Bringing AI in to the physical world; working on software to make autonomy a part of our everyday life
Working with and adjusting technology in a project
Testing and getting the business models right
2020 Autonomous Vehicles Readiness Index:
How we work in testarena Kongsberg City & Lab ecosystems, https://cityandlab.no
Del denne Casen
En LØRN CASE er en kort og praktisk, lett og morsom, innovasjonshistorie. Den er fortalt på 30 minutter, er samtalebasert, og virker like bra som podkast, video eller tekst. Lytt og lær der det passer deg best! Vi dekker 15 tematiske områder om teknologi, innovasjon og ledelse, og 10 perspektiver som gründer, forsker etc. På denne siden kan du lytte, se eller lese gratis, men vi anbefaler deg å registrere deg, slik at vi kan lage personaliserte læringsstier for nettopp deg.
Vi vil gjerne hjelpe deg komme i gang og fortsette å drive med livslang læring.
En LØRN CASE er en kort og praktisk, lett og morsom, innovasjonshistorie. Den er fortalt på 30 minutter, er samtalebasert, og virker like bra som podkast, video eller tekst. Lytt og lær der det passer deg best! Vi dekker 15 tematiske områder om teknologi, innovasjon og ledelse, og 10 perspektiver som gründer, forsker etc. På denne siden kan du lytte, se eller lese gratis, men vi anbefaler deg å registrere deg, slik at vi kan lage personaliserte læringsstier for nettopp deg. Vi vil gjerne hjelpe deg komme i gang og fortsette å drive med livslang læring.
Velkommen til LØRN.Tech - en læringsdugnad om teknologi og samfunn med Silvija Seres og venner.
Silvija Seres: Hello and welcome to a case with LØRN and BI Norwegian Business School. This is a part of the series we are creating together to explore the main challenges and opportunities to create good AI as part of by course responsible AI leadership. With me as the co- host, I have Christian Fieseler, professor of Communication Management and VI and our guest is Gunn Drogset, the CTO of Applied Authonomy in Kongsberg. Both of you are welcome. And we will start this conversation as all the others in the series where I will try to force Christian to tell me something a little bit personal about himself, and then we will Gunn if she can introduce herself as both a developer of autonomous transportation systems, but also as an interesting person. So we'll start with you, Christian. What's your exotic hobby?
Christian Fieseler: My exotic hobby? So we argue, discuss or kind of like we're going back and forth, kind of like that Silvija wants to know my most exotic hobby. I think my hobbies are not that exciting. But you will hear actually in a later podcast that we have some of our guests who are, for instance, inspiring writers. I'm an inspiring artist or aspiring artist that I was very, very close to actually entering art school. I'm actually very much interested in combining AI and the arts. It's not not much time for hobbies, but that's the one thing which I really would like to do whenever.
Silvija: You are sneaky. You have talked with me so many times and you have been hiding this all along. This is an amazing hobby Christian.
Silvija: AI and art, that's what we need to make a podcast about that. So, Gunn who are you?
Gunn Drogset: Yeah. I'm a 54 year old woman working at Applied Autonomy, which is a startup in Kongsberg, and I've been there since the start in 2018. and previously I worked in a big company, Kongsberg Gruppen, for nearly 25 years. I'm a Master of Science, cybernetics or control theory, and that's my hobbies. I have family and I like outdoor life skiing and hiking. I try to learn new things. That is something I do. I've started with trying to learn to see, and the last thing I started with was to volunteer for the Red Cross. The last thing I did and also into arts, but that is music. I love music and going to concerts and such things.
Silvija: What music do you prefer?
Gunn: Oh, it's Funk, rock, everything. I love music.
Silvija: I kind of keep rediscovering music because for a very long time, I've been listening to books. Every time I'm in my car, I turn on a book and focus on that. And lately, I've been really, really tired after two years of Corona and all kinds of other troubles. I've noticed that I really need to turn on music and I need to turn out music I like. And it could be Bob Dylan in my case, it could be Doors, it could be Dire Straits that that works really well. It has an effect on us humans that shouldn't be underestimated in these digital times, I think.
Gunn: I agree.
Silvija: Very cool. So, Mr. Professor Christian, why are we having this conversation with Gunn?
Christian: I think what is really interesting, what we found really interesting with your work is that you deal with when it comes to AI, and doing that well with actually something which is very challenging, right? Because I mean, it's one thing to essentially have something which and I'm now using very imprecise language, but it's one thing to have your AI somewhere in the computer but what you and your colleagues are developing, right, is essentially a AI in the physical world. You're working on software to make autonomy a part of our everyday life, but that then also happens where we are also in our everyday life, right? You essentially bring AI to places, to the streets, to the real world. What I found interesting talking to you is the many thoughts that go into the regulation aspects, the talking with community aspects, the getting the technology right, getting the business models right. So everything which actually makes bringing AI into the real world complex and how you deal with that and what type of thoughts you put into doing that well.
Silvija: AI in practice in a sector that is a very frontrunner in the field. Right?
Gunn: Yes, we work with automated and autonomous vehicles and look towards both public transport and logistics, transport and road maintenance. So we have different kinds of domains that we work in. And the vehicles, either they are manually driven or more autonomous, are digitized and have functionality inside which can produce more and more possibilities, services, more safety, more new utilization for everybody. It is also important that the vehicles nowadays are able to deliver data and function as sensors of a vehicle fleet can use to monitor, improve the transport processes and ensure a more sustainable operation because that is something that is important in the future. And that is something we do. We collect data from vehicles, from road infrastructure, from logistic systems, from different public transport systems and combine this to create a safe and optimize the operation of the vehicles. So that is why data is becoming so important and then you can build on the smartness.
Silvija: We need to give people a few pictures. So there are lots of very advanced concepts here and even the concept of automated or autonomous vehicles. So I want us to go back to an autonomous vehicle, but the picture I'm having in my head is either a self-driving minibus or a self-driving plow that might be cleaning the roads or it could be cleaning the tracks at an airport, for example. So those are the sorts of vehicles you are not necessarily developing, but you would buy vehicles to create this necessary integration and then roll them out, right?
Gunn: Yes. We do integrate different kinds of autonomous vehicles into our platform and support creating services for the customer because we are not the end customer. We have the customers that use them. And it could be a shuttle which you have probably seen in Oslo or in Kongsberg or in Drummond. And we are also having smaller logistics operations.
Silvija : That means trucks in a mine, for example, or harbor?
Gunn: Yes, harbor or airport and moving goods. The important thing is we have an autonomous vehicle. If you have an autonomous vehicle, you still have so many processes that should be connected to move these goods. And that is what we call looking at the automated process of transportation because it's important to think about what kind of process you are optimizing.
Silvija: I can see that the applicability of autonomy on some of these service vehicles or logistics vehicles or even on these shuttle vehicles, better than in most others. So we talk of services or needs. We talk a lot about self-driving cars, but, you know, for a car to have to be able to go everywhere on any kind of a road, it's a very, very long path to get there in a really safe and user-friendly way. But if a shuttle is always going the same round and it can learn to know that route extremely well, and the route can be even adjusted a little bit to that shuttle with sensors, for example, or if you need to drive several snowplows, for example, in a row next to each other in order to get the job done, then it's perfectly suited for automation or autonomy.
Gunn: Yes. That is the most important thing is to find that use case that these kinds of vehicles are suitable for and, and look pilot and learn how the how this is working and what kind of data you need to support them because they're not able to perform everything themselves, even though they are called autonomous, because then they need to have a task to do. Then need to be told what to do and they need to be monitored maybe because someone is is wondering where their vehicles are and somebody is going to have some KPIs, deliver KPIs to the boss saying how well they perform or the business is going because this is a part of how you are creating your business, so does it fit in the process and does it fit in the service that you will provide.
Silvija: Can I ask you, what is the most difficult thing to do here? What is the real crux of responsibility in developing these systems?
Gunn: We thought that the autonomous vehicle should be more mature. And this in 2022 then it may be AR. They are still struggling a bit handling the situations around them because they are changing so much. It's complex and they mitigate the complexity by driving slow or stopping. And if you drive slow or stop, often the service is not very good. So that is one thing that hasn't come that far as we hoped for now but that even though they are able to drive faster, they still need to have connection to other systems, either other vehicles or infrastructure and other: your task, the logistics systems, for instance. We have not yet all these things are not standardized, how you integrate it and communicate with it around it. So that is a thing that we work with and try to try to do that for the customer so the customer doesn't have to integrate and take care of every aspect that is needed around this fleet. So that is the work. And then you have the regulations in Norway today we have laws that say that we are allowed to drive in mixed traffic, but we need to apply for each, each type and we also need to have a safety operator on board. As long as you need to have an operator on board, it's not cost efficient. And then the business part of it is not accelerating as we could want to.
Silvija: Can I ask you and I'd like Christian also to help me in this direction, I'm wondering if our regulations are in some ways also stopping the development here, because, you know, I'm sitting on the board of a public transportation company Ruter, and I notice that both our goals and our bylaws and also the regulation that deals with us is very precise and very much based on 150 years of experience with vehicles going on tracks and the old idea of what transportation is and it's actually stopping us from being able to do efficient systems with this completely way of doing transportation.
Gunn: Yes, I think so also because the regulators are used to approve conventional vehicles and then I checked the brakes. It's important to have good brakes. The digitized vehicles are having a lot of software onboard. They have software that controls the brakes. And what happens when you change the software? What happens to the functionality of the vehicle? And that is something that the legislator and the regulators need to learn how software controlling the vehicle is can be a possibility and also be a risk for safety and operations. We work together with the government and the municipality and the country to innovate both on the technology, the business models and also the regulations. So it is very important that we work together with that. So because everybody needs to at least face, learn and also improve or change the process.
Christian: I like the idea here of working together with regulators. Because I think maybe a lay uninformed opinion can be that these regulators are holding us back. Right? Why can't we just learn by having that out on the streets? But you can equally as well make an argument that through regulation you learn, right? At least if regulation is open enough to treat this somewhat as a sandbox, something to figure out something together and which might even be a some form of down the road competitive advantage. Right. Essentially building safer solutions than just building it on your own, or at least what I would imagine.
Gunn: I think it's important that we work together and that is also something that Norway has been in the forefront of working on. And I think it's maybe our possibility here to to make new businesses.
Christian: May I ask something? As a layperson, I don't know whether it's kind of like your corporate secret, but how does that look in practice? Because I think the ability to build software for autonomy is something very specialized? Something where there might not be that many designers, not that many programmers? How does a regulator work together with you? How can they understand how the systems work? Is that something which only you do, how do you essentially create an equal footing between the people that need to create some type of oversight or understanding? Just for my understanding, the regulator, do they also have software engineers, so how does that work in practice?
Gunn: No, I think when we apply for driving in a route, we need to provide documentation of what the capability of the vehicle is, what kind of weather and conditions and routes it is allowed to drive and that might be test documentation, design documentation from the OEM provider and together with the plant area and the route. In the risk assessment done in that area, we provide information to this legislator about the regulators. So that is the process, but they should, of course, bring some software competence into their organization because they are new to the things happening all the time as you said, you have a modern EV, you get new software over there at night and in the morning you have a different car. What does that do with your training, your set of certification, what kind of things happen there? So it's not only an issue for autonomous vehicles, but it's also an issue for modern manual vehicles that are connected. I think it's important.
Silvija: I have to ask you one question in in consequence of Christian's question, and it is ownership of data for what I'm worried about is a situation where, let's say I'm driving my Tesla and Silicon Valley servers know eventually more about my both driving habits, but also about the movement of people in Norway, our own regulators and our own traffic institutions. This is very important for both development of our roads, development of our infrastructure, but also for regulating our traffic in the future. Basically, I just want to underline how important it is and what do you think about it in terms of that we have to develop some of these vehicles and some of these systems so that we also own the data and keep in the driver's seat as to speak for the future development of these things and society.
Gunn: It is important that we have access to to the data and have the possibility to share it. The road authorities are very interested in in looking at this data. They are working a lot with data management and and how they should rig the system for the future. If we have only a few OEMs that is is from abroad and not willing to share the data, that they will probably kill the business for us. But I think coming from a base company that was very protective of of information, this is a new world for me because we are talking about data sharing all the time. And that's its difficult. But more and more OEMs are seeing that this is necessary and their own business models is changing. They are not car producers anymore. They are software companies with new business models. So they are extremely both worried and of course, looking for new possibilities in this new market that think so. That is important for Norway because we don't have any vehicle productions. So we need to grab this data part. I think that is important for us.
Silvija: I'm going to jump in with one more question, and it is: Christian, when you're talking to your students who are nominally studying something related to management, is it difficult to explain to them that the companies they're going to be managing are more of software companies than, let's say, a car company? What I just heard John say is that, you know, what they're building is more of a software thing than of a traditional vehicle. Right? How does that change what managers need to be thinking or doing?
Christian: I think maybe by virtue of being younger than me, I think our students most likely understand it better at face value. Because what I find interesting is that when you talk to younger people that, for instance, the decision to buy a device or a car is also by now governed by what the technology, what the software is in that car. I think for people in my generation, not that old, but older, I wouldn't bother so much. For instance, kind of like how my navigation system feels or not. I think the younger you get, I think these components make more of a difference. And I'm saying that now because I think there is maybe more of an intuitive understanding of the role and the importance of software. However, I think that is independent of whether you're a student or a professor or a professional. I think what we often tend to forget when it comes to managing these systems is that managing software is quite complex and that gets more complex than we would initially imagine. I mean, it's kind of easy to say, oh, my car now has a software layer, but it comes with so many things and maybe even more than just a physical product. Then dimensions, which you also discussed about data sharing and data governance, come into play the things of upgradeability about the product essentially being a constant quote unquote beta, something which develops over the air software updates and all that. I think to answer your question a little bit long, while lately, I think there is understanding but I think we all are a little bit, quote unquote, challenged in understanding the very dynamic nature when products or could become essentially not services, but something which is more or less constantly in flux and has many, many different code dependencies.
Silvija: I guess this goes to Gunn as well. You know what defines a good service? What defines a good vehicle before it might have been its hardware, robustness or feel or you know, and now it might be its constant and continual upgradability or its ability to learn quickly from the data or you know, what, what is a good vehicle in the future Gunn? What is the best vehicle?
Gunn: I'll tell you what the best vehicle is. I see from my vehicle, my private car. I have a fairly new car and I see and get updates once in a while. But the Mini is very updated with the app, how you can control your vehicle from the living room, not driving but controlling the heating and the charging and such things. So I am not sure. I think we will choose cars that have provided a good service for what is important for me. Someone is choosing shared solutions. They can share a car when they need it. Someone wants to have a private car. Cars that are smaller or more if you're talking about these buses could be important for developing services for people who haven't had the possibility yet to take a bus, because there are straight lines services today. I'm not sure what the best car will be, but it certainly would have to work and give the impression that I am moving for instance, if it's me, it's not stopping. It has good service. It doesn't have to go very fast, maybe, but it has to show progress. But I'm not sure what the best car is.
Silvija: It's really interesting to hear your reason as you're reasoning now, because I think this is the way we are going to try to figure out really the opportunities as well in this new identity of a car or actually Christian for new identity of many of the services that have become digitized or AI driven even more? And I heard you say, you know, maybe some of us will like to have these shared solutions, maybe some will be driven by sustainability and so on. Some of us will want as much flexibility as possible. So personally, I live somewhere where it takes me at least 15 minutes to the closest bus station, actually 25min now with all the building going on. And for me, that's too much, right? So what I need is this capillary transport that will bring me to the big bus or to the big train station. I like the idea that my car is developing as quickly as AI is developing. So that concept of continuous improvement is important to me, maybe not to somebody else. So I think there'll be many ways of solving this future. And I also heard you say fun. You didn't say fun, but you drive a mini, you know, you so some people will want that and maybe they will view their car as their health station or their entertainment station or their social moment of the day. I think exploring these opportunities, both in industrial settings like your types of vehicles or in personal settings, is going to be a necessary part of market development going forward.
Gunn: To know what value is for the customer is always important and it's changing and it's difficult to hit the target because people are different, businesses are different, what they have as value for the transport and requirements. So I guess that private transport is the most difficult or the worst for that because we are quite different. But for logistics, they have transport from A to B with some goods for instance. It's important to customers typically, but they have different ways of doing this process. Some can plan long ahead to do it, starting early with the transportation. Someone has urgent needs so the systems around need to be fit to that as well.
Silvija: Christian, towards the end of our conversation, what is your main takeaway and what, what's the one idea you really want to extract from Gunn's head?
Christian: What I get a better understanding now after having talked to you is the complexity that goes into what you do, right? The many different constituencies that you have to talk to the interplay between what is commercially viable, what is socially viable, what is technically viable. I also, by the way, find the discussion which we're having about not necessarily building vehicles. Because we always assume this is one vehicle builder and that is it. But that essentially somebody also takes transportation hardware and makes it into systems. That was very instructive for me, this idea of that, there's that there's an additional layer of activity of work, of design journeys, right? Because if you are obviously a systems integrator, you deal with systems, which makes it that's the information point about constituencies and all that very complex. I find this complexity of your task, your job what you're dealing with really interesting also tells us, of course, a lot about when you want to do that responsibly that you have many touch points that you want to do. That was really interesting to me.
Silvija: For me as well Gunn, I have to say, this idea that you are not just developing vehicles, you are developing systems of transportation, and you have to think about both the roads and the sensors and the vehicles and the users and the use case industrially and balance effectiveness or the attractiveness of the service with the available regulation and keep developing that regulation so it actually allows the necessary growth and development. Super interesting this connects also to a conversation we will have later related to how does regulation and law provide space for innovation?
Silvija: Is there anything else Gunn you think we should underline for our dear students that need to understand what's responsible for digital innovation and development for the future?
Gunn: I think it's important to ask when you're doing what kind of value I want to create for a customer and what kind of questions you want answered for when you go into digital types and collect data because then you can select the correct systems and data systems, data quality and such things. That is important because you have to have a kind of vision of what the value is and what kind of answer you want from your data. That is also to the AI part and the machine learning part. I am a technician and I love the technology that is not the most important, it's just a tool to create systems for the customer that is attractive and safe.
Silvija: Christian, do you have some famous final words?
Christian: Famous final words? Hopefully not yet. I don't want this conversation to end, but I think we're nearing the end. Thanks for sharing all your experience with that. That was really wonderful to actually learn as a layperson how this actually works in practice and what type of decisions you need to make.
Silvija: Thank you both for a very inspiring and educational chat.
Christian: Thank you.
Gunn: Thank you.
Du har nå lyttet til en podcast fra LØRN.Tech - en læringsdugnad om teknologi og samfunn. Nå kan du også få et læringssertifikat for å ha lyttet til denne podcasten på vårt online universitet LØRN.University.
Du må være Medlem for å dokumentere din læring med å ta quiz
Allerede Medlem? Logg inn her:
Du må være Medlem for å kunne skrive svar på refleksjonsspørsmål
Allerede Medlem? Logg inn her: