TÜV, like DEKRA like others. ⁓ And what they do is they're of like the technical arm of KBA and trying to ensure, in first level, find out what you did with your system. And this means... They are actually joining you while you're actually producing or developing such system. It's not the way that you just show up at the end and it's like, look at this, it's an autonomous car, look, I can drive around the corner, it seems pretty autonomous. Give me just like, sign this piece of paper. It's not that it does work this way. Germany, think, and especially also Europe, has a long history in the automotive industry of making things safe. And making things safe means like we have to measure and to benchmark and to really validate everything we're doing here. And this is like also what accounts for autonomous driving. Meaning the process to bring an autonomous vehicle onto public streets is really like showing all these different sort of layers of the system, showing that you can actually detect an object, that you can detect an object on certain weather conditions. And when you are claiming for having an autonomous vehicle driving in certain operational design domains like ODD, meaning I'm having a vehicle driving in winter, in snow, in rain, heavy storms, or driving 100 kilometers per hour, et cetera, et cetera, then you have to prove that you can first of all detect objects and that you are able to steer this way. ⁓ at the end, that you have redundancies ⁓ your system ⁓ really like... show that and to ensure that there's always like a fallback solution even if one system or like one part of the system fails and This first of all is a huge difference to the US So when you take a look at vehicles, for example in the US they don't even have like redundant drive-by-wire systems within the vehicle They might have something which steers it but it's not redundant Yeah, so obviously the vehicles you see in the US right now driving autonomously will be not the ones coming to the European Union the vehicles you see in the US right now driving autonomously ⁓ will be not the ones coming to the European Union Maybe people talk about like, you know, in London, there's something different. I saw the same sort of vehicles they have in the US. Yeah, but London is UK. UK is not European Union. It's like totally different safety standards. They don't follow this sort of standard. But this is like what we have here. We have these standards. We have to follow them. And you follow the process. And one part of the process is that you then have like some sort of like a testing permission driving around to test vehicles. ⁓ with this testing permission, you normally can't actually commercialize vehicle, meaning ⁓ You have to test, that to the KBA, show that to the TÜV, to the DEKRA whoever it is. But can't bring people into the vehicles. Croatia went a different way right now. Now they opened it up and said, now you can drive around with a test vehicle and commercialize it. it's a test vehicle with which you can actually commercialize. We don't have that in Germany. But you have the same sort of thing in the US. These are not type-approved vehicles. Worldwide, there is no type-approved vehicle right now for autonomous driving. But this is what we're doing here. And this is the step and the final achievement we're working for is doing the type approval. And once you have the type approval, you can commercialize. In between, it's all testing. And this is like, I read a lot of things about different bloggers and whatever, and they always claim, yeah, Europe is so slow, we're always so bad, whatever. No, it's not true. It's actually not true. What we do is we have a proper testing process. We follow that. We obviously can't commercialize. When we don't commercialize, we don't put 1,000 vehicles on the streets because it doesn't make sense, you know, because you don't make money with it. what we do is like we put vehicles on the street in different sort of like, ⁓ they always claim, yeah, Europe is so slow, we're always so bad, whatever. No, it's not true. actually not true. What we do is we have a proper testing process. deployment statuses, different sort of like software statuses to really test things. And basically we take then these results of the tests and the systems and then we show it to actually the technical ⁓ assessments companies ⁓ at the end then to the authorities or let's say to the regulator ⁓ this is then the final step. But what it means is it means the regulator and the technical assessors. what means is it means the regulator and the technical assessors. are part of the whole development process. They are looking inside your systems and they're also looking inside your code. mean, we a situation in the US where Tesla doesn't even like show or provide data ⁓ about testing kilometers and whatever, yeah. And here we even show the code. That's a totally different understanding of like how to build such a system, but that's how ensure safety. are part of the whole development process. They are looking inside your systems and they're also looking inside your code. mean, we a situation in the US where Tesla doesn't even like show or provide data ⁓ about testing kilometers and whatever, yeah. And here we even show the code. That's a totally different understanding of like how to build such a system, but that's how ensure safety. ⁓ And we're following process. there's still ⁓ way to go. ⁓ the good thing is everyone has to follow the process. It doesn't matter how much money you have, doesn't matter how your PR is, at the end of the day, everyone needs to follow the same safety standards. ⁓ And I think this is really good because, first of all, this creates some sort of like equal ⁓ level playing field everyone following the right technical approach. And second, ⁓ it's pretty for ⁓ at the end the customer. Because we're talking about liabilities, we're talking about potential failures and ⁓ which could happen ⁓ shall not happen. That's basically the problem we have. ⁓ at the end, ⁓ we understood and what we think it is like, ⁓ is much easier for you to really like then operate an autonomous vehicle because you don't have to be afraid of ⁓ the legal claim or ⁓ being by... whatever if something happens because you can really like follow your process and ⁓ having a safe system and provide that to the public. And I think therefore ⁓ crucial to understand ⁓ there are different markets. European Union is a totally different market than the US ⁓ a lot of companies ⁓ even like start and try to build systems within the European Union. ⁓ did, we always... build it in the way that it really meets the EU criteria therefore we claim that we are Right now I think it's really the only German full-stack autonomous driving company level 4 here the market. Daniel Abreu Marques: It's super interesting. you're saying this is not necessarily a competitive disadvantage for Europe to have these regulations like type approval But it's more like an advantage because you then afterwards have the possibility insure the vehicle and so on. ⁓ Roy Uhlmann: Absolutely, it's actually, this is some sort of a moat It really, at the end, helps you as a company. And I mean, we're doing that now for a couple of years. Some would say like, oh, why does it take so long? No, it takes so long. That's the point. And these years, which it took so long, are actually our competitive advantage also in time, because everyone else has to follow these steps. There is no shortcut. Daniel Abreu Marques: And I also read that your CTO, Adam Bahlke ⁓ said on the DIN Commission that helped shape the German autonomous driving standards. So how much of an advantage did that give you in understanding what the final regulatory requirements would look like? Roy Uhlmann: Yeah. ⁓ Yeah, mean, the way you say it, it sounds like we shaped it, but it's not true, to be honest. This was like early beginning 2021, 2020, 2021. And for us, it was really important because we understood how the system looks like, the safety system and how this legal framework will look like. And it was some sort of advantage for us because it gave us not only a hint, rather like really a direction in which we have to develop. but it's not the way that we actually actively shaped it in the way that it wasn't our favor or something. in this commission, were like sitting ⁓ a of other companies, the big OEMs as well, but it was pretty cool. for us a small company, ⁓ also some of like a recognition to sit at table where the big OEMs sit as well. Daniel Abreu Marques: Sure. I want to lead also ⁓ conversation to a bigger question that goes beyond just MOTOR Ai as a company and touches also something political, technological ⁓ MOTOR calls itself, as you said, the only German company developing level for autonomous driving. ⁓ does it actually mean to be the only German company doing this? Roy Uhlmann: Aha. Mm-hmm. Mm-hmm. Yeah, mean, ⁓ hurdle for others, but it can also be a burden. ⁓ I mean, ⁓ is no real meaning for us behind that. When I'm going back in the years and when we decided to build such a system, we learned something. And this is you can't build such a system with the automotive industry, which doesn't mean that you build it against it. ⁓ It just you're building it without it. And ⁓ during all these years, we always faced a lot of problems and we had to solve them on our own. There was no specific hardware just somewhere in the shelf and we bought it and dissolved it. No, we had to build a lot of things because they didn't exist. And ⁓ the only one at the end means ⁓ had to build it. There was no market for a lot of solutions which we just built. ⁓ the end right now, it ⁓ ⁓ like it means or like led to the point that we are. ⁓ full stack that we build everything in-house, that we don't have like open source software things from someone and this would then, by the way, even create problems afterwards when you commercialize it. And it also means that we occupied, I can say it like this, some crucial points in the value chain, which we need to bring a vehicle on the street, just as I said in the beginning. ⁓ there were no HD maps. mean, there were HD maps out there. know TomTom, et cetera. ⁓ But at time we needed them and we wanted to test things. The quality was a bit different ⁓ when comes to highways compared to like the last village somewhere in Mecklenburg-Vorpommern. So we to come up with solutions for that. ⁓ Same things some hardware components, et cetera, So basically ⁓ it was pretty tough, but it to the point that had to solve it. And this is also some sort of like a culture we have in the company, like we have to solve it. is no one there just like offering you something, solve it. And basically it's some sort of resilience ⁓ have to establish then in the company, also like in your own sort of like mindset. And resilience led to the that we, and I just have to say it this way, ⁓ that we still because I mean, there were a couple of other companies, but they don't exist anymore. And we were always short in money, to be frank. We never had the billions. But it led to the point that we had to build a system with much less money. We are absolutely capital efficient. And right now, what we think, this leads to the point that we can really bring an autonomous vehicle on public streets with the type of approval based on a financial model no one else can. Because we are just so capital efficient and learn to be capital efficient. in the last nine years and like possess or like own the neural points in the value chain like no other company out there. And I think this makes us a bit unique. Daniel Abreu Marques: Yeah, so I mentioned earlier you raised 20 million in seed funding, I guess. then if you, yeah, plus a bit more. But for context, you see Waymo burned through billions, ⁓ Wayve just raised over a So I think some would then question is 20 million plus something more enough to reach that type of approval and commercial operations or is a series A already being planned? Roy Uhlmann: Plus a bit more. obviously planning the next financing route. It's always the case when you're having a startup. ⁓ I think when you have a lot of money, you spend a lot of money. There's obviously certain amount of money with which you can't really build such a company. ⁓ We're able build it. think Jan Thomas in another podcast ⁓ The time will come where we'll have also like much, bigger funding rounds when you want to fund a couple of other things then. However, we're driving on public streets. We're driving autonomously here in Berlin on public streets. You can see our cars. So I know what we reached. And obviously there are some other steps you will have in front of us. And this is obviously the type approval. And since no one has the type approval worldwide, you see that the type approval is not the question of money. It's a totally different question. Daniel Abreu Marques: Sure, makes sense. And maybe just double clicking on the technological sovereignty theme here. So let's say if Waymo or a Chinese technology provider can then deliver a safer or maybe cheaper level 4 system tomorrow, so why should a German transit authority still choose Motor AI over them? Or should they still choose Motor AI? Roy Uhlmann: Mm-hmm. you mean i'd like to have a lot of sales perspective like out of the sales pitch Daniel Abreu Marques: like out of a sovereignty perspective, do you think it would be necessary that ⁓ Germany still ⁓ a German virtual drive instead of a US or Chinese one? Roy Uhlmann: Mm-hmm. Mm-hmm. Yeah, I mean, to be honest, what is sovereignty Sovereignty means being independent. I mean, we learned that the hardware in Germany, the hard way, sorry, the hard way in Germany in the last couple of months and I would say in like one and a half or two years. And independent means, well, we need to have our own technologies. We need to have our own offers. We are one of them, or like actually we are the one in autonomous driving. ⁓ is maybe not like really the USP an let's say, transit provider or like ride-hailing on demand provider. But at the of the day, we're sitting here. We built for this market here. I'm not the general manager who's being flown back to his mother country if something didn't work. ⁓ I go to jail. you can really grab me. That's one of the big things I think which are important when you want to really bring something like critical infrastructure and autonomous driving is critical infrastructure onto public streets. So ⁓ this is not only about data and I think a lot of companies just claim, yeah, then we have the data and you hope it's on. No, it's not what it's about. ⁓ built the system the way that a European ⁓ system should be built and the European way of understanding such a system. ⁓ And ⁓ to be frank, it to the market right now, we see a lot I say, dynamics. ⁓ So there's not only the AD providers, ⁓ You have like the right hailing companies, et cetera, et cetera. And everyone now claims to be like in the autonomous driving market and being the European autonomous driving company wherever you want to like encounter that. the end, it's about some crucial parts. It's like, what do you actually own and like developed in ⁓ the chain to bring an autonomous vehicle on the street? ⁓ There a lot of pieces in there. And I think ⁓ When it comes to the price, it's really important that you do know and do understand and in most ways do occupy or like possess these parts of the value chain. Then you can really like ensure a pretty good pricing to your customers because we are not the ones who have to earn money on HD maps. We're not the ones who have to earn money on the drive-by-wire system. We're not the ones who have to earn money on the technical surveillance because we are the AD company. We are having like everything in one place, everything in one product. Now, and this is, think a big, big difference. And second, it's ⁓ when you think about, let's say 10 years in the future, and we have an understanding that a lot of vehicles will be driving autonomously. It means that there is some sort of like autonomous infrastructure and someone will manage that. And we have to ask ourselves who shall that be, and what kind of values shall that happen. And basically this is something where we think as European ⁓ company being of a really European country as well, are one ⁓ the first you should pick. We have the goal to bring an autonomous fleet onto public streets. We have a pretty good goal and ⁓ of how want to do that, what we want to achieve on that. And ⁓ I think ⁓ one of the business is like, first of all, developing the AD system. ⁓ Second is then the operations of the system. ⁓ We a lot of layer in the operations and a lot of layer you have to provide ⁓ also on a technical and software side to operate something like that. And we're building on a lot of things. lot of people don't even know that, but I can't say everything. But it comes to an operations of an autonomous vehicle, ⁓ it's, I it's an infrastructure. It's an infrastructure layer laying there. ⁓ And we provide that ⁓ and lead the point to really launch AVs, purpose-driven. ⁓ for let's say on demand and ride- hailing up. There are a lot of plans out there. Daniel Abreu Marques: All right. So we've talked a lot about technology and strategy now, but as you already touched on a bit, MOTOR Ai isn't just planning anymore. You're actually on the road. Maybe take us through what's happening on the ground right now. So walk us through your current deployments. Where are your vehicles driving today? How many are out there? And is there still a safety driver in the vehicle? assume yes. Roy Uhlmann: obviously there's a safety drive in the vehicle because we have a testing permission for that. testing permission, ⁓ pretty simple to get ⁓ understanding of the autonomous driving market in Germany ⁓ at the website of the KBA, you can actually just get a list of like who has a testing permission. ⁓ whenever you take a look, you can just like count one, two, three, four, five and see like how many vehicles out there. So right now we have for a testing permission for our generation one ⁓ We're ⁓ the generation two vehicle the next couple of weeks. We see how fast this is. ⁓ This not a question of the team rather at this point a question of the authority, but they're doing good. doing good. And then like then the final sample which we call the C-sample which will be the generation three vehicle which we plan this year as well already and working on it. So ⁓ to give bit an answer on your question actually. It's not really about how many vehicles you have, rather like what you can test with a vehicle. And it's not that you're trying to drive as many kilometers as possible. We rather have the so-called autonomous vehicle capabilities. we divided that in certain capabilities, which the has to encounter or does encounter on public streets. ⁓ And give you, for example, thing, or to explain one thing right here around our office. There is a roundabout, this is nice, but above the roundabout is a huge bridge trains. It's completely made of steel. ⁓ you don't have any GPS when you're driving with a car into the roundabout. There is no GPS anymore. ⁓ you really have to create solutions which help you even to localize your vehicle then without GPS. And this is only one part. Second is we have cobblestones here in front of the building. Looks so nice, looks so nice, but there are no lane marks, just a couple stones. So a car might think, I have like the whole street for myself. No, it's not. And you have to encounter all these things and they have to be part of your development and part of your driving capabilities that the car actually can localize itself right. And everything kind of like is redundant at this point. And we're testing these things layer by layer by layer, meaning it's not about like you're just driving around. So like, you're worked. It's rather like, oh, now we're just testing this sort of, let's say, localization layer, or this sort of detection layer, or whatever. And this is what we're doing here. And I what also did is we ⁓ thought it's always to ⁓ test on the customer side. So therefore, also have contracts with public transportation systems on which we're working strongly towards test ⁓ in their areas then as well. This is going to happen the next couple of weeks and then. ⁓ You obviously always need vehicles for that. We're also waiting for the testing permissions. Meaning at the end we are at urban ⁓ suburban areas. Not really highway. That's a totally different thing. ⁓ So that's what we're working on. And the goal is really like ⁓ being in the cities, being in the urban areas, ⁓ plus being in the suburban areas of like and so on. ⁓ A village is a village. Berlin at the mostly is a village as well. But connecting these is also important for us. Daniel Abreu Marques: And when do real passengers get So is there a roadmap ⁓ that you look like from Roy Uhlmann: Yeah, we have. But I wouldn't shoot it out here right now. But yeah, there is a roadmap for that. Daniel Abreu Marques: All right. Then maybe let's ⁓ zoom out a bit Motor.ai exists in a global AV market that's moving fast. And you already mentioned Waymo several times. So they did over 40 million rides last year and are heading towards a million rides per week. How you think about the gap between where Waymo is and where Motor.ai is today? Roy Uhlmann: Mm. I don't even think about that to be honest. First of all, ⁓ the US market. think they ⁓ a lot of things to do on the US market. Yes, is that they are ⁓ aiming for whole world. Well, this will years. I not only think, do We see what's going on with right here on the German market, which is nothing. and basically also know what the process is to actually launch in Germany or even launch in the European Union ⁓ and this takes time. It's ⁓ absolutely okay. there are a couple of companies out there. mean you name Waymo but there's also Pony and others and they're also aiming for these markets so ⁓ that's why don't really think about the gap between Waymo and their ⁓ of rides rather like i see it complete landscape of autonomous driving companies as you can't landscape off companies who want to apply it rather think about that to be honest since i think will be a bit of a question like who really makes money with it who really makes money in autonomous driving ⁓ and for example when take a look on on ⁓ ride-hailing companies like Uber and so on i mean they're pretty dynamic on now how they are encountering this market. ⁓ a lot of fear actually right now. It's like not being too late. ⁓ is really ⁓ and I think one of the more important parts ⁓ when I take this I would say we're not too late. We're right in the middle. We're doing our job. We have to do it safely. Don't rush. ⁓ Rather do it as you have to do it. And ⁓ we see what's going on with all the others. As I said a bit in the beginning, think what's really, really important is you need to get your operations right. The value chain, not only of the AD system, also of the operations afterwards, is the important part. this takes time to build up. for example, taking a look at the ride-hailing companies, a lot of them never had operations. So a completely new market for them. I'm really curious to see how they will act within this However, other companies, ⁓ the car sharing space, also AD companies, they have much more experience right now with operating vehicles on the ground. mean, car sharing is obviously much better than AD companies since they have like less cars. But at the end, the operations of autonomous vehicles is different. And think these are important things in the future we have to take a look at. And will also differentiate the market a Daniel Abreu Marques: And when you now five years ahead, where do you see Motor AI standing? Roy Uhlmann: Yeah, it's a difficult question. like it's always like, I mean, you're asking that because I can just shoot out something really, really huge. Yeah. We're doing our job. We're doing our job pretty good. I think that's why we are doing it for about nine years. There will be a lot of other years in which we will really grow and put our system on the road. So in five years, we have a type approved vehicle on public streets. I think in five years we see of my perspective ⁓ totally different market and autonomous driving than now. I think there will be a huge shift in ⁓ the, let's businesses we see right now, ⁓ how vehicles being used, maybe also like how public transport is being used. And ⁓ I'm curious see how... many AD companies will then still exist to be frank. ⁓ I said, value chain is everything and ⁓ some fast successes ⁓ autonomous driving a lot of ways, ⁓ sometimes paid just without ⁓ having a let's standing in the value chain and you will pay for that ⁓ in the future ⁓ Maybe it will you from the market. And that's why we rather work slowly and having a strong footprint on all these different things in the market to have a standing then for the future. Daniel Abreu Marques: ⁓ Roy, ⁓ thank so much for being the today. was a really interesting episode and was a deep dive into the part of autonomous driving world doesn't get nearly enough attention. It was very insightful, as I said, and I'm really looking forward to have you back on the show, maybe with then the news from Motor.ai that you currently aren't able to talk about. ⁓ But yeah, thanks Thank you so much for joining me today. Roy Uhlmann: I thank you.