Wie weit sind wir von selbstfahrenden Autos entfernt?

Anja Hendel: Hallo und herzlich willkommen zum Deep Dive Mobility Podcast von Digital Kompakt. Mein Name ist Anja Hendel. Ich bin Geschäftsführerin bei Dekonium. Bei uns dreht sich alles um digitale Transformation im Bereich von E-Commerce und Mobilität. Heute spreche ich mit Sheila Nidelko von Here Technologies über autonomes Fahren und welche Rolle dabei Karten spielen. Da Sheila Amerikanerin ist, werde ich das Interview auf Englisch führen. Hi Sheila, thanks for being here. You are the Senior Director Automotive Products at Here Technologies. Yes. Ich bin sehr, sehr gespannt, mit Ihnen über Autos zu sprechen, Autos zu fahren, aber ich bin sicher, dass nicht jeder, der uns hört, alles über hier weiß. Können Sie uns eine kurze Vorstellung geben, was Hira Technologies genau macht?

Sheila Scanlon: Hira Technologies ist eine leitende Intelligenz- und Technologie-Plattform-Firma. We are focused on providing location data, tools and also services to better power real world outcomes. Our core competencies really are maps. We've been in the map making business for a long, long time. But we are really evolving all of our capabilities to become more of a location platform company. That allows people to not only enjoy the products that we create, but also to bring their own data and create new products on top of ours.

Anja Hendel: Super interesting. I know that here's vision is to enable the world for autonomous driving. So how would a world like this look like?

Sheila Scanlon: Yeah, and I think our vision is to enable the autonomous world, so not just for driving, but for all levels of autonomy. So if I were to imagine an autonomous driving world, what would that look like? You know, it's going to essentially start small and expand and increase capabilities over time. So I think we're going to see applications that are very controlled at first rolling out and then ultimately get to the full driving, personally owned cars in the future. But I see it. I think the word that comes to mind for me is stress-free and safer. So to be able to have a transportation system that's fully autonomous from end to end in a multimodal world that allows us to get to where we are in a very stress-free way and in a much safer way. And I think The focus on the way I perceive transportation today is how do I get to the end point as fast as I can. I think it redefines our use of time, that if we don't have to drive and pay attention every second of the journey, that we can actually repurpose those minutes while we are in transportation to be more enjoyable for, you could think of entertainment or reading or getting work done. So I think it's really going to transform the way we think of transportation and what we do while we're being transported.

Anja Hendel: Ja, super interessant. Ich denke auch, dass vielleicht die größte Sache, wie du es vorhin erwähnt hast, der Zeit oder die Dauer des Fahrtens sehr wichtig ist im Moment. Vielleicht wird das sogar verändern, oder? Ich denke über das Schlafen. Vielleicht will ich acht Stunden schlafen und dann habe ich keine Ahnung, ob how long the car will drive or the device will drive me, because I won't wake up earlier anyhow, right? So maybe this will not be the most important thing. then, because you have your conference call, for example, and it takes an hour. And even if you are at a destination in 40 minutes, you will not step out and end it, right? So it will really redefine many things. Do you agree?

Sheila Scanlon: Ja, das ist genau so, wie ich es ansehe. Ich habe Kinder, die in der Schule etwa fünf Stunden von mir entfernt sind. Und zu denken, dass ich dort fahren kann, ist alles darum, wie schnell ich dort kommen kann. Aber wenn ich das ehrlich machen und schlafen oder mein Shopping machen oder mich auf mein Lesen befassen, während ich das tue, I wouldn't care how long. And I think we'll plan our travel completely differently, you know, once we move to a full autonomous world.

Anja Hendel: Sure. It will also maybe disrupt other industries than the automotive, right? I mean, like hotels and things like this. So we really have an impact on many, many things probably.

Sheila Scanlon: Yeah, I didn't really think about that, but you're right. It will totally impact a lot of those existing businesses. And we consider location data to be obviously really important in the automotive space. But if you look at, especially in times of COVID, transportation and logistics, und wie wichtig sie sind, um uns das Essen oder Packungen, die wir brauchen, weil wir nicht in die normalen Storchen gehen können, wie wir sie normalerweise gehen, brauchen. Denken wir darüber nach, wie die Autonomie diese Industrien auch erhöhen kann, ist eine Art lebenswerte. Ich sehe Autonomie-Trucking als einen wichtigen, wahrscheinlich ersten Roll-out der Autonomie, wo wir Autonomie-Trucks haben können, Das ist ein super interessanter Thema.

Anja Hendel: Ich meine, wenn man über Trucks spricht, Many people think that this will be one of the first use cases, which will really be live, right? On the highways, trucks, because you can separate them from the other cars or from the traffic. And probably also the way they route through the streets is easier than in cities, for example. Do you agree on this? Do you think that trucks will be the pioneers of autonomous driving?

Sheila Scanlon: Yeah, I think there's going to be two initial launches and we're seeing it already today, right? We're seeing places where you can have very slow moving, fully autonomous vehicles, you know, serving a purpose on university campuses and retirement villages. And so, and we can control those operational design domains very closely and also provide a service that is really required. But when you think about trucking, it just makes so much sense. And even from a mapping perspective, it makes so much sense. Because usually they're on highways with really, really good lane lines and not pedestrians and not cross traffic. So the challenges an autonomous vehicle faces are less on highways. And I absolutely agree with you. I think you're going to see, just like we have carpool lanes today, you might start seeing some dedicated autonomous driving lanes so that we don't have to worry about how an autonomous vehicle, when they first start launching, interact with passenger-driven vehicles. So, yes, I 100% understand why you're seeing a lot of folks in the industry start seriously looking at that space. Given the fact that we know there's a huge lack of truck drivers out there today, there's an absolute need. You can see that they can say, okay, even if I just drive from point A to point B fully autonomously, that's going to reduce the stress on my driver, reducing my requirements to have so many drivers. On top of that, it's actually really good for fuel efficiency. When we're driving in a fully autonomous way, there's less braking, less speeding, which definitely has a direct impact to fuel efficiency.

Anja Hendel: Oh, I never saw this point. The sustainability point, it's really a good one, right? But I always saw this point on the work-life balance of the truck drivers, right? Because it's not a dream job to be around all the time. It's much better to To pick up the trucks on a certain hub and do the last mile with them than being on the road for days away from your family and friends. I am having an expert like you in this podcast. I really would love to go a little bit more into other technology topics because many people think that the role of 5G is still a very important point. So there's a big controversial discussion on this, how much of 5G is needed for autonomous driving. Do you think this is an important technology to make it happen or Would self-driving cars go along also without this technology?

Sheila Scanlon: I can understand why this is still a contentious conversation. I think 5G is super exciting and what that's going to bring to the market. But I think there still is The verdict is out of what role 5G will actually play for autonomy. And the challenge there is an autonomous vehicle needs to be able to react to every situation at a given time. And if there's any loss of connectivity How does the vehicle do that? So the challenge or the question continues to remain of where does that compute happen? Does it happen in the cloud? Does it happen on the edge or does it happen in the vehicle? And I think those conversations are going to continue for some time. I think there's some obviously low hanging fruit for 5G. When we think of V2X application or V2V, given that those are going to be very dense networks, cars talking to each other will absolutely augment autonomous driving capabilities. But I think the question will remain for some time of what compute can you actually do on the edge or on the cloud or which absolutely in terms of redundancy of an autonomous vehicle has to remain in the vehicle. So I don't think I can answer that yet. I think we're going to see that evolve over time and I'm absolutely part of some committees that talk about where they think 5G will play in the autonomous vehicle space.

Anja Hendel: Ja, super, super interessant. Sehr gute Insights. Du hast gerade Car-to-Car und Car-to-X-Kommunikation erwähnt. Das ist wirklich auch eine der wichtigen Dinge, wie diese Autos sich mit einander kommunizieren und auch mit ihren Umgebungen kommunizieren. Ich denke, China arbeitet auch viel mit smarten Städten und Autos, die mit der Stadt kommunizieren und viel Technologie aus dem Auto, mehr in den Geräten um sie herum. Also mit dir als Datenexperte, was für Informationen und Daten benötigt werden, um diese Fahrzeuge zu teilen? Worüber kommunizieren sie? Was ist das Bedürfnis, für das 5G wahrscheinlich eine Lösung sein könnte?

Sheila Scanlon: Ja, und ich denke, Datenschutz ist ein weiteres Thema, wer die Daten besitzt, all diese unglaublichen Daten, die aus all diesen Autos kommen. Aber ich denke, es ist ziemlich stark akzeptiert, dass wir denken, dass wir Daten für sicherheitskritische Anwendungen teilen sollten. Also ich schätze, die Daten, die aus diesen Autos kommen, die zu anderen Fahrzeugen geteilt werden, wären, es gibt einen Unfall vorhanden, es gibt eine Ambulanz, die um die Kante kommt, es sind die Augen, Du weißt, dass man die Autos nicht um die Kante sehen kann, aber ein anderes Auto um die Kante sieht, dass die Ambulanz kommt und kann das an alle Autos in der lokalen Gegend broadcasten. Und wenn du dir denkst, wie wir als Menschen fahren, wenn eine Ambulanz kommt, langsamen wir und wir gehen weiter, richtig? Autonomische Fahrzeuge müssen im Endeffekt besser als Menschen fahren. Und ich denke, diese Art von Kommunikation, was mit Sicherheit zu tun hat, you know, there's an oil spill ahead and somebody's traction control goes off in their car, which would detect something slippery on the road, you could broadcast that through 5G to all the cars in the vicinity and that can help them slow down and drive more appropriately in those areas or avoid them altogether.

Anja Hendel: Super interesting. You mentioned some points which I'm really interested in, one is the safety aspect, but I would love to Deep dive a little bit more on the data topic first. So here, I think here needs an enormous amount of data to create these maps, right? And how do you make sure that you always have the most accurate digital representation of the world? And where does all your data come from? I mean, it must be an enormous task to have all this data collected and shared and also keep it up to date. Yeah.

Sheila Scanlon: Ja, das stimmt. Eine perfekt akkurate Repräsentation der Welt ist wahrscheinlich unmöglich, weil die Welt jeden Tag verändert wird. Jemand kann einen Stop-Sign überwinden und wie schnell kann ich das in meinem Map updaten? Hier liegt es wahrscheinlich auf Tens von Tausenden von verschiedenen Datensourcen, um unsere Maps zu updaten. So we rely on government sources. We actually troll media sites to understand if there's construction coming or new regulations happening in countries. We obviously have our own fleet of survey vehicles that provide super high precision accurate data. We also ingest vehicle sensor data from adas systems that are out there on the roads today. so there's a lot of cars already and many more coming out with essentially front-facing cameras to enable some adas functionality and we leverage those data streams as well to make sure our maps are super fresh. So we build them incredibly highly accurate and then we use this fresh data coming off vehicles that are driving all the road networks all day long to make sure our maps are up to date. And we call that self-healing. So we don't only use vehicle sensor data, but of course we use crowd sourcing, government sourcing, there's overhead imagery as well that we can use to create and maintain our maps.

Anja Hendel: So you mentioned many things in your last answers. dass man nie eine digitale Repräsentation der Welt haben wird, was ich glaube, für alle stimmt. Und alle verstehen das. Und auch der Punkt, den ich super interessant fand, ist, dass wir wirklich denken, dass Autos besser fahren müssen als wir. Weil ich denke, dass mehr als 1 Million Menschen Autofahrer-Aufgaben jedes Jahr. Also brauchen wir natürlich Autofahrer, die bessere Fahrer sind als wir und mehr von diesen Aufgaben vermeiden. Aber es gibt immer eine große Diskussion, wenn es um einen Autofahrer-Aufgabe geht. Und die Leute haben viele Sicherheitskonzerne, weil ich denke, sie haben andere Erwartungen an Maschinen als sie an Menschen haben, richtig? Was ist Ihre Antwort auf all diese Konzerne in Bezug auf Sicherheit bei Autofahren?

Sheila Scanlon: Yeah, I mean, I think if you look at the accidents that you referred to, they're usually down to human error. So the idea of robots driving our cars, which is what autonomous driving is, would be eliminating the human error. And I do think ultimately it will be safer. I think we have a lot more work to get there. And I think the tricky part is we're mixing today human driving cars with autonomous driving cars. And so we have to get smarter about how that happens. And I think we're going to rely on government infrastructure to be put in place to enable that. But I do think ultimately it will be safer by taking the human out of the loop. Und wenn wir dann wirklich auf einige der Verletzungen schauen, die passiert sind, waren sie nicht unbedingt in der Funktionalität der Autonomie. Also, ich denke, man muss das fast auf eine Fall-zu-Fall-Basis betrachten.

Anja Hendel: Ja, absolut richtig. Also, Anfang dieses Jahres hatten wir das erste total virtuelle CES dieses Jahres. Ich habe gesehen, dass hier Technologien die High-Fidelity-3D-Modelle für 70 Stadtzentren weltweit eröffnet haben. Können Sie uns ein bisschen mehr darüber erzählen, für welche Anwendungsfälle solche Modelle hilfreich sind?

Sheila Scanlon: Ja, absolut. Wir haben mehr als 70 Modelle von Stadtzentren weltweit geschaffen. Es ist wirklich beeindruckend, um zu sehen, was es ermöglicht. Es ermöglicht AR-Ausgaben in Automotiven und auch in der Unternehmensindustrie. There's been customers coming to us wanting to use our city meadows for last mile supply chain delivery systems. And then urban planning and 5G planning actually, because they're such precise, you know, die präzisen Locations der Gebäude, und es gibt auch Eingriffspunkte und Fassaden, wenn man das glauben kann. Es ist also sehr interessant, all die Applikationen zu sehen. Und der Fakt, dass wir eine Plattformfirma sind, wir haben all unsere Daten da, also unsere 3D-Metalle sind da, die es ermöglichen, Programmer zu kommen, und sie expositieren Anwendungsfälle, die wir noch nicht einmal gedacht haben. Aber jetzt sehen wir viele urbane Planungen, 5G-Planning, AR-Ausgaben dafür.

Anja Hendel: Also haben Sie auch einige Ideen über Smart Cities, richtig? Das bedeutet also, Sie werden nicht auf dieser autonomen Welt bleiben, aber diese autonome Welt beinhaltet auch das Konzept von Smart Cities und das ist ein Feld für hier?

Sheila Scanlon: Ja, ich meine, absolut. Wir sind, The city planners for smart cities utilize our location data for a lot of their planning activities. And if you think about the data coming off these vehicles that we can provide, it allows them to provide more services to their residents. I think also what we're seeing in just city planning is the need for the new infrastructure that's going to need to be put in place as we move to a more autonomous world. You can imagine, like right now, we have bus stops, for instance. How do those bus stops have to change when it's a fully autonomous bus, right? I mean, I would imagine these pick-up and drop-off locations are going to have to adapt to fully autonomous buses or cars. So I think all of the information that we provide of here is Ja, super interessant. Ich denke, das ist wirklich ein sehr wichtiges Feld für alle Transporte, Logistik oder Transporte.

Anja Hendel: und Mobilitätsaspekte, die immer benötigt werden. Je besser du die Welt siehst, desto einfacher wird es für alle sein. Also ein großartiges Visionarium, auf dem du hier arbeitest. Als ich einen Artikel von The Guardian im Jahr 2015 sah, They already claimed that there will be self-driving cars for customers in 2020, which obviously was not the case right now, at least not on a wide range. What is the roadblocker here? What makes the development? auf self-driving cars so complex from your point of view?

Sheila Scanlon: well i mean it's a super complex problem to solve. i think some people have equated it to putting a man on the moon. but it's not just the technology that has to be fully solved and all the edge cases that come along with it it's also the matching regulation. We are very proud to say that our HG Map is actually powering the S-Class L3 function. that's going to be delivered in mid-2021 through Daimler. And that's going to be based in Germany only right now, and that's because of the limited regulation. As regulations improve and as they expand, Since it is a connected product, Daimler will be able to expand the functionality dynamically for these vehicles. But I think we have a long way to go as far as regulation. And because it's all done kind of based on country by country or municipality by municipality, it's going to take some time. And I think we're going to see it roll out. in different geographical regions at different times.

Anja Hendel: What is your impression, which region will be first? Who is the frontrunner in terms of this?

Sheila Scanlon: Well, I think that's an interesting question, right? I've been in the automated driving space since 2015 and I will tell you that there's interest from every country and every city around the world to have autonomous driving pilots going on in trials. They all see the benefit of how it can feed into their transportation systems to make them more efficient and more cost-effective. I think the US has been a big leader in this space. There's definitely countries in Europe that are completely leading the way. But when we look at places like Singapore and China that have a little bit more, I would say, broader laws and regulations and their countries are a little bit more active in putting the infrastructure in place quickly because they want to be those leading smart cities. So I think you're going to see it kind of from everywhere. But I think China and Singapore are places to watch as well as obviously there's so much. I really consider the birth of autonomous driving to have started probably with the DARPA challenges in 2007. Lots of US companies leading the way in the technology.

Anja Hendel: Super interesting. So you think it's really a race because everybody wants to be first, right? That's what I understood.

Sheila Scanlon: It's kind of like putting a man on the moon who is the first to do that as well.

Anja Hendel: What I wonder is a little bit of, because you mentioned that the contents are also a little bit different. Does this also mean that the devices, so the cars will be different? Will we have different autonomous vehicles in China and Singapore than in Europe and in the US due to these regulations and also to the way how

Sheila Scanlon: Yeah, I think the interesting thing is we work with so many OEMs and Tier 1s to provide them maps for their autonomous driving vehicles. And through that work, we can see that almost everyone's autonomous driving system is slightly different. and relied slightly differently on their vehicle sensor setups. So I think they will be different everywhere. I also think that you can imagine the form factor of the vehicles could be different geographically in what's important for the user experience. You know, there's been tons of work done understanding what the user experience of these vehicles should be. Because it's not just about removing the driver for the driving function. The driver does so much more in that vehicle than just driving. They're really the authority of the vehicle as well. So there's security aspects, there's comfort aspects, there's safety aspects that have to all be considered to provide that ultimate user experience. Based on the use case each AV is serving, and it could be mass transportation, it could be ride hailing, it could be private usage, it could be trucking, the vehicles

Anja Hendel: could

Sheila Scanlon: have a different driving style, they could have a different user experience in the car, and they certainly could have different autonomous driving systems.

Anja Hendel: Wow, that's super interesting. I never thought about this one. So having in mind that we always overestimate what happens within one year and always underestimate what will happen within 10 years, technology-wise at least, I would love to hear from you a little bit what you think. five years from now, where are we in terms of self-driving cars?

Sheila Scanlon: Okay, so that puts us in 2026. So I agree with your earlier statement that there were bold statements that we would have autonomous vehicles. I mean, I even heard one company say by the end of 2021, they were going to launch their own shuttle service. And that clearly hasn't happened. What we're seeing from our perspective is that full autonomy is being pushed out in the future a little bit in terms of technology and regulation catching up. But the interesting thing is we're seeing a huge press for increased ADAS functionality. So over time increasing levels of autonomy. and obviously the great news for us is we're seeing that all of our customers that we're talking to are seeing the value of the map in all of those situations. So for me, I feel like we're going to see autonomy slowly growing over time. in the private car space, right? So when we buy private cars, we're going to get increased levels of autonomous functionality or, you know, assistance in the vehicles. And I think when we get to the non privately owned vehicles, you're going to see in autonomous driving, fully autonomous driving in very limited, what we call ODDs, operational design domains. So they're going to pick areas that are much easier controlled, dedicated lanes, places that don't have cross traffic, places that don't have a lot of pedestrians, and then expand that functionality and that coverage over time. I also think you're going to see a lot of focus on, you know, more increased focus on autonomous trucking. Because, you know, there is absolute value for the trucking companies to save money on the drivers, increase fuel efficiency and reduce costs in that way. And then leverage autonomy on highways and like hub-to-hub applications.

Anja Hendel: I want to deep dive a little bit in one thing, because I'm really excited about this one. So you talked about autonomous driving cars, which are owned privately. So you think that we will really buy cars and own them? with autonomous driving functions or do you think cars will or self-driving devices will always be managed by fleets?

Sheila Scanlon: No, I think ultimately the future will have us owning self-driving cars. I think that's way out in the future. I think it's definitely still in my lifetime. But I think the challenge of having self-driving cars right now is really the cost. When you look at the cost of the sensors that need to go on these vehicles, since they're new sensors and they're not in bulk quantities, they're just too expensive to think that we would all own them. But I absolutely think over time that will happen. I agree with you, the initial use cases will be fleet-owned cars though.

Anja Hendel: Okay, so it will start with fleets, but later on we will probably also own self-driving cars. Interesting.

Sheila Scanlon: Yeah, and you're already getting some versions of that today, right? With Tesla's Autopilot or GM's Super Cruise. I mean, the way the OEMs are looking at this is just increasing functionality over time. And as those prices of sensors reduce, then you can imagine increased levels of autonomy even in privately owned cars.

Anja Hendel: Und gleichzeitig erhöht das Sicherheitsniveau unserer Autos, weil mit mehr Technologie, die in die Autos kommt, auf dem Weg zum autonomen Fahren, macht unser Fahren immer sicherer und sicherer.

Sheila Scanlon: Ja, ich stimme 100% dazu.

Anja Hendel: Sheila, this was super interesting. I learned so much about this and got so inspired by all your things. So to come to an end, unfortunately we have to, because people don't have too much time to listen to podcasts. I have one last question for you. Can you please complete the following sentence for me? And the sentence is, for me, mobility means

Sheila Scanlon: I think mobility means to me getting from point A to point B in a stress-free way. Stress-free and safe way. Very nice.

Anja Hendel: Thanks a lot for the nice conversation, Sheila, and your time. Thanks so much.

Sheila Scanlon: I really enjoyed it.

Anja Hendel: Thank you so much.