Panel Discussion: The Future of Autonomous Driving Technology

So hello, everybody. I hope you've enjoyed the presentations that've been given by our panelists, many of our panelists today. We are now actually entering the live part of the event.

So obviously, if this technology is going to go wrong, this is where it will happen. So I apologize in advance. Now, if we get any glitches, but so far so good, everything seems to have been working very well.

So this first panel is the future of autonomous driving. And we're really going to be looking at where the industry is going and how we need to create innovative quality software to impact that journey. So I'd like to firstly start off by just introducing our panelists, some of which you've already seen today in presenting and others may be new to you.

But firstly, by way of introduction, once again for myself, I'm Steve Harris. I'm the Best Practices Director here at CloudBees, and I'll be moderating the event. And we also have on the panel Buffi Gresh, who is the Area Vice President of Product Business Team at CloudBees.

Andreas Dharmawan, You've actually seen presented earlier. He's the Area Vice President of Software Delivery Transformation at CloudBees. Fernando Buck, who joins us from ZF, and he is the Team Leader for Continuous Integration in the area of Autonomous Driving.

So welcome. Welcome, Fernando. And then Rainer Schroeder from Google Cloud, who looks after the manufacturing and industrial segments, and he may be joined depending on whether Emil finishes his questions with Emil Siemes who is actually the Hybrid Cloud Specialist at Google Cloud.

So that's our panel for today. And obviously, if you want to ask questions, please feel free to do so. There is actually if you go into the auditorium and into the part of the the auditorium where the agenda is, there is actually a place there where you can post questions.

So please feel free to go and do that there, not in the general chat, not in the general chat that's on the front page because I won't get to see those, unfortunately. So I'll try and mix in those questions in with the questions, some questions that we have already pre prepared and we'll see how we get on. So we have 30 minutes for the panel and welcome once again, guys and thank you very much for your energetic participation.

So before we get going, one big question while I've got you all here because you're all experts in the car industry, when am I finally gonna be able to buy that car That drives me to work on my favorite next Netflix shows, update me on current affairs and makes me my morning coffee? I've been hearing about it for years and I need a prediction. So, Fernando.

I think your best place to give me that. What do you think? Yeah.

Let me watch my crystal ball and tell you. I mean, first of all, I think we will see that car. It will be like a better Internet device, let's say.

But maybe let me start with where we are at the moment. I think we already see the first, like commercial autonomous driving vehicles, but we see them in enclosed areas, and the key here is the ODD, the Operational Design Domain so we start off in places where we can control everything already and already have a real reliability we need. So, for example, in airports and company premises, and I think from there it will move and more into mixed operation and finally into that what what you are talking like really the level five autonomous driving everywhere.

I think there's still a long way to go. I personally don't expect to see that before 2030, but we will see more and more going into that. So I think there will be like an evolution, not the one day where it comes out of nothing, sort of.

And there was me thinking I was going to learn that they'll be available next year for me to buy. So that's been dashed now by your by your feedback. But, hey, it's really exciting.

It's a really exciting area. And obviously, it's not just about autonomous driving technology and cars these days is about connected vehicles, electrification of the powertrain, shared mobility. There's a whole bunch of stuff there that's coming into play that is all impacted by software.

In fact, McKinsey is saying that the electronics, the electronics and software market is going to grow seven percent compound every year over the next 10 years in the automotive space, while the car business itself is actually only going to grow three percent. So it's obviously an exciting area to invest in. And a lot of people and a lot of companies are looking to invest in those critical technologies.

So let me ask the first question then. So who delivers these technologies of the future is that the car manufacturers? Is it technology companies?

Is it the tier two, tier three suppliers or even a combination of all of them? Andreas, what do you think? Yeah.

In the past, innovation actually happened in certain regions of the world or a certain type of company or industry. Right. But thanks to the Internet, innovation is actually happening all around the world.

More importantly, tier one, tier two, tier three, they are working totally at different level of technologies and components. So I believe innovation is actually happening at different levels, oo it's a combination of all. So, you think it's going to be a combination of everybody playing their part and delivering those pieces?

That's correct. Yes, definitely. Do you think every theater has a certain focus area of expertize.

Right. And domain. Right.

And even innovation is still happening on Windows regulator in a car. Believe it or not. Right.

Even as simple as tail lights, there is so much innovation now because we are employing LCD with all of the different patterns of lighting, the different components of LCD. So innovation is not just happening at the car as a whole, but also at each component in the car. Absolutely, I think that's you know, that's an important fact that that everyone's got a role to play and it won't just be one particular company that is actually doing this.

So we see autonomous driving, as I say, is one of the exciting parts of this industry movement. But software seems to be the connecting factor for all of the different technology areas. How do you think that software delivery automation is going to play a part in that transition to support the development of the advanced software needed?

Maybe, Buffi, maybe you could have a stab at answering that. Yeah, I think just a key off of what Fernando said earlier, we don't know. Obviously, I think at some point autonomous capabilities will find their way into mainstream cars, but more importantly, high tech in general, like Andre, I said, is going to be mainstream.

So bottom line is, consumers want it all. You have the millennials that are coming. You know, they're they're graduating college.

And as we know, they you know, they are high techie lovers. They're going to become key consumers that autos are going to fight over. But even the previous generation expect a generous amount of high tech in everything they buy.

So we do see a trend where people love this whole retro styling, but it must be delivered in the highest tech fashion possible. So the key is consumers are becoming less loyal to brands and more loyal to innovation. So what happens is autonomous driving is just one of the many tech features that are going to be mainstream in the near future, and capabilities like feature flagging, canary testing will allow you to get those features out super quickly, roll it out and roll it back and even target certain people.

But so the answer is yes. Software delivery automation doesn't just play a part, it's actually essential, essential to getting your high tech. The very strategic by the sound of things and very, very foundational in terms of people's ability to produce the massive software that needs to be created.

Right. Exactly. So I assume that people's decisions about how they buy cars in the future are even changing then based on what you said.

Yeah, I mean, I think the industry in general has got the whammy, unlike other industries, it's very unique, it's like got the triple whammy. You have to digitally transform not only your product, but how you manufacture the product. So the product itself in autonomous driving and other capabilities, how you manufacture it, being really agile and being able to do custom custom building, but then also how you sell it.

It's like the triple whammy. So in that it's extremely unique industry that has to transform all three legs. Yeah, absolutely, I think.

I think it's gonna be it's gonna be exciting and there's gonna be a lot of change for the industry over the coming years. Given this is your area of expertize, Fernando, what do you consider to be the biggest challenges around creating reliable software for autonomous driving systems? Yeah, as we said, like the car of the future will be like more software system than ever and it will be incredibly complex.

And so I think that's the key word here. How do we deal with that complexity? That was never seen before.

We're developing one of the, if not the most complex software system that was ever developed. I saw a presentation, I think it was from Remo where they estimated this system, software system to be to have like in between one and 10 billion lines of code or something. I mean, no matter how big it will be in the end, it will be very, very complex.

It will consist of many parts developed by many different teams in different places. It will incorporate a lot of third party libraries. It will be incredible, incredibly hard to develop, integrate and test such a system.

On the same time. We need obviously need huge reliability. We cannot afford to have errors in that system.

So one way of coping with that is a very good DevOps practice is to have a working software system all the time and to only introduce small incremental changes into that system. So that, I think, is one of the keys to to cope with that. And I mean, yes, we have to become...

We have to be top notch in software development. And I think one of the things we saw in the last decades that that really transformed the whole software development is DevOps, agile methods, and finally continuous deployment and so I think that's the way to go. Absolutely, and I think, you know, Steven, met may I add something?

Yeah, of course. Fernando, I'm fully with you, but also I think we need to extend the whole process and DevOps area into the data part. Right.

Because building these machine learning, deep learning models involves these tons and tons of data, which together, combined with the software, will deliver the autonomous drive. So it's really also bringing these parts better together. Yeah, I think Emil, you are right on topic that my next point was really exactly that, that I think all of this software is going to come from a broad range of different sources.

It's not just gonna be stuff that's going to be built in-house. Andreas talked about how you have to deal with all of these different sources of technology and incorporate and integrate them together. I think that was very important in your presentation.

But Rainer, maybe from, you know, as you're one of the representing one of the companies that may be providing a large part of that software, what do you see companies like Google and Google's role being in that, in the whole process? Yeah, that's a great question. So companies like Google or Alphabet in whole, we have a lot of different areas where we contribute.

We've heard about Waymo, we've heard about Anthos even entering the cars. But there's still one other thing. Just look at autonomous driving.

This means constant training and adoption of the ever changing landscape out there. So this means if we have then these billions lines of code of the software being able to to maneuver a car through the real world, there might be changes just think about the E-Scooters that appeared in only three years ago around our cities. So typically driven by, sorry for that, drunken people who don't care about what what is allowed or not.

So you have to train against these new threats, perhaps in future its low flying drones, whatever, and training in these new threats might be best done in virtual environments rather than reality. So that's exactly what we are offering to automotive. We had before in our presentations is General Motors crew's example, testing 30,000 virtual drives every day within our Google Cloud and never a physical damage within the physical world.

So that might be one thing. One underestimates that a Cloud provider can can help all the industry, just virtual training of the ever changing landscape. Yeah, thanks, Rainer.

I think that is very important, that there is new capabilities that you're bringing that really nobody's ever thought about in the past. If we think about it, we think about the software production line, you know, the automotive industry basically introduced the world to the production line. It was the biggest sort of innovation that happened at the beginning of the 20th century, I guess.

But given that we're trying to do the same thing in the future software factory essentially, Fernando, maybe you could comment on what you think the current state of CI/CD is today for the automotive sector, is it well used? how do you see it advancing? I remember a presentation in the last year, Jenkins Embedded by Harald Göttlicher which will be in the next panel.

And the title was Embedded: Get out of bed. I agree on one hand. On the other hand, I think we have to see wider than this.

I think it's obviously much harder to to to apply all those practices from DevOps continuous integration, continuous deployments, especially in the area of embedded and even more so in the area of autonomous development. but I see even, since last year, I see things moving, I see things changing. Like nobody doubts that anymore that we have to do things differently, have to change, we have to to be more agile, more flexible.

They have to be faster, we have to reduce development cycle times. So I think we are on the way and they are making fast progress But I think that's one of the key factors, is to combine those modern software development practices with the good old, let's say, automotive grated system development. I don't think we have to throw everything overboard.

V. proved to be able to develop complex systems in a very reliable, predictable way, but now the question is, how do combine the new and the old world? Sort of.

And I think I am on the way, I think we are in a good way, but on the other hand, it won't get boring. So one of the chat from you, Daniel, thank you. Is he's asking whether or not the autonomous driving technology is being held back because of these constraints of implementing software as embedded solutions, essentially, and all the safety regulations and hardware constraints.

Do you see that, that if there was less of that, it would it would accelerate things faster? Probably, but on the other hand, we cannot just get rid of it, so... I mean, there's there's a reason why we have those regulations, those processes.

And, yeah, it's not easy, But I think we have to work on that. And then to have it. How do you see the you know, the architecture between the software and the electronics sort of changing?

Are they becoming more commoditized in the future or is it becoming even more complicated and more bespoke, shall we say? It's a good question. I think we will see that like the so-called the software defined vehicle in the future and software will become ever more important And so just like with our set of smartphones, we get updates over the air and so on.

I'm not sure what what's the consequences that will happen in the end on the architecture. But a lot will change. Definitely.

Yeah. I certainly hear a lot that there are new partnerships and cooperation models for software emerging in the industry, and that's creating opportunities for more standardization. I just want to add on that one of the things that I see in IOT and autonomous vehicles and robotics and drone, it's not just software running on a computer.

Right. Packaging is important because if you just use a standard hardware, then drones will not fly because the laptop will be too heavy to fly. Right.

So it's about miniaturization of the component and packaging. So this is why embedded software is becoming even more complex. You cannot just take the software practices only.

You also have to understand that the hardware best practices as well so we need actually a combination of both. Yeah, yeah, it makes a lot of sense. Anyone else got any thoughts on that, on that sort of general topic before before we sort of move on?

OK. Right. Well, let's change gears slightly.

Using that's a pun, of course. Let's talk a little bit about people. The people that create the software and make a difference.

And I think given that we all agree that, you know, we are requiring a lot of new technology, a lot of new techniques. We're going to have to change the skillset of our workforce to cover software engineering and cloud and big data and digital content, et cetera. How do you see that that's going to create challenges in terms of your company or other businesses hiring people with those skills?

That question to me. Well, Fernando, or anybody you know, is it a problem you see today or will it become increasingly a problem in the future? I mean, it's definitely already a problem today, I would say it's a huge challenge, as you mentioned, many new technologies, cloud technology, big data, things that are, the automotive world is not so familiar with, at least in the past.

They have to deal with that. As I mentioned before, I think the trick is to become things together. We need the automotive expertize.

We need those people there with automotive experience, but we also need software developers. I think most important is we need open minded people. The we need to go in new way where it's necessary and keep good practices there where they make sense and find a good balance on that.

Yes, I agree. I agree with that. I mean, I see in general trend in hiring.

We used to hire in depth, right. So someone who has real depth and not really look at the breadth and I see much of a shift towards less hiring of a resume, a more hiring of a person and their ability to have breadth and span other areas. I mean, even just the combination of Dev and DevSecOps and even some of the things you were talking about of the combination of software and hardware requires that talent have a little more ability to move to the left and right and recognize not just their silo.

And, you know, so even us as a company, we're recognizing that we want to holistically hire people who have ability to shift. So it's a different, different way to hire. Right.

Than just like that techie techie expert. And I want exactly to underline what you said. That's also a big shift at Google since in some years that we that we concentrate on hiring the right people with the right, what we call "Googleyness", that we believe they fit to us, that they are open minded.

And you would be amazed to see some of the move from those people through our company, from software developer to product manager, perhaps even sales then going back to software development. Yes, you need a certain set of skills, but I think the most important skill is to wish to be open minded, flexible and yeah engaged. Yeah, absolutely.

I would agree with all of those points. So thanks. I think I'm one of the questions that I know you and I have talked about before, Buffy, is, we're talking about how automation and software standards can actually lessen the burden of having to find such skills where you're having to then train people on bespoke technology.

What is your view about about how automation potentially might help here? Yeah, I think one of the things you and I talked about earlier was in the early 90s and even 2000s and even post 9/11, we had this real shift where disaster recovery was like this all encompassing importance. And now we have this other thing that's super important, business continuity planning.

And part of that needs to include the who and the where. So the talent, but also the location of that person. So jobs have to be able to be done without having dependency on employees location and software has to be able to be delivered even in the event that the person becomes unavailable.

So when we talk about kind of that that whole tribal knowledge and people say if you add automation, maybe you lose the tribal knowledge or you put the tribal knowledge into automation. I would rather say it that the automation allows you to publish tribal knowledge such that you lose the dependency on the individual. Super important.

And the other thing is when you automate those tasks so that tribal knowledge or the administrative overhead or even the delivery of software, what you're doing is you're shifting the developer from doing odd, from doing administrative tasks to innovating. So you're shifting your people from lower value tasks into higher value tasks. So automation, a lot of times people think that it means that people are expendable.

It's really just the opposite. It means people shouldn't be doing repeatable tasks. Right.

Let the computers do that and let you innovate. Right. That's how you're gonna move the company forward.

Super important. Yeah, I know. One of the senior directors, HSBC, I was talking to a few days ago, really emphasizes how important it is to keep developers happy because hiring them is hard and keeping them motivated is hard and making them do the right things is Obviously, the you know, the valuable things, as you say, is obviously the the emphasis that the management want to have.

So, so good, good point. So I have one final question because we're sort of getting close to the time now, again, which has come in from Louis. It's actually, I think in in it may even be in Spanish or Portuguese, I don't quite know, but I get the gist of it.

And really, I think it has something to do with, you know, how do you see, given the COVID situation? How do you see post COVID we're going to have to change the way that we work and the way that we create new software value? May I add my perspective on that, a lot of us, we grew up in times where we had to be in the office for doing your work.

So these were the times of the big workstations and the low bandwiths and so on, So you had to be onsite. And now the whole industry, everyone, not only, by the way, our digital industry, every industry is doing now a proof of concept. Are we able to say we don't care where the people are located?

We just look on what they are doing. So a real KPI driven experience, and this is great. And if we see on all the bad things that happening, the good thing is I truly believe that after COVID we will have 80 percent of the industry can say, well, I can also work at least part time at home and avoid the traffic jams.

That might be the good thing. Do you agree? Got any views on that, given it's so topical at the moment?

I know. I'm sorry. I know it's it's hard to get good engineers in Germany and I guess and other places, and that's a new chance of like, really, if we are able to develop distributed, it will open up totally new possibilities, too.

I think that's one very important aspect and the big chance. Great. OK.

Well, look, I think we're almost out of time, so I'd like to thank everybody for their contributions and their inputs. Thank you very much for being active and involved in the discussion today. I think from my point of view, I really want to summarize this as saying, look, software is coming and it's going to become increasingly a major part of what the automotive industry is doing.

And companies like CloudBss, these are in a position to help automate and deliver faster, higher quality software and reduce the risk of doing that. So thanks, everybody. I really appreciate your inputs.

And we'll draw this chapter close. Thank you. Thank you.

Bye bye. Thank you.