Constructed Futures

Tessa Lau BIM Driven Robotic Construction Layouts with Dusty Robotics

Episode Summary

How will robotics impact construction? One answer is by choosing tasks that are narrow, useful and loaded with details - like drawing layouts. Dusty Robotics chose this starting point because it is perfect for a robot - Tessa Lau, CEO of Dusty, explains why, and how the team chose layouts. Tessa's explanation of their decision process is a masterclass in what robots are good at, and what constraints they face.

Episode Notes

Learn more at: https://www.dustyrobotics.com/

Follow Tessa on LinkedIn:  https://www.linkedin.com/in/tessalau/

Episode Transcription

Tessa Lau

[00:00:00] Hugh Seaton: Welcome to Constructed Futures. Today I'm here with Tessa Lau, founder and CEO of Dusty Robotics. Tessa, welcome to the podcast. 

[00:00:10] Tessa Lau: Thank you. I'm happy to be here. 

[00:00:12] Hugh Seaton: Hey, so let's start with what Dusty Robotics does. 

[00:00:16] Tessa Lau: Sure. So Dusty Robotics is a construction robotics startup that's building robot power tools for the modern construction workforce.

What that means is that we are building a robotic layout technology. That's our first product that actually takes BIM models and prints them full-size on the deck of a construction project, effectively creating Ikea instructions for construction. 

[00:00:40] Hugh Seaton: Ikea instructions for construction. That is amazing.

So talk to me about what that practically means. So I've got a BIM model and I feed it into your system. What does it do? 

[00:00:53] Tessa Lau: So the observation is that there's so much information contained in those BIM models that never makes it out into the field. So we are trying to make it so easy that every single job site is going to take a lot more detail from that BIM model.

Bring it out into the field where it can be used by all of the men and women who are out there in the field, working with their hands and building those buildings. Intuitively right now, when teams are taking BIM models and marking them out, doing layout. They're marketing the bare minimum of information, right now. 

They're taking paper plans, sometimes measuring tape, using chalk lines, snapping those lines, and they are then moving on to the next thing. But we see the opportunity to take loads of information from those fully detailed models, things like part numbers or the orientation of fixtures or the color of the paint or the color of the tile and bringing that information out to the field.

And because we're building a printer that automatically prints all of that information full size on the deck, you can do that in a fraction of the time it would take you to do manual layout. 

[00:02:00] Hugh Seaton: So wait, you've got this Roomba like thing and (please don't be insulted by that analogy) that is kind of zipping along on the floor. And it's writing out this level of detail that now is, instead of it being...you need to squint to see it. It could be whatever size you need it to be, right. So that must be amazing. How do you guys think about gating, how much information is the right amount? Are there, you know what I'm saying?

Is there some process where either the author of it or the consumer of it or the manager of it, or somebody says, I think this much is going to help add this in or gate it here. 

[00:02:36] Tessa Lau: That is a great question. So we started out just by printing the bare minimum of information, just replicating what crews are doing today.

And so typically when you get started, you would start using Dusty for wall layout. So we'd be printing framing lines. And then we started printing finished, you know, the drywall finish, typically located five eights of an inch off of the framing lines, if it's all single layer drywall. And then we started getting asked to do more information like soffits.

And in order to do those, we have to distinguish those from the framing and the finish lines. And so we developed a new feature that allows us to print lines with embedded text in it. We call it rich text line styles. And so now we're printing soffits and they say, soffit, they say the word soffit at every 12 inches along that line.

And so as we're, we've been going and developing, we've actually been starting to do multiple trades layout at the same time. And we're using that same idea of rich line styles to actually print mechanical duct work to print electrical fixtures, plumbing plumbing lines. All with very distinctive text that shows exactly what it is you're looking at when you're looking at the floor.

So it's really interesting to see how clients are going to start using this, but we're starting to see more and more trades pick this up and make use of it and seeing more information set down on the floor. 

[00:03:49] Hugh Seaton: So I'm thinking you're solving a user interface problem measured in meters. Everyone else thinks in pixels and you're like we can only do this if it's so big. 

How do you think about different trades in the same place then? Do you think about fonts? Do you think is this a problem that you guys are working through or you've kind of resolved or is it exactly where you are right now. 

[00:04:14] Tessa Lau: So we actually have the capability of printing with multiple colors. That's something that our clients are really excited about.

So for example, you could imagine doing all the mechanical lines in red and then have the electrical stuff in blue, and then the plumbing and green on the drywall and black. And so every trade walking the floor can see at a glance where there stuff is. 

[00:04:34] Hugh Seaton: I'm thinking about whiteboards and how blue always runs out. Sorry. That's right. “We got that one solved. That's problem number one.” 

So how, how do you guys prep the scene or does the machine or the robot itself kind of prep what's in front of it or does it need to be, I mean, obviously you can't have big, heavy things in the way, but how do you, how do you handle the messiness of a jobsite.

[00:05:00] Tessa Lau: Yeah. So, you know, job sites are messy and that's why we're called Dusty Robotics. Right? We know that we're coming on to a site that's not as clean as you might like it to be. What we asked for before we can deploy a printer, a field printer onto that site is just a broom swept floor. That's all, it's the same thing that a layout crew would be asking for before they snap chalk lines.

And the reason why is because if the floor is too dusty and we print on top of that dust, as soon as you sweep that dust away, the ink is going to go with. Right. 

So not too bad. 

[00:05:30] Hugh Seaton: So you're not asking for anything the existing layout people[don’t] ask for.

[00:05:35] Tessa Lau: We just need the exact same conditions that those layout crews need today. 

[00:05:39] Hugh Seaton: Probably said better that way. That's cool. So how does... the ink gets laid down a lot, most of the time, you're assuming that something's going to be put on top of that floor anyway, but are there instances where the ink itself has to be taken off later?

I know that sounds like a crazy downstream question, but I'm just curious about how you've got multiple lines that have been put down and they're presumably somewhat scuff proof. What's the downstream. If it happens to be an exposed area of concrete or whatever, and the final building. 

[00:06:07] Tessa Lau: That is one of the most common questions we get.

So our ink is it's a liquid it's delivered via a print cartridge, just like you have in your printer at home. If you still have a printer at home. And if you need to swap colors, you just pop out a print cartridge, and you put in one with a different color ink in it. But that ink it's a liquid and it soaks into the top layer of the concrete.

So immediately it's scuff proof. It's weatherproof, it'll last for months without any clear coat. It'll just stay there, embedded in the concrete, but it's only on the top surface of the concrete. And so if you come and grind down that concrete, like for example, you're polishing the floor, turn it over as a polished concrete surface, and it's the ink is going to come right up.

[00:06:45] Hugh Seaton: So it's really a non-issue. Yeah, that's great. So now that we've gone through the kind of ground level function of what it does, I'd love to talk a little bit about how you, how you got here. Like what inspiration led to this need. 

[00:07:01] Tessa Lau: So about four years ago, I started this company before this, I was working at another robotics company.

I had co-founded a company called Savioke, which was building robots for the hotel industry. So you could imagine like a three foot tall R2D2 taking the elevator in a hotel and bringing you a sandwich and your room. If you ever saw a robot like that, it was probably one of mine that was the Savioke Relay.

But after five years of that, I decided I wanted to do something different. And so I decided to start a new company. And I wanted to take my expertise and robotics and all the learnings that I had from building my last company and use it to do something successful. So I asked myself, well, what should I do?

My co-founder and I were brainstorming ideas. And, and at the time I was remodeling my house, it was my first time doing a major home remodel. And I had a general general contractor and and workers just coming to my door every day with power tools and they were building this by hand. And I was horrified because as a roboticist, I think that, you know, the entire world is supposed to be automated.

And then I discovered that construction is still heavily dominated by manual labor. And so I decided, well, there's gotta be something here for me to do. So I bought a hard hat and I bought some steel-toed boots and I started walking construction sites, talking to everyone and trying to learn what all the problems were. Proposed a bunch of different ideas to people and got their about their impressions on whether that would be a good idea or not. And one day I realized that this problem of layout of actually taking the drawings and bring them out into the field. There weren't really any solutions for it. And it was still just people using measuring tape and string.

And I thought that's something I can build a robot to do. And everyone got really excited. And so that's why we're here. 

[00:08:41] Hugh Seaton: And it sounds, I mean, as look, I want it for the listeners that may not check LinkedIn like I tend to, Tessa's got a little bit of background on this. And I'm being facetious... there's an enormous amount of depth in terms of understanding computer science and AI and what things can do.

So I wanted to get that out there, but the question I would ask is thought of from that perspective, some of the things that went into you making this decision are, the ground will be easier to navigate. So that's less uncertainty for the robot to navigate, right. And the need for precision that humans are not always so great at is also present.

Is that some of what got you, where got you here other, I mean, there was a need for sure. There's a lot of needs on a job site. Most of them involve, you know, uncertainty or danger or this or that, where in this case, it seems like it was a good collection of the right things in the right place or the right number of factors to make this a good idea. 

[00:09:38] Tessa Lau: There were a number of things that were guiding our search for the right product to bring to the construction industry as our first flagship product. One of them was size of robot. My experience is that the cost of a robot goes up exponentially with its volume. So imagine trying to build a small Roomba beside a robot versus building like an excavator.

You need a lot more space and a lot more resources to build and test excavators or automated excavators. So just for example, I was trying to find a use case that would be solvable by a small robot. Another thing I was looking at was can it... does it have to be done by a robot, right?

If there's a solution that doesn't involve robotics, well, then we shouldn't be building a robot. Only those solutions that really, really need robotic automation were going to pass my bar. And then the third thing I was looking at was, is there an opportunity to do this job 10 times better than it's currently done today?

And a lot of the robotic ideas don't pass that test, right? If you're just trying to build a robot that does something as good as a human then you know, at best your value is going to be limited to how much you pay people to do it today. But if you can do something that no one on earth can do, then the market is yours.

And so that's what led us to this idea of layout automation, because it's something that matters so much to the success of a construction project, but it's so hard for people to do it today. 

[00:11:00] Hugh Seaton: And we're on what basis is that 10X better? Talk a little bit about what dimension or what dimensions it's 10X better. 

[00:11:08] Tessa Lau: Yeah. So there's a couple of different dimensions that we're improving on. And I would say that we're part of the way to 10X, but that's our goal and we're going to get there pretty soon. So one of those is speed. Obviously in construction, time is money. And so the sooner you can get that construction project wrapped up and turned over to the owner the sooner you get paid, so you can move on to the next project. And there's very few technologies out there that can actually speed up construction. But ours is one of those. So we're taking this critical path layout task and doing it somewhere between five to 10 times faster than people are doing it today.

And so that number is just going to keep increasing as we further refine the technology. The other metric is accuracy. People make mistakes, not so surprisingly, anytime you have a human in the loop, people are fallible. And so robots don't make mistakes. They just do the same thing that ...they do exactly what they're programmed to do.

And so we can get a 10X improvement in the accuracy pretty easily through the use of robotic automation. 

[00:12:06] Hugh Seaton: Are you able to do it overnight? 

[00:12:09] Tessa Lau: You know, that's a really interesting question because when we first started designing this solution, we imagined that it would have to be done overnight that that's what our customers would want.

But then we started talking to people about whether it would be feasible to operate overnight or not. And You know, the reality is our robots are so much faster than people that by the time you finish it, you know, maybe it's a section of floor. Maybe it's like an hour or two later and you need to move to the next section of floor.

Well there, if the robot was slower, then having it work overnight would make sense because you set it up once and then you come back the next morning to pick it up. But because it's so fast, it's actually doing the job in a fraction of the time people would do it. So typically the way we operate is we set up in a corner of the floor and then a couple of times through the day, as soon as it finishes that section of floor someone comes and moves the equipment over to the next site and a floor and clears it out. For the robot and make sure that, you know, the, the site is clear that workers are, are not in that area and so on. And all that coordination typically is supported by our human operators. 

[00:13:12] Hugh Seaton: Wow. That's really cool. And so, you made a big statement a moment ago that robots don't make mistakes.

And I was like, well, they don't if they're in the right track, right. But if, if they're off a bit, they're going to perfectly execute off a bit. You know what I mean? Like they perfectly execute what they've been set up to do, but that implies some things about calibration and setup and that nothing, nothing you didn't expect happens like presumably it's on wheels.

And if there's a divot in the, in the floor you didn't expect, and it turns in a way that it didn't know how to. I mean, I'm assuming some of that is corrected. Do you know what I'm saying? So how do you, how do you think about QA and calibrating? 

[00:13:56] Tessa Lau: Yeah. So the saying that we have a Dusty is garbage in garbage out, right?

And so robots are only as good as the as the data they're given. The two main reasons, actually the two only reasons why we see errors on, on job sites is because of bad control or bad CAD. So control, you know, if you're using a Total Station, they're set up based off of points, you know, grid intersections that are marked on the floor and our system works off of the same principles.

It requires control points to be clearly marked on the floor at known locations, but then compares those locations against the same points in CAD. And that's how it aligns the coordinate systems in the field versus the coordinate systems in the CAD. If those are off, then we can't guarantee accuracy. That's just the nature of it.

Everyone understands that from their use of Trimbles. And so we're no different, 

[00:14:47] Hugh Seaton: The only thing I would say, though, is that when a human is doing it, they'll be able to course correct, because there they'll recognize, Ooh, someone got that off a little bit. So do you guys find that because of the absolute accuracy or anyway, very close to absolute accuracy of what the robot can do that actually the control points now become a constraint and you have to go make sure that those are done as well. So is there a scanning process or something you do to make sure that what's going because once the ink is down, presumably you've got yourself an issue if you're off.

[00:15:19] Tessa Lau: Yeah, absolutely. And you know, what we teach our clients to do is that the very first thing we do as we're setting up the system is check the accuracy of that control. Right? And so within 10 minutes of showing up on site, we can tell you whether that control is accurate. And if it's inaccurate, how far off is it?

We've been on sites where, you know, a typical site we'll have control accurate to within an eighth of an inch. And that's perfectly reasonable. The worst site we've been on had control off by a couple of feet. Yeah. And you know, what do you do in that situation? You don't want to print something down because it's, there's no way it's going to land where it's supposed to land.

There's just not enough accuracy off of the incoming measurements for us to guarantee any kind of precision on the output. So at that point we would stop. We would work with our clients, whether it's the surveyor, whether it's the general contractor to actually improve the positioning of those control points before we start printing.

[00:16:10] Hugh Seaton: Yeah, I can imagine. Really interesting. So in terms size of space that you tend to work in, is there a sweet spot? There's presumably there's a lower limit below which it's like between setup and all the rest of it, you know, it's just take the chalk out, but is there a sweet spot that you're finding is most applicable.

[00:16:29] Tessa Lau: So, I mean, I guess the bread and butter kinds of projects that we've been doing lately are multi-family multi-story multi-family residential properties. So, you know, we've done a couple of nice towers. We did a 13 story at 20 story. We're now on an 18 story building in Sacramento, I believe. And you know, that that kind of construction is just perfect for robotic layout.

It's basically the same layout all the way up the building with small variations. And it's so repetitive and it's very dense typically, especially in urban areas. And so that's been the, the bread and butter of the kinds of projects we're on. Lately we're also doing some TI office renovations, things like that.

We're building out a new headquarters for some of the tech companies in the bay area, doing some office complexes you know, multi building campuses. So it's, it's kind of all over the map, but I would say anything upwards of about 10,000 square feet is fair game for Dusty. 

[00:17:23] Hugh Seaton: That makes sense. And you guys right now are primarily focused on the west coast.

Actually. I think it's just California, because demand is so high your like, we got to handle what we can handle. 

[00:17:34] Tessa Lau: So we're based in the Bay Area, San Francisco we're actually in Mountain View. And so we started our service in, in and around the Bay Area, but we've recently started expanding.

So we are just about to bring up a branch in Seattle providing some services up there and we have an LA expansion in the works and we're also starting to lease our robots to customers all across the US. And so that's how we're actually going to get our way east of the Mississippi. 

[00:18:03] Hugh Seaton: Very cool. So you're learning all sorts of new muscles in terms of just supporting things that aren't right down the road.

I mean, that's, that's, that's gotta be a nerve-wracking at first and then I'm sure exhilarating a little bit later. 

[00:18:16] Tessa Lau: We've gotten really good at shipping robots around the country now. 

[00:18:19] Hugh Seaton: I'll bet. Do you find it... just crazy question, but after shipping, is there something you do to make sure that nothing got funky when they were in terms of calibrating, or am I over thinking it?

[00:18:30] Tessa Lau: Yeah, we learned so much the first couple of months we did this. So we started shipping them around earlier this year. And we had some shipping disasters. I'll tell you this, cause it's, it's, it's funny and embarrassing in hindsight, but you know, we're past this now, so I can tell you this story.

There was this one week where we were it was a really busy week for us. And we had three customers that all wanted to use our robots in different cities. And so we shipped out three different robots with three different cities and every single one of them arrived, broken. That was a terrible week for us.

So then, you know, disappointed customers, very stressed field team. We discovered that it was a problem with our shipping method, not method, but the packaging that we use. One of the things that we discovered was that when the box goes into the airplane and the airplane goes up into the air, there's lower air pressure.

And so the box is like a...it turns into a vacuum and it sucks all the air out and it compresses the contents, compressed our robot in a way that we weren't expecting and it shattered one of the parts. And so we had to learn that the hard way unfortunately. Now we've learned that doesn't happen anymore. And so we've learned a lot.

That's what I mean when I say we learned a lot about how to ship robots. 

[00:19:37] Hugh Seaton: Oh my gosh, what an expensive lesson, right? I would never have thought that, I would have thought the opposite, that it would have the cold or not the opposite, but I would've thought the cold or something else, but you know, compressing in, you just can't know, can you? 

So, when we were talking prior to the podcast, you, you had this kind of a vision that you started talking about this, you know, what you're doing now is, is like, is as, is often the case, you're establishing a baseline product and really learning how to execute well, product market fit, all the things investors want to hear, but the bigger vision it relates to BIM and completeness and digital models.

Let's get into that a little bit. Talk to me about what you meant and where you see some of this kind of knitting together. 

[00:20:19] Tessa Lau: Yeah. So, one of the interesting things that we're seeing in the industry is that, we're coming into the industry at this really interesting time where it's in the middle of this digital transformation, BIM has been... the transition to BIM has been underway for maybe 10 years now.

And it's starting to take off. We're seeing pretty much every single commercial project that we're on being modeled, at least partially in BIM coordination is happening. People know how to use the tools. They're comfortable with them. But there are still limitations that we're seeing. And, and the nature of our product is actually exposing some of those limitations in how BIM is being modeled on a lot of jobs.

And, one example is that we will often get onto a job and discover that the architects have modeled out the walls. And they've just chosen the default wall thickness, which is 4". And if we were to print that we would be printing two lines of framing, four inches apart, and any carpenter will tell you, framing does not come in four inch widths.

It's like three and five eights or it's six, two and a half. And so the architect's model is not constructable. And that happens on some number of the projects that we're on because they're not being modeled out in enough detail. And not just that, but we're finding that every trade has their own model.

Right. And if we actually want to realize this vision of doing multiple trades layout at the same time, working and giving our robot just a single file that has all of this information in it. That information is being pulled together from all these different files and having to be glued together and cross-checked and points aligned and sanded and, you know, made sure that those are the right that all the coordinates are correct.

So that is a challenge that we're facing. It's one that I hope the industry rises to the challenge of fixing because the more information we can get in the same place and have everyone look at the same model, rather than everyone working off with different models and then trying to fix it later, right?

I want to push the industry towards working on every, having everyone work off the same model, having the same model in being used for design and for construction and potentially someday for operation as well. Once you turn it over to the owners. 

[00:22:29] Hugh Seaton: And what do you think is the path to get there? And how are you guys, how do you think you you're going to play?

I mean, I don't want you to give away too much of what's in secret pitch decks, but how do you see your role in some of that? 

[00:22:41] Tessa Lau: Well, I think the first step is identifying the problem. And so, you know, I think there are a lot of opportunities to make things better in this space. We are starting to explore moving upstream and enhancing tools like Revit.

To actually give architects and designers more capabilities so that they can design robot ready drawings. Right. Wouldn't it be cool. If, if you could take your 3d Revit model and you can hit a button and then it goes straight to a Dusty printer and then it prints exactly what the crews need to build.

So we're trying to really streamline that process, make it so that there's less human error or human work involved in getting from that design intent out to something that's constructable in the field. 

[00:23:25] Hugh Seaton: See a robot ready model. That's a bumper sticker if I've ever heard one. 

[00:23:30] Tessa Lau: Tagline for the future. 

[00:23:31] Hugh Seaton: Totally. No, I'm kidding. I'm only half joking because you know, all the effort and attention given to DFMA and you say, well, there's going to be an equivalent, right?

That someone is going to have to say, how are we building in the right metadata or the right instructions? Are those the right, whatever it is. And I would ask you, at least conceptually, what is the, whatever, what is the other bit what's? What would, what would robot ready mean other than more accurate?

Are there things that you would want them to include that would make it easier for a Dusty robot to work? 

[00:24:04] Tessa Lau: So you know, I would love to see we're starting to talk to architectural firms because what I'm seeing in the industry and the broader industry is more of an interest in going back to that concept of the master builder.

And having the architect get more involved in the actual building process. So rather than just handing over a design and saying, here's, you know, this beautiful image I have in my head of this building, we want to build, now you go build it. You figure that out. 

It's starting to become a more collaborative effort between the architects and the builders to actually define what it is that building is not just the design intent and how it's going to look and feel, but actually how it's constructed. And so the way I see this going is models starting to encapsulate a lot more information about constructability.

So instead of an architect just modeling out how I want the walls to look and where they're going to go. They're going to start incorporating information about the materials being used to build those walls locations with the studs, how those are all going to get framed out because that matters in terms of the constructability of it.

And so all of those considerations are going to start being a part of the design process. 

[00:25:11] Hugh Seaton: One hopes, I think liability and, you know, and flow of information is there's still a lot there, right? Is that you're actually not asking about the general contractor. You're really saying the, the trade contractors connecting back up... Or over, however you want to say it to the design team so that they're sharing gets learned in the field and how to, how to construct and execute with the original, which I think is great. And I think you're right, whether it's design build or some of the other delivery methods that are out there. There's a lot of hope for it. I think, I feel like the legal side of things is, is still, you know, catching up. 

How do you guys...is there a, you know, when you think about that side of it in terms of, you know, you've got a robot on a job site is there a legal and, and kind of risk and liability set of hurdles that were, that you guys had to get through early on?

Were there things you had to reassure people of or any of that? Because certainly with some of the more mobile ones and by that, I mean, spot that hops around and makes, you know, people freak out a little bit. Was there, was there a bit of a hurdle in the beginning that you needed to go look at. 

Yeah, absolutely.

[00:26:18] Tessa Lau: I mean, when we started, we were actually really worried because you know, when you're starting a company, you try to think focus on the positive, but you also think, you know, what's the worst that can happen and can we prevent the worst from happening. Right. And one of those worst case scenarios that we considered early on was what if we print something in the wrong place and then someone builds off of it and then you get to the end and you discover that, you know, your bathrooms are not ADA compliant or something, and it turns out to be Dusty's fault. Right. And so we spent a lot of, a lot of sleepless nights thinking about that, but what we ultimately realized is that, is that what we're building is a printer.

And the same way you don't expect your Epson printer to spell check your document. Right. You know, if there's a typo in your, in your word document and you print it, you know, the printer is going to print exactly what you gave it. And so that's exactly what the line we're taking now with our field printer, it's a printer and it's going to print exactly what you give it.

And so if you're in your CAD, if those bathrooms are not ADA compliant, that's how we're going to print it. We're not going to correct that error for you. It's not our responsibility. 

[00:27:22] Hugh Seaton: I love that you put it in those terms, it's a printer. And I think the reason why that, that matters is. I worked with a guy named Josh Levy, who's got a, company Document Crunch, but he's a lawyer. And we put a paper out that related to risk of sharing models. The reason for that was some of what you mentioned earlier is that a lot of trades are nervous about sharing models and scans that they've done, but the point I'm making that comes back to what you just said is, the law doesn't cover most of this stuff. It's what's in your contract. So in your case, you make clear that that we're liable for accurately printing what you give us and that's it. And the same thing is true across a lot of products, right? Is that we're responsible for making a safe product that won't, you know, do dumb things like blow up or whatever.

But other than that, the execution of your designs is... we're only responsible for the piece that we say we're responsible for, and defining what you're responsible for is really, on both sides, right? For the buyer, also to understand that they're only going to get this done and, and we'll make sure that that that's it really, really interesting.

[00:28:28] Tessa Lau: Exactly. And this has been an ongoing discovery process for us to figure out where is the boundary between what we can reasonably guarantee and what we push back on to our clients to get. 

[00:28:41] Hugh Seaton: Really cool. That's funny that and, and some of those questions you could have thought of early on and you just figured them out and others you're like, I never thought that'd be an issue.

The robots themselves are not very tall. Right. So is that another thing, have you thought, do you guys like sometimes put flags on them or is it not really an issue. 

[00:28:59] Tessa Lau: We've had that request. They're short for a reason, right? Because the closer that reflector is to the ground, the more, more stable it is, and the less error in measurements that you get, so we can be more accurate. So there's a reason they're there about, I think they're about a foot tall right now. And, and also the smaller they are, the lighter they are, the easier they are to transport. So it's all, it's all part of our design process, was coming up with something, but, you know, typically when we're on a floor, we're one of the only trades on that floor because we tried to get on that floor first thing is as soon as it's available. And so we don't tend to have too many problems with visibility or people not seeing it and tripping over it. 

[00:29:33] Hugh Seaton: So the last thing I'll say is I love that you said one of the first we're one of the first trades on the floor that you really are thinking of this as integrating as a trade into the industry. Very cool. Well, Tessa, this has been so much fun. I've just, I keep imagining the robot zipping along in this amazing kind of graphic on the floor. Can't wait to see it someday. 

[00:29:55] Tessa Lau: It felt if anyone's interested in seeing more videos, you can check us out. We have a website Dustyrobotics.com .

We also have a channel on YouTube. So just search for Dusty Robotics and you can pull up some cute robots. 

[00:30:06] Hugh Seaton: And I'll put them both actually in the notes of the podcast. So Tessa, thanks for being on the podcast. 

[00:30:12] Tessa Lau: And thanks Hugh. This was a lot of fun.