In this podcast episode, Angelo continues his discussion with Myk on the use of artificial intelligence in building materials, focusing on AI Timber. The conversation explores the potential of AI in creating new materials and the benefits of assembling building components off-site in factories, which can improve quality and reduce construction time. Myk highlights the importance of technologies like augmented reality (AR) and digital twins in modern construction. Digital twins, which provide a real-time virtual model of physical objects, can enhance efficiency by tracking components from production to assembly. This approach is compared to the aerospace industry, where parts are manufactured in different locations but assembled seamlessly. The use of digital twins extends beyond construction, aiding in building management and sustainability by optimizing energy usage and facilitating future recycling or upcycling of materials. The episode also discusses mass customization, which allows for the creation of unique building components at scale using CNC machines. AR is emphasized as a crucial tool for on-site assembly, with applications ranging from remote assistance to fully automated guidance systems. This technology can significantly improve the speed, accuracy, and safety of construction projects.
Host: Angelo Kastroulis
https://www.linkedin.com/in/angelok1/
https://twitter.com/AngeloKastr
www.hermes-nexus.com
www.carrera.io
Guest: Mykola (Myk) Murashko
https://www.linkedin.com/in/mykola-murashko-041b47156/
https://www.maestro-tech.com/
AI Timber Pt 2
Angelo: Welcome to part two of the topic of using artificial intelligence in our
building materials. This is going to continue last episode where we talked about
AI Timber. I'm your host Angelo Kastroulis and this is Counting Sand.
So going back a little bit. So we talked a little bit about the details, I think, of
how to make new materials with AI timber, and that is a great idea and very
fascinating. And I'm sure over time, the constraints of the market forces will
drive costs down, make it easier and cheaper to do.
And even finding new opportunities for some things that were. More expensive
to build structurally. Now because of its structural nature, once that gets figured
out and easy to predict, can now be different. You know, we use, instead of
laminated beams, we might be using something like this. So there may be
different applications for things that may be discovered.
But, backing up just one second to some of the other things you talked about
that I want to make sure that I understand the computer science behind them.
Being able to build components to assemble them on site is a really interesting
idea. I mean, we get that a lot of times with, in the United States, with truss
construction. Trusses are built in a factory, effectively, delivered and then
installed.
We don't build trusses on site, but the rest of the building pretty much gets built
on site. And so, this idea that you could kind of design the building and it comes
out, you get it in factory, send it to onsite. And then you use, you mentioned
several pieces of technology, you mentioned AR, you mentioned digital twin.
Let's talk about of those just for a second, because listeners might not
immediately get this connection. That could change the entire way that we look
at construction in terms of quality is much higher when something's built in a
factory, but also it's much easier and faster to put together a building.
I think you had mentioned this that instead of months you're talking weeks to be
able to put together a building. That’s very interesting because it's not ideal
conditions to fabricate onsite.
Guest: You know, if any of the audience have ever been on a construction site,
especially in January or February in the Northern Hemisphere and walked
around you would realize that this is not the place where you make Swisswatches, right? This is a place where the tolerances are not in millimeters, but
rather are in centimeters or inches.
And as a result, you know, you have lots of scope for mistakes. Mistakes equal
delays and rework. And in the end, the whole package comes out costing more
than expected and later than expected. And so we were trying to target this and
we'll get to digital twins in a second.
But maybe if we follow through the logic behind how we got to digital twins as
a solution. The first place we started looking at is, you know, buildings are not
the only things that we physically create in the world, right? We make planes,
we make cars, and we seem to be better at making planes and cars than we do at
buildings, right?
If you look at Boeing or Airbus, the way that they construct the plane being a
much more complex machine than a building is in factories across the world. In
the case of Airbus, its engine is made by Rolls Royce in the UK, gets shipped to
Spain where it gets assembled with a wing that's made in France, and you know,
all of these individual components, the engine, the wing, the body, they're
complex, they have their own expertise that's required to create them.
But in the end, the final assembly comes together quite easily within a matter of
weeks, right? And the first kind of question that we posed to ourselves is, well,
how can the building site look a bit more like a plane assembly. And here, kind
of the first moment was, well, we would need to manufacture parts in factories,
right?
But the building is very different from the plane in that each building is unique.
Nobody wants to live in a cookie cutter home. Homes across the planet have to
kind of negotiate different circumstances, both environmentally, but also
culturally. And so each one is kind of unique.
And here is where the first piece of technology comes in, which is mass
customization, right? The fact that a CNC machine that runs on code is able to
produce two different pieces of steel from the same sheet of metal, which can
perform two different tasks, but the cost of producing either one is mostly the
same, because the machine is quite agnostic to the final geometry of the shape.
If it uses the same process, it can have the same cost, right? So, the first, kind
of, thought of, here is something that can unlock scale of building this way is
the fact that mass customization has become so commonplace with CNC
machines kind of found in basically every machine shop today.You can think about scaling that up to the scale of construction. And then the
second question comes, well, okay, I can produce different components that are
all customized for my different buildings, but I need to have a degree of control
over that because if you think about the scale of a building you have tens of
thousands of component parts, each one might be different.
And here is where digital twins come in, because you need to be able to control
not only the design of these parts. The design of these parts, we can control
through a 3D model through something much simpler, like a building
information model. But in order to actually get these things made, you need to
go from your BIM, your building information model, to instructions for the
machine.
Then you need to be able to track the physical component part in the real world,
in the factory. You need to be able to link it to your digital twin of that part and
track it as it gets produced, as it gets transported, and as it gets assembled. And
this is really where the digital twin, as a concept, is the thing that can drive
innovation and construction.
Because if you imagine that tomorrow, with, you know, a bit of software, you
can take an architect's model and you can transform it into something that can
track components, that can produce components and that then can help workers
on site assemble those components, you know, it doesn't take much to imagine a
completely different future of the way that we built.
Angelo: That's amazing. So that's key. We hear the word digital twin all the
time now. It's used very often. Companies are looking for digital twins to digital
twin everything. And we might be tempted to think that the digital twin is the
design, is the CAD or whatever. That is not the digital twin of the building.
That is just the reference design. The digital twin, in order to be a digital twin,
you have to account for the information of its entire life cycle. Whatever you
consider the life cycle, if your life cycle is through the construction, you would
create a digital twin that would be able to mimic that building all the way
through its construction.
Alternatively, you know, there could be more room for future digital twins, for
example, what happens to the building in 20 or 30 years, right? How do you
maintain it? Those are all other digital twins that could be useful. But what
we're talking about here is a digital twin through construction, right?So, some of the information you mentioned, was IOT contributes to this digital
twin. Because we need the data, we need to know how this works.. How does
that contribute to the digital twin, the kind of putting it together and making it a
reality?
Guest: Yeah. So we've been testing different approaches from RFID and other
kind of tags on our parts.
And in the end simplest way we found that works pretty well is each one of the
component parts. Imagine steel, timber, glass, gets tagged in our partner
companies with a simple QR code. That QR code is scanned the moment it
leaves from the factory and the moment it arrives on site.
It's effectively inventory management on the scale of a construction process. So
what this means is, first of all, it feeds information into the digital twin about
what has arrived on site and it allows us to program what needs to arrive next
because the next step from the part arriving on site is the site worker, the install
crew, scanning the component part, and getting a live visualization of how that
needs to be assembled into the building.
And this is really where the digital twin is very different from a 3D model of a
building. The 3D model is static. Whereas the digital twin is a representation of
a process, right? In this case, we're talking about the building process. We can
then in a second talk about the building management process once it's complete.
But in the case of the building process, you know, the digital twin helps us kind
of do production. So that means translate things to CAD/CAM, to computer-
aided design, to computer-aided manufacture. Get code out into the machine.
Have physical components produced, tag those components track them through
the life cycle of getting them from production to site, and then use the
information in the physical component to help the worker assemble it in the
correct way.
In the end just like before, kind of when we break it down to this it's quite
rudimentary, right? Like, all you need is a laser cutter. And you need a printer
for a QR code and a camera on your mobile device in order to scan it. And that's
the simplest version of how digital twins can power construction.
But even this being deployed on some live projects right now, we're seeing huge
difference. We're building a small construction in Palm Beach, which is a
building that connects two existing buildings, and it's made from CLT panels,
and those CLT panels have arrived with these QR codes that enable you seewhich one goes where and in what sequence. A nd it basically enables us to have
a crew of workers that are experienced people that have done construction for
years and years, but have never seen a CLT panel or have never assembled CLT,
be able to do this flawlessly because in the end it's as good as putting together a
Lego, right?
Imagine if you had Lego, but with live instructions for each block. That is the
kind of future that digital twins can enable. Then another kind of interesting
moment of research is what happens when the digital twin is used to help the
building during its occupancy, right?
Because as we said at the very beginning, the sustainability question and
construction can be broken down into two groups, embodied carbon, so
everything to do with the carbon that goes inside of the building in order to get
it built, and the carbon footprint of operating the building, so heating, cooling
the building during its lifespan. And the digital twin can then be used to help the
second group of questions as well by tracking occupancy.
So imagine if you are in an office building tracking the presence of people in
different rooms and switching on heating and cooling according to that. And it's
something that my partner and co-founder of the firm, Carlo, has done 10 years
ago. The pioneers of doing this in an office building in Italy for the Agnelli
Foundation.
Basically tracking occupancy in the office building and using that to control
thermostats in the building. And as a result, you know, they made a double digit
percentage saving in the energy consumption of that building. So I think digital
twins as a topic is something that can power industry, not just in construction,
but you know, it’s used widely in manufacturing and I'm sure others and, you
know, I think it's really the string that can tie together these different
technologies and that can tie together the physical and the digital world.
Angelo: Yeah. So in this case, we could use a digital twin to assemble the parts,
make sure the supply chain is sending us the right things, we're putting together
in the right order in the factory.
We're cutting the parts in the right way. But you can see that there could be
many digital twins depending on perspective, right? An energy company may
make a digital twin of the building to try to optimize efficiency, to try to figure
out how can we build this based on how the house is facing and what the
property is like and what's the climate conditions and optimize, because as we
know, small things like the direction the building faces and where the windowsare and all of that makes a huge impact on the efficiency of the building itself
throughout its whole lifetime.
And so we can use tools like digital twins and simulations to try to produce
optimal buildings, not just during construction, but throughout its entire life
cycle. Even managing them, how do we keep it in good shape for less cost?
Because those are resources we're using continually to fix things in the building,
right?
Guest: Yeah, and about kind of, the lifespan of the building and its occupancy
and then its end of life. This is another interesting moment where thinking of a
building as an assembly of these complex parts means that at the end of life, you
know, 70 years after you build or 150 years after you build, when you need a
different function on the site, or maybe the building has degraded, it's very easy
to disassemble those component parts, know what the component materials are,
right?
So thinking about recycling and upcycling the building, the digital twin can
make that much simpler. But another thing that you touched upon, which is very
important, is simulation, right? Because as we said, there are two moments in
time when you can make a difference in construction.
The first moment is design and the second moment is when you translate that
design into physical parts. And simulation of buildings that are built using
digital tools is something that's super important because what we have to do is
effectively take an architectural intent as an input and translate that into a set of
machined components.
And this process of translating an architectural intent into instructions for the
machine is one that requires a lot of design work, but also a lot of optimization
and, you know, at this stage, a lot of the optimization we're doing is manual, but
we're looking into injecting AI into making that work more efficiently.
One of the big, kind of, metrics that we're trying to optimize for is what we call
constructability. So that means, what does this component look like? How
difficult is it to manufacture? And then how difficult is it to bolt together into
the final building part? And this is something where we're experimenting.
We're using some degree of artificial intelligence in order to assess how
constructible is a part. You know, it can start from something as simple as, here
is this sheet of metal that needs to be bent to create a custom bracket for my
structural component. How many bends are currently in this design?What kind of angles are they at? Is this something that I can do automatically?
What kind of machinery does this require? Do I have this machinery in my
supplier network? And a lot of these questions can be answered through the
digital twin if you feed it the right data. So this is sort of what we are working
on right now is making this kind of comprehensive layer on top of our modeling
environment, such that our designers can basically use all of this information
that can stored based on the physical places in the world where these
components are produced and use it to inform this design for manufacture
process.
You know, and it's yet another application in this building life cycle of how
digital twins can make that work.
Angelo: I imagine it'd be useful also in the design for assembly part of it, where
you'd have a component and it could work perfectly, you can build it, but then
you might require a crane for one part, and there could be another way to build
this to design it where you don't need a crane, and it can be entirely done with
the resources and the team there so you have to think about assembly as well,
and maybe some of the AR, the augmented reality stuff you were talking about
comes into play there, because I imagine the team gets all of this material, and
even though there's QR codes and instructions, you know, some of it's probably
pretty big and there's a lot of trial and error, you know, to kind of try to figure
out, even though you have instructions, sometimes it's hard to know. Is that why
this idea of augmenting reality was born?
Guest: Yeah, absolutely. So the first buildings we're building right now are
using kind of more static ways of describing the assembly instructions. The
worker scans the QR code, they get an animation of how they need to bolt two
pieces together, but in trying to make this faster and more efficient.
We started looking towards the automotive industry to see how they use AR to
both assemble and service vehicles. And there are kind of two different
applications that we think are crucial. The first one is the simpler of the two,
which is Volkswagen during COVID got Microsoft HoloLens for its mechanics
across the world, such that when they encounter a problem with a car that they
don't know how to fix they get the engineer sitting back in Germany to look
through their goggles.
Look at the problem and then display on the screen, the kind of solution that
one might need, right? This is quite a manual process because you need to have
a person on the loop, but having this kind of visual interface where you can
sketch and draw on the space in front of you makes a huge difference.And I think this was the first kind of moment where we thought, you know, AR
can be a game changer because if you have inexpensive goggles on a foreman
then the engineer can provide input live in a way that the traditional
construction site would require, you know, site visits and inspections and so on.
Most of these things can be solved with a quick sketch and if that sketch is done
live in front of somebody's eyes, that’s even better than being on site with a
piece of paper. So this was the first kind of moment that we are testing out on
our construction sites using AR specifically for this.
Then the second more, kind of, technologically advanced version of this is
taking the engineer or the technician out of the loop and letting the digital twin
do this instruction itself so that would require awareness of where the operator
is in the building, awareness of what is the current progress of the building site.
And, you know, overlaying on top of the real world, the information about what
is the next component to bolt? How does it come together? What are the pieces
that you need? What is the equipment that you need? And I think that this can
be kind of a really powerful tool to yet again, make construction faster and more
efficient and safer.
And there are a number of existing companies that are looking at this, that are
focused just on AR in construction. And I think that's almost an industry in itself
because there the same technology can be applied across, I think, all of
manufacturing and all of assembly not just in construction.
So, right now we're looking at the first of these two scenarios, but I think AR is
going to be huge in the assembly of everything, you know, from your phone
down to your bridge, building, airplane. Because the moment that we can
overlay for the human operator, the digital information with the physical world,
you completely change the game in terms of what kind of skill (and) experience
the person needs to have. (It) makes learning on the job much, much faster.
Angelo: Well, I couldn't agree more. That's a very fascinating and interesting
approach to AR. And I just want to thank you for joining us today. We've talked
about a lot of different things. You guys are truly working on some cutting edge
stuff and you're on the edge of technology and computer science in
construction.
But also I appreciate the fact that it's practical, that it isn't just trying to kind of
revolutionize and upset everything. It's trying to actually get it adopted and
move forward. So thank you for being with us today and looking forward toyour future stuff that you guys invent. And then maybe we'll talk about those
one day.
Guest: Thank you, Angelo. It's a pleasure talking.
Angelo: And that's our two part series on this topic. As always, thank you to our
listeners. We couldn't do this without you. Please subscribe to the podcasts
available on your favorite platform, and you can follow us on LinkedIn at
AngeloK1 or AngeloKastr on X. I'm your host, Angelo Kastroulis and this is the
Counting Sand Podcast.