Urban Innovate TALKS

Henriette Cornet (00:05)
Welcome everyone to Urban Innovate Talks, our webinar series where we explore the topics we care about most at Urban Innovate, transportation, technology, sustainability, and of course, people. Urban Innovate is a strategic consultancy based in San Francisco, helping cities, transit agencies, and mobility providers design inclusive, future-ready mobility systems. You can learn more about our work.

on our website and follow us of course on LinkedIn and YouTube. Today, I'm excited to welcome Marek Van Zura, an expert in autonomous driving technology. I had the pleasure of working with Marek on a large European project that I was coordinating during my time at UITP. And this project was about deploying autonomous vehicles in like 15 countries all over Europe. And Marek led the deployment in the Czech Republic.

which by the way was the first to go live in the country. So big kudos for that Marek. And since then Marek has continued to explore the world of AVs, testing various technologies and services around the globe and sharing his insights through some great LinkedIn posts. So you can also follow Marek on LinkedIn to have more of these insights coming. My goal is really to get a snapshot where we stand today in terms of autonomous vehicles technology and driving behaviors.

We've really focused on the vehicle and we can touch upon the various brands that are out there, but really kind of get a better understanding of what is an EV and what is happening on the market now. And for those of you like me who are super interested in the services that EVs can offer, stay tuned really for the services because there will be other episodes coming on that matter. And for that, don't forget to follow the channel.

Enough of the introduction, let's get into it. Welcome Marek and thank you so much for joining us.

Marek Vanzura (02:07)
Well, Henriette, it's a pleasure to be here and thank you so much for inviting me.

Henriette Cornet (02:10)
Yeah, would you like to maybe introduce yourself for those who don't know you yet, what you are working on this day and what's a bit your background?

Marek Vanzura (02:19)
Okay, yeah, I can try to do that. Yeah, my name is Marek Banžera. I'm originally from the Czech Republic. Yes, Henry already mentioned I worked as the, let's say, the director of autonomous driving research at the Republic. We deployed the first autonomous shuttle in the country. Then I did several similar works in software development company, company that actually works on autonomous driving.

Then I did some research in the United States and now I'm back in the Czech Republic and we actually make components for autonomous cars. So that's where I am right now.

Henriette Cornet (02:55)
Okay, okay. So do you want to describe a bit more your company? I would like to when you say components for autonomous car, what do you mean with that?

Marek Vanzura (03:03)
Yeah, I can look more into it. Yeah, the company is actually called University Spinoff. It started as a research project and it turned out that the product itself is such a great thing that there are actually customers for it. they decided to turn it into a spinoff company. We are called Otino, which is short for Automotive Innovation. And the product that we currently provide or offer is so-called the drive-by-wire interface.

which is, let's say, a communication gateway between the systems in the car and external systems, such as autonomous driving, stack, and sensors, and all these things. So basically, by using our system, you can communicate with a car and control it however you want. we're basically allowing you to convert conventional cars into autonomous cars.

Henriette Cornet (03:51)
Wow, okay, interesting. I would love to dig into that more later when we go into the various way a vehicle can be automated, so to say. Is it like purpose-built or like this retrofitted aspect? But maybe before that, how would you describe an autonomous vehicle to a four-year-old?

Marek Vanzura (04:10)
that's challenging question. Yeah, let me do it this way. Basically, everybody knows a regular car, a regular vehicle, which means there is this metal box and a person is sitting inside. And this person is responsible for all the tasks, like looking around the vehicle, checking the surroundings, making sure that they follow the traffic rules and all this stuff. they're like they're pressing the pedal.

they're moving the steering wheel and doing all these sort of things to make sure that they navigate through the world in the vehicle. And the autonomous vehicle is basically the same stuff, but instead of this human being who is sitting inside, there is a computer. And this computer is doing all the same stuff that the human being would do. So it's controlling the steering wheel, pedals, it's checking the surroundings, it's making sure that it won't hit anything, it won't hurt anyone, and it's doing all this stuff by itself.

Henriette Cornet (05:02)
So that would be kind of like what you just described would be the various components that we have in the AV. Can you elaborate on what is made of, know, because here in San Francisco, so Urban Innovators is based in San Francisco, we can see the Waymo on the street where we see all the sensors, the lighters, et cetera. But I'm not sure everybody's familiar with the different sensors that are on a vehicle, on the AVs and also

What's behind is not only vision, right? There is much more behind AVs. Like, would you mind describing a little bit? different parts.

Marek Vanzura (05:38)
Absolutely. I will probably stick to that metaphor with human beings sitting inside a vehicle because it's probably easier for us to imagine what it takes. So when we take the autonomous vehicle, we have several tasks, several things to figure out to make sure that the car can really work perfectly or can work with no issues. And it means like we have to check the surroundings. We have to have some sensors that will make sure that we see everything around the vehicle.

And these sensors might be of several different categories or types. The most common sensors are cameras. We know them from digital cameras. That's basically quite a simple sensor that just detects the light coming from the world and it creates some image from it. So that's the most typical sensor. It's the cheapest one. It's very easy to work with. And there's plenty of these sensors around the car. But it's not the only one because what we want in these cars

is reliability and redundancy. We want to make sure that the vehicle will work under all the different conditions and circumstances. So it means that we need more sensors, we need additional sensors and in this case usually what we use is LiDAR. LiDAR is a so-called laser radar, which is an active sensor that shoots laser beams around it and it detects the time, know, how quickly the...

the beam gets back to the sensor and it calculates the distance and so on. And you can basically recreate a 3D model of the environment around the vehicle. And this sensor has its advantages and disadvantages. But when you use one sensor and the other sensor, you can compensate for shortcomings of each other. So in this case, sometimes, or quite often, we also use different sensors like radar, which radio frequency sensor.

So it again has some shortcomings, but it also benefits. But when you put all these sensors together, you can also use Ultrasonics for close distance measurements. You can use GPS or GNSS global navigation systems for positioning to make sure that the vehicle really knows exactly where it is. You can make it even more precise by using some other devices. And you can also use...

sort of sensor is so-called HD map, high definition map, which is basically a model of the environment where the vehicle navigates and it sort of compares to the map that it has in its memory with the sensor inputs that it gets from all these different sensors like cameras, light arts, radar, et cetera. So this is the inputs that the vehicle gets in a single way as we get inputs from our eyes, ears, and so on.

and it puts it together. It's called sensor fusion. Basically connects all these different sensor modalities, sensor inputs into one coherent piece and it has the representation of the world and it can basically estimate where it is. It can sort of understand the environment to some degree. And once we get this, we have to process it. It's like when we see the environment with our eyes,

our brain processes it in a way that we get some output, we get some decision. We know, okay, we have to press the brake pedal now because we have to stop in front of the other vehicle or traffic light or something like that. So once we get the sensory inputs, there is some processing units, some very powerful computer that processes all the input data and decides, okay, we have to turn left, we have to turn right, we have to slow down, we can speed up and so on. So it's like the equivalent of human brain, it's processing data.

and giving some commands for the vehicle and the vehicle executes these commands and navigates through the environment. So in the end, we have this vehicle that moves ideally, very smoothly in the traffic. It obeys all the traffic laws and it behaves like an experienced driver.

Henriette Cornet (09:40)
Yeah,

that's a great description. And you touched upon some topics I would like us to dig more because of the map, the HD map. For instance, I think there is some difference also in the market in the different brands, how they use these maps or if they use any of these maps. But when you say, way you describe it and to make clear to people here that listen to us is like...

The vehicle is in a way independent, it's not like teleguided or I mean, I know the answer, but I would like also to touch this topic of teleoperation. you know, I've heard different views. Maybe people think it's on a rail. It's kind of a virtual rail that the vehicle is following. And we know in the past, it was the case, right? Or maybe there is someone behind the scene, manipulating the vehicle and teleoperating that. So the most...

Advanced technology that we have now is not the case. Can you elaborate on that?

Marek Vanzura (10:40)
Yeah, I think this is a great point to bring up because there is sometimes a little bit of confusion about these things. And because on one hand, we have autonomous vehicles, and on the other hand, we have teleoperated or remotely controlled vehicles. And in both cases, they can operate without anyone on board. So they might appear as the same category or as the same kind of vehicle. But their functioning is quite fundamentally different because autonomous vehicle, as I described before, is controlled by this, let's say,

onboard computer that does all the compute, all the decisions and ⁓ sensing, cetera. But the remotely controlled or teleoperated vehicle is a vehicle that is similar to these RC cars that we can have or RC planes. It's controlled by somebody who is not sitting in the vehicle, but is sitting somewhere else and is using wireless network for controlling the vehicle, making sure that the vehicle navigates. similarly, there are some usually cameras on this vehicle.

The video picture is transmitted through the wireless network to some remote station. And this remote station is equipped with steering wheel, with pedals, basically recreation of the interior of the car. And this person can control the vehicle and is doing it through the wireless network, LTE, 5G, and etc. And from the outside, the vehicle appears to be the same as autonomous, but actually it's not. So these are two different

ways of operating vehicles, but they significantly different. But there is another thing that's probably worth mentioning and that's when we talk about autonomous cars, autonomous vehicles, there is sometimes this talk about teleoperation also included in it. But in this case, it's usually not the case of remote driving, but we usually speak about remote assistance or remote assistance provision or something like that.

which means it's not direct remote driving, but it's like providing contextual information to the vehicle through commands, let's say, for example, and the vehicle basically receives this information, this command, contextual information, and it processes it inside the compute and decides how to work with it. So basically you tell the vehicle, okay, this is tricky situation. You have to cross the, you know, double yellow line.

and go around the obstacle because otherwise you will be stuck there forever. because the vehicle wouldn't do it by itself, but once the human being, the human reasoning provides information to the vehicle, like this is temporary obstacle and you have to drive around it, the vehicle will do it. But the person is not really driving it. So this is the difference that's important to mention. When we can see the discussion about it in public space, usually these different ways of using telepration get confused. It's like, okay,

remote driving or remote assistant. was that?

Henriette Cornet (13:36)
Yeah, yeah, yeah. And I can see how it's also, it has huge implication in terms of liability. So if suddenly someone take control of the vehicle, you know, elsewhere, and there is an accident, who is responsible, right? So what I hear mostly from autonomous vehicles operators is they prefer the vehicles to stay in charge all the time. And the human who is in the supervision center side gives like,

indication to the vehicle and say, if it's safe, cross the yellow, like in the US, the central line is like the yellow double line. You're not supposed to cross it, but if you're completely blocked, okay, you should go cross it. But the vehicle will still have to decide if it's safe, if nobody's coming, et cetera. And I think this is a huge, huge difference and it will have consequences on this liability.

Marek Vanzura (14:30)
Absolutely, because the remote driving has this one huge disadvantage, which is latency. The speed that the command travels between the vehicle and the control station. So you usually want to avoid too much latency, but sometimes you can't really do that because of distance and stuff. So if you use the remote driving, you're very prone to having challenges with having lower latency.

But if you use remote assistance, then the vehicle does the compute by itself. And it's like everything is real time. Whereas with remote driving, there is always some delay. It's not really real time. ⁓

Henriette Cornet (15:11)
Yeah, yeah, yeah, no, that's a very important point. And also, yeah, to mention that usually the, maybe we can dig in already into the topic of mapping, because I think it plays a huge role in the driving behavior of the vehicle, because it.

do this mapping beforehand and it enabled them already to know where there is a traffic light, where there is a stop sign, a crossing section. So what can you share about your view on like HD mapping? Because it's also very expensive to do. So what's your view on that and the challenges, let's say around all the mapping.

Marek Vanzura (15:50)
Yeah. I think, you know, when we look at these things, can almost always see that there are different camps of people who believe in something or they're fans of something, proponents of something and then proponents of something else. And it's the same with HD maps. have these camps of people who are saying, like, HD map is totally necessary. We have to have it because it's critical and so on. And then we have this other camp that's saying like, yeah, you know, it's too expensive and really...

If we won't operate everywhere, would take forever.

Henriette Cornet (16:23)
entire world, right? That's what they would say.

Marek Vanzura (16:26)
Yes, so these are two camps that are sort of fighting with each other for a long time.

Henriette Cornet (16:31)
are

the pros and cons? Can you summarize what you are hearing?

Marek Vanzura (16:35)
⁓ Yeah, I think it's important to mention here that with autonomous vehicles, we are still in quite an early stage, which means it's not the product that's finalized. It's still something that's underdevelopment, something that is being improved every single day. Therefore, we don't really know the perfect setup, the perfect combination of stuff. And companies are usually

by experimenting with, let's use this or let's use that and see how it works, how it ends up. And usually there's this goal like, okay, let's have the system which is the most, the safest possible with the current available technologies. So we see that usually companies which take the safety aspect rather seriously, they try to implement as many technologies or as many sort of things as possible into their system. Therefore they implement

HD maps because it provides another layer of safety, provides another layer of confidence for the vehicle to operate safely. So I think if we talk about HD maps, one of the main benefits is that they increase safety because they provide another layer of confidence for the vehicle. And it's important. I mean, I can imagine that the vehicle can operate without the map completely, but the problem is that, you know, there is suddenly

there are many more variables that vehicle has to take into account and has to calculate what to do about them and needs to work more. It has to work more to figure out all the situations, all the stuff. So I think the HDMap, even though the entry cost is higher in the end, it allows you to operate much more reliably and in much more

is much easier, I would say. Whereas with no mapping, it's tempting because it's like, OK, we can enter anywhere and we can do it very cheaply. But it might turn out that really operating in some large areas, it's not really that easy for the vehicle if it doesn't have the previous map and so on. when I look at the current state of the industry, can see that the consensus is, OK, let's use HD map because

it allows us to operate as safely as possible. And over the time, it will be getting probably cheaper and easier to decrease the definition of the map or decrease the coverage and so on. So I think we are progressing towards something.

Henriette Cornet (19:09)
Yeah, you mentioned the cost and I know that it's a huge burden for this company doing this mapping beforehand and being able to map really all the aspects of the city, the neighborhood, where they want to deploy. I've heard some companies mentioning that they want to map at the almost as the centimeter level, which I found out, maybe a bit what you say, like maybe it will not be necessary in the future to be that precise because I can imagine the computing behind is huge and

Yes, the investments that you need. I like that you mentioned as well that it's still not, despite what we could see and mostly the people living here in San Francisco may think, oh, okay, that's autonomous vehicle technology. It's done. I kind of like that you mentioned now there are still aspects and maybe, maybe mainly around the business models that still is not clear to all the AV providers so far on what they should do.

how much they should do in the entire kind of what we call the autonomous stack and like on this mapping. So on the mapping maybe to conclude, and I don't know if you want to share something on this very topic is that I was wondering when I was working in Europe, if there would be a way to create like this digital twins, you know, you map the cities, but like it's kind of shared, like the city would be the one hosting.

this digital twin, this mapping, the city could provide the information to any providers that would like to deploy. like, would be like, maybe it's a bit naive, a bit like a really very optimistic in the way, ⁓ public, like public private collaboration could happen. But I, over time, I lost kind of faith in that, that, that it could really happen. And because of the topic of obsolescence and costs for the city would be maybe too big, but like.

What's your view on this idea of having a mapping not done by every mobility providers, but like something more central by the city?

Marek Vanzura (21:12)
I actually encountered here in Europe some ambitions or some plans for cities to do this mapping and provide it to companies. The reasoning behind it was like, okay, we will create this hdmap for you or we will create this digital twin for you. So you can train your models, you can train your vehicles in simulations based on this data we provide you with. And then you can come and the reasoning was like, we want to attract these companies.

test in our cities or we want to attract these companies to even deploy in our cities. So that will be like the added value that the city would provide them to attract them. But yeah, as far as I know, I mean, this didn't really progress very much because, yes, you said the cost of making this map and also the cost of maintaining it because there are so many changes around the city. And if the city decides to provide this thing, they would have to have some dedicated team to maintain, you know, the

up-to-date of the map and all the stuff around it. It's like maybe once they did some math, like, know, cost and pros or, you know, cost-benefit analysis, they figure out like, okay, this is not really something that we can afford or, would be beneficial in the long run. So I don't think there are cities really doing this, but I may be wrong, but I don't think, you know, it's something that in the end turned out to be something related.

makes too much sense for a city.

Henriette Cornet (22:39)
Since we are talking about cities, think what would be also interesting to mention now already in our discussion is a topic of connectivity because there is a level of automation, like everything on the vehicle. And I'm personally very interesting on what happened on the vehicle side and how can we have really a level of independence on the vehicle side. But I know that mainly in Europe, by the way, there are discussions around.

adding a layer of connectivity where the vehicles could interact with the infrastructure, for example. Can you tell us a little bit your view on that matters and what you think is still ongoing or what aspects are we not considering anymore? So what's your view on the topic of?

Marek Vanzura (23:24)
There are so many things that I'm thinking about right

Henriette Cornet (23:26)
We

can make an all episode just on connectivity probably.

Marek Vanzura (23:30)
Yeah, well, where to start? Well, connectivity is kind of important because once when we talked about this in the beginning, it's like, when we have this remote assistance or remote driving, then it totally relies on connectivity. Although, you know, when we have autonomous vehicle, usually those developers aim to make the vehicle completely self-reliant. So the vehicle doesn't really rely on some external help. But actually, you know, it's not possible or at least, you know, right now.

So there's always some corner situations, some small thing that we cannot figure out and needs some help. And you need connectivity for that because we need to call some operator remotely, get the information back. So this is sort of the important element of having deployment of autonomous vehicles, having some connectivity. But when I look at the current situation, especially here in Europe, I can see that there is a lot of

work being done around so-called 5G corridors. usually like cross border highways between countries. And there is this idea that, let's put some 5G communication units there so we can communicate vehicles. And this is related to so-called CITS, Cooperative Intelligent Transportation Systems. it's like usually we don't even need autonomous vehicles for using these things. It's like

providing information to drivers about what's ahead of them. For example, in the winter, you can tell the driver by using this information or these systems that there is icing on the road, like five kilometers ahead or five miles ahead, or there is a crash 15 miles ahead of you. So it's better to just exit the highway and take some detour or something like that. So this is the thing that you also appreciate, having some connectivity.

communication between the vehicle and the infrastructure. So these are the activities that I can make.

Henriette Cornet (25:24)
It

could be in the city as well, The city could send information on road constructions or a ⁓ demonstration happening somewhere and the AV providers would know how we avoid this road.

Marek Vanzura (25:36)
Absolutely. The city is a good example because for some use case that's actually quite common, especially in Europe, is exchanging information between emergency vehicles like ambulances, firefighters, police cars, and so on with other vehicles. So you can know in advance, even before you hear a siren, that there is an ambulance coming and you can adjust to it. Or there's communication between

public transport vehicles, public transit vehicles and infrastructure like traffic lights. So the vehicle can get priority on a traffic light. Therefore, you know, it's much, much smoother ride, much more efficient ride for people in this public transit vehicle. So these are all the examples, not all, but these are some of the examples of how you can utilize communication or connectivity, not just in autonomous vehicles, but basically any vehicle that's equipped with some sort of device.

Henriette Cornet (26:29)
Yeah, yeah, Okay. So it's clear and I like how you present this aspect that actually all these sensors, all these technologies that are developing, sure, they will be central for autonomous, fully autonomous vehicles. And I guess when I say fully autonomous, you see already where I'm heading at. But we have also what we call in the jargon, in the area, lower level of automation.

where you still have a driver in the car and a so-called advanced driver assistance system, so-called ADAS, where all these sensors that we have described, or maybe some of them at least, could support a human driver driving the car. So can you explain a bit how you see these two different markets? Because they are very different. yeah, what are the main differences? What people shouldn't?

distinguish when we are talking maybe on one side, the Tesla versus on the other side.

Marek Vanzura (27:29)
We got a huge topic there. There are so many things to talk about. Yeah, I think I would probably again use some like example with two different camps. We have one camp, you know, we can talk about, have this autonomous vehicle as a holy grail where we want to get. And one camp is like, okay, let's take current vehicles, let's improve their technology and let's slowly get there through, you know, assisting drivers until we get to the point that the vehicle can drive itself. It's like this evolutionary.

kind of perspective. But on the other hand, we have this camp like saying, OK, let's skip all the assistance and let's just go into a vehicle that drives completely by itself. It doesn't require any human input. And it's like the fully autonomous car. So these are these different categories. And when we look at the driver assistance systems, it's actually something that is pretty common these days. Almost like, I would say, every modern Every new car. Yeah, new car that's being sold.

at least one of these systems on board. and these systems are not supposed to replace the driver. They are supposed to help him or to assist him with certain driving tasks. So, for example, one good example is cruise control or adaptive cruise control, which is the system that allows you to move your hand, remove your feet from the pedal.

And it keeps the speed for you. And it also keeps the distance from the vehicle in front of you. So it sort of cruises for you on the highway. And you don't have to care about your feet. Or we have the lane keeping distance, which means the vehicle basically moves the steering wheel for you according to the curvature of the road. It follows the lane markings and so on. So you don't really have to do too much stuff with your hands.

⁓ But these systems are not really supposed to replace you or to do the verb instead of you. They are supposed to make it easier for you to be more relaxed and I would say they more like comfort functions. And yeah, so they always expect that there will be a human being sitting behind a string wheel and sort of like paying attention to the environment around the vehicle, paying attention to...

all the stuff that's happening there and responding appropriately. So the person is always responsible for the outcome of the vehicle. So if the vehicle makes a mistake, the driver should compensate for it and make sure that there is no negative outcome that is undesirable.

Henriette Cornet (30:00)
liability,

right? Back to there is an accident and someone was using too much the cruise control and looking elsewhere. The person would be responsible and not the provider, not the car manufacturer or the sensor manufacturer provider.

Marek Vanzura (30:14)
And that's actually like the, how to say it, the drawback of these assistance systems or the tricky part about it because if they work too well, then the driver tends to rely on them too much. And I think it's called automation complacency. So you expect a vehicle that it will be able to do it more than it's designed for.

you put too much trust into it and then it leads to bad situations because you just stop paying attention or you do stuff that you shouldn't do. And that's different from autonomous vehicle because in autonomous vehicle, the default situation is like there is nobody in the vehicle. Therefore, the vehicle is completely capable of doing all the things that it should do. And then there might not even be a steering wheel or pedals or anything like that. So that's the difference between these assistant systems.

expect you to be there and react, whereas autonomous vehicles, don't expect you to be there at all.

Henriette Cornet (31:12)
And another difference I can see between these two paths, what you say these two camps or these two teams, I don't know how we call them. First, it's usually like, yeah, very different markets, different players in these areas. One is the view I have, and I would love to hear if you have a similar view. The ADAS, it's still very much towards like private car ownership. So you still own your car.

and your car is becoming more more more autonomous to this, how you say, the holy grail, the time where you could just go in your garage, go in your car, push a button, and the car is taking you by itself to your workplace. And you can do, you can sleep, you can do whatever in the car, but it's still your car. You are still kind of responsible for it, in a sense.

The full autonomous vehicles that we see out there, for me, there is more the concept of fleet and more the concept of services. That's for me also very different kind of approach. And if I add a lay on that, one would be maybe that the first one, the private cars, we see mainly on the highway, helping people, supporting people for this long distance ride or in congestion.

still like highway or unidirectional, let's say, a type of driving, whereas the other one is more in a confined area. Do you see that these markets will meet somewhere at some point?

Marek Vanzura (32:42)
They might. I think there are at least two reasons for this distinction or the differences. One is cost, because developing ADAS is relatively more affordable or cheaper. Therefore, integrating it into a private vehicle is cheaper for the manufacturer. Therefore, there is more chance that the end user will buy it. There will be a customer for it. Because autonomous cars are really expensive because of all the sensors and...

and computers, And the second reason is, vehicles are still, as I mentioned, in development stage. They're not final products. They're flea. Yes, yes. So that makes sense that there is a flea that's owned by the developer, because the developer is continuously working on improving these vehicles. Therefore, they own them. And even though they're available for people to use them for their everyday tasks, they are still being maintained and improved by the developer and manufacturer.

Henriette Cornet (33:19)
autonomous one.

Marek Vanzura (33:37)
Therefore, this is the difference why you can't buy a fully autonomous vehicle because it's still under development. But you can buy a car that has automated functions and, as you mentioned, can help you with driving on a highway, for example.

Henriette Cornet (33:53)
Yeah, support that. And I know that there is also like a cost factor that one market will help the other because all this ADAS development, since the scaling is huge of private cars, millions of cars being sold, there is kind of a sense of this, all the sensors that tend to be very expensive. few years ago, the cost may drop because of the scaling, which may benefit the other.

team as we were calling it. But also what I realized when we're talking now, even more to make more clear is on one side is really selling a vehicle, the ADAS one, whereas the impression I have with the fully autonomous vehicles is more like selling a driver. So the vision I have about the deployment of Waymo, and I think many people associate Waymo with Robotaxi with this idea of,

let's say autonomous Uber or Lyft or whatever TNC you have in your city. I don't see that as what defines them. For me, what defines Waymo is their driver, their capability to have all this technology and all the understanding on the driving behavior.

Marek Vanzura (35:07)
Yeah, I think that's a good point that this, I mean, this is is, I would say a good point for comparison of the approach in the US and in Europe. Because I can see, you know, when I'm in the US, I like or enjoy taking Ubers or you know, I'm just a very lame or praise, I always enjoy taking Lemos. But when I think about these things, I usually don't take too many

Uber or taxis in Europe because of prevalence of public transit. Therefore, see that the robotaxi concept makes a lot of sense in the US because of how people move around. They mostly move around in their own cars. Therefore, you're providing some replacement to some degree of their own cars in some areas. Whereas in Europe, don't really see that need for that as much.

in the US.

Henriette Cornet (36:02)
Yeah, No, no, I agree. I see that exactly. how maybe also comparing to the people from Europe coming into US will only see way more as a robot taxi, so to say. Whereas we could see in the future if maybe the technology will be used for other types of vehicles. And we have now also in San Francisco coming new platform, new form factor, having larger vehicles that maybe in the future can even be shared.

But this again, it's going in another area. But since I mentioned already some names, like in San Francisco, the big one now is Wemo. How do you see the different vehicles and brands on the market? You have tested so many of them. And I'm the first one liking your post and reading them very carefully on LinkedIn where you describe the different driving style and how it feels and what you like about one or the other.

Is there something you want to share with us on where are your hopes and expectations with these various brands and where you see a bit where the way is going? Yeah.

Marek Vanzura (37:06)
Yeah, mean, this is sort of like my passion for trying out different autonomous vehicles. It initially started as sort of like, you know, work-related stuff because, you know, visiting different locations or different developers and seeing what they are working on and trying to see how it performs and how well it works. So it started as this sort of work-related stuff, but over the time I became sort of interested in doing it just for fun. So that's the reason why it's sort

I'm trying to experience as many of these vehicles as possible. I wouldn't say I tried all of them. I haven't. There's still plenty of them that I am looking forward to be onboard, hopefully soon. But yeah, I think the form factor is important there. And as you mentioned, we relate this autonomous driving usually for the robot axis. But what I'm enjoying to see that there is this

bigger Zika vehicles, know, higher capacity vehicle or more special vehicle. And also what I like to see, for example, the main mobility company from the US, they started collaboration with some European bus manufacturer. and I think autonomous buses is something that I'm really looking forward to seeing more and more.

Henriette Cornet (38:22)
So, shuttle buses, which size are you? Do you have in mind?

Marek Vanzura (38:27)
are more like regular size buses. I'm not a huge fan of autonomous shuttles to be honest.

Henriette Cornet (38:32)
Tell us more. I want to hear both. What do you like about autonomous buses and what's the thing with shuttles?

Marek Vanzura (38:38)
Yeah, let's start with buses. think that's probably because where I live, so we have a lot of public transportation, we have a lot of buses, and I see that the automation of buses is just the next step or something that would make a lot of sense because there are already people used to using them and there are being happy to use them and they're just so common.

And we also seeing the shortages of drivers. would be quite beneficial thing or quite useful thing to have automated buses. Therefore, they could serve more areas. They could serve them more frequently.

Henriette Cornet (39:16)
night, right? I'm thinking like at time where it's difficult, even more difficult to get drivers on the wages, maybe even more impacting the costs. That could be a good compliment.

Marek Vanzura (39:27)
Yeah, because of, you know, if they serve cities, there is usually higher demand. So it makes sense to have large bus instead of 15 robotaxis or 15 robotaxis. So because of, you know, all the capacity stuff and so on. So yeah, that's the reason why I think automated autonomous buses would be great thing. And I'm a huge fan and proponent of them. On the other hand, I'm not a huge fan of autonomous shuttles, at least in their current form, because I think

their main selling point is the futuristic look. like, know, okay, let's design a vehicle which looks futuristic. And it sort of, you know, benefits or builds on fact that it's autonomous. But the performance of these shuttles, I'm just having such a hard time figuring out what's the added value of these things. It's like, you know, once you start researching what are the usual operational conditions,

or areas, was the performance of these vehicles. It's usually not really that impressive. It operates on very fixed small routes, like circular routes around some very isolated places, which is very important.

Henriette Cornet (40:40)
campuses or like ⁓

quite of semi-open areas.

Marek Vanzura (40:45)
Yes, yes, semi controlled. And the seating capacity is sort of like very low. We can sit four or five people there, which is not very different from a passenger car. The other thing is because there is this issue with sudden breaks. They don't allow people to stand there. Therefore, you're losing some capacity that you are used to having in a public transit vehicle. So yeah, I feel like the main selling point is the looks of these vehicles.

but the performance is really underwhelming and I still haven't seen, well, maybe with one exception and that was the ⁓ VRI turbo bus.

Henriette Cornet (41:22)
I wanted to get there because I tried it in Paris when it was raining around the Rolongeros and I kind of liked it. I found the view very nice and there was more seating space inside. It was a bit bigger. So it's a shuttle, but it's kind of a mix, I would say. It's not really this old shuttle that we saw at the very beginning of this topic.

Marek Vanzura (41:43)
Exactly. I think the best thing about this is that they're using the autonomous driving stack and they're taking it from the robot axis and integrating it into the robot bus. it's like they're making it to work in more challenging situations or conditions and then moving it into this situation. It's much, much, much functional or it works much better. Whereas with autonomous shells, I feel like, you know, I don't want to offend anyone or I don't want to be too...

route or something, but I sometimes feel like there is this sort of like laziness behind it. It's like, okay, you know, let's design it in a way that it goes like only 10 miles an hour. It goes on this like, you one mile long route. And it just goes and goes and goes around with no traffic there. So it's like, yeah, it's like, you know, like low hanging fruit. Let's do it in this way. But there is no ambition to move it forward or further into more difficult conditions or situations.

So maybe I'm wrong, maybe you know this is my wrong assumption, but I feel like...

Henriette Cornet (42:41)
I can see that. maybe a way to put it, a positive way to put it, would say, maybe they did their job into generating curiosity at least. know, attention in this idea that it's not a passenger car that just get like something you get, you know, they brought for me the positive things of this deployment that started actually almost 10 years ago with the shuttles. It's already put a mental model, created a mental model in the head of people.

that autonomous vehicles should be shared. Like the ride should be shared. for me, I really, really appreciate that. I I think it was super important. We did an interview in Singapore about if you think of autonomous mobility, what comes to your mind? And they kind of picture this type of vehicles, which I found successful in a way to not say, yes, of course it's my private car and it drives me everywhere.

That would be something I think they did a good job, like creating this mental model of shared mobility, but maybe it's time to move on on something often exactly what you say, new form, new technology and go away from the low-income fruits the way you said it and this closed areas, et cetera, and move more towards something very useful for transit and for the people.

Marek Vanzura (44:00)
But this is a great point. I have to say I haven't really considered that it provided this sort of like notion of autonomous vehicle that is shared. Yeah, I think that is quite significant contribution.

Henriette Cornet (44:14)
I'm trying to be always optimistic about everything I see. yeah, anything else, any other brand or development, because we mentioned, the Wemos, the Sparse Shuttle, the Rewrite coming, which I found interesting also because they collaborate. is kind of a partnerships happening in the deployment, at least that I could see in France. And I think there will be more coming. There will be some Chinese company coming in Luxembourg also.

in Switzerland, but always very strong in collaboration. And in the US, I think we can also mention that Wemo is collaborating with Uber and with other players really too. Like we can see that there will maybe not be like one player doing it all. I don't know if you want to share something on that and if you have some expectations for the future on how partnerships look like.

Marek Vanzura (45:04)
Yeah, I mean, as a proud European, I always, you know, support local companies and local developers to work on these things. So I'm always happy seeing, you know, if there is some company from Europe actually working on these things. And that's also the reason why I joined the company because we're providing the enable for working on autonomy. So that makes me really happy and proud about the work. yeah, so I think what we can see these days is like,

Even though the autonomous vehicles are still in the stage that they're not final products, they are still in the developmental stage, they reach some level of maturity that they can operate more or less reliably and safely. Therefore, we have several companies from the US, Waymo and from China, in this case, this might be V-Ride, might be Baidu or some other ones. they are trying to work on

they're trying to sort of establish their presence in other parts of the world. And I can see as you mentioned Luxembourg that here in Europe or in France, I can see that this sort of clash of these, you know, companies from different parts of the world, and they are trying to establish their presence. And I would really love to see more action from actually European companies trying to, you know, push back and see like, this is our territory. And this is where we work. And

This is where our vehicles are deployed and where we are.

Henriette Cornet (46:26)
That's a very good point that you say, and I think there is a lot of effort also from the European Commission working on like competitiveness of Europe. then what you know, the view I have on that, and please contradict me if you disagree, tell me if you disagree, is like, if the tech doesn't exist in Europe, I prefer to see like foreign tech and we can develop in Europe other part of the autonomous tech, so to say, because there is not only

driver, we don't only need an autonomous driver, we also need all this supervision center. We need the integration within transit. We need the booking platform. We need to deal with the data privacy. mean, there's so many things to be done, so many responsibilities around a full service, full autonomous service. I would prefer to see foreign tech included in that, but then we move forward rather than blocking and not dealing with like very, like you say, this tech that is not satisfactory.

that we had with other companies. And that may be even worse because it will also badly influence the market. So how do you see that? Where to put the limit into how free the market should be and how open should we be to these aspects?

Marek Vanzura (47:38)
What I definitely see is that when it comes to expertise, talent, capabilities of people, we have that in Europe. It's definitely here. What we usually like is support, and it's support in terms of regulatory support, providing conditions that you can actually do these things in public settings. It's the technological support, something that we do.

we're trying to do to provide the necessary technology to accelerate, to enable, to make easier for these companies or people who want to work on that, to work on that actually with a very low entry bar. And support financial, it's like maybe because it hasn't been considered to be such priority, therefore there was not so much money put into it, in comparison with the US or China.

where it was obviously set as a priority. So I think these three things are sort of increased. Then we can develop something quite significant here. And when it comes to using some tech from abroad and trying to build something around it or maybe... I mean, it's a functional thing and we can see it in France. They're doing something like this with Chinese.

I mean, I'm still slightly hesitant to go in this direction. I feel like I'm more patriotic in this way. I see that we can do something like that with help from the outside. But for example, just a few weeks ago here in the Czech Republic, the first autonomous train was put in a regular service. So we have actually quite

significant progress and achievements that you know, are ahead of the world of the rest of the world. But it's not like in the like, road autonomy, or, or these sort of things. But there are some some some domains where you're actually like, it's leading the way, which is great. But I mean, it will probably end up in some some like, hybrid sort of and or hybrid situation where we will be incorporating these technologies from

the US from China and building something around them and making them like localized, yeah, making them according to our settings. For example, I would love to see like, like bus with the way most autonomous. That would be cool. And I think that would be even more interesting for European users than robotics.

Henriette Cornet (50:03)
Yeah, yeah, I'd to see that too. Let's see what's ahead and what new collaboration and how the market will evolve both in US and Europe where we both worked. So we are approaching the end of this episode. Two questions, a bit kind of challenging just as a teaser for the end to end. Let's start one a bit more negative and one positive. What...

could go wrong with AVs, what is the worst scenario you could think of? Speaking of AV, either on the driving behavior or the evolution of the market, we can see around the world now, we can think about the evolution of the world in a dystopic or utopic way. Let's look at the dystopic way for AVs. What would that mean?

Marek Vanzura (50:50)
Well, at the end of the day, the success of AVs will depend on how people will accept them. So, you know, if people actually don't see them as something that's bringing value to their lives and refuse to use them, then it would be bad for AVs. I mean, it wouldn't necessarily be bad in general, because if people have different preferences, then why to force them to use something they don't want. So, but for the AV industry, that would be bad. And yeah.

related to that, when it comes to this public trust or interest in these things, I think if there is some bad accident, maybe like, you know, killing somebody or doing something really bad, then I think that would probably make people less interested or enthusiastic about these things. So it's like very fragile sort of situation. And I think most companies are aware of this, like, you know,

They're walking on very thin ice and they have to make sure that they won't screw up. They won't make some bad thing that would really harm the public.

Henriette Cornet (51:54)
But something

that I would love to address also in this webinar series is the topic of why, you know, people, there is a world accident every day on the world, killing people, literally killing people every day in every country around the world. And when there is an accident with an AV, there is such an emotional response to it that I would like to dig to and to speak maybe to some psychologists on what's in our brain that make us so

so sensitive about that. when we hear about car accidents, we are very sorry, but we will not think about it for very long time. It will not come in the press at all. So there is something spectacular about the machine being responsible for death of human. I think that's another topic. I'll keep that for another time and you can join again if you want to talk more. But let's say a final positive word about what would be the perfect development, the perfect, like the utopia.

How would the city or where you live now, how could Evie help it make it better in a way? Is there a place for Evie in your ideal future?

Marek Vanzura (53:03)
Yeah, I think this is somehow evident or visible from what I said. think what I would like to see or enjoy seeing is having more automated or autonomous buses or public transport vehicles. Because I can really see that it would increase the quality of life of a lot of people because of increased service hours, increased frequencies, capacity, etc. So I think that would be really beneficial.

So I would like to see this move from small personal passenger cars and turning more into larger vehicles like buses and so on. So this development automation is something that I think would bring a lot of joy and benefit to a of people. yeah, I I'm enjoying seeing how Waymo is expanding across the US and now even into like Japan.

And, you know, places. So I think if they continue doing what they're doing, then what they have been doing so far, that would be great. And I think they're on the right track on the right direction. So I think the future is bright in this.

Henriette Cornet (54:11)
Nice!

That's a perfect way to end the discussion unless you want to add something on this topic, something that the public should know around EVs or we end up with this bright future that may come.

Marek Vanzura (54:27)
Yeah.

think one thing that I would like to add is that if you have a to use this car or to try it out, then do it. Because there is nothing better than real world personal experience with these things. Hands-on experience. if I look around myself, there is still 99 % of people around me, they're not aware of this. Such a thing exists. So if you can do it, if you can take a ride,

two, three, and see for yourself, then I cannot recommend it enough. It's really the best way how to get, you know, obtain some knowledge about these things and see how you feel about these things. And that's something that we should promote.

Henriette Cornet (55:10)
Yeah, yeah, yeah. That's a good point. yeah, they're more and more depending. The listeners, we don't know where they live, but like depending where they live should see or maybe on vacations, can go try and maybe it's worth it. I guess you will meet what we are doing, right? Because when I tried the one in Paris, I was actually visiting a friend and I heard about Roland-Garros deployment and I just do it on my vacation time. So I kind of have this habit as well. Thank you so much, Marek. It was a

Fantastic exchange. I would say everybody stay tuned on the channel for there will be more topic coming around automation about autonomous vehicles, but urban innovate talks is beyond that. It's all it will address mobility in a wider sense. So stay tuned and we will continue the discussion. Thank you very much. Bye bye.

Marek Vanzura (55:56)
Thank you.