Underground Utility Verification and Coordination Using Drones

MAN: You guys can fall asleep to John's beautiful voice on the recording later.

JOHN NILES: Better?

MAN: Yep.

JOHN NILES: That's a little better?

MAN: OK.

JOHN NILES: Do I need to start over?

AUDIENCE: Uh, well, people--

JOHN NILES: You guys don't me starting over, do you?

MAN: No.

[LAUGHTER]

All right, I'll spare you that. So I just want to make sure that everyone understands, there-- are-- is there any surveyors in the group? OK. I want to make sure that everyone understands I'm not looking, by presenting this, to make the surveyor obsolete because absolutely what we're looking for are coordination issues for our design team, for our subcontractors to look at, and additional deep diving into some of those locations. We just want to flag those and then we get a surveyor involved in that.

So the presentation you're going to see was created on the laptop that we're using here. If you guys want take a picture of this, great. I'll hold it up here for a second. All the animations that we're doing today, all the work, all the processing that we're using was done on a laptop. No big, huge workstations, just a basic laptop.

So this is our workflow and we've got a-- we've got bookends that I call the capture process. Once you fly once and some of you guys already know this that once you fly once, you want to fly all the time. We found, oh, we're just going to create one flight. We started flying twice a week. We could fly every single day if I could. That's how much change occurs on the site and how much we would like to capture. So we put capture at the bookends, Jackson is going to talk in great detail about that capture process because there's a lot to that.

Our workflow, I've broken it down into three basic components. That's going to be the process, the fuse or the fusion, that's bringing the models together with the point cloud, and then the visualization, sorry. And each of those are broken down. We'll use ReCap for the process portion of it and then Navisworks for both the fusion and the animation portion.

So let's talk about the process piece. We're going to assume that we've got a good LAS file, a laser scan file, from our drone provider and we're going to download that and we're going to put it into ReCap. We're going to take that unstructured data and index infrastructure data and we're going to update some origin points on that. So again, we're just-- we're going to show some quick videos on just that process. All right, now we're going. OK, so one of the things that we want to do is we want to go ahead and pull the LAS file into ReCap. You don't have to have ReCap Pro to use this. You can use plain vanilla ReCap, which is free, actually.

We'll download that point cloud. We're going to save it into a point cloud folder. I put it on the desktop, I don't want to put it on the network because I don't want to hog up the bandwidth on the network so I'm going to drop that on the desktop and we're going to name that file something that's intuitive for Navisworks. So you can sort it and then you can start to look at the laser scans or the point clouds over time. You can flick them on and off, toggling them on and off, and then we're going to index-- so I've pulled those files in, we're indexing them, we're basically taking that unstructured data and putting some structure to it. And that's going to go through-- this usually takes about five minutes. So through the magic of movie editing we can make that process happen a lot faster.

We've now pulled that point cloud in and what you'll see is this little origin point grid. And I'm not sure why, if anyone knows why I would love to talk to you after the class, but it comes in and what we want to do is adjust it. And the reason we want to adjust is so that it makes a little bit more sense when we pull it into Navisworks. So we're just going to pick this corner point over here on the building that seems to stay fairly static. Early on, before construction starts, you may take a construction trailer and use that.

And then we're going to go in and we're going to change the point density on here just to fatten that up a little bit and make that just stand out a little bit better so we can see those slabs. And by doing that we can see our second floor. What we're going to do is we're going to adjust that and set that origin point on the bottom corner point. We basically want to get that, you know, that x, y, z section that we want to see.

That's updated, it's-- now move this over. You can do this all in one command. I chose to just break this out a little bit. So now I'm going to update that z-origin, that z-axis, and what I'm going to do here is I'm going to actually pick the second-floor corner point. And again, if you're using ground control points this gets a lot-- a lot quicker but it also allows you to dial it in a little bit faster. But I know most of us are not using ground control points.

So we've updated that z-axis and now we want to update the y-axis and the x-axis. And so that will now update, that grid's in the right location, we want to go ahead and save it and then we're going to jump out of there.

So Jackson, you might want to talk about the ground control points for a second while I get this back up.

JACKSON DYRE: Yeah. So I think what John and I discussed earlier also was, you know, this is the reality of it and so they are obviously workflows for establishing ground control on site. I'm not a surveyor, I don't claim to be, either. We-- you know, typically by the time John hears a request-- and we now have fallen into a process with this-- we've captured the site.

We'll talk a little bit more about how we generate point cloud, but ultimately, you have maybe color arrow points or you have monuments or just spray-painted control points on site. We do have one of those for tagging those but-- my mic is not on. How's that? Is that better? Thanks, Michelle, appreciate it.

So we do have these workflows. We establish ground control, we can absolutely talk about that after. In fact if you're a surveyor I'd probably like to learn from you after this. But what we have here is non-ground control LAS. We process that using imagery without knowing that we would need a point cloud later. It was just a standard job site capture for progress documentation and communication with a little bit of context.

So it's the reality of the situation when you don't-- this workflow is really handy as an ad hoc workflow when you did not expect to have to do this, orienting that data and aligning the point cloud with the actual BIM is, I think, a handy ad hoc tool. But I'd love to talk to you about the ground control, establishing those. It is a little bit more complex and it does take time on site. You've got to verify them. I think the ground control-- the arrow points that we use take about 45 minutes to register. So happy to come back to that in my portion.

JOHN NILES: Thanks, Jackson.

JACKSON DYRE: Yeah, you bet.

JOHN NILES: So just to recap what we did in the process portion of this is that we've used ReCap to turn that point cloud into something more useful in Navisworks. A couple of lessons learned early on that we didn't realize is make sure you actually specify an LAS file if that's what you want to use. ReCap is free, which is great. Look at your point density, that's helpful when you get in there and make sure you know you're going to fidget with this a little bit when you're doing this manual process. You're going to do it a couple of times until you get into it a little bit more and get familiar with it.

So the next portion we want to talk about is how do we actually get that Navis-- or that point cloud into Navisworks and then how do we move it around a little bit so that we could make sense of it in the field?

It does not like my videos. OK.

They're doing a great video here.

JACKSON DYRE: And I think you just duplicated the screen, right? Instead of extending?

JOHN NILES: Yeah, I don't know where it went. Yeah, same thing. There we go. Thank you, Michelle.

OK, so we're-- let me back this up just a little bit. OK, so we're going to just fire up Navisworks Manage. I'm using 2019. We didn't do anything special to it. We're going to go ahead and open up that federated model, it's also got all the utilities laid in, and we're going to scroll out and we're just going to pick that site scan file that we had. Again, we date-stamp that.

AUDIENCE: These utilities came from Civil 3D model from the Civil channel, or you guys modeled it in Revit?

JOHN NILES: So we modeled it in Revit. We actually worked with the civil engineering files to duplicate that and then pulled that data and that's one of the things that we're looking for are points that cross each other where we have zones of influence in the foundations and out in the site. So you're going to see that we've pulled that point cloud in. It looked pretty good until you actually brought that elevation into view and so the point cloud came in a little low. By having set that-- reset the origin in ReCap, this gets us a lot closer to being able to line those models up right away.

And we'll go ahead and get those set.

AUDIENCE: These are both existing, as well as the plans to do this [INAUDIBLE]--

JOHN NILES: So, and this is a-- it's a Greenfield site so a lot of these utilities were already-- not-- well, they didn't exist. So that's the opportunity. Some of the ones that did exist were moved out of the way prior to our work.

So we're able to now line that up and immediately you can start to see that there are some things going on that, for us, are flags. So immediately I look at this and-- I'll just back it up for a second here. All these along here are of interest to me. We need to reach back out to our electrical team and say, hey, what's going on here? We're way high. What-- what's happening? Do we need to lower those? Same thing with some of the storm-- maybe we just don't have all the final grading in. It could be as simple as that, but it's asking those-- it's starting to show those-- bring to light questions that we need to answer. And that's where I think the value of this is. And it starts to answer the question, you know, can we use drones for this stuff?

So we're just going to run through this. We're running around the edge of the building. We see a lot of manholes that are too high so there's obviously some work that needs to be done there. We also can see that a lot of people are not parked in the right place, that's another issue. We also have sewer-- or manholes that are in curbs so those are things that just start to pop out at us that normally wouldn't probably pop out without that data.

AUDIENCE: Did you guys try to model the proposed grid?

JOHN NILES: So that's a good question. That's-- that's something that-- an ongoing workflow that we're starting to incorporate in, yeah. So one of the ways that we can do that is we can-- yeah, it is.

JACKSON DYRE: I can speak to that a little bit more at the end, too.

JOHN NILES: Yeah. I was going to say--

JACKSON DYRE: Really what it comes down to is making sure that you have accurate vertical accuracy so the horizontal axis is typically comparable to a traditional survey, you know, we do this when we generate the point cloud. When you establish ground control on site you have to get better positioning and accuracy in the vertical direction. So what we typically say at Hangar is that we can get within about six inches vertical accuracy and it's kind of a CYA. You know, it really depends on a lot of things like lens distortion and the lighting conditions and the quality of the stitch and processing.

But again, ground control can solve a lot of these problems and so for recurring-- we'll do cut fill analysis and we'll do all kinds of grading. The ability to fly and establish a baseline and then measure from that and then run sort of comparison of two different surface models allows you to get some pretty interesting insights. And that's a whole other seminar, it's next at 12:30 PM with 3DR Site Scan. I think we have someone here. But definitely can talk to that.

JOHN NILES: All right, we switched back OK. All right just to recap what we did in the fusion portion of this, we imported the point cloud into Navisworks. We also did some measures using transform and then-- and then a couple lessons learned that we came to find out as we started flying more drone flights is we needed to have a naming convention that made a lot of sense. So, you know, as you get into this a little bit more, name those files. They will get big and then at some point you'll want to get rid of some of the older ones. Make sure you save as an NWF. I know that sounds kind of crazy but it'll collapse all the files down if you save it as an NWD file.

Also, if you're going to share your files with someone else, your partners or anything else like on a box account or Dropbox or 16-- or 360 or something like that, make sure that you copy all the RCP files, all the folders off your desktop and onto that drive. For some reason it needs it and if it doesn't find one of those files it'll create what they call voxels, and it'll look like a porcupine exploded inside your model. You'll get a lot of questions right away from your team members, what's going on? Why is it not working? It has nothing to do with the display, it's just you didn't have the right file pulled over.

AUDIENCE: It's interesting because Andy [INAUDIBLE] took the lead--

JOHN NILES: Exactly, yeah. Yep, absolutely. OK, so the next portion of this we're going to jump into is visualize. So, great, we find all these issues but how can we share them without having to actually share the whole entire model? A quick way to do that is through visualization. So we're going to create some animations using Navisworks. We're going to create an animation folder and then we're going to create some viewpoints associated to that animation folder and then we're going to import those camera viewpoints and export them out into Movie Maker.

JACKSON DYRE: Will it play from there?

JOHN NILES: Yeah, now it'll play. Seems to have frozen.

Michelle, do you want to come work your magic again here? I should probably pay attention this time and-- I--

JOHN NILES: Next year, I'll be doing a class on PowerPoints.

AUDIENCE: When you have that problem, do you model these--

JOHN NILES: I just need to see the desktop.

AUDIENCE: --even if you have not [INAUDIBLE]--

JOHN NILES: Yeah.

AUDIENCE: --just a standard practice, or it was done only because you were trying to compare with the drone [INAUDIBLE]?

JOHN NILES: Can you say the question again?

AUDIENCE: I said that modeling [INAUDIBLE] utilities in Revit [INAUDIBLE], is that a standard practice that you do on every project even if you don't have the raw data, or you model them because you were trying to do this exercise?

JOHN NILES: No, we're actually modeling them as a course of, we're trying to make sure that we've got all those utilities coordinated beforehand. Yeah, absolutely. Yeah, that's weird. Usually I can just--

JACKSON DYRE: On this project, too, we've also [INAUDIBLE] process. So John brought us in here, ultimately we started just with project documentation. The team-- this is-- they got hooked on this and they-- I guess, how often are you doing this now? Twice a month, I believe you said?

JOHN NILES: Yeah.

JACKSON DYRE: So we have fallen into a process now where they are continuing to do this for the several-- the five data centers, I think, that you guys are building out where we continue to process that point cloud. So it's, you know, John can speak to how they're modeling it more but it's something that we've fallen into a process and at least for these projects we've built out our own workflows, knowing that the product team continually asked for it.

JOHN NILES: Yeah, we-- we're finding that we could be flying daily, I think, at this point. It-- sometimes, you guys know, there's-- on the site it changes daily and so the more we can fly, the more we can capture, the more we can see, the more we can document for future use. So back into our facility management team that's going to inherit all this data, they know where all this stuff is buried at now with pretty good accuracy. That means they're not going to be digging into something that they shouldn't have.

So we're going to go ahead and create this animation and there's a couple of steps that should be followed on this. The first thing is make sure when you create your animation or you're going to start with it, right-click on your saved viewpoint and create that animation folder. And then what we're going to do is we're going to just pan and zoom and fly to a couple of location points that we want to showcase and we'll go ahead and save those viewpoints.

And we'll just move our camera angle again to a couple new locations and we're just going to create some additional viewpoints to save views till we get about five or six. I don't like to have anything more than that because of the speed it takes to export. You can break these into segments and then put them back together in Movie Maker.

So one of the things to be really cautious about, and I learned the hard way a couple of times, was to make sure that you actually, when you're capturing those viewpoints you'll notice that those views-- just move it back there. Notice how they are right underneath the little film roll? Make sure that they actually are in the animation folder. This will drive you nuts the first time you try to make an animation and nothing comes into your animation. I've got a few of you guys smiling because you've seen that before. So that was definitely a lesson learned.

We'll go ahead and open up the animation button and the animator tool box and we can import those camera points. Yeah, they're going to be those black diamonds. We can adjust those out a little bit to where we want them at and then we'll go ahead and export the animation. Now on the handout I've given you settings for the export animation. These by no means are the ones that you should follow. These are the ones that work on this computer. You may have to tinker with it a little bit to get it just right but I did find that making sure that you save it out at 60 frames per second helps with the clarity of the file as it's pushing out so it's not pixilated.

And now we'll just use Movie Maker to pull that in. Movie Maker is another free piece of software that you can utilize. Because they knew I was going to be speaking this month, Windows announced a new Movie Maker for Windows 10. This will work in Windows 10 as well so you can use the old version or the new version. Both of them are equally as good. So then pull that in. We'll just import that file in and it'll take a couple of seconds.

It's now in and what I want to do is I'm going to run through that and it seems like it's too fast, right? It's just burning through it. So what we can do in Movie Maker is we can slow that down. We'll go back up and we will change that playback speed to quarter speed and that looks a little bit better for our purposes. And we'll put a-- we'll put a header and footer on here and some additional videos into this if we need to illustrate some additional things and then we have something that we can post out or we can send to a client or a surveyer or anyone else that may need to see this data so we can act upon it.

So to go back to the original question, what if we could use drones to verify our infrastructure and we could use it for coordination purposes? Do you think that some of this would be able to do that? I think so. So I'm going to turn it over to Jackson now. I'm going to get the PowerPoint back up for you, Jackson, and--

AUDIENCE: Quick question?

JOHN NILES: Absolutely.

AUDIENCE: Real quick, did you ever think about actually embedding the point cloud and providing that to the client so the whole file-- [? the WD ?] file-- and you can embed the point cloud. If it's a big file, and you have to have all sorts of [INAUDIBLE]. But do you do that just because [INAUDIBLE] provide a [INAUDIBLE]?

JACKSON DYRE: Mhm. I'll defer to John.

JOHN NILES: I'm going to say-- here's what I'm going to say. I'm going to say no because I've got a better-- I've got a better idea and Jackson's going to talk to it a second here. It's pretty compelling.

JACKSON DYRE: I'll do my best. We have some ways of previewing it and we have some R and D that I-- I hope next year I'm here in front of all of you guys announcing that with Autodesk together, but we'll show you kind of how we present it and deliver a web shareable link with that point cloud modeled out.

We're doing some really cool things so after remind me to tell you about some of the Revit models. So we have-- one of the things that we've sort of made a name for ourselves in the initial-- we're a two-year-old company, by the way, I'll give the formal introduction. But we call these 360 is the Hangar 360 so these are just like a typical 360 camera from a Rico Theta or whatever you might be using. And we'll drop in Revit models and then do 3D magic is what I call it but reflection mapping, materiality. We'll match that, orient it, work the model to have the vanishing points line up and look photo-realistic.

So it's almost-- what you end up with as an output is a web shareable link, and you have a branded viewer essentially with the model, the Revit model, dropped into these 360s. We've gotten a really good response from those. So it's not an off-the-shelf, turnkey product, you know? And we also don't want to compete with Autodesk.

You know, there's a lot of tools that are fantastic at this now. But we've done some really cool stuff just to throw at least Revit models for developers or real estate brokerages. But I think we can start to do that with BIM in some more interesting ways and have these workflows be, I would say, somewhat automated in a sense. But that helps an unsophisticated user quickly bring it up in a meeting and navigate the same way.

AUDIENCE: What was your guys' role on this particular project?

JACKSON DYRE: Yeah, so I'll speak to that. And I can go ahead and just get started. You know--

JOHN NILES: Yeah, I'm going to restart the computer real quick.

JACKSON DYRE: OK so again, my name's Jackson Dyre. I started-- well, I work at Hangar Technology. I'm the director of customer success.

That means anything that happens post-sale, I ensure that there's a smooth transition from the salesperson who promised the world to the actual delivery of the client. I used to be that salesperson. Now I got frustrated with things that I would pass on. You know, there's a lot of complexities, a lot of-- every construction site is different.

And I think that leads to system of the industry-- you know, the intractable productivity gap that exists today. I try not to talk about that because everybody talks about that. But I really like getting out in the field and spending time with our clients, rolling my sleeves up. And I see a couple of folks here that I've been on sites with before. So that's cool.

We're a drone technology and data management company. So ultimately, my path to the company was through a company called Autopilot, which we acquired back in the-- and I have a slide for this. But back in the day, we had a consumer app.

We wrote our own flight controller for DGI drones. It we had full control over every component, modular component, of the drone. You could tell that drone what to do, put it anywhere you'd like in space and time, hit a button, and it'll execute a flight path that can be shareable between users.

And what that provides is the ability to capture consistent data consistently over time and, regardless of the skill level, of the pilot, take an off-the-shelf drone and capture data like this. So our role, to answer your question, was to capture-- so John and Hitt Contracting hired us to help build and scale their drone program. So we offer both the service-- so we have a pilot network that we've curated.

They're licensed, vetted. They have the same hardware. They're geographically distributed around the world, mostly the US. We have a technology platform, which is really what we want to be. It's a product company and sell the technology platform.

It's very agnostic to who flies if you have your own drone pilots That's why we ask that question. We're totally fine with you guys flying it. We'll just give you the apps and train you just like a normal pilot would.

So we're providing both the pilot and the technology platform that helps collect the data consistently. On these five projects, we're flying twice a week for a few years. So we're getting constant data, and we'll show you that. We automatically process the data. So we capture it using autopilot.

An operator shows up and hits a button. It captures its data, uploads to the cloud, processes, and then we deliver a web hosted link. And that's where John would actually download the raw files and the processed OBJ, the mesh, or the LAS file. So you have some interesting ways of interacting with that with the actual images. But you also have the processed assets as well.

And then what we'll do from that point on is take those back into ReCap. Either you can reprocess it in ReCap by taking the zip drive, just the raw photos from the folders, and then generate the point cloud there. Or you can upload the LAS file, convert it to different file formats-- same way. So-- awesome, OK.

JOHN NILES: There you go.

AUDIENCE: Let's start with the model.

JOHN NILES: Sure.

AUDIENCE: So when you process the data and they get the link, is that providing a service of [INAUDIBLE]?

JACKSON DYRE: Yeah.

AUDIENCE: Or it's a software that they subscribe to?

JACKSON DYRE: It's a subscription, and it's a bundled service component baked in. So the variable cost just depends on the labor-- so are we supplying a pilot or not, what tier pilot do you want, you know, what-- are we establishing ground control points on site? That might be a tier four pilot, which is the highest tier.

It's a little bit different pay. If you're flying for yourselves, just a weekly subscription, monthly subscription-- we try to be focused on projects, but we actually work with enterprise. So one of the things that we're super fortunate, two years-- we started Autopilot in 2014.

We raised venture capital and rolled up a bunch of other service companies and smarter people than me to build out what is going to be Hangar. So super fortunate in two years to work with a bunch of really great companies. I think we have 26 of the top 100 E&R at this point we've done business with and are on some kind of regional or multiple projects.

It really depends. We start with one and we expand from there, sort to prove ourselves out. I don't think anybody-- it's safe to say that I don't think anyone in the industry has fully scaled the drone program. So what's really interesting to me is I've been in-- I came from the medical world.

I was in medical school. I dropped out because I saw a cadaver and the eyes and the fingers get you. So I said, hey, all my friends are making tons of money with health IT implementing electronic health records. And I was here spending money.

Got out of that, find out medical billing was a little bit too mundane for me. I got into drones and photography as a hobby. But that was around 2010, 2011.

Was building my own RC helicopters and putting DSLR cameras on them. Bought my first DGI drone. The operating manual was in Mandarin. It didn't have positioning.

Drone did not know where it was in time space. So just in a few years-- I need to fact check this. So if you guys see something wrong with this, please let me know so I can correct it for the next presentation. But you know, there's three main areas that I see really in the past six to eight years that's a complete change, a complete paradigm shift from what it used to be back even 2012.

So I remember being super excited about eight megapixel photos. We now have 20.8 megapixel photos on a Phantom 4 Pro. The battery tech, really the convergence of multiple technologies-- battery tech, RC technology, extended range, obstacle avoidance and vision systems, positioning. They just came out with real-time kinematics and RTK Phantom that will help with-- we'll see with [? PBK ?] processing.

JOHN NILES: Just to interject real quick.

JACKSON DYRE: Sure.

JOHN NILES: This is the reason why we have Hangar on board with us is because they're dealing with the technology changes that we just cannot keep up with. We need to go build buildings, not fly drones.

And so that's where this natural partnership has happened. We've had plenty of people on site that say, hey, we're going to fly our own stuff. And then they crash it or they just-- there's issues, right?

JACKSON DYRE: Yeah.

JOHN NILES: And so it just-- you guys worry about some of those things that get updated and--

JACKSON DYRE: Yeah, and believe me, it's a chore to keep up with these things. I mean, every three months, DGI releases something new and we have to go buy new hardware and test it. It's an entirely different software development kit that we have to program.

Sometimes-- the Mavic 2 came out. It's got a Hasselblad. It's a different field of view and all kinds of things.

So yeah, there's three main areas. So hardware, the convergence of all these different things, sensors on the drone, three axis gimbal stabilization, manual flight control, [? two ?] complex automation solutions-- so there are some really awesome companies. We're friendly with everybody in the industry.

I support all of our competitors because this is such a new budding industry that I want only positive things being said about everyone. It'll only be good for everyone else. So there are simple waypoint mapping missions-- you know, drop a couple waypoints, flight that mission.

There's a lot of free tools for doing photogrammetry, which is a little bit of about what I'm going to talk about. We've seen operational management tools like Skyward, which was bought by Verizon-- really great for tracking airspace. Air Maps-- other one that's in our venture capital portfolio and integrated with our product.

And then what Hangar is trying to do is end-to-end enterprise tools. So have what we call the request-and-receive workflow. Any business user can right click on a map, upload a site plan, generate an autonomous mission, and execute that in the field, upload process for folks like John. And then what's really interesting is the regulatory environment.

So commercial flight was virtually unregulated prior to-- and this is where you guys can fact check me-- I think 2011. 2012 is when the 333 exemption was passed, so a big stack of paperwork. You know, James Cameron and DARPA were the first to take advantage of that.

The part 1 of 7 was passed in 2016. That was kind of unexpected. And then now there's something like 100,000 part 1 of 7 operators. So essentially, you can go take what is the equivalent of an FAA ground school course.

And now licensed you can get licensed to fly commercially for UAS under 55 pounds. And there's obviously restrictions there. So to John's point, we bookmarked capture at the beginning and the end.

We're doing this iteratively. We're flying twice a week on sites, uploading that day to day later usually. And we break down capture into all these steps.

So to John's point, the complexity is what we solve for. So to generate a point cloud and scale that across your business and call it a solution, you really have to be able to do all these well. You could easily go get an off-the-shelf tool, buy a drone, learn how to do a point cloud, process that, and get something done. But you want to implement this as a workflow as part of your business and rely on that and really fall into a process, you really have to be able to scale it. And so we're building technology to help do that.

So I'm going to cover-- so this is basic workflow. I'm going to cover this workflow. This is a 360 zoomed in. So we start with bear earth site.

We right click and request and order and drop what is basically the site boundaries and the points of interest for what we're doing. We select a few assets. We'll generate a complex autonomous mission. And that could take four or five batteries.

And we're tracking hundreds of data points every millisecond. So when you pause a drone to do a battery swap, it's going to go right back not just to the waypoint that it left off on. So you might have to do a whole ther leg of it worth on [? Mosaic. ?] You're going to go right back to exactly where that left off, the exact camera angle and everything.

So just automating simple things that people probably take for granted. So I'm not going to cover all of these. I'm just going to touch on a few.

So planning is what am going to start with and talk about the execution. If we can show you a video, I will. If not, I'll send you the link.

How we ingest that-- there's kind of handy trick that we developed to do that-- How we process and transform that into other assets and then how we view and deliver that. So I'm going to start with planning. How much time do we have, by the way?

JOHN NILES: Not as much as we think.

JACKSON DYRE: Cool. So I'm going to buzz through this and happy to answer questions at the end. So it all starts in the right-- so thank you, John, for letting me use your account. It all starts with a Hanger account.

We call this Hangar World. It's a pretty simple tool, just a base layer map. These are a few of John's projects in the Virginia area.

You start it with a right click on the map. You search your site address. You pick the-- you right click to set up the GPS pin. And then you fill out a form with the project information. So this would've been John's information here, maybe the point of contact for the PM on site.

JOHN NILES: Yep.

JACKSON DYRE: So what Connor here-- our operator knows to call him. We can title it whatever we'd like, select the type of mission. This also checks airspace for you. So we have an integration with Air Map and Lance.

So now through Air Map, we're able to get an instant authorization. In Sterling, we actually-- we're in the direct approach for-- I forget what airport that was. Dulles?

JOHN NILES: It was Dulles.

JACKSON DYRE: So we're just outside of the no fly zone in DC, but we're in the direct approach to Dulles Airport. We-- I think we called the tower and got an authorization. But in any other perfect-- we'd prefer not to do that. But in any other situation, you know, you can right click and get an instant authorization through Lance and be able to fly.

AUDIENCE: So real quick.

JACKSON DYRE: Yep.

AUDIENCE: With Dulles Airport parking, [INAUDIBLE]--

JACKSON DYRE: We were, at a ceiling. 100 feet, 200 feet maybe.

JOHN NILES: That was our original challenge that I threw out to Hangar. I was like [INAUDIBLE], if you think you can fly it.

JACKSON DYRE: That's how we earned John's business is-- he said he gave us the hardest problem to solve, and we can't always do it. But we were able to that time.

JOHN NILES: DC is probably one of the most restrictive places that you can fly drones.

JACKSON DYRE: It's a fun place to be with [INAUDIBLE]. So this is just an example of the order request. So I could have had a base layer map with satellite imagery to see the existing conditions as of, what, six months ago. And we can import other base layer imagery as well.

We have like 13 options down here. Basically, what you want to do here is just identify the site perimeter. We'll set our 360 locations. You'll set the frequency and duration.

So you set your project schedule. We can upload documentation, like site plans, or they call-- the schedule itself. So here what a lot of company with a lot of our clients do is just mark up a site plan. And we'll attach that to the order request.

What actually happens here is this goes into our system, and then we have a whole process and a room full of people that review it, ask questions. And then people like my team, actually, the customer success team, will work through the client requirements and make sure that they have the appropriate experience. The autonomous mission will get generated.

And so that's where I once spend most of our time. So how do you generate a high quality point cloud? Autopilot is the autonomous flight engine that we built out.

Essentially allows us to take full control. So this is overkill. That's a real project where we knew that we wanted to do a densified point cloud, multiple 360s, video, maps, elevations.

So this is kind of the full soup to nuts capability of autopilot. We're also doing QA/QC stuff vertically. So I'll show you that interface after.

But there's a couple considerations. So simple mapping missions in nadir view-- so nadir being straight down photography and images. Super helpful-- you can still generate point clouds from that.

But you also want oblique imagery to do-- have it ultimately stitch better. And you want a certain level of overlap. So I would recommend doing both nadir and oblique images.

And there are programs now. We highly recommend using flight automation-- if it's not ours, than somebody else's. And again, there's free tools that you can do this if you just want to go try it out.

Flight automation allows you to get proper overlap. 80% is kind of what I use as a rule of thumb. It can be less and still work. And it really depends on the site. You just want to make sure that you're getting enough points ultimately to provide surfaces.

JOHN NILES: This certainly means a lot when we're getting that point cloud because the more flights, the more passes we have on it, the better the imagery, the deeper the imagery.

JACKSON DYRE: Exactly.

JOHN NILES: Just-- that capture.

JACKSON DYRE: One thing to be cognizant of, though, is that the more-- so this was overkill. I was on site in Virginia, Richmond, for probably something like four hours on site. And the more images that you're trying to process, the turnaround times exponentially increase. And you guys are probably familiar with that as well.

So you don't want to capture too many images. So you'll start to learn-- this is what our team really provides us as added value. We've seen so many sites now, we can say, all right, we might need some oblique images to get the bottom of the actual structure, you know, where that meets the ground, to get a better idea or sense of how to coordinate utilities from that point.

So ultimately, what happens here is we have multiple assets. And we are able to execute that mission with the touch of a button. It's going to take off and run its cross hash. So what that cross hatch is actually doing is a double back nadir view-- so mapping mission.

the camera is facing straight down. And it's going to run and cross that with oblique images as well. So that does a pretty good job of giving you great coverage on the site.

We also had the building pad circled with an orbit at a certain altitude. And that just gave us more confidence that we have images to process if we needed them. So we actually didn't process all of the images here.

We just kind of stored that. But we did do is process those into individual assets. And we hosted those for photo documentation.

So we had every angle of that site captured. But we do that every time because if we do-- in the event that we do want to process a point cloud, we know we have that imagery. And we have no problem having that pilot fully dedicated and spending four hours in some cases. So the capture component-- I don't want to go into the video. I think we'll skip it.

JOHN NILES: Yeah.

JACKSON DYRE: But I'll send this to you. Ultimately, you probably find it on Vimeo. This is the preview of the flight plan for another job. It's actually our office.

You have certain things here like the time it takes to run the mission. It's very similar to the DJI GO interface. But all you have to do is hit this button. It's going to take off.

And all the pilot does is become the visual observer. They monitor battery health and make sure that they're swapping that to finish the mission. You could check altitudes and do all kinds of preview here.

The pilot would have originally claimed this mission. So the pilots that are out in the field, just like Uber, you know saying, hey, I'm going to pick up this rider and claiming that ride, they would claim this mission and load that mission. And that's our operational tool for tracking what's happening.