Uh, maybe I would like to touch the elephant in the room first, um, because to most of the people when you speak about drones, probably the first brand that will come to mind is DJI, okay? Yep. Uh, we know what's happening in, in, uh, the United States. Uh, I think we can, uh, look at it from two different angles. One is like the legal stuff around it, and the second is like, what do you think is happening in the background? So, maybe if you can just, uh, walk us a bit through, like, what's happening there, like, in a legal way.

Yeah, well, I'm going to take a little bit of a different approach towards it. Um, some may take it as a bit controversial, but I think it's more of a logical approach where, look, I understand why certain industries or certain companies are pushing forth certain legislation to block or restrict or just make it a little bit more difficult for our foreign competitors to compete within our own market. Now, am I for that? Not necessarily, right? Because we're basically grown on, uh, like, [clears throat] a capitalist market, right? We like to have a fair and open trade. Um, however, when it comes to, we'll call it, the critical infrastructure, military, and really anything that has to do at a high government level, I'm not talking about doing just basic mapping of streets and everything. I'm talking about things that have to do with saving human lives, uh, doing short-range reconnaissance, doing any military operations. We should not be dependent on a foreign entity or a foreign company's components, uh, to be alongside with us. Are they a bit more superior in terms of their capabilities? Yeah, absolutely. Are they a bit more cost-effective? Actually, a whole lot more cost-effective. But, when it comes to my industry, with the commercial industrial side, it does make it a bit of a challenge, um, for us to do that.

And over the last, we'll call it, two years, uh, there's been a lot of change, and there's a big piece of legislation that was basically a big bill that just got pushed through on the FCC side back in December that, quote unquote, bans DJI, right? This is the main brand in the drone industry. I mean, they control about 85% of the global market, uh, when you think about a consumer or an enterprise drone. But then when you think about the military side and the private sectors, right? There's a whole lot of other competitors in that space, and DJI is still very dominant in that, not just in the United States, but globally. Um, so, a lot of that has been a fight towards bringing a lot of, well, one, manufacturing back to the United States, which I think is good. It's a good thing for us to do that. It creates more jobs, creates more opportunities for people to get into robotics and really the manufacturing side of the industry, right? That was something that we were so proud of, uh, when we were manufacturing our own vehicles. And we're trying to get back to that with all of these new AI data centers, even though that's not manufacturing. There's a lot more manufacturing that's coming onto the tech side. So, that's one side of it.

The other side is, well, what about us, right? What about people like you and me that don't care about that or don't even operate, um, in that regard? It puts us into a very tricky position. And as it stands right now, you still are legally able to purchase a DJI system as long as it has its FCC certification and fly it on projects. I'm sitting on over three or 400,000 dollars worth of DJI equipment here that I am using weekly with my clients. And it's 100% legal to do. However, we're starting to face some of these restrictions coming into play where I am now having to outsource, um, or even source specific US components or the companies here in the US that have built US drones or US sensors to go and complete some of these projects. And that creates a huge price barrier. And in some cases, it's a four to five X price barrier. So, I know I went a little bit of a tangent here, but I'm kind of explaining how all of these things work on the back end, right? 'Cause on the front end, you see all of these articles, whether it's on LinkedIn or a news article, "DJI drones or Chinese drones banned from the United States," "It's illegal to fly." And there's a lot of misinformation that exists. I get hundreds of comments every single day across all of my content saying, "Oh, well, you're flying something illegal. You can't do this. You can't do that." When in reality, there is a legal path that still exists.

If I, if I understand right, the FCC certification is about new technology incoming, right? So, so like the current drones and current approved, you you can fly them, yeah?

Yeah, it's all it is, it's a licensed certification for the radio signal, so the radio waves to work between the controller and the drone. That's all that it really comes down to. It has nothing to do with any of the other components on it. Even though they are extending it towards that, which I don't think is right, but that's just kind of...

Is there [laughter] Is there a difference between, uh, commercially used drones, so probably like the division of, uh, DJI Enterprise compared to the consumer drones or for the FCC licensing?

Yeah, yeah. Uh, no, it's just a blanket, right? So, DJI as an entity as a whole, whatever their FCC licenses that are currently standing, whether it's something like a DJI Mini or all the way up to a DJI M 400, Mhm. they all have to go through the same certification process. Um, so there is no difference between the two. Obviously, the difference is a $1,000 drone and a $10,000 drone, but that's it's really a price factor in that sense.

Is there, is there currently, cu- cu- currently, uh, drone from DJI that you already, uh, think that would be good to have and is not approved in the United States?

Well, currently, everything that's come out from DJI is approved here. Even the new Avata 360, you can purchase it here and legally fly it, as far as I know. Uh, but if there are any new DJI drones that are coming out, which we know that they are, um, 'cause they're just like Apple, they have something new every single year or at least four new things every single year. Um, there's things in the pipeline. Right now, it's uncertain, right? There are current certifications that exist for drones that haven't come out yet. So, you know, a lot of people are banking on the fact that, all right, well, those certifications are going to be applied towards these new drones for the next two or three years. But, again, we don't know that.

I have seen also some discussions or or like conspiracy about, um, Sky, um, Oh, was it Skydio? Uh, Skydio company doing some, uh, behind it and just lobbying, uh, like, uh, against the the DJI. Do you think it's it's it's really possible? Because I think that the DJI DJI is too big, like, they can pay whatever and and deal with whoever, I think, all over the world, so.

I'm going to try to not sound like a politician here, um, 'cause I've got friends all throughout the industry. And, you know, a lot of them are in certain positions that kind of bring unfortunate, uh, attention. I think indirectly, there have been things that Skydio has done, um, that a lot of the commercial industry is just not fond of. Um, whether it is backing a bill or paying a lobbyist, right? But, I actually made an entire video about this that I posted back in November. It's about a three and a three hours and 45 minute, uh, we'll call it a documentary that goes kind of behind the scenes and looks at all of the US drone manufacturers that exist. But I also have right in the beginning of the video, I talk about what things are and and and really who the proponents are. I also made a second video talking about the whole DJI ban that goes a little bit more in depth about how and who and what and where the money has been flowing. And in fact, DJI has actually put in way more money than Skydio in the lobbying efforts. So if people want to play the finger-pointing game at, "Oh, well, they spent this amount of money." Well, guess what? So has everybody else. That's just the game that is played on the global economics when it comes to companies that are manufacturing goods and even services, especially in the tech industry. So, kind of yes and no. Skydio, I wouldn't say is the ultimate or the only company to quote unquote point a finger at that has put efforts or at least money towards making it difficult for DJI to operate here in the United States. There's a ton of other companies that have also done the same.

Now, it's not [clears throat] as easy as it seems on paper. In the beginning, you mentioned like the approach or the point of view from the case of the defense and let's say the ability to not rely on on foreign entity and foreign technology, right? So, do you think that especially with what we can see all around the world with wars and everything else, do you think that especially now this should be like a primary line of how to look at this and just do it no matter what? Because like if the US just bans DJI drones at all, uh, will probably move prices up, uh, technology for a few years, maybe a little bit down. So, how do you see the approach here?

Well, here is the thing. If, for whatever reason, the US decides today, "DJI drones no longer. We're pulling all FCC licenses. We're doing all that." It will single-handedly collapse the entire drone industry in the United States. And the way that I actually explain the drone industry, we're not a primary industry. We're a sub-industry of many primary industries, right? We're just a tool. That's all we are.

[clears throat]

But in the not efforts, but in the conversation about war, right? War is something that you, it's unavoidable regardless of where in the world you are. And it's not an area that I, you know, particularly enjoy, uh, talking about, but it's also one of the biggest factors in innovation. If you look at the decades of war that has happened and the tools that have come out of it, I mean, GPS is one of them, right? So, that's one side of it. There's always going to be constant innovation towards it. But when we're talking about using foreign companies' systems for your own efforts, that gets a little bit of a sensitive topic. It would be like the US purchasing a bunch of, we'll call it, Russian drones to use for our own efforts. That would not make any sense at all.

Or But is is this the same because like I I don't think that the, like at least like openly, there is a very different approach to China and to Russia, right? Uh from from the United States, so.

Well, yeah, of course. I mean, that it just has to do with the diplomatic relationships that they have and, you know, the the decades and and centuries of...

But but but you you think on the background it's it's all the same, yeah?

To simply put it, yes. I mean, but it goes much deeper than that and honestly, I don't even think we'd have enough time to to really [laughter] cover all of that. That's all like a whole political conversation that can happen in a more friendly conversation, but it's things that you and I would never have any say or control over. Right? These are people at the absolute top of every single country. These are the ones who make the decisions on that, not us. We're just the, you know, the pawns in in in in on the board. We just move whenever we're told to move.

I would like to know how you, how you came to the drone industry. How, how did you start? What's, what's your background?

It's funny. Um, I was just having this conversation literally the other day with actually a police officer who is now using drones for their own efforts. And I actually started in video production. My whole background comes from video production, shooting music videos, TV shows, doing a bunch of directing. And I was using a drone at the time as just another video camera, just one that flies. And, uh, honestly, did that for a few years. Had an opportunity that fell into my lap to do something with commercial real estate where there is a brand new warehouse that was being built, and they wanted to see a unique way of presenting it for obviously the parties that were invested in it, the brokerage firms, the investment firms, all these different companies that are, you know, part of a project like that have hundreds of millions of dollars invested. They want to know where their money's going. And so I asked the company, I said, "Hey, have you guys ever seen a drone scan before?" And they sat there and they're like, "No, we've actually never seen that before." 'Cause they're so used to people just flying drones, capturing photos and videos, and then just sending it on a Dropbox or throwing it on YouTube. I also hadn't seen a drone scan before either that moment when I asked them. And so that day I went back home and I got a trial of DroneDeploy back when DroneDeploy was very cheap and, um, went and scanned my parents' house about 17 times that day just to make sure I got it done right. Went out to their property the next day, scanned it a few times, and then went and did a full presentation, um, at their office, uh, at the end of the week. And, uh, by the end of the presentation, the entire office was in that boardroom. And people's jaws were on the floor, the light bulbs were going, and everything was happening all in the right aspect. They're like, "Hey, can you zoom into that spot? Hey, can you do a volume calculation, um, of that dirt stockpile over there? Oh, hey, look, that's Tim's tools in the back of his truck." Like, "Ah, that's going to get stolen." Like, conversations like that, but then people started realizing, "Oh, wait, there's actually a lot more benefit to this 2D map and really 3D model that they were looking at, a lot more insight into their project and where the money was going." And in fact, that moment, I secured a contract right on the spot.

I dare I want to stop you because I was especially wondering about that. Uh, we are in a very sexy industry. So, like the fact that everybody was like drooling and like staying in awe and just just watching at what you were showing, it's quite a common situation in presentations, okay? Because whether it's a laser scanning, photogrammetry, whether it's a drone scanning, and everything else, because people are not used to be looking at at the 3D models of their like buildings or lands and whatever. But I wanted to ask especially that, like after that, were they able to purchase something to lock you for some jobs, or they just stayed like, "Yeah, it's very nice, but what should we do with this, and who will pay for this?"

No, no, they, uh, on the spot, they realized that there was a, there's a value to it because they were able to not have to send any of their team out there or rely on project managers to update them on where the progress was of this project. I mean, this was a five and a half million dollar project that they were invested in, and they were responsible for all five and a half million dollars that were invested on this project. Um, and so that has to do with every component of the project. By doing that for them, they were traditionally relying on just some photos and some videos and like, "Okay, well, that looks good." But that was it. It just looked good. There was no actionable data off of it. It actually wasn't until three months later that we had started that contract that they really saw the full value of it, where the owner of the company, he said, "Hey, Dylan, can you can you go and look at the pile and can you give me a volume calculation of that one stockpile on this specific part?" I'm like, "Yeah, sure." Took me 30 seconds, sent him the report. I get an email later that day saying, "You just saved me $60,000 from that." I said, "How so?" He said, "Well, found out that the contractor was lying about the material, the volume of dirt that was on site, and in fact, we went and looked back at some of the videos and saw that they were sending the dirt off to another contractor." That $12,500 contract that I locked in right there saved them $60,000. It paid for the entire program forever, pretty much, 'cause I still do work with this client seven years later. Um, which is which is crazy. Like that was the moment that they realized, "Hey, this this is actually a very valuable asset." Um, not just the asset that they were building, but also the data that I was providing to them. And it expanded beyond there, too. I mean, they mostly care about, uh, dirt volumes for all of their sites, but that right there was the pivotal moment that I realized, "Oh, wait, there's actually a lot more utility with drones than just taking pretty photos and videos." So, that was really my start.

I I think that this, uh, volume measurements of the dirt on construction sites, this this is a pretty common use case because this is happening almost everywhere that they are either not able to to calculate them accurately or they don't want to calculate them accurately, as you mentioned, right?

Exactly, yep. But, uh, especially in my region, I have seen one more situation. Uh, it was the first two. And in addition, the construction manager was the one that didn't actually want to know the truth because he was involved in that.

[laughter]

So, you don't have a partner for this, okay? So, there, there is nobody to check accurately whether this is okay or not. So, the subcontractor and the site manager were going like together on it, and there is nobody to check this accurately. Ah, crazy.

[laughter]

But it happens. It happens literally in every continent. Like, there is no, well, maybe not Antarctica. They don't have dirt out there, but...

[laughter]

But no, that that as simple as it is, just being able to track material. I mean, there's this whole saying that the dirt, dirt work is a dirty industry, and it really is because you've got contractors that have been doing the same old method of, "Yeah, looks good to me," and they just measure with their thumb and call it a day. Um, where in reality, they could have been losing tens of thousands, if not hundreds of thousands of dollars a year on miscalculated material. And with the rapid pace that we are starting to build nowadays in terms of scaling, at least here in the United States, we are building a lot of data centers, and there are companies paying hundreds of, honestly, billions of dollars right now. Uh, a lot of money is being invested into developing warehouses. And a warehouse is basically just one big flat building. So, they need to move a lot of dirt to make the building as flat as possible for the foundation. Um, but yeah, like simple as it is, that right there was the pivotal moment in me getting into the commercial drone industry. And from there, kind of, I would say the rest is history, even though I've got a lot that has led me up to the conversation that we're having here today. Um, but if it wasn't for that opportunity, then I wouldn't be where I'm at right now.

But it's very interesting and like Scott is just saying that, uh, that, uh, yeah, he's just confirming that any like case that can demonstrate the quantifiable ROI is a win for the industry. And, um, it's it's funny because it's not only easily quantifiable, but it's also a long-hanging fruit because you can be almost certain that this is happening on every site. Yeah.

Right? Oh, absolutely. I mean, ROI doesn't have to necessarily be measured by, uh, dollar amounts either. There's many ways to measure ROI. Is there a time savings? Which, yes, so directly that also correlates to a towards saving money. Um, are you taking a human life out of a very unpredictable situation and putting a robot in place? Right? That is a huge ROI. There's no amount of money that you could place on a human life. Right? So, there are multiple different factors. [clears throat] At the end of the day, if you are solving even the simplest of problems, it takes drones out of it, just with 3D data collection that we do, whether it's terrestrial, whether it's, you know, under the surface of the water, or whether it's geospatial. Someone is going to be able to pull and extract some kind of value out of it. You just have to understand what that value is to them, right? What problem is it solving for them? And then there's practically endless opportunities and things that you can do with drones, with 3D scanners, with, you know, USBs, like so many different things that you can do.

Can you just for a second travel in your mind back to the times when you realized that drone can, or it's not only about the photos, images, and videos, but you can actually like recreate, uh, 3D models and something. Like, when was your first realization and how did you feel about it?

It was that actual project that I did for that client that won me the $12 and a half thousand dollars.

But but but you you must have heard about the digitalization before, right?

Yes. Uh, through YouTube, funny enough. Uh, the platform that I that I love and obviously that I'm on now. Um, I had seen a demo. Um, I don't know where or why it popped up, but I mean, YouTube works in mysterious ways. It just wanted me to see DroneDeploy being done. Uh, "Hey, you can take a drone up and do a 3D scan and create your own models." Like, that's cool. But that actually had sat in my mind for two years until I actually went forward with doing a trial and going out and providing this deliverable to the client. Granted, I didn't know what kind of deliverable I was doing. This is, this is 19. I was 19 when I did this for the client. Uh, so I didn't really understand the industry to the extent that I understand it now. Um, but it was that moment that I saw and then two years later, I was like, "Oh, let me just try this out." That was that was a very pivotal moment for me.

Uh, how long after this, this, this like first project, uh, how long have you still been build, still been doing the like pictures and videos? Or did you just cut off the business and stayed with 3D models?

Um, I still dabble in it from time to time. I mean, I transitioned a lot of it into the YouTube that I have now. I mean, I'm in this studio that I've built out, something that I've actually dreamed of doing back when I was first doing the photos and videos. I mean, I had a cheap $500 camera and, you know, a $1,000 drone, and I thought I was the coolest thing on the block. Um, but, you know, now fast forward seven, eight years now, um, I still leverage a lot of, you know, my knowledge from then, my expertise in doing the photography and videography, and it has actually allowed me to build my own personal brand, um, which is obviously myself and then, you know, my company Pilot Bite. Um, so, I still technically use it just for my own efforts now, not for other people's.

I'm wondering about one thing, uh, Dylan. You moved from, uh, doing pictures and videos into doing full 3D models. I'm getting quite often and, like, disturbingly too often, uh, this question, like, "My cousin has a drone, can he do the 3D models?"

Yes and no. It depends. If you look at how 3D modeling was done back in the day, which this is actually [clears throat] something that I teach. We didn't have all these tools of using RTK and ground control points. We used a lot of photos, stitched them together to, in a sense, photogrammetry to create and render these high-resolution, very dense mesh maps and models. But, you know, you don't have to spend thousands of dollars to do it. Literally, I can do it with this little drone that I have right here. So, yes, you can absolutely create a 3D model with it. There's a lot of effort that's involved in doing it, but it doesn't mean you can't do it.

Uh, the problem is that the my question is not aiming to whether you can do it with the like retail store drone or the smallest drone or whatever. The problem is that the question is usually about, "We need to scan this building or we need to scan this site."

I see what you're saying. "My cousin has a drone, can he do it tomorrow?" And that's a very different question to what what you... That's a very different question. Yes.

Um, you know, that's funny because there's a lot that people don't know in terms of what they are asking for. It's like, it's like I get a client that asks, "Hey, can you go and do a scan for me?" Okay. What are you doing with that scan? Right? What are you trying to extract? What value are you trying to get out of that scan? Are you just trying to look at it like a pretty model? Cool, that's easy. I'll use this to go do that for you. Or are you trying to do real-world measurements? Like, is this centimeter accurate? Do I need to tie it to control that you have on site? That completely changes the way that I'm going to pitch the service or the way that I'm going to collect the data and provide them a deliverable. Because a model is not just a model. A model can be just a model, but a model can also have a lot more intrinsics behind it, a lot more data behind it. And it really, it's really up to the end client or what the end client's expectations are of what they're going to be able to do with that model. And I think there's a big misrepresentation in terms of the accuracy. And, you know, I'm going to vouch for my surveyor friends here. You know, there, yes, it is too easy nowadays to just go pick up, literally, I keep pointing to this, but you can literally pick this up, and there's people that are going to go out and say, "Oh, I can do a point cloud with my Mini 3." Technically, yes, you can, but how accurate is it, right? What are you trying to do with that? And I totally understand the frustration on their part, and I'm with them on that, because if you don't know what you're doing, then you're going to set yourself up one for failure, but also put a bad taste in the client's mouth. That way, when an actual professional comes into the position to say, "Hey, I can actually do a scan for you." They're going to be like, "No, no, no, no. I don't want to do that. I've already spent thousands of dollars on it, and I was disappointed." And it was because of someone not fully understanding what it is that they're providing or what they can't provide with the tools that they have, um, at their disposal.

My response to that is very, very often like, it's it's like comparing ice hockey to figure skating. Like, yeah, you both have ice and skates, but it's like totally different sports. And it actually goes the other way around. Uh, just because I'm able to create like a perfect 3D model of a building, that does not mean I'm able to shoot a wedding, for example, with the drone or some cinematic pictures, right? So, uh, it, it's going both, both ways, and it's not about like diminishing the capabilities of those that are doing just pictures and videos. People simply need to understand that these are different sports and different applications, right? You can either use a hammer to build a house or a hammer to smash your hand in, right? It's still a hammer at the end of the day, uh, but if you don't understand, you know, the full capabilities or what it can't do, you know, then that's going to put you in a position of, you know, not being able to provide what it is that you think you're providing to your client.

Now, look, we all start from a place of ignorance, in a sense, right? We don't know what we don't know. But that doesn't mean that you can't learn. I've spent over 12 years in not just flying drones, but learning about industries, learning about how to use these tools. The first four years I was using it as a cinematography tool. And, but then also had that transition for the last 10 years that I've been doing this in a commercial sense. So, I've been able to slowly and gradually graduate to that next step, that next evolution of understanding what the tools and capabilities are to the point where now I'm working with engineering and surveying firms not just all over the US, but all over the world to help them implement the technology. Sure, I'm not a surveyor. Sure, I'm not an engineer, but I also know enough and I've worked with a lot of my colleagues in the space to understand where their frustrations lie in the technology and how to also re-educate them on their actual capabilities and the nuances behind the tech so that they can actually leverage it to the extent that they need to leverage it to and not listen to a sales pitch or, you know, somebody that just picked it up off the street and said, "Oh, I can go do this for half price." Like, no. There's a lot that goes into understanding the tools.

I would like to dig deeper into this because you mentioned that you are not a surveyor, me neither. I'm not a surveyor as well, and we are doing like scan-to-BIM projects and laser scanning and photogrammetry on objects. Still delivering valuable assets, survey-grade assets, but we are not surveyors by education and by profession. You just muted yourself, uh, Dewan. Yeah, okay. So, I just wanted to ask you, how do you tackle the question when you come to a new client that doesn't know you or just very briefly, and he asks you, "Oh, so you're a surveying company, right?" So, how do you tackle this question?

No, I am very explicit in that, and I'm also very explicit with my students, too. Look, if you're operating within the United States, 'cause I know what the rules, regulations are here, and also the fines that exist here if you misrepresent yourself or misrepresent your data. I'm explicit in how I pitch and sell my services. I never use the word "survey grade." Now, if a client's going to come to me and say, "Hey, I need a survey grade this." I'm like, "Look, I can't provide that, but I can provide you an accurate model for your reference only." Like, keywords, "your reference only." And I also have a bunch of context that I provide in the contracts and a bunch of disclaimers, letting them know that if they want to use this for survey purposes, they can't. Or in front of the jury or...

Yep, if there's any legal, legally binding document that goes with it, I'm like, "I can't I can't provide that." Now, with my established relationships that I do have in the space, I can go to a handful of my surveying buddies, have them do the bid for me, and I'll be a subcontractor for them. Now, would it be smarter for me to go and get my PLS, my professional license or surveying license? Then, maybe. But I don't want to go back through school for that, and I'd rather just leverage my relationships for that. But also, not every single project needs survey-grade quality.

That's what I wanted to touch. Like, we are like right now we are not discussing how to do this without the educational, without a license. We are just trying to explain that there is a plenty of jobs that does not need the the education, the accuracy. You don't you don't need to tie it to the control. Maybe they just need a reference, right? If you're just scanning dirt, do you really need to have 1 cm accuracy on dirt?

Absolutely not. Yeah. Because there's no surveyor that's going to go out there and get you 1 cm of accuracy. So far, they they were having the error of several trucks, so it doesn't really matter whether you are...

[laughter]

Exactly. And I'm not, and I'm not trying to, you know, downplay a surveyor's role because they are still a very valuable asset in the world of construction and it's not just construction, just everything that has to do with land and building. But you don't always need a surveyor or you don't always need to provide survey-grade or survey-quality deliverables to a client for everything. That, and I firmly stand behind that, and I know a lot of my surveying friends will also say the exact same thing, too.

And the trend is there, I think, like most of the people that are coming into reality capturing and creating 3D models or mapping, they usually are not coming from surveying companies.

No, no, they're not. They're coming from other industries that they're looking for a career change or they're looking to add another service or tool to their, uh, to their tool bag and provide that as a service to a client. Architects.

Do you have any idea why like typical surveyors which are on the market, let's say, 15, 20 years? Why they are not like eager to learn something new and to go into this new technology and just staying with the old GNSS sticks and...

Well, understanding the cost of this... The surveyors are going to hate me right now, but I just wanted to ask you.

No, no, actually, I think, um, it's a very valid question. But if you've looked at where they've had to come from, right? Say 15, 20 years ago, GPS was new for them. Using the GNSS receiver, they were like, "Ah, that thing's never going to catch on. We're just going to use our tried and true method of using a laser." Then they saw, you know, the capabilities of a GNSS. Sure, it's not going to be, well, maybe nowadays, but back then it was nowhere near as accurate as some of the tools that existed. But it also allowed them to produce faster results because they're able to collect data faster, right? And that's what we're in right now. We're in this race to get more data, more data, more data, faster, more accurate, right? But not everything needs to be that millimeter accurate for every single project. And I think there's kind of two parties, right? There are those that are willing to adapt and get onto the bleeding edge. And then there are those that are like, "We're good. We'll just subcontract somebody out." But then you also have to look at the cost of the equipment. Some of those total stations are $50,000, and they're expected to continuously keep paying for that. Meanwhile, the cost of surveying jobs are going down. That's a hard ask on their part. You know, maybe we were to blame because we're using all these new cool tools and cutting them out, but we're not cutting them out. We're just doing something that doesn't necessarily require the tolerances that surveying deliverables have to abide by.

Mhm. Mhm. Uh, we have an interesting intake from AG Creek, and he is actually very right, but, uh, he's just saying that after the rain, day or dirt measurement is no longer accurate. But I actually think that this is to confirm the importance of measuring with drone or with whatever, because the dirt, that's true, is changing all the time on the on the on the construction site. It's one thing that it's like changing the volume because of the time, because it's settling down. Then it's the rain, then it's whatever, right? Uh, so, but you can re-measure it again and again to just be sure.

My counter is that are you going to go walk out there with your GNSS and do measurements on that stockpile while it's raining or after it's raining? No, you're not, because it's just going to take you a whole lot longer. It When we're talking about these tools, it's and this is actually how I measure it. It's the level of effort to get that data. All right, what is that level of effort? Are you really going to go out there and map traditionally all of these stockpiles? Actually, I've got a great example. This is about four years ago. I was on a, [clears throat] 300-acre project. They were redeveloping the entire thing. It used to be an old mall and they were redeveloping it into a mixed-use commercial industrial site, whatever. It was $17 million worth of just dirt work. And in total, I think there is a good 95 stockpiles of material all over that site. Um, I flew the entire site in one hour with lidar. In that same one hour, the surveyor was only able to capture one stockpile. Now, granted, yes, it was the biggest stockpile that was there. But if you look at the level of effort that it took to capture that one stockpile versus the level of effort that it took for me to capture the entire site and then provide them an accurate deliverable, and I'm talking a 2-cm accurate deliverable by the end of the working day for them to then go the next day and say, "Oh, wait. We're $200,000 short of extra material that we did not account for." Who cares if it rained that day before? And actually, what's funny is it actually did rain that day before [laughter] when I went and did that site. The level of effort to go and get that data again is a fraction of what it would take to do it the traditional way. Even if you're not, you know, a centimeter accurate, right? Within doing dirt, you got to be a tenth, which is what, three centimeters, four centimeters?

This I think very much correlates to what we were discussing just before, like not every job is a survey job. Okay? And and this just confirms that there is often times the speed beats perfection. So you don't really need like absolute accuracy. You just need it freaking fast.

Yep. Well, especially with the demands from the higher-ups now. They need, they want everything this fast. That's why, for example, like every point app is doing great because you can just pull out your iPhone and you can measure your stockpiles, right? So, yep. This is really exactly the case. Who cares whether it's absolutely accurate? You just need like rough measurements of that.

Yeah. Uh, Dylan, when we were discussing a few topics, you came up with the term KISS method. Yes. Does it correlate to what we were just speaking about? Like, just just speaking the easiest way to how to do it, like good enough?

Yeah. Yeah, literally. The KISS method, there's other two ways to go about it. Keep it simple, stupid, or keep it stupid simple. Right? Either works. Um, Keep it simple, you stupid. Yeah. [laughter] Seriously. I mean, I think sometimes we just get so wrapped up in all of the new technology, especially 3D Gaussian splatting, different LiDAR. I just literally saw a video last week of Ouster. They're announcing a new LiDAR that can also collect color data at the same time without an RGB camera. Wild. Right? That I can't even wrap my head around that. But especially in my position, I know I get wrapped up in all the technology. Yes, I get access to all these cool tools all the time. But you got to also think about the everyday individual or the everyday company. They are still using tools that they've been using for the last five, 10, 15 years. They keep it simple. Because if they start overcomplicating it, they're going to start falling behind on deadlines, they're going to have to build out new workflows, which you do have to do that. You have to evolve. I'm not saying you don't do that. But sometimes even just being as simple as possible for your end client is going to win you the job. Often times I see some of my competitors in my region here, they like to overcomplicate what it is that they're doing and what it is that they're providing. I'll come and say, "I'll just go scan the dirt for you and I'll just give you the output." They're like, "Just like that?" "Yeah, just like that." Keeping it simple. They like simple. They don't like overcomplicating because as soon as they feel overcomplicated, it becomes overwhelming. They're like, "How much is going to cost?" And then it just keeps going down that rabbit hole. Now, there's certain instances where you can't keep it simple because it really is a complex problem that you're trying to solve, but even in the most complex situations, there's a simple method to approach. Um, and I typically use the KISS method in almost every single project with almost every single client, and it's still working for me. Um, and it's what I advise for a lot of my students to do whenever they get out into the industry.

I I think people are overestimating the intelligence and the knowledge of their clients and underestimating the the strength of simplicity if you can deliver it. The other day, I was speaking to one of my partners, and they were asking whether they should be speaking with developers about digital twins, about the measurements of different sensors and IoTs and everything else regarding the laser scanning. And I was like, "No. No, please don't. Just come with the easiest way." If they have like already built building, it's like an office building, for example, and they are renting their spaces. The easiest low-hanging fruit is, "We can measure your office building and we can tell you whether you are renting more or less square meters than you actually have on your invoices, right? On your billings." That's like the easiest way how to use laser scanning for the first time with the building that is in operation right now. Don't overcomplicate things. Just go with the easiest one and with the most simple one.

You know, it's so funny that you say that 'cause I was literally in a conversation earlier this week. Exact same conversation. Um, we have, we have a lot of airports here in the United States. Not, I'm not talking about big airports. I'm talking about a bunch of small municipal airports. And there's this one big company here that owns about 240 of these, um, they call them FBOs. So, the big hangars where the private airplanes hang inside of. Um, they don't actually know the square footage of the hangars. They bought them. They have no idea. And they are a business that sells leases based off of the square footage of these hangars. And they have no idea. They're just throwing a number out that sounds good. And we have gone in and convinced them saying, "Hey, look. We can simply go in and within less than an hour, we can scan your 40,000 square foot hangar and get you an actual estimate of it. Give you an LOD that is more than sufficient enough, a level of detail, level of design, depending on how you want to go about it, of a full as-built for the site. That way you can use it to understand what the actual square footage of the hangar is and then you can bill per square footage for the hangar." Like, "Oh, that's simple?" "Yeah." And this is a company that has hundreds of millions of dollars of hangars all throughout the United States, over 240 of them, and they have no idea what the actual square footage is. They have a rough idea based off of what the previous owner told them, but there is no factual data to back it up with. There's

No scan. You don't need a surveyor to go and give you a centimeter accurate, even though the tools that we're using are going to give us millimeter accurate, um, with but But it's not what you're selling. It's not what you're selling. No, it's not. It's not what we're selling. We're We're solving a very simple problem for them and we don't have to overcomplicate. Oh, we don't have to go in with a drone and do this. No. Scan it, done. Send it to our BIM team, done. That's it.

Uh, this This might be actually, I think, uh, even though I I don't want to position myself into this, but I think this is a good business, uh, tip as well. Like, don't try to oversell what what you're doing and just to overcomplicate what you're selling. Uh, if you are in on intro call or intro business meeting and and you're trying to explain what your company is doing, just pick one simplest thing that you can help the client with because once you're trying to explain two, three, four, five different things, you lost them.

Yeah, yeah. >> goes over their head. >> They're already thinking about their problems, not not about your solutions. Or they're thinking about, well, how do we integrate all this now? Yeah, perfect. Uh, we have some comments. Uh, could you please describe case, please elaborate? Yeah, we already mentioned that it's keep it uh, simple stupid or keep it stupid simple. Uh, Scott is saying that they were using this in in army. Uh, I got it from my dad. My dad My dad was in the military, so >> [laughter] >> he he he taught me that one. Yeah.

Absolute accuracy is really hard to achieve, especially for a long-term assessment. I think this is this is also the case of of what your client actually needs. Uh usually people that don't understand what they need, they are usually asking about the accuracy and the best accuracy possible and whatever. Can you support this or do you do you have any other experience? Yeah, I mean, for long-term projects well, the some of the long-term projects that I typically work on are for warehouses that are being built. And for the most part, they're just tracking the dirt. Um now they still have surveyors that go in on site and they're still tying down to control and using their total stations. Um to get that millimeter accuracy cuz that does matter, right? That that's the difference of a fire escape being you know, a a few centimeters off and a few feet off. And that's a big violation. That's going to cost them tens of thousands of dollars to fix. Where for dirt, as long as I'm I mean, I'm still providing absolute accuracy to them, but in a sense, the initial data is relatively accurate to the site. Um having all of these other methods of control, right? At the end of the day, control is is the most important factor here. That's how you you maintain a long-term um effectiveness of your data. Otherwise, if you're just relying on relatively accurate um ways of delivering your data, you're going to see those shifts. Now, again, it always comes back to what is the purpose of your data? What problem is it solving? And you kind of work back from there, which again is why I also emphasize if a surveyor needs to be involved in a project, I'm going to get my surveyor buddies involved in the project. Cuz I'm not going to sit there and pretend, "Oh, well, I can just do everything myself." No, absolutely not. There There's a time and a place where I can provide even absolute accuracy without using a surveyor. Again, I can't guarantee that, and I have all of those disclaimers in my contracts with my client. But for the most part, once I actually kind of realize, "Oh, I don't really need absolute accuracy, relative is more than enough, especially on 300 acres of land, you don't need absolute accuracy um on that in terms of, let's say, earthwork or just understanding what the land looks like.

Yeah. During the design and build phase? Yeah, that does matter. Uh you have like 20, 30 seconds. I just uh wanted to give a shout-out to to our sponsor of of this episode, uh which is uh Nubigon. Uh Guys, I've been working with the Nubigon for quite some time. It's uh software to just animate a and render your 3D assets. So, it doesn't really matter whether it's coming from lidar or terrestrial scanner, or it's a photogrammetry, or or is it 3D Gaussian splatting, you can all put together, even beam or DTM models, whatever. You can put it all together and uh create nice uh animations and and renders. The most important thing is that you have to understand that usually the one that is paying your bill is not the one working with your data. So, if you want to wow your client and use your data as a marketing, just uh render like 30 seconds video for your client and deliver it alongside the the other deliverables, and you will win the next job right away. I guarantee. Uh nubigon.com. If you guys want uh discount for that, I have some code, so reach out to me directly.

Right at the beginning, we we touched the the company uh Skydio. And when I think of them, I always think about AI, even though it actually, at least from the beginning it wasn't about AI at all. It was just like great algorithm that was able to to like calculate the the paths and everything correct for the autonomous flight. Uh is there already AI involved in Skydio? Oh, yeah. Well, AI is actually kind of the forefront. Um It's actually really interesting the history of Skydio. >> I mean I I want to speak about AI and drones in general, but this was something that [clears throat] I wanted to look back. No, AI has actually been the forefront of the technology on Skydio. Skydio, they started out as a college um a project. It was like a capstone project where they were trying to figure out how do we how do we create a programmatic way for vision to automatically detect obstacles and then find the least path of resistance. Basically, how do how do you traverse that without hitting any objects? And it was an algorithm that they had built. And they're like, "Well, what can we put this on that moves in a three-dimensional space that isn't on wheels, that isn't underwater? Oh, drone." So, they actually put it on a fixed-wing drone and they were flying through parking garages with all the columns and everything. That was their initial start into drones. They They're They're basically a a software company that bolted on to a drone and they became a drone company. Um which I mean that's kind of how a lot of these tools get built out, right? You got to find some kind of vehicle to put it on. Drones are cool and sexy. Um so, yeah, no, AI has been the forefront for for uh They were AI before AI was AI essentially or before what we think about AI. Cuz AI has been around for a very long time. This is not anything new. We're just in a rapidly progressing um state of it that I didn't even think was even possible.

Are you using AI in any way with your >> Every single day. Every single >> jobs? Yes. How? So, there's [clears throat] a lot of um there's there's a lot of repetitive tasks that I have built out for for automation in terms of doing calculations. Um in terms of uh doing the actual data processing itself. They're going to spoil the secret sauce on my end, but um I've built a few uh workflows that once I get the data, it's going to sort everything through for me. It's going to do some error calculation uh based off of what surveyors give me, and then it'll do some basic processing for me. Um but, you know, we're starting to see I mean, AI has been such a big buzzword since chat GPT-3 came out when everyone was able to get access to it. And now you're seeing every single company on the face of the earth, technology-wise, even not technology-wise, integrate uh AI, whether it's on a software front, whether it is on a hardware front, um whether it's full autonomy. I've got friends that are building these crazy algorithms to do change detection on transmission and distribution lines through I mean, it's mostly just uh vision, but it's still AI at the end of the day. Mhm. And it's rapidly progressing. Uh but, you don't use it in terms of flying and planning of uh uh >> Well, more recently, um I actually have been. So, funny enough, I've actually got this guy right here. So, this is the the Manifold 3. It's a I'm going to be sim- simplify it here, oversimplify it here. It's an it's an AI computer. It's just a computer Keep kiss. Yeah, it's a computer that bolts on to the uh Matrice 4 Enterprise from DJI. And it gives expanded capability. One of the we'll call it models that they integrated or that they allow you to download and run onto the drone is their smart explore feature. So, let me kind of describe how this works. Let's say we'll just pretend that this is the the Matrice 4 Enterprise and we're going to bolt it on and if I got some tape here. There we go. We'll just we'll tape them together, right? We'll just pretend that that that's working. So, >> [clears throat] >> ultimately what's going to happen, let's say my microphone is is what we're trying to to do our scan of. Traditionally, I would have to manually fly around this and collect all the data. Now, what it can do given that these drones also have additional cameras around for obstacle avoidance, it's going to use all four cameras plus the main camera and then just kind of fly around the environment building a real-time point cloud with and if you're using RTK, it's going to get you centimeter accurate position of that point cloud. It's going to generate that point cloud in real-time and then process it into a model on that Manifold 3. Once it has that model, it's going to build an automated flight path based off of the parameters that you want. So, what is your distance off of it? What is the resolution that you're trying to get, right? I can get all the way down to 6 mm of ground sampling distance or actual resolution of the model on there. And it's going to automatically build that flight path while keeping the entire physical model of the real world intact and understand where obstacles are to build that automated flight path. This is something that would have taken Actually, a great example this office building that I'm in right now. If I were to capture this by hand or even using some of these smart capture methods that fortunately that the Matrice 300 Enterprise has built in, it would take me probably about 6 hours to do a thorough and detailed collection of the entire building. And probably about 14,000 photos. In 1 hour and 20 minutes including the time that it took for the drone to fly around the entire building by itself, I was able to do that and run an automated flight path and collect just about 6,000 photos that I think produced a uh one or like half a half a centimeter um ground sampling distance model. All by itself. All I had to do was just babysit the drone to make sure it wasn't going to crash into anything, but it it didn't because it knew where every single object was. Which is just wild. And that's just one of the things that this thing can do. I actually have a call next week with another company that's building specific models to run on these drones uh to do object detection. Whether it's animals, whether it's cracks on the asphalt, um whether it is something with a transmission and distribution line. Let's say a thermal heat signature that is throwing off a a flag. Right, there there's so many different use cases that are starting to come out now. And granted, DJI is, you know, not I wouldn't call them at the forefront of integrating AI. There's actually a company here in the United States called FlyBy Robotics. Um in fact, that 3-hour uh uh documentary that I posted last year, I went to their facility in California and they are putting Nvidia uh AI chips inside of their drones. And there's some really cool stuff that they showed me where they're doing full drone swarms that they're all talking to each other. And I can't really go much further than that, but they're they're doing they're doing some just wild things with these drones with real-time processing. Stuff that would have taken going out, collecting it, coming back into an office, processing, getting an output, they're doing all of it in real time. And we're just at the start um of of where AI is going in the space.

Uh, let me get it back for a bit to that autonomous uh flight for for the 3D model. It It must be the stupidest uh question to ask right now, but uh it's it's it's similar to to taxi drivers and uh autonomous cars like Yep. What's What's going to a drone pilot do in 2 3 years? There's still going to be jobs here. There's a lot of highly technical skill sets that are required. And I'll give you a few examples. So, in the transmission and distribution space, right? Actually, one of the things that we're still having to deal with is the part 107 regulations. Here in the United States, you have to maintain visual line of sight unless you get a waiver to go beyond visual line of sight called BVLOS, or you have to get a second person or a third person to be your visual observer, your VO, to watch over that drone. And with how things are here, at least in the United States, there's a lot of friction when it comes to just letting the drone do its own thing. Granted, there's companies here that are becoming successful in letting the drone do full automation. But, there's still hiccups that have that happen, right? There's still anomalies that are going to come into play, such as a bird attacking the drone, or somebody shooting down the drone, or the drone having some kind of magnetic interference, or an RF interference, right? There's all of these things that we think in a perfect world, you know, AI is going to run everything. That's not going to be the case. You're still going to have to have a human to intervene when a problem arises. And maybe the skill level of a drone pilot is going to have to go up in order to maintain your job, which I mean, look at every single industry that has introduced automation, and we've seen that the skill level in order for you to have that higher-paying job or that higher-level opportunity, it needs to go up. It can't just be a subsurface level of I got my Mini 3, let me go make $100,000 doing nothing. You you you probably cannot stay in a position of being an operator. Uh and I think this is a cross-industry like it goes for for every industry, I think. Like being just the operator of anything uh will not keep you the job. Like you you have to be able to to bring on some deliveries, some some like critical decision-making in the future, and and some some other skills that you will have to bring to the table.

>> Yeah. I mean, as as much as we I'd love or not I would love as much as some people would love to live in a world where computers just run everything. I don't want that. Like I still want to have fun. I still want to fly my drone. But there's going to be a lot of very low-effort jobs that are taken by automation. That's inevitable. That will always happen, no matter what industry you get into. So, I think that's just a reality that people have to unfortunately accept. But it is a reality that's going to happen, and you're starting to see it. There's a lot of these newer dock operations. So, drone lives in a dock, and the drone will come to and from, but what about the one-offs, right? You're not going to put a drone dock in one place just for one operation. Drone docks are made for repetitive tasks, right? And that's where the automation is really going to come into play. If something is not a repetitive task, you're still going to have to have an operator go out. So, for example >> It It It might depend because in case of fully autonomous autonomous capabilities of these docks, you can imagine that the robot will bring the dock on site. Just pull it out of the car, put it out there, or keep it on the trunk and just let the the drone do the autonomous on itself. You know what? That absolutely is a possibility, but >> [laughter] >> look, as as much as um Just to give you some context here, Michael. The The United States is pretty big here. You know, it takes me 9 hours to drive from one side of Florida all the way to the other. So, and you know, we think about, you know, electric vehicles and autonomy and all these things, great, right? That That works fantastic in larger cities or when cities are very close to each other, but in the reality of things, especially if you look out in Texas, Texas is a massive state. I mean, if you throw Texas on Europe, I mean, it's pretty big. And if we're just going to rely on automation and robotics to go from station to station, eventually will that happen? Yes, but we're still at a crossroads when it comes to technology, like tech like battery technology. We're We're at our limits right now.

>> Absolutely. There There is a There is a big There is a big difference between something being able to be done and something to actually being able to replace something else, right? There is a huge gap between those two. And there's always going to be friction um that that's met with it, too. I I wanted to Yeah, sorry. I was just saying that there's always going to be friction that that's met with it. As as great as all this technology that exists today, and and I have this conversation all the time with my peers, and and I'm very entrenched into the AI world, not on the drone side, but just AI in general. I'm listening to every single podcast I I can. I'm watching every single YouTube I can to learn about all the new tools, all the new things that are coming out. And there's all of these conspiracy theories that are like, "Oh my god, your jobs are going to be lost. This and this and that." And then I go back into the real world and I drive from my house to my office and people can't even drive their car. How are they even going to be able to use AI? You know, I always go back to that. It's like, "Okay, you know, I can I can breathe a little bit. There's still a lot of opportunity that exists and there's going to be opportunity that exists indefinitely."

I I wanted to ask uh regarding the AI, like maybe one last question uh because we uh already discussed like the the capturing of the data and and drone uh autonomous flights. Uh I'm uh I wanted to ask also about outputs, like whether you're using some AI-generated outputs or some AI automation or I don't know, have you vibe-coded something to help you with the outputs? One thing I I do want to make a clarification on um in terms of the wording, uh AI-created deliverables. There I actually got this comment the other day saying, "Oh, well, doing drone scans is going to be irrelevant because I can just type into my chat and have AI create that model for me." Absolutely, you can do that, but what about the real asset, right? What about tracking what's actually going on in that real asset? So, in especially let's say transmission and distribution or a bridge, yes, I can have ChatGPT create me a 3D model of a bridge. But, what's that going to do for the actual bridge that has been there for 100 years that needs to be that needs maintenance, that needs to be fixed, that needs to check the cracking? That AI model that I'm asking to generate is not going to solve that problem. So, I do want to want to clarify that, but in terms of AI generated outputs of leveraging AI or we'll call them um models to help with the uh analytics and analysis of an actual uh object or asset to then get you the information that you need in a more um streamlined way. Yes, I have actually been using that um more recently. Uh there's a new tool that I just started using called uh Pix4D Element. They have this AI stockpile. Realistically, it's just automatically drawing a bounding box on the struc- stockpile for me. It's just looking at a height map to be able to generate that. So, and that's actually part of that KISS method, right? A lot of these AI models that are being built out or AI automation is using the most simplest way to extract data out of these or create deliverables out of the data that we've already collected. So, they they are basically just creating contour lines. Mhm. So, you don't really need AI for that, right? No. Software Software's have been doing this for years. Exactly. That That That's That's why I laugh so much about oh, well, AI is doing this. It's like, well, actually computers have been doing a lot of automation for us. We just never called it AI. We never have until AI became the big thing. And so, we just slap AI onto everything when in reality it's just a procedure that it's running through. That's all. That's That's That's true.

Uh Norman is asking they are going to to try to combine drone data with the ground scans. Uh so, what they should be aware of. And whether AI tools will eventually handle this alignment automatically. We have a good news for you, Norman. Uh Dylan can start and I will finish. Yes. Well, the good thing is if you have good control on site, then you can align your terrestrial and your aerial data together. And this has been done for a very very long time. Um and that I mean >> your Norman is asking about that. Uh I think Norman is asking about the actual combination of laser scans with the photogrammetry data. Oh, well, I mean I know about the rules and everything. Yeah, I mean Not only about the alignment, but actually using the the pixels for geometry uh reconstruction. So, this we have been doing this for years with reality capture now called reality scan, where you can actually uh upload or or um you can take your laser scanning data. It needs to be uh structured data, so from terrestrial laser scanner, you can input them into reality scan um software. Uh then you can load your images from the site. And if you have done the good job, it should align uh all automatically. If not, you will have to use control points and manually just try to put uh things together. Then you can actually mesh them together, like the point cloud data from the laser scan, you can mesh with the with the photogrammetry um calculations. And then you can use textures from photogrammetry onto on the uh 3D mesh model. What what about outside of reality scan? Cuz I I know how to do that method too, and I've done that for for years with some of the terrestrial scans that I've done to mesh with uh my aerial capture. Um I mean reality scan is such a great tool, and it's free. Uh That's That's That's why I tried to uh like emphasize that if you want to use the the photogrammetry data also for the the geometry and for the mesh creation. Got it. Because there are ways how you can take also like less data from from lidar or whatever. Then use uh photogrammetry just to create uh texture and reproject the texture on those data. Yeah. >> [snorts] >> What >> But this this way you are not actually combining data for the geometric creation. What's your experience with uh processing all this through the 3D Gaussian splatting algorithms? Have you done aerial and terrestrial together in that algorithm yet or no? Uh you mean like from pictures? Well, from pictures and also with >> about the laser scans. Both. Together. It does not really make sense because laser scans uh you cannot really use those data for for Gaussian splatting because they are too too sparse. Too sparse, okay. >> There's there's not not enough data. You would be you you would need to take laser scan every 1 or 2 m, you know, it doesn't make sense. Got it. Even if in a spaces where I was doing quite dense uh scanning positions, I was trying to like uh export 360 images from from like panoramic pictures from from the scanner, but uh it looked like so I don't care about it. Uh but we have done multiple uh jobs where we combined aerial pictures with ground uh pictures and just put all together in in Gaussian splatting. That works perfect. Oh, okay. Yeah, that that is an industry not industry. That is a technology that I think is has been the most rapidly evolving tech. Um at least what I've seen in my time in this space. And if I feel like it's changing faster than just traditional photogrammetry and LiDAR um outputs. Uh you're speaking about the Gaussian splatting? Yes. Yeah, yeah. Uh before we jump into that, do you usually use LiDARs for your mapping projects or is it a combination? It really depends. I'd say actually now over the last 2 years it's been probably 50/50 where I'll do lidar or I'll do mapping. Usually if I'm doing lidar I'm also doing mapping at the same time. But yeah, it's for me it's So you you have both both sensors on on as a payload on your drone? A lot of a lot of lidar sensors actually have built-in built-in cameras. So like this one here that I have from Inertial Labs you can see there's a an RGB camera here and then the actual lidar itself. So it does it at the same time for me. Mhm. [clears throat] Mhm.

Uh, so please tell me you first like how do you have implemented 3D Gaussian Splatting into your deliverables? I actually haven't yet for my clients. It's an area that I'm doing a lot of exploration right now. Um funny enough about 2 weeks ago I posted a video using the Anti-Gravity A1, this guy right here. Yeah. And I was able to actually work with their team to produce a 3D Gaussian Splat model of this water like this pretty famous water tower that's here um in my city. And you know, at first I thought it was a gimmick. I was like, okay. I thought this was a gimmick um and until I saw the final output where a we'll call it a 3 and 1/2 minute flight around this tower generated I'm not going to call it a model cuz it's not a full model, but the 3D Gaussian Splat is like, wow, that's actually kind of insane that I'm able to get that kind of perspective in 3 and 1/2 minutes of flight where I've also at the same time have done a full photogrammetry scan of that and it took me 45 minutes and almost 3,000 photos. Now granted, the output of the model beats the Gaussian Splat because I'm using almost 3020 megapixel images versus we'll call it 3 and 1/2 minutes of two you know, sphere spherical cameras. There's obviously going to be more fidelity in those single images versus what I get out of this, but it goes back to that level of effort. What is the purpose of me generating that model? Am I just doing it to document an asset? Okay, maybe a 3D Gaussian Splat is the best way to do it. Or am I doing any kind of maintenance or am I doing an inspection? Okay, then photogrammetry is the way to go because I need to be able to see that detail. I need to be able to see the cracks and actually do measurements. So, that's kind of where I've been experimenting right now with with 3D Gaussian Splats.

>> That's That's, you know, that's quite quite easy question because what you need What do you need to focus on? Do you need accurate geometry? Yeah. Go with the photogrammetry. Do you need accurate visuals? Go with the 3D Gaussian Splatting. Once you are doing the the photogrammetry project and you are taking 1,000, 2,000, 5,000 pictures the easiest way is to use the very same data set on a 3D Gaussian Splatting software. Just to process it one more time. You have two layers. You have two different outputs and you can [clears throat] use them wherever and however they are need needed to to be used.

>> Well, it also depends on what the client If they just need me to go out and and document it so that they know that it exists and it's there, then I'll just use a Gaussian Splat cuz they're not doing any measurements. They just need to know that it exists and they just need to see it. But, if they're doing measurements, then yeah, photogrammetry all day long. Yeah, because 3D Gaussian Splatting is not not meant for for measurements. Like the the the the principle is very different.

>> apparently it's changing. That's what I've heard. Uh there's some new algorithms that are coming out. Um but, we'll see. It's not there yet. Uh the the thing is that uh very easily, the basic principle is that with the 3D Gaussian Splatting you are placing 3D blobs or 3D objects in the space and they don't care about the the surface, about the edges, about the corner corners. It's just placed in the space and where it's very dense then the the geometry appears appears where it's very sparse then there is like like just just the free space. Uh the thing is that instead of blobs you can use basically any geometry for for these splats for these Gaussians, okay? Uh the things you are speaking about is that people are trying to use like triangles instead of blobs. They're trying trying to use different other shapes and yes uh probably continuously we will be getting more and more accurate measurements on Gaussian splats as well. For me personally I think uh this is something that we should not be running for because you will be doing some visual compromises with this, okay? And the 3D Gaussian splatting is perfect because it's it's brilliantly visually accurate. Like the the main goal is to get as good reproduction of visual reproduction of the of the space or the object as possible. If you need accuracy just do the laser scanning or do the photogrammetry. Why do you want to replace one with another? No, I mean I I I do see that um but I think to counter you know, as we continue to evolve with the technology and we can have more efficient means of data collection and then if we're able to embed high accuracy with it in terms of compressing it down and again going back to that level of effort I think it's a win-win across the industry. Now, is it a replacement? No, not at all. Do I favor 3D Gaussian splats in certain instances where I want a visually accurate representation of the environment? Absolutely. I mean, I think the biggest use case is the reproduction of water surfaces and reflected glass. Those two right there break every single time in photogrammetry. But then when you look at a 3D Gaussian splat, it's like, oh, wow. And you you kind of move around in the model and it shows the same reflection. I'm like, that's That's a more presentable call it a model, I'm using quotation marks here, um of a data set or of an asset or of an environment than using photogrammetry, cuz the trees looked intact. The the glass, the water, all of that looks visually pleasing. That's the goal. Yes. Like this this the main feature of the Gaussian splatting. It's it's not about the accuracy. And once again, even if we did development and and great coding and with help of AI, even if we get like good enough geometry, like why? Right. Why should do this if we already have photogrammetry which can achieve sub-millimeter accuracy if right tools are used, right? And the right control. Right. Why would we need to develop other way how to reconstruct the same right? Right. Right right now that there is clear clear importance for 3D Gaussian splatting in terms of visuals. Why bother with the geometrical accuracy? Right. But maybe that's a discussion for Another time. >> another day. >> [laughter] >> Uh so, but I wanted to uh get back to why are you not using it yet with your clients? Because I don't think you should be like selling it the 3D Gaussian splatting, but at least just just like try the waters, just Yeah, I'm doing some experimenting with it. Um I actually was having a whole conversation yesterday with a uh engineering firm. Um and I get into rabbit holes with them all the time. And they're now starting to test out Gaussian splats. I said, "Hey, I've actually got some access to tools um uh Insta360. They're actually going to be sending me some more cameras to do some testing with because there's actually been a huge interest uh from my audience to learn more about it and dive deeper into the Gaussian splat world." And so I said, "Yeah, of course, I'd love to do that." And then the discussion that I started having with this engineering firm is they actually want to start using Gaussian splats for their walk-throughs um on their sites because they're getting sick and tired of just using the 360 camera just from place to place to place, and they want to be able to have a more fluid uh walk-through of these environments, but also track it over time, have an archived history that they can go back and forward just to see where things were. They're not doing measurements, they just need a visual documentation. And I see that as actually a huge benefit for that. Um Now, uh an industry that I think would greatly benefit from 3D Gaussian splats is also the real estate uh world. That's actually the the industry that my parents primarily uh work in for almost two decades. And so I'm actually starting to have some discussions with them on how, "Hey, rather than using something like a Matterport, right? Everyone knows what Matterport is or at least has is aware of what Matterport's outputs look like. It's that circle on the ground that you click to go see that 360, but it visually doesn't look that good in my opinion. It It was good for the time when it came out, but they have not evolved to where the technology is today." Totally agree. And I think where where we're headed um with companies like X Degrees, with companies like Insta360 that are developing these uh cameras or these tools, scanners to collect this kind of specific geospatial data, and then having these new advanced algorithms for 3D Gaussian splat, I think it's going to open up a new world in opportunities for uh visually displaying assets or environments that commonly would not have been able to be achieved with just simple 360 photos. Or walk-throughs, I should say. Yeah, let's see. I'm I'm I agree with you about the Matterport not not being too too evolved. Uh some some I have heard some some rumors about their being into Gaussian splatting and and exploring this space. Uh I think the biggest problem they have is with the quality of the output data because if they want to rely only on the stationary images from from the Matterport cameras, uh while so far they were doing their best for the least amount of position to create good enough model, you know, >> [clears throat] >> you will not have enough uh pictures and enough overlap for for to to be able to produce a a nice and pleasing free Gaussian splatting from from that. Right. >> So, this is probably something that they are struggling with. That would be just my guess. I'm right there with you. >> [laughter] >> Yeah.

Uh, I wanted to touch because we already mentioned the the lidar and that you are doing lidar maybe 50% of of the time. Where do you see the development of of uh this? Is it going to be more accurate or is it going to be like what's the future of lidar scanning with the drone? Yeah. Um that's actually something I am I'm going through I I was going to rush it out, but then I decided not to because I would have done a big um disservice to the evolvement in lidar, but um I'll explain what I mean here. Uh I started working on this research and development project to compare all of these different LiDAR sensors not against each other, just just in general. Um of what their capabilities are, what their accuracies are at different output or different altitudes, different parameters, to really see, right, what is the value that you're able to extract out of the money that you're investing into a sensor. So, for example, this sensor that I have right here, this is about a $35,000 sensor, LiDAR sensor. Now, DJI, they have their we'll call their $18,000 LiDAR sensor that they just came out with not too long ago called the L3. Wouldn't the $35,000 one be better and more accurate? No. But also, yes. Right? So, that's a very confusing answer that I'm giving here, and it has to come down to a lot of different factors. One, what kind of laser is it using? That new L3 is using that 1535 or 1550 nanometer wave, where this uh XT32 that I have uh from uh Inertial Labs or really from um Recepi. No, not Recepi. Who Who's it Who makes it? Um Hesai, sorry. Hesai that makes the the LiDAR sensor. Right, there it's either a 905 or an 835 nanometer wave or an 805, right? The These are different kinds of lasers that are being used um in these scanners. And what we >> Uh can you So, sorry. Can Can Can you tell the audience like what does it mean? If If they have different wavelengths, what does it actually mean for the outputs? Well, no, that's exactly what I was going to get into, right? All All All of these different sensors, if you look all the way back from where we came from, when not aerial LiDAR from manned aviation, but aerial LiDAR in the introduction to drones, so think of companies like Regal and Phoenix that were putting out those sensors that cost $250,000. That was such a huge barrier to entry, but also it was relative it was a relatively sparse point cloud that was being generated. Not that many returns, meaning like how how frequent the lasers are coming back to the sensor. They were accurate, but they were not as high accurate. You had to fly them low, you had to fly them slow. They were clunky, you could only fly them for 12 to 15 minutes. Where now as we have gone up um or evolved in the technology to something like the L3, I could fly that if I were to get the legal requirements to do that 900 m in the air and still get 3 cm of accuracy. Where with those other Regal sensors that cost you $250,000, your sweet spot was a 160 ft or what's that? About 50 something 55 m um above the ground. That right there is a huge factor in terms of speed, the cost of operation, and the effective data, right? The accuracy that you're getting out of these sensors. So, to kind of go back, right? Why would you get a 1550 nm wave lidar sensor versus an 805? Different kinds of penetration, different kinds of accuracy, different kinds of capability that you can do at different altitudes, different speeds, different parameters. And what we've seen is the it evolved from $250,000 all the way down to about $20,000. And that $20,000 sensor is outperforming that $250,000 sensor sensor by a factor of like 25 to 1. Which is insane.

>> This This is giving very different perspective on my next question because I wanted to ask you like how would you compare and where would you draw the line between using on a drone a lidar airborne lidar compared to those slam solutions that you can actually put as a payload under the drone and using them airborne. Yeah. But these slam solutions are their costs are going like 30,000 plus, right? And now you're mentioning the the lidar system that is actually around 20k. Yeah. And I'll be a little bit more realistic. Yes, it only works on DJI. So, if you wanted something a little bit more comparable, um you would spend probably about 40,000 US dollars um for a comparable sensor that you can put on to any any drone cuz a drone is just a vehicle. >> in a very very similar still in a very very similar price tag, yeah. >> so I still think there's quite a ways for these hybrid slam {slash} aerial lidar sensors to to to go. You know, one of the biggest problems with slam is drifting. Um that happens over a longer period of time. Now, granted, that's more of the algorithms at play, not necessarily the sensor, right? The All of these slam scanners are using a Hesai lidar for the most part. Um it comes down to the algorithms and every single one of these sensors have different algorithms that they go through for processing. Now, when you're doing airborne lidar, you already have some kind of known position. That's why they have the GNSS receiver on it. That's why they don't need to run through a a slam algorithm to figure out what was captured. When you get into that hybrid approach, you're mixing you're trying to basically use a Swiss Army knife to solve a bunch of problems rather than using a a tool that's intended for one purpose and one purpose only. No, I'm not saying that things aren't going to change over time. But to kind of answer your question, you know, why why wouldn't I just use an Emesent um scanner over, let's say, the L3, right? Well, Uh but to be clear, not for mapping, okay? We are not speaking like large-scale mapping.

>> Okay, well, that's what I was going to go towards. If it's large-scale, I'd much rather use a purpose-built lidar scanner, specifically one of these long-range lidar systems, which that's what the L3 is marketed as. CHC Navigation also has long-range lidar systems that you can be 900 m up in the air and get that few centimeters of accuracy. That actually helped me big time on a massive project that I worked out in the Midwest. 700 acres that I captured in 57 minutes and had 2 cm accuracy. I would not have been able to do that with that other sensor that I have next to me. That would have taken me 5 hours of data collection. That's where that kind of comes into. Again, it goes back down to that back to that level of effort. What's the level of effort that it's going to take for you to capture all of that all of that data? And that's how you start measuring all these scanners and sensors. And as we continue to evolve with the technology, it gets cheaper, faster, better, more accurate. And that that that theme is going to continue on. And that's what we're kind of seeing with aerial lidar. It's starting to get cheaper, smaller, faster, better.

At the beginning, we were doing the poll about uh about how many people are using the the drones in the day-to-day um jobs, and 2/3 are already using drones. 1/3 was not using drones. So, in the next few minutes, I wanted to discuss like the low-hanging fruit or the easiest way how to enter the market or uh as as you mentioned earlier, like how to make big money with small drone or or just just give me the simplest way how people could start with the with the with the Yep. doing models with from drones. Look, I I I think it it goes back down to that KISS method that I was talking about at the beginning. You don't have to go out and spend tens of thousands of dollars in gear to be able to compete at the highest level. Because reality, you're not going to be competing at the highest level yet. You got to use these opportunities as stepping stones. And what I mean by opportunities is involving yourself and and and putting yourself out there. As simple as approaching somebody in real estate, right? Using your existing contacts and relationships. Theoretically, I can use this Mini 3 to do an inspection with. Right? It [snorts] doesn't need to be positionally accurate with using technology like RTK. But, maybe it's just a roof inspection. Right there, that's a $150 job that you can work with a roof contractor. It saves them from having to go up on the roof and walk the whole thing. You can go and do spot checks with them. Or maybe you're doing some marketing, >> [clears throat] >> right? Again, a drone as simple as a Mini 3 can can get that done. Or hey, maybe somebody just wants to understand what a small plot of land looks like. Again, you can use a Mini 3 to automate collecting the photos in a systematic way and running it through an application like WebODM or even Reality Scan because they're both free and you can generate a 2D output. It's going to be relatively accurate, not absolute, but maybe that person doesn't care. Maybe they just want a simple map of their property. You don't have to go super complicated. You don't have

To go down the rabbit hole of, all right, I need this GNSS receiver, I need this RTK, I need this thing, this this no. Get comfortable with it and also involve yourself in conversations that's going to uh encourage I don't want to call it innovation, but just having interesting conversations with with individuals.

Like, one of the biggest advices that I give to people that are just getting started, if they They know where to go, go to those town hall meetings, and I'm this is more of like the construction approach, right? Go to the town hall meetings and understand, hey, what permits are being pulled for this new commercial development that's about to happen? The people that are going to be at those conversations are the developers themselves. Those developers need to have insight on their properties, even if it's as simple as just some photos and videos. You start there and use it as a stepping stone, right? You get your foot in the door, you start providing them an asset that that they like. Hey, have you thought about you know, I I see all this dirt coming in at a site. How are you guys measuring that? How are you guys validating that the contractor is actually providing, you know, accurate results? Oh, well, he just says that he is. Would you mind seeing the scan that I did for you, right? Providing as much value as you can, but also understanding how these certain industries can leverage the technology. You can't go in to a roof inspection and say, hey, I'm going to throw lidar on this because because lidar. No. Or go and scan a stockpile with thermal because the colors look cool. No, you got to do some due diligence as well. You got to understand the industries that you're trying to sell to because at the end of the day you're providing a service, and if that service doesn't solve a problem or doesn't provide some kind of ROI, remember ROI doesn't always have to mean money, then the service that you're providing is completely useless to that end client.

It's important to say that buying any gadget is the easiest thing to do on the whole business you are going to do. Everybody has a credit card that they can use to go spin it and put it on there. Yes, exactly. There is a whole bunch of procedures and skills and knowledge that you have to learn or gain or buy or have to be able to do this business. And buying the the most expensive drone is the most not the most important one, not in the list, I think. I I Even. So, just to be aware that if you want to start with simple photogrammetry, with simple scanning, with a drones, even the least expensive drone, the the most cheapest one which actually have the camera, can be used for for photogrammetry. Yeah. The same way how you can use your smartphone for doing a ground photogrammetry, it's it's the very same. So, just to enter the industry, just to be able to do your first model, just use whatever you have right next to you and and the the the easiest cheapest way.

I agree. And that's what I told everybody. You know, you don't have to start with the latest and greatest in technology. There's people that still ask me, "Hey, I have a Phantom 4 Pro." Which you can kind of see it up in the corner, if I can point, right over here. Wait, over there. Yeah, there we go. Um, that drone is still fantastic to to use. Still 20 megapixel camera, it can still provide reliable results. You just got to start somewhere.

Dylan, I wanted to ask you one last question because, um, beside the fact that you're entrepreneur, you are doing great jobs, you are uh influencer, you are creating a lot of content on a YouTube. You have almost 140k subscribers on YouTube. Guys, subscribe to Dylan. He's doing amazing uh, content. You also are doing educational or how would you how would you call it like you have a Pilot Institute company that is also teaching people how to do the right? Yes. Yeah, so so Pilot Institute is primarily a training, education, and consulting company. So, we work with companies all over the world. And companies can be even just one person, right? So, we individuals that are looking to get into the space, companies that are looking to leverage the technology. I do everything that I can to provide all the tools and resources, whether it's through my YouTube or through my platform Pilot Bite, through courses, through online training, through uh in-person training. I actually have an entire facility uh that I do it all at uh for both mapping, LiDAR, and we're even expanding beyond that to give a good foundation on what this technology is and what this technology can do and provide as a reliable uh deliverable or just a reliable service to to to clients.

Um, Now, well, a lot of people ask me like, "Why are you training your competition?" I'm not, honestly. Uh I actually collaborate with a lot of the people that I work with. And this industry is built on a lot of relationships or building strong relationships, uh building reputable uh individuals that you can rely on, uh alongside with a reputation that you can deliver what it is that you're saying that you're delivering. But also, it for me, it's helping set a proper foundation for the right tools, the right assets, the right approach, the right workflows to deliver consistent reliable data. Because I think like what we talked about at the beginning of this, anybody can go out and buy a drone and just say, "Hey, I can do a model." Okay. And that's where I think education is very important. Not in my own bias of okay, you know, yes, I that that's where I make my living of of educating, but educating the the masses as best as I can, alongside not just me, but you know, I my my other we'll call them competitors like Pilot Institute, Drone Launch Academy, um UAV Coach, right? They're also colleagues of mine that also do education. And our goal is to make sure that people are doing things the right way, the legal way, and you know, not putting out misinformation of what a a tool or a software can and can't do. Uh because at the end of the day, that hurts us, that hurts the industry, that hurts people like you and me that are trying to go and sell this service or provide you know some kind of ROI to a client. So that that's why I have built and and continue to run Pilot Bite for the last 4 years now.

Thank you so much for sharing your knowledge, your experience. Thank you for coming to this live event. I wish you a good luck. Good luck with your projects. You said that you have a lot to do in next few months. So, just just wish you wish you all the best. And again, thank you for coming here. Thank you guys for watching this, for spending with us 100 minutes. You guys are amazing. Thank you. Thanks. Have a great day.

Get Our Mobile App

DJI Ban, AI Drones & LiDAR Mapping: Dylan Gorman on the Future of Drone Reality Capture

Transcription