50: Robots in Construction

We, you know, at CES in January, I saw robot orchestration for the first time. This is a real important new new product, not even product category, a service category to run manufacturing. And this is real important for making America manufacture again. Not that we ever stop, but we want to bring more and more manufacturing here to America and to the western world. And this is going to be interest one, talking about how robots are being run inside factories and other places. Who are you?

My name is Thomas Sherwood. I'm the CEO, co founder of Civ Robotics. We're a construction tech startup based in San Francisco and glad to be here, Robert. Thank you for having me.

Yeah, I should come over and see you. You're only an hour away. You know, did I lay that up right? That you're doing robot orchestrate? Well, tell us a little bit about what you're doing. Let's go to 30,000ft.

So high level. We are automating construction surveying. Majority of our customers use our robots in infrastructure project like solar farms and data centers and road construction and ports. In essence, what our robots do and perform is markings. So we mark the blueprint in real scale outdoors in construction sites. So it's an all terrain machine that drives around the job site and points on the ground coordinates based on the blueprint where to build, if it's a foundation of the structure or a stop sign or a curb or a solar array that is being marked.

And what's the breakthrough here? Why do we need this at this time?

So surveying is being done mainly for a long time. It's a very repetitive task and it's required at every job site pretty much. Construction faces huge challenge when it comes to labor shortage. And in order to get the job done, we need to find talent. Skilled labor in particular is even worse. So our product is basically empowering surveying professionals to get the job done about four to eight times faster, depending how you measure it and without much skills need to be done like any can operate by a machine. It's very straightforward. We've trained customers in less than an hour sometimes how to run the machines and perform much faster than people who've been doing this for years.

Yeah, they used to use sextants to figure out the construction site. This does it real really fast. What does the machine look like? Do you, do you have a video or picture of one or.

I have one right here too.

Okay, great. Yeah. What is that?

And a bunch over there.

All right, so this thing rolls around. What is it? Does it have a bunch of lidars and cameras on?

Has A survey grade GPS technology called GNSS or RTK that provides the positioning of the machine within less than half an inch. So it's within like a third of an inch positioning. We are also using an IMU sensor for like, for knowing the terrain conditions. Because if we are driving on a slope, we need to mark on the slope also. So we are need to be aware of this conditions. We do have some obstacle detection avoidance in some of our machines, but most our customers display disable it because it's a cobot. So there's still a person involved in the loop and therefore we do not really need the fancy lidars or cameras on this machine in particular.

Okay. And at a construction site, do you roll it around in a grid? Does it automatically map out? How do you set it up and use it?

Great question, Robert. We keep it very simple. We take the same file that they use today manually by surveyors, select the area on the map, literally just draw a polygon, like four buttons or six, whatever the shape of the area you want to mark and press start. That's all you have to do. Follow the robot. Either we'll spray paint on the ground the markings or if you want to be really, really, really precise, it will shoot a laser pointer on the ground, as you can see on our website, and will verbally tell the person, mark. Really very straightforward, simple operation. Just follow the machine and do what it says.

It knows where it is in 3D space really well and it helps mark out a job site. So you get started in the correct place at the correct angles and all that.

Exactly, sir.

What's the hardest part of building this robot?

Ruggedness, I would say. You know, today we stand here, you know, we've been at it for over five years. Deploy the first product in the field in 2021 and keep on iterating and improving. You know, construction sites outdoors in the desert, in the mud is rough and these machine needs to survive and mark thousands of points. We have a machine that marked over half a million points and it came back. It's actually the one behind me. It's one of our oldies and we keep it here as memoir of how much we've come and we've continued to iterate and improve our product to make sure that it's robust and can handle the different challenges that we'll face in the field.

Yeah. How much time does this thing save? You know, over the old way of doing it.

So I'll give you an example. There was one job site they used for over a year before us, they had 18 people doing construction marking. 18, wow. With us three.

Okay. So that saves you a lot of labor costs. And how much does this robot cost?

So depending on the model and if you choose to buy or rent, give or take, I would say eight grand a month.

Yeah.

And that alone is actually equivalent as if you are replacing the manual handheld surveying equipment. So there's cost savings, of course, of labor, but actually balance out equipment to equipment. So the main equipment, if you need three of those or four of those, it equals to like using one of our robots. Wow.

Okay. So it's very efficient at saving you money. Where is AI used in this? Is it just for navigating it around or is it actually doing some extra work?

So there's multiple layers of using AI in this product. Most of it is in being smart and how it's not mapping, but more like how it's planning its mission. You know, you could dump the machine wherever you want on the job site. It will start marking and you don't have to do much. Say, okay, I want to mark this area. Go market. And even determining. In some cases you can set it up that way. That. Let's say you need to mark a fence.

Yeah.

And a pile. A fence does not need to be that precise. That's. It could be within an inch, but a pile has to be within a third of an inch. So we can able to. Based on the layers the AI could use to identify and mark the different accuracies, that will affect the output as well.

Yeah. Why did you get into this? You know, because you're a smart dude. You could build robots of all sorts of different kinds. But you know, I saw a robotic lamp yesterday. Right.

I'm not a roboticist by trade. I'm actually construction guy. So I worked in construction before starting this venture, and I spent on one of my job sites about $4 million on surveying services. So I spent $4 million people walking around marking coordinates and measuring. I'm like, this is silly. There has to be a better way. And actually use a different product. Basically a blueprint on an app. So instead of having the big piece of paper. This is awesome. But I am out in the field, acres of land in front of me. What does this map help me? I want to. I want to select here and see in the real world. And that's how everything came together.

Yeah. For people who don't know the construction business, why. Why do you need to do all this surveilling at the job site? You know, before you build a skyscraper or something like that.

So I'll give you an example. The designers have a skyscraper has foundation, let's say. I'll give you an example. Here in the Bay Area, we have a, you know, the, the Chase Center 10,000 foundation points designed in a particular way and in order to, you know, really support the structure as designed, but to build it accordingly in the real world. So we take that design and bring it to the reality and marking it one by one. Because they give it in a steel structure, there's not much wiggle to room. If the points are 10ft apart, that beam is 10ft, maybe 10 foot and 2 inches. So you don't have much space for error. And in order to connect the beam, your piles have to be bullseye. So it's very important to build according to plan. Otherwise you can go and do reworks and that's pain in the butts. And also, you know, someone design structure in a certain way, taking all the engineering background and you build it differently. You don't want that risk on you.

Yeah. What have you learned by doing and building this kind of company? Because this isn't the usual kind of company I'm talking to.

So I've learned that many things along the routes. But I think the most interesting thing that I've learned is that we're working with people and people have, are cautious sometimes when it comes to robotics and automation. And it's all about how you phrase what you're providing and, you know, try to focus on the values that you're creating for those individuals that you talk to. Because if I speak to an operator or a superintendent or project manager or procurement, it's a very different discussion. So how do you tailor the discussion to the individual that you are working with or communicating with to touch their pain points and make sure that we address them and show them the value of using our technology.

Are you hitting the same kind of resistance that others are worrying about jobs going away? Because you laid it out. 18 people used to do a thing, now three people do. Right? Word gets around pretty quick. Oh, this little box is going to replace my job.

So there's two things I'll point out about that. One, the construction market has a huge shorter labor talking numbers. Over half a million people. Jobs open, available buy two. For every three surveyors that retire, only one joins the force. How do you mitigate that gap? And projects are getting bigger and bigger while the workforce is getting smaller. We're not trying to replace. We're trying to empower the talent that we do have. To get the job done faster. And the one who were supposed to do it should do things that are more smart.

Yeah.

More sophisticated. Because if our machine could do it really, really well, really efficiently, that person could do a task that robots can't do, can perform. And so use that workforce to do tasks that are. That are harder in reality.

Yeah. Does this increase the accuracy over doing it the human way? Because there's. Where there's humans. I know when I drive my car, I make mistakes. The car doesn't make mistakes anymore. Right.

So it is more reliable. It's one of our core values, being reliable, being simple, being efficient. And it's reliable because we're literally squeezing all the specs from the technology that we use. So we are using the same surveying equipment technology that you have in a handheld device on the robot. And the robotic arm can move within 0.1 millimeters. A person holding a pole seven foot tall, looking at a bubble or a screen, telling him how much more left, how much more right. That's. And we're using the same data sets is less precise and less, you know, you don't know how many points they'll do. One day, one person, one crew can do, let's say, 300 points. The next, we can do 500 points. The next 250. Our machine does between two to three points a minute, depending on your accuracy requirements and spacing consistently. And knowing how to project your outputs and knowing how many resources you have to allocate to perform a task is also valuable to do risk project delays, which is the biggest problem we have in construction, I would say.

Yeah, what, what's going on in San Francisco from your point of view, because you're in the robotics industry there, there's a lot of different kinds of robotics that's going on there, you know, or within 60 miles of San Francisco, you know, from landscaping robots to humanoid robots. But I'm sure you taught you, you get together at meetings and conferences for robotics, but what's going on?

So first I'll take a step back. What is a robot is way more a robot?

Is Waymo a robot?

Yeah.

To me, my Tesla is a robot. My little vacuuming matic robot is a robot.

The Tesla, before you had the autonomy of Stack, it was not a robot.

That was a car. Back and forth, right from robot mode to carbo. Not soon there will be one with no steering wheel. That's a total robot, you know, no humans, no human in the loop.

So I would say that there's different robotics solutions out there. One Is even just more softer oriented. One is more driving oriented or automobile. What is more in agriculture? What is more in arm picking. There's a lot of different robotic solutions out there. You mentioned one of them like mowing and, or cleaning or. I've seen also solar panel cleaning robots out there. There's a lot of robots even cleaning walls and inspection robot is a drone. A robot also.

Yeah, I, I count it as it. I put drone companies on my robotics list. Yeah. Because they're, they're, they're flying robots, right?

Yeah. So there's a lot of robotics systems out here. Actually we're, I'm part of a few meetup groups that discuss robotics and hardware, share your knowledge and experience and network. You know, we are, you know, many of us don't compete with one other at all and share the knowledge and share the experience that we have. And it is a nice community out here in San Francisco area for robotics and hardware in general.

Yeah. And the guy who runs factory AI has taken on Waymo to work for a year and a half. Waymo's not even on my street and it was invented here.

You know.

We saw in the early days I think of robotics in America.

Yeah. I think San Francisco in general is a great hub for robotics and AI autonomy in the different industries from. You mentioned some of them, you know, from operation maintenance, construction, agriculture and many more that we haven't even spoke about from manufacturing. It's not, you know, robotics has. Robots are huge part of warehouse automation and manufacturing.

How did you convince a venture capitalist or somebody to give you capital to start this company? Because I don't think many of them are in the construction business. And so they're like what did you do a lot of education about this business, how big it is.

Yeah. So one of our first investors in our pre seed was actually a construction company. So they saw that they understood the value and they know how much to spend on surveying services. So it's very clear in our seed around we got some call it real veteran money. And construction as a whole is an industry that needs disruption. So they saw the opportunity in disruption the construction industry and our bolder vision of automating. Not just surveying, but also construction tasks as itself. And I would say there's no point to automate a task if you're not doing it. Precisely. We got the precision nailed down. Now we're working towards more autonomy construction solutions out there.

And give me an idea of what, what's next for your company. What, what's, what's the impulse here?

Yeah. So I'll give you a two, two, two parts of it. In our cedar route, Trimble invested in us. Trimble is a large surveying equipment manufacturer and also machine control company based also here in the Bay Area with headquarters in Denver. And in our last round, on top of the venture money we have from Ali corporate, other of our strategic investors, Bobcat invested in us. And as you can imagine, that's where we're heading. We're heading towards automating larger machines like Bobcats to perform tasks without or with less human intervention. So instead of having an operator sitting in the cabin, you have an operator sitting in a truck overseeing multiple machines working in tandem.

That's an exciting future. What does this mean for manufacturing and construction in America? I'm seeing that there's a huge amount of pressure to get buildings built fast. Right. Elon just built a huge data center in 19 days. You know, that's incred incredible what's going on in construction. Where do you think this is going for construction and manufacturing in America?

I think construction is a whole depending on what you're building, there's value of getting the job done sooner. A data center like you mentioned, I believe for every day that you finish earlier from contracts, sometimes I've seen it's a million dollars. So get if you could shave it a day, doesn't matter how they cost you less than a million dollars, you should make that investment or do risk delays as well. So there's a huge incentive contract wise to get the job done on time, if not sooner. Especially with assets like data centers or solo farms that produce millions of dollars by selling the energy to the grid, clean energy. So Silverbox is helping our industry to get there sooner. And you know, being able to build a Data center in 19 days, Elon definitely used more autonomy and robotics there.

Yeah, we're seeing a lot of communities get burned or destroyed because of climate change. Is this real important for rebuilding those communities and them back up, up and running?

So we haven't had much many customers using our machines for the housing mostly for infrastructure. If you're building a house, you're maybe marking 10, 20 points. That's not even six minutes for our robot. So they will really use a machine to mark six points, 20 points a day. Yeah, we're being used to deploy on large assets like solar farms. You know, I mentioned earlier we had a job that had about 300,000 points, 300,000 points marked across. I think it was like 5,000 acres.

Wow.

Crazy wow.

Is this a good way to watch subsidience you know, ground going down because water is being pumped out of the ground. That that's happened in my house. My house is probably about 6ft lower than when it was built. Right. And you can see that crack. Cracks the roads and changes the terrain a little bit. Is this something that I'm going to see a city surveyor rolling one of these around?

What you would see is probably construction companies are using to rebuild some of these infrastructure assets that are. That are suffering from challenges. Like you mentioned, the soils going down due to drainage or other damages you may have seen in the cause. I would not per se say it's climate change. I would just say it just. We live in a world in America that needs a pump of capital to build infrastructure. Again, if it's highways, trains, whatnot.

Yeah. Here in my neighborhood, we live on an earthquake fault too. So there's a lot of little USGS markers that somebody surveyed and put down in the concrete. Could this machine be rolled around the streets to watch how earthquakes are? The fault lines are slipping or anything like that? Or is there, in other words, is there other uses other than just building skyscrapers and solar farms and stuff?

So there's one more use case that we see customers use a machine is. Is it's called topographic measurements. Yeah. Basically, the robot will drive on its own and just collect data of mostly positioning X, Y and z. So if there's a. If a surface that went down, it will be able to detect that and monitor that progress. Our world per se is not doing much of that work. We have two of our customers use the topo feature for checking the work of civil contractors on their job sites.

Yeah.

But not as much for operation maintenance. Yeah.

Wait. For people who might have seen a construction site, you see a lot of little flags, and sometimes you see this gray paint. Is that what. What you're laying down? And what do those flags mean?

Robert? I think it's best I just show a quick video.

Okay.

That be cool.

You could do that, or you could send me the video and I can cut it in right now. Or. Or just show it to me. Either way, it works awesome. It's not like cuc me days back in the 1990s where nothing worked.

So, robot drives around the job site, person walks behind it, and the machine will literally show them with a laser beam on the ground where to put the marker. You mentioned earlier, your flag could be a nail with a feather, could be a flag. And the robot communicates. Like I said, it's a cobot. Communicates with the operator telling them the point description, in this case, a yellow marker, because yellow means a certain type of foundation. And sometimes you use a spray paint because you don't have to be that precise. So you can use our base model, which is spray paint. It's not as precise. And we can mark a point or a dash showing the contractor where, let's say in this case we're doing a fence also where the trench to dig a hole, a line. So this is a quick video of how our machine work.

That's awesome. That's awesome. Tell us about the machine itself. Is it just off the shelf parts or are you building your own electric motors? You're going to go into competition with Tesla.

You know, as an early safe startup. We are not building our own motors. We, we did design a platform ourselves. We actually design our market mechanism that's not off the shelf, but the motors and the motor controllers are off the shelf. And of course, the GSS technology that we're using is well off the shelf. And we put everything together, designed, connect, and to be honest, most of our developers are software engineers who write code for autonomy, stack and precision. It is a geospatial product. We took all the geospatial, call it brain and math and simplified it. So, like I said, anyone can operate this machine.

Yeah. I assume it has an Nvidia card in it or something like that.

Of course, of course.

And so you're doing a lot of machine learning to figure that's all out.

Especially with the new products.

Yeah, it's cool. Customers must really love it because it's. It looks like it saves them a lot of time and a lot of hassle of getting things wrong. Right.

Yeah. We're very fortunate that today Superbotics has over 100 robots deployed in the field. Some customers start with 1, they get 2 more, 3 more, 5 more, 10 more. Our largest customer, I think has right now 17 or 18 robots deployed. I. I lost count. And we have customers who started less than a year ago, now going up to 10 already.

Yeah.

They see the value very quickly. The return on investment is immediately because they usually rent our machines.

Yeah.

And it's, it's very simple to scale it because if it works for one job and they have five more, they need five more.

Since I'm not a robot assistant, I'm not in construction. Is there a question I should be asking you? Maybe, maybe on behalf of the construction people who might be watching this?

I would say that in a construction, people should try things and be more open to change And Superbox not trying to change dramatically how they do the task. We just try to change slightly the resources that you need to allocate in order to get the job done.

Yeah.

And one of the things that I think our customers love the most about us and people that for instance, feared robot taking our jobs, we are able to convert them to our biggest advocates because we're making their life better. So instead of carrying 20 pounds of gear on their shoulder, you just have to take a tablet that's like three pounds and follow a robot. Much more enjoyable work and also getting their steps in.

Yeah, that's a good point. On a big solar farm, how long is the, how long does the battery last? How far can it travel before it needs to be charged up? You know, do you have to have a cybertrack on site that charges it up or something?

So we, we design with that mentality in construction, we have lunch breaks. During that time they will swap the battery. It's a hot swap. It's a very straightforward process. Takes less than five minutes because you boot it up again. If you don't do it, hot swap because it might take a 30 minute break. So in this case I will turn it off, swap the battery and five minutes before you kick off, you start your back. After lunch, turn back on and continue working.

Very cool. Well, I'm out on stupid questions. It's an honor meeting you and thanks for what you're doing for America getting, getting us building better. Right and faster.

Thank you, Robert, for having me and pleasure being here.

Where do we learn more about your company and where can people get a demo or so?

You can look online@silverbox.com to get more information. We're also very active on social media, on LinkedIn, but we do attend industry conferences quite often. I think the next one is coming up in September in Las Vegas called RE Plus. We also are international so we have robots already deployed in, in the Middle East, Europe, Australia and we also attend conferences in those regions as well. It's great to be here. It's great to build America. Made in the usa. It's really important for us to make our machines here locally and not have to be put into tariffs or supply chain challenges. It's made here in San Jose, delivered to us just last yesterday. We got another batch of 12 more brand new robots and next month will be 30 more. Market is growing. Very thankful to be here and continue building this company.

Well, thank you so much.

Thank you, Robert.