Exploring the Possibilities of Digital Twins & AI at the Edge

Tom Soderstrom:
Welcome to the Executive Insights podcast. My name is Tom Soderstrom and I'm an AWS enterprise strategist. Today we have Burkhard Boeckem with us. He is the CTO of Hexagon, a very interesting company, and we're going to talk about a trend called productive at the edge. Welcome, Burkhard.

Dr. Burkhard Boeckem:
Thanks for having me, Tom.

It's a pleasure to talk to you here. My name is Burkhard Boeckem and I'm Hexagon's CTO. My background is actually in engineering. I'm a surveyor by education and then I had the honor and pleasure to do my PhD at the ETH, the Federal Institute of Technology in Zurich. I started with a company that is called Leica Geosystems. And there I was responsible for bringing laser tracker technology into metrology, industrial metrology applications. And in 2005, Leica Geosystems got acquired by Hexagon. And then in 2020 I had the honor to become the group CTOs. And with this I'm overseeing the innovation of the group so that we can really make an impact with our products, with our solutions, with our software on the market.

Tom Soderstrom:
Hexagon is a really intriguing company and this trend right now, you're one of the world-class leaders in this. Where do you think that this industry is going? And would you explain what digital reality means for our viewers?

Dr. Burkhard Boeckem:
Digital reality is you have the physical world where we all live in and with our sensors and a lot of edge computing. At the edge in our sensors, we capture the real world, the physical world and transform them into very accurate digital twins. And then basically we take these digital twins, which I define as the exact replica of the physical world in all the dimensions. And then we can simulate things, we can design things, we can do as if scenarios, we can get insights. And this is where AI comes into play out of the data that we gathered at the edge. And then if we have the data, the designs, the simulations, all the what-if scenarios and we come to a design that we think is the best and can be the design of a bridge, of an infrastructure, of a mine even, it could be the plant, the manufacturing line or part or a phone, for example, a smartphone.

Then basically we take these sensors, load the data back to the edge and bring it into reality. And what is important, and I think where digital realities will go in the future is that this is a wheel, this is a flywheel or this is a feedback loop. There's something that needs to be closed. So basically it's the physical world is the real world into a digital reality. Some say digital twin, doing something with this and then bringing it out into reality. And I'm very optimistic in terms of innovation. I think these digital realities have the power to make the world a better place, to make the world at least work better.

Tom Soderstrom:
I think you're right, and I think it's important here to really distinguish that this is not the old metaverse where people get together and sit and have a party or they have a meeting. This is a place that's difficult to get to and it's like you said, it's an actual replica of the physical system. And my father always used to tell me, "Son, measure twice, cut once." Well, with what you're doing and what we see the future here with digital twins is you can now measure a hundred thousand times and deploy once. I think that's the thing that gets me so excited about this digital reality.

When we were at Hexagon Live, I remember you showed me how you can drive a truck through a city and now it has sensors that measures underground, measures inside the buildings. And after you've driven through it, you now have your digital reality of that physical space and now you can start changing the design and all that. That's pretty powerful.

Dr. Burkhard Boeckem:
It is, it is. And the things you can do with this I think are just amazing. I always think if you have a digital twin of something, a digital reality, and then you do simulation, you do designs, you use AI, generative AI even to find the best solution to a problem, make your mistakes, make your errors in the digital world and not in the real world because they're far, far less costly if you do them in a virtual environment and not in the real world.

Tom Soderstrom:
I think that's really key is in this culture of experimentation that we're seeing our executives be interested in is to limit the downside. And if you do it on the digital twin, there really is no downside. 

Dr. Burkhard Boeckem:
I think curiosity is a really big part and it's don't stop innovating. Be really curious how you can improve things, how it can make things better. Don't be afraid to find a very different solution because it's done before, it did not work, that doesn't mean that you cannot do it. And I think some of the products that I could help develop with teams during my career just were based on this principle. We just thought of a problem. We thought of something to be solved a little differently than has been done before. And also it helps if you have a great team of various diverse views and various backgrounds as you did for example in Hexagon, a product like the BLK360, the little scanner that quite frankly changed the industry of reality capture overnight. The team was composed of not domain specific people that came from the scanning background. It was game designers, it was people with an architecture background, people with a physics background that thought about a given problem in a very, very different way. And I think this is where innovation meets curiosity. And I think if you pair this with the commitment and the accountability that this needs to become an innovation, a product that really makes an impact, I think then you can go very far.

Tom Soderstrom:
I think you are echoing what we're hearing from other successful executives. In Amazon we call it don't fall in love with the solution, fall in love with the problem because once you fall in love with the solution, you stop. And as you said, don't stop innovating because things progress. So obviously you have a team that's excited to do that. 

How do you measure success? How do you measure return on investment and innovation? It's hard. I don't expect you to have the perfect answer, but what do you think?

Dr. Burkhard Boeckem:
No, I think it is very hard. I mean you can measure it of course with commercial success, you can measure it with revenue figures. I like to look at it also in terms of growth. Is there a market that is growing through your solutions, through your success? And I think then you really do the innovation right, but I also look at things that give purpose to the team. I like to work with very small teams where every individual in the team makes really difference. 

Also you can measure on how long do the people like to stay with you? Can they grow in their roles, but do they really find the purpose in what they do? For me also quite a measure of success is we have a quite large poster in our offices, if you come in-

Tom Soderstrom:
All your offices or in-

Dr. Burkhard Boeckem:
Yeah, it's pretty much the same, but we try to distribute them in the offices and it only lists the world's first.

Tom Soderstrom:
First?

Dr. Burkhard Boeckem:
The world's first. And I think it is a really strong feeling that a team gets and a really huge sense of achievement for a team. If you list it on this poster, you are around the one that  did the first world first, for example, laser tracker that could measure in six degree of freedom, the world's first digital level.

Tom Soderstrom:
That's great.

Dr. Burkhard Boeckem:
And I think the sense of accomplishment by being on that list is quite high.

Tom Soderstrom:
And also builds that pride in the team.

Dr. Burkhard Boeckem:
It does.

Tom Soderstrom:
How do you deal with failures? Because obviously if you're innovating all the time, some things don't pay out. How do you deal with that? Do you get smacked around or is that expected? How do you set the vision for that?

Dr. Burkhard Boeckem:
Failure is part of the process to be honest. I see that if teams have a track record that they hardly ever fail, they probably don't innovate. They just do incremental improvement, I would say. 

It's also the mindset, go as far as you can and then fail and then go a little bit back. You have to fail. But if you fail, fail fast.

Tom Soderstrom:
As I talk to enterprises across the world, the word failure is very loaded. So what I'm recommending is don't fail fast, experiment quickly. But some companies, failing fast is startups, that's what you do. What do you think about your company? Are you comfortable failing fast or do you need to experiment quickly? It's the same thing.

Dr. Burkhard Boeckem:
It's a good question. I never thought about this. I was more on the fail fast side at the time.

Tom Soderstrom:
Something to think about.

Dr. Burkhard Boeckem:
But as you mention it, I think the ability for a team to generate quite quickly ideas, try to test them, experiment, evaluate them, are they good, are they not good? And then come up with something else. Maybe something that is 180 degrees completely different than the first idea. I think that shows a little bit the quality in innovative teams. That said, there's a time for failure and there's a time for experimenting. And there's also a time, and this is also part of the DNA I wish to see in the team where if you are at a certain point in the project, then it's engineering par excellence, then it's something you, I like the term productize, then your idea is set. And then basically it's a quality also to come into the best execution of making a product. And this is also something that teams have to learn, teams have to get used to, and it normally takes two, three product life cycles until it's really ingrained in their DNA. And for me, maintaining the experiment fast or the fail fast culture, but having then after a certain point reach the equilibrium where we say, okay, now this is our best idea or this is the best idea we can come up with the foreseeable technology that we have available today. That is a huge benefit.

Tom Soderstrom:
I like that. It fits our model that we're seeing successful on the translating to the cloud. Think big. So you come up with all these big ideas, then start small. That's all those experiments and the ones that pay out, you scale fast. Now in the cloud, if you build a cloud native, it scales up and down. I wanted to ask you about a few examples. You and I have talked a lot about HxDR, but if you could explain what that is because our viewers probably don't know what it is.

Dr. Burkhard Boeckem:
It's a cloud platform that is made for digital realities. Digital realities contain petabytes of data in order to, as I said, to mirror the real physical world into a digital twin. And all of our sensors are gathering or capturing this data need to have a place in the cloud where basically to first of all store the data, then process the data, and then gets insights of data. And HxDR is all of this. It's a cloud platform for collaboration and almost equal as collaboration importance is visualization, be able to visualize your data. We are visual people, so a digital twin that is very abstract is not that approachable to a almost photorealistic digital twin. And then basically you want to have an almost infinite storage, you want to have AI pipelines that can basically run autonomously over the data of the digital twins. And yeah, we packed all this up in a cloud offering called HxDR, and HxDR stands for “Hexagon Digital Reality”.

Tom Soderstrom:
I think what gives this legs and hope is that it's built on the cloud so you can scale infinitely, more and more data is coming and you're able to deal with it, but you don't have to be the world's expert in data. You leave that to the cloud essentially. No, you're doing that well.

Dr. Burkhard Boeckem:
And there's also, Tom, there's a nice orchestration when we were talking about the productivity at the edge, I see all the sensors that capture the data. Now these are basically the sensors that have a lot of computing power, a lot of edge compute. And while you're still out doing the job, while you are in a plant, while you are in a field, while you are in an infrastructure project on site, you scan, you stake out, you capture the data.

Another benefit for the productivity at the edge is the small latency you get if you do it in the field, but it's almost instantaneously in the cloud. 

So when we look at this trend, productive at the edge or productive at the extreme edge, collect lots of data from the sensors, put it in the cloud, run whatever machine learning you need to at the day, and now you have a digital twin, so you can now experiment on the digital twin all the things we've said. For this to really work with all of this new data and sensors, I think we're going to need to be able to run AI at the edge. Are you seeing that too? And how are you preparing for that? 

We are investing currently a lot of research and development resources exactly into this because a lot of insights and decisions could already be gained in the field. I'll give you a very simple example. A product that is called a laser tracker is doing in very high accuracy and very high speed following a target, which is a retro-reflector. And normally these retro-reflectors are for the manufacturing world, are installed in little 1.5 inch steel balls.

And this is basically your reference of a target, could be at a wing, for example, if you do a wing to body maneuver with an airplane to get the wings in the right position. But for example, using edge compute here is the laser tracker always needs to follow these reflectors and they attach to the wing that is positioned to the fuselage, to the body.

On the shop floor for example, or in a big hall where these airplanes are produced, there are a lot of people and there are a lot of obstructions. Sometimes the laser tracker loses, it loses the target and using AI at the edge to predict where the target went. Was it handheld by somebody or was it on a robot? That basically then has a feedback loop within or at the edge that does not need to have a cloud connectivity to complete this task. 

And this is where basically where a lot of vision technology, vision-based AI is in the system. And then also I think using smartphones in our everyday life and social media. What current smartphones bring out for image quality is quite incredible. However, everyone also looks with these eyes towards imaging sensors that are maybe just used for computer vision in sensors at the edge. So also using here something called computational photography in the system that use neural networks in order to improve the image quality beyond what is currently accessible with the lenses and sensors I think is another nice example for AI at the edge.

Tom Soderstrom:
Yeah, I think that's really good. So  you started as a hardware company. Are you now a software company or are you a software and hardware and how do you see the progression?

Dr. Burkhard Boeckem:
If I look at our people in R&D, even if we looked at predominantly hardware, 75% of our engineers are software engineers.

Tom Soderstrom:
That's what we're seeing from across the world, that the hardware companies have to learn software.

Dr. Burkhard Boeckem:
I see a trend going back that you have a lot of advantages in making the perfect solution in software, in cloud architecture, in productivity edge, if you built your own hardware. If you tailor built your own hardware, that doesn't mean being exclusive to yourself and exclude anyone else from the data, but it allows you to build the solution a little better, to build it a little faster and also to decide constantly where you want to have more compute. Is it at the edge or on the cloud or somewhere in between?

Tom Soderstrom:
Good point. We saw the same thing at NASA when we started by being scrappy and prototyping and experimenting with the hardware and software, you dare to change it and you can break it and you can try it again. Then once you perfect it, you can figure out how to get it manufactured. But it's a good point. So to wrap up, three things that you recommend to other executives.

Dr. Burkhard Boeckem:
First of all, I would say lead by examples. People look at you and it starts with you, so lead by examples and if you're curious, I think probably your organization will be curious. If you're innovative, then people will follow you. Then basically in today's technology world, nobody can do everything by himself or herself or themselves. So it's important to build great teams, build great and diverse teams to allow as many perspectives as possible and collaborate, collaborate, collaborate. We spend a lot of time in our jobs and work, so it needs to be fun, it needs to be a purpose with them. And yeah, stay curious as day one in your job at the beginning of your career and if you can maintain this curiosity, the fun and the pleasure to go to work every single day, then I think that's something to start with.

Tom Soderstrom:
Great advice Burkhard, and obviously you've been very successful and will continue to be even more so. I want thank you for these thoughts and for participating in the Executive Insights.

Dr. Burkhard Boeckem:
Thank you so much. Thank you, Tom, for having me. 

Listen now

Listen now

Listen now