Using Drones, Laser Scanning, and InfraWorks to Get Design Buy-In

KEVIN FOSTER: Can everybody hear me OK? Is it good? All right, sweet. We'll wait another minute or two to get started. So we'll see who comes in. Just out of curiosity, second day in Las Vegas, third day in Las Vegas. Who's tired? Who's just a little bit tired by this morning? Yeah. Who's a little bit hung over this morning? Anybody?

And I apologize. If you guys were in the last session they gave out chocolate at the beginning of the session. You guys aren't that lucky. You just get to listen to us and I'm not sure how exciting that is.

I'm Kevin Foster. I'm the business development manager with Truescan 3D. With me today is Brendan Welsh. He is our group leader at Truescan 3D. Whoops.

So Truescan 3D, just to give you a little background on us. We do 3D laser scanning, modeling, and drone services. We have 12 reality capture employees. We have three UAS drone pilots. We're based out of Cincinnati but we actually work all over the country. We do a lot of very large and complex projects. We do a lot of smaller type stuff.

The project we're going to talk about today is actually not one of the large and complex projects. It's actually a smaller project that we've done. But I think it has some real interest because of what we were able to do in the time we were able to do it and the value that we were able to bring to the project.

So just a little bit of background on this project. And also I know that we are going to talk a little bit about InfraWorks in this presentation. If you guys are coming to this presentation thinking that we are experts on InfraWorks, you're going to be very disappointed because we're not. But I think part of the draw of this was we actually learned how to use this program very quickly and somewhat efficiently in order to do this project. So I think that has a little bit of merit in this whole thing.

So what I'm going to do is I want to start out and I'm going to play this video. And this is kind of the end product of where we ended up. And then we're going to talk and tell you how we work back from it. And when you look at this video you're probably not going to be just totally blown away by it. It's fine and it got the point across, but it didn't--

I guess what I'm trying to say is we didn't spend a ton of extra time on all the little details and all the frou frou stuff but it did get the point across. So we'll show you that. And then we'll start to talk backward and show you why that makes sense. And of course--

BRENDAN WELSH: One more click.

KEVIN FOSTER: And this video, this is about a three minute video. And this is the piece of the trail that we really focused on with our reality capture, with our UAV work and our laser scanning work.

BRENDAN WELSH: It's a shame the sound's not coming through.

KEVIN FOSTER: It is kind of a shame the sound's not coming through. That would've made this a lot more interesting.

But again kind of the feel of this video. This is an InfraWorks video. It's taken from all of the reality capture information that we collected in the field and then essentially just showing the proposed trail path and kind of helping some of the users or potential users understand this.

So with that is this is kind of wrapping up here for a second. I'll tell you guys that Brendan was our project manager on this project. And he physically did most of the field work on this himself. We had several guys busy. He doesn't typically do a lot of stuff in the field but he physically flew a drone, did the laser scanning, collected all the data. So after this video's done he's going to start to talk a little bit about details of the project and kind of how we went about it and the why and all of that.

BRENDAN WELSH: OK. So as Kevin said, this Mill Creek trail project was born out of a group of a whole bunch of different organizations in Cincinnati, in the greater Cincinnati area. There's currently a very piecemeal trail network and the intent is to connect all those trails and have an interconnected multi-use trail system that serves the entire greater Cincinnati area.

So we were actually contacted by the Green Umbrella Group and they started working with our engineering department and they said, look we want to focus on a couple of parts of this trail and we need to go get funding. But to get the funding we almost need a set of DD drawings. We need to at least get a design that far along that we can tell people this is how much money we need. This is where our critical design points are going to be. And we've at least done enough planning to know that this trail is feasible.

And so our engineering group said we'll help you that far. What we need to do is figure out a way to get to that sort of design development stage without putting a lot of money into it because at this point there's no funding. So the map on the left is basically the greater Cincinnati area. As you can see the big green spots are parks. And the intent is to connect all those different parks, providing a recreation destination for community bikers.

On the right, you see an inset of the Sharonville area and that's where our project was. The dashed red line is a proposed connector to Sharon Woods, which is one of the bigger parks in Cincinnati. And the communities of Reading and Evendale have already approved the funding. But they don't want to build out their trail until they know that the users can make it all the way to Sharon Woods, which is the ultimate destination.

So the city of Sharonville basically said, look, we need to develop this plan for this trail so that we can go out and solicit funding not only for construction but then find funding sources for maintenance and operation during the life of the trail.

And so this is our focus area. As you can see by the dashed red line on the right, the area is pretty heavily urbanized. The dark part in the middle of that is the Mill Creek. And the idea would be to put the bike path as close to the creek as possible, not only for the experience of being close to the creek and having some trees, but also because the rest of it is just developed.

And so what our engineering group came to us with was how do we survey this? How do we get accurate topography? How do we accurately measure the location of existing buildings, features, pavement without doing a full blown survey? A full blown survey is going to take weeks and cost tens of thousands of dollars, which we don't have either one of those.

They said can you fly the drone over this and make a map. And I said maybe. I can definitely fly the drone. But there is a lot of vegetation covering that stream bed. And so they said, well, what if we wait until the leaves are off. And I said, well, to be honest I haven't really done this type of project in this length. So let's try it and see what we get.

And so basically we flew about a mile here, from the Sharon Bridge all the way up to Sharon Woods.

And so why use reality capture? Historically topographic survey involves a guy out cutting brush, cutting line, taking a shot on the ground at critical points. To do a creek bed like this you would basically traverse up one side of it and cut sections through the brush, down into the streams so that you could profile the shoulder.

In my estimate, as a professional surveyor, this mile would probably take one or two weeks to complete. Then you would also need to pick up all of the existing features, the pavement, the parking lot. There's playgrounds, there's bridges, there's roads, poles, fences, all that kind of stuff.

Currently our surveyors are about four to six weeks backlogged so you'd probably be looking at six to eight weeks to get the survey done, another couple weeks for the base map, two or three weeks for the engineering. You're looking at a two to three month process and anywhere from $20,000 to $30,000.

And at this point in the design we don't really need to go to that level of detail. Number one, because we don't have the money to pay for it. And number two, because we don't even know if this trail's ever going to get approved. At this point we're just asking people for money. So we wanted the quickest way to get some accurate data to do a preliminary design and reality capture is really the best method for that.

KEVIN FOSTER: I think the other thing to add to that, too is a lot of the user groups, or the stakeholder groups of the trail partners or whoever was involved in this, a lot of those people didn't have experience reading a set of plans. So if we did a survey, we end up with a set of plan and profile drawings that most of those people wouldn't be able to read anyway.

And they're not that visually interesting. And they may not understand it that well. So being able to use reality capture and create a model at the end of the day gave us that visualization element, too that we needed to kind of help sell this to the community.

BRENDAN WELSH: So field data capture surveying has many different instruments. Reality capture has different methods. It's really important to plan out how you're going to do it based on what your expectations are. So what level of accuracy do you need? What level of visualization do you need? How rich does your data need to be? That's all going to be a balancing in between accuracy, expectations, and basically time and money. The more time and money you have to spend, the better data you can get.

So in looking at this segment of creek, and the segment of trail, we had to identify what are the critical points. There were two bridge crossings. There were two buildings that would impact the design. There were three or four parking lots and utility structures that would have to be accounted for. So we kind of went through and made a game plan of where we needed to focus a lot of data capture and where we just needed general data capture.

And I can't overstress the importance of making a good field plan before you go to the field. It'll save you a lot of time and a lot of headache.

So our data collection equipment, everything starts with control. You've got to have good control to bring reality capture data into a common format, whether it's Civil 3D or Revit or InfraWorks. You need to make sure everything is going to line up. Otherwise, it's not going to be very useful.

So set and control. I went through the project and developed a plan for my control spacing. I set the control using a survey grade GPS. Survey grade is critical because-- There's a lot of equipment that has a mapping grade GPS but the survey grade GPS is going to be necessary for the resolution of the point cloud, and the orthophoto, and the laser scan.

I set about a dozen control points on the mile-long corridor. And then I set another four points as quality control so I could check the accuracy of my adjustment.

I used the Leica scanner. At Truescan, we're pretty brand agnostic. We don't support or sell any particular brand. In fact, we believe that each brand of equipment has its own strengths and weaknesses. We have several different types of scanners in-house.

I used this particular scanner because of its survey workflow. Using this scanner, it's very easy to identify targets that are on a known control point from a long distance away. And basically, using this survey workflow in this scanner I can save myself a lot of time by tying into control and having it registered when I get back to the office.

Also it happened to be threatening to snow the day I was out there and this particular scanner is probably our best to use outdoors in weather. It will work in the rain and snow. How well it works is depending on how hard it's precipitating at the moment.

The drone down at the end is the drone that we happened to use on this project. Again we're not here to promote any one particular brand. They all have their strengths.

This is sort of our flagship drone. It has an ingress protection of 43 so it is resistant to some weather. A light dusting of snow or a mist of rain isn't going to hurt it. It also has the capability of carrying numerous different sensors. With the big batteries it has a flight life of-- they say 38 minutes. I'll give them 30 on a good day. So given the length of the scope that I needed to fly, given the weather conditions, and given the camera that I wanted to use I picked this drone out of the aircraft that we have in-house.

So again the whole corridor was just shy of a mile, 4,500 feet plus or minus. Once I had the GPS controls set I was ready to fly the drone. So basically your control points on the ground need to be aerial, need to be visible from the air when you take the photo.

So we use a combination of premade targets, just checkerboard. It's actually vinyl flooring that I found that I put on a piece of sheet metal. I put those in situations when I can't make a spray paint mark on the ground.

I like spray paint marks on the ground because I don't have to go back and pick them up. But depending on the landowner or the visibility of the place, I don't always like to spray a big X. So I used a combination of those two to mark my control points and then I was ready to fly.

The certification process and the FAA rules for a commercial flight of unmanned aircraft state that you can't have your aircraft out of line of sight. So no matter where I fly and what I'm doing I have to be able to see it. Given the length of this project and the maximum altitude that I'm allowed to fly at, which is 400 feet, I wasn't able to pick any one spot on the corridor that I'd be able to see the whole drone flight from.

So I wound up doing six different flights. So basically I would drive the truck to one portion of the corridor, get out, set the drone up, fire it up, turn it on, do a small flight. 10 minutes were most of them. Break the drone down again, move the truck, repeat the process. I did six flights in under two hours, which is incredible for the amount of data that was collected for the amount of acreage that was flown.

I flew at about 200 feet. We found that 200 feet is our best balance. For our sensor for flight time data collection, the data sets are huge once you get it processed. And then the resolution that we need to get the accuracy for the point clouds and the topographic products that we deliver.

In flying the drone and producing the point cloud and the orthophoto, we're picking up topography, but we're also picking up limits of improved surfaces. Parking areas, buildings, utility lines, fences. We get an immense amount of information through this process that would otherwise take weeks to deal with in conventional surveying.

KEVIN FOSTER: There you go.

BRENDAN WELSH: So there were two bridge crossings. And basically the trails folks that came to us said, can we get a bike trail underneath this bridge or do we have to go around it. And as you can see here, there's a road up front, and there's a driveway that connects the landscape company that goes underneath the bridge.

Vertical clearance on this bridge wasn't a problem. Everybody knew there was enough. And horizontally, there was probably enough. But what they really wanted to understand was how do we route the bike trail through the bridge and cross the road in a manner that's going to provide enough site distance around the hairpin curve so that pedestrians and bicyclists are safe from traffic? And drivers have enough sight distance so that they don't have an accident.

So we did a scan underneath the bridge-- because you can't see underneath the bridge from the drone-- and created a point cloud. And by tying in the same control that I used for the drone, we were able to bring that point cloud into the same coordinate system as our drone-generated point cloud.

The scan setups were pretty quick, pretty easy. I did maybe half a dozen scans at each bridge location. It was honestly like two scans to get the bridge envelope and then another couple of scans to work my way out to where the aerial target control points were.

I also did three transits down the stream bank just to confirm that the photogrammetry process was giving me good contours on the ground. So we used ground control to QC everything. And then we used our scan data to QC the drone data.

KEVIN FOSTER: So when we created this model after we collected all the data, the next step was obviously to create the model. So when we created this, we needed a program that we could use quickly and efficiently to try to put all this data in a format that was very easy to understand for our users.

So we looked at some different things but we ultimately ended up deciding to use InfraWorks on this. And the reasons that we decided to use InfraWorks were, first of all, we were able to create that model pretty quickly. We were able to pull some things in and create that quickly. And working on a very limited budget for this project, that ability to move through that was a big advantage for us.

We also-- as I said at the beginning of the program-- we didn't really know InfraWorks that well when we started this project. So within a day or so of playing with the program and going through it-- we had a couple of guys that knew it a little bit-- we were able to really kind of learn that program very quickly and use some of those capabilities.

And one of the great things about it-- if you're familiar with InfraWorks, if you know-- is it has the ability that you can use existing GIS information. And you can pull the existing GIS information in to help create the base of your model. So we used that as kind of the initial first piece of it. We were able to pull some of that existing information in.

But then it was also very easy for us to overlay our reality capture information on the top of that. So those two things, working in tandem, were very nice and it gave us exactly what we needed for this project. And then obviously it gave us that ability to graphically visualize this project for some of the users.

So those were kind of the reasons we decided to go with InfraWorks. And it was the project that seemed to make the most sense. And it turned out to be, I think, the right decision for what we were trying to do on the project.

And then, ultimately Brendan showed you a list at the beginning. I think there were about 15 or 20 stakeholder groups that were involved in this. But there were really seven key ones that we were working with and they were government groups, local trail groups, community groups. And again that ability to visualize and see what's going to be there ultimately or at least have an idea.

And I don't think, if you're doing really detailed design, InfraWorks is probably not that program to use. But at the stage we were at in the process, the DD/SDDD type phase, this was exactly what we needed to get that visual across. So that led to being able to transfer this information, show where conflicts were, show where potential issues were, and be able to talk about them. And to have people visualize that was really the key to gaining some buy-in for this project.

So as we were going through with the model, there was four different phases to how we went to set this up. We started out, and again, like I said, we took that existing data from Hamilton County. we're in Hamilton County, Ohio. So we pulled some of that Hamilton County GIS data, some basic topos, some roads, that sort of thing.

But what we found was we looked at it and we realized it's not accurate enough to show what we want to show. There's a lot more detail that needs to be shown. So the phase two was to take that UAV data, the LiDAR data, and import that into the model. And we'll show that in a second, that process for doing that.

And then once we had that-- If you're also familiar with InfraWorks, when you bring the GIS information in, it gives you a lot of generic building shapes and all that. So to go in into detail, to put a little bit more specific detail on the buildings to make them look like, hey, these are the real buildings that are there. So people can identify with them. And again, not a great deal of detail but enough to let people know that they're looking at a building and they can familiarize themselves with it.

And then our final piece of that, once we got past that, was to physically go in and add some of our trail elements in of the proposed design, which was the trail, the vegetation, things like light poles, and bike racks, and just some very simple amenities that we were able to show on the models. So that was kind of the process of it.

So what this video is, it's just a real quick video, and it just kind of shows the process of taking the base and-- That's not going to work. Or maybe it is. And again it's unfortunate. There's some cool music that goes under this. Our guy that put these together had some really cool music selections. But unfortunately you guys don't get to appreciate that, I guess.

BRENDAN WELSH: So this just kind of shows the 3D surface that was created. And you can create this in a number of ways. You can do it from, like Kevin said, GIS data. The limitation there is-- I don't know your experience but my experience is GIS data lacks sufficient metadata to really rely on it as a survey source. I might be overly picky because I'm a surveyor.

But I want to know when that data was collected. I want to know how accurate it was. I want to know if the GIS says there's a light pole there, is that light pole still there because I know that road was worked on last year.

So using either GIS contours or the point cloud that we collect, InfraWorks will basically scan that point cloud for terrain elements, bare earth elements, and it'll create a surface. And it'll use that surface to update the model that it already has. So when you get to that design phase you know that you're working with good topography. You know that the ground that you're seeing, that you're designing on top of, actually looks like that.

And so this is kind of a flow chart put together to attempt to convey how we did it. The top is the drone workflow and the bottom is the point cloud workflow. So basically you can bring a point cloud in whether you generate that in ReCap Photo or a number of other different products that are out there to generate a point cloud from drone photos.

You bring that point cloud in. You're going to clean out basically the points that are going to be duplicates. You're going to clean out points that aren't good. If you're going to scan with-- If you're going to use the scanner for buildings or bridge decks, you want to make sure that you're not duplicating those efforts because it'll throw off the surface that you're building in InfraWorks. So there's some cleanup to be done.

You also need to make sure that everything you bring in is on a common datum. Otherwise it's not going to line up. And at the end, you process that point cloud to a terrain surface in InfraWorks. And as I said, that terrain surface then is used to update the model. So once you go into your design, you know that you're working with good data.

There's additional cleanup that can be done. There's additional design elements in InfraWorks. It kind of depends on how far you need to go with it and how tightly detailed of a design you want to do.

InfraWorks is never going to get you to construction documents. You're going to need to ultimately go to Civil 3D if you want to put all the plans together. So you kind of have to balance your time and your energy with how much you want to get out InfraWorks. But there's a lot to it and there's a lot of capability.

KEVIN FOSTER: So we're going to walk through. There was three critical areas that we were going to look at little examples on. The first one is what Brendan talked about. It was a Sharon Road bridge area. And there was a little video. There's about a minute long video that goes with each one of these. So we'll start to look at that.

Really what this video is showing is it's showing you the area but it's also just kind of jumping back and forth between the point cloud data captured and the model and showing that they're pretty much lined up right on top of each other.

BRENDAN WELSH: And I think you get a sense of the rich visualization capabilities that this will give you when you're going to ask people for money, versus just having a 2D set of plans.

You can see here how the bike path was routed around to cross the road in a place that offered a lot of sight distance for bicyclists, pedestrians, and drivers. Like Kevin said, you can see the terrain elements have all come together. This is an area where we had GIS data, scan data, and drone data all in the same very small space.

The Sharon Road overpass was a critical concern for everybody. How to get people using the trail safely through and across the street. And so it was and it is a very big focus of the project.

KEVIN FOSTER: Again, this next area, the community center area that we were working in. It kind of provided a different set of challenges. Again, Brendan can probably elaborate on those a little bit more than I can.

BRENDAN WELSH: So you see the green building up there. It basically encroaches out into the stream bank. The trail would have to go around that building. That would require us to build a retaining wall. Part of the retaining wall is to get around the building but a good portion of it is to get the trail far enough away from the building that we could maintain adequate site distance.

The Ohio DOT regulates bicycle path widths, slopes, and site distances. So we had to produce a trail that conforms to those specifications. And InfraWorks was a really cool way of doing that accurately, and doing it visually, to make sure that you're maintaining those design criteria. But you can also visualize where those critical points are.

KEVIN FOSTER: And I think one of the things that the data that we collected helped us understand in that area was there's going to be a need for a retaining wall. But how tall does that retaining wall need to be? And how long does it need to be?

So it allowed us to make that cost estimate. Or allowed our civil engineers to do that cost estimate to more accurately create the overall project estimate. So it was little things like that. None of these, in their self, are huge details but combined all together they gave us a lot of information for the project.

And then the third area--

BRENDAN WELSH: And it adds credibility to our design to know that we've identified these critical places and that we're going to have to account for them. Sorry, I'm talking into your video.

KEVIN FOSTER: No, go ahead. So what this helped us understand too was-- You'll see in here there's kind of a little playground that's circled there. And you'll see the trail runs right through the middle of it when we flipped to the model and it's proposed. But one of the things it helped us see too, was OK, where are property lines? And what features encroach into the area of the trail that we're proposing?

So do we need to approach people about variances? Do we need to approach people about land acquisition? So if we do, can we start those conversations early in the process? And can that help us maybe get ahead of the game a little bit?

So we may be able to go talk to a property owner and say, hey, we want to encroach 20 feet on your property. Are you willing to sell? And to have that conversation up front. And know if that's going to be a big issue if the trail goes through or if it's not. So being able to understand those issues and get ahead of them a little bit were critical for us too.

BRENDAN WELSH: Yeah, because some of those can be showstoppers. I mean, if you need to take somebody's parking or if you need to tear down their playground, those are things that can completely shut down a project. And it's good to know about those well ahead of time so you're not way into a design and realize that your projects just can't go forward.

KEVIN FOSTER: So I think the really cool thing, and the big value part of this whole thing is, is what we were able to do in the time that we were able to do it in. So if you kind of look at this for the field time, Brendan basically spent a long day in the field to collect all this data. So he actually spent longer than anything mobilizing and setting his control for survey.

The drone, like he said earlier, was in the air about a total of two hours. The laser scanning piece was two bridges. I think you did, I don't know, 15 or 20 scans overall, probably. Maybe not even quite that many-- But that was a total of about three hours.

One of the things he did do was did some extra checking with the scanner. He kind of scanned down the stream bank which wasn't really necessary but it was kind of to check the drone data and compare it. So we had a total of about 10 hours in the field.

Office time to register and process everything was about six hours in the office. The model, it took us about three days to do the model. And again, this was one of those things where we could have gotten a lot fluffier on the model and could have put a lot more detail into it.

But we were actually working-- We were kind of doing some of this work pro bono in the hopes of getting the project down the road. So we didn't want to spend any more time than we had to, up front, if we're giving some of this work away. So we were essentially able to create this, capture the information and create the InfraWorks model in a total of about 40 hours, which works out to-- if we were doing that on a project-- would be about a $5,000 fee, more or less.

So to be able to collect that amount of data and to process them and do it and create a model for $5,000, I think, is kind of the very interesting part about this project. So I don't know if you have anything to add to that, that's but that's--

BRENDAN WELSH: Yeah. I mean, we started on a Monday. We were done on a Friday. If we would have surveyed that and done a preliminary engineering design, we're probably looking, like I said before, two to three months and $20,000 to $30,000. So to use this as sort of a marketing tool, as a business development tool to go after work, this is much more appealing to my boss and the decision makers then spending $20,000 or $30,000 in hopes of securing some engineering contracts.

KEVIN FOSTER: And I don't know how many people are familiar with the requirements for federal trail funding but you essentially have to have a set of-- Your drawings have to be at kind of a design development level. And you have to have a cost estimate to be able to apply for the funding.

So while I would say what we did probably wasn't good enough to do CDs off of, it was very good for what we needed it at this point in the project. And to get everything to a DD level to allow the engineer to understand quantities, to make cost estimates, and all of that. That was kind of what was--

BRENDAN WELSH: Yeah. And our engineers will tell you some ballpark numbers for constructing a bike trail. I mean, they know about how much per mile that it costs. And you can go out and tell people these things.

But if there's factors in there that I call showstoppers, acquisitions that can't happen or the stream bank just doesn't have enough width to hold the bike paths. The bike path has to be 10 feet wide and have two one foot shoulder. So you're talking about a 12 foot profile, basically.

If there's not enough land there to do that and you go out and tell people that you can build it and give them an estimate, then it really reflects poorly on you. So this is a really quick and accurate way to confirm that the design is feasible and it can get built.

KEVIN FOSTER: So go ahead. I'll kinda--

BRENDAN WELSH: So a couple of things we learned. The stream bed itself was heavily vegetated. We waited until the leaves were off. We still weren't entirely sure how well the ground could be topo'ed underneath all those branches. On the day I happened to be out there, there was maybe an inch of snow on the ground. It was just a dusting.

Typically doing photogrammetry on top of snow is a really bad idea because the digital photogrammetic process doesn't work well with really homogeneous scenes. It needs a lot of texture. It needs a lot of identifiable points. Having a field of white just doesn't work well. So I was a little apprehensive about flying the drone with the snow.

As it turns out, the contrast of the snow underneath the dead branches produced really great results. It worked awesome. We verified it with our scan data. Everything lined up really well and was well within the expectations we had for contouring a stream bed.

We learned that it's a very quick and easy means of doing that type of survey. You definitely need to do it in the winter months. We learned that InfraWorks is incredibly user friendly. It is a great way to combine all of our reality capture data sets. You can bring these things into other formats but InfraWorks, I think, handles them the best. Point clouds, orthophoto photos, GIS data.

And the learning curve was surprisingly small. We got up to speed pretty quick. And I think on the next one we could probably do it in half the time.

KEVIN FOSTER: So with that, that's the end of what we have. Are there questions or anything? Yes?

AUDIENCE: So, I just, really, just did the very exact same except [INAUDIBLE] trail. [INAUDIBLE] works. Really surprised at how well the drone data integrated with the laser scanner data aligned. I mean, it was kind of a great check to see that perfect [INAUDIBLE] but I didn't feel the same process [INAUDIBLE] That's why I processed it. [INAUDIBLE] processed my digital scans there. [INAUDIBLE] whole lot of issues with the georeference data for [INAUDIBLE] the whole process.

BRENDAN WELSH: Oh.

AUDIENCE: So I don't know. And again, it's all new to me too, but I was kind of curious because it seemed like ReCap was causing that snafu. And I even took two scans, made two copies of them, modified them separately [INAUDIBLE] with the georeference information. It was like ReCap was shifting in its interval. It was something that I [INAUDIBLE].

BRENDAN WELSH: Was it a vertical problem or horizontal?

AUDIENCE: Horizontal.

BRENDAN WELSH: Really? Were you working all in state plane coordinates?

AUDIENCE: Working in all state plane, yeah.

BRENDAN WELSH: Did you have the same state plane definition?

AUDIENCE: Yeah. I even rebuilt everything [INAUDIBLE] I went through the process twice to rebuild it and make sure [INAUDIBLE] ReCap [INAUDIBLE].

BRENDAN WELSH: You do.

AUDIENCE: [INAUDIBLE]

BRENDAN WELSH: You have to tell it. And were you on ground control? You had survey ground control?

AUDIENCE: Yeah.

BRENDAN WELSH: And was your ground control surveyed on the same datum that you're--

AUDIENCE: Yeah. Datum was the same. It was something weird ReCap was doing. There's blogs about that out there [INAUDIBLE].

BRENDAN WELSH: Oh, it's--

AUDIENCE: Well, is it not only my issue?

BRENDAN WELSH: No it's not.

AUDIENCE: [INAUDIBLE] kinda good it worked for you guys.

BRENDAN WELSH: It worked. It worked completely seamlessly on the ninth try. So we had the same problem. And I'll admit we brought it in half a dozen times.

AUDIENCE: OK.

BRENDAN WELSH: And I told you I'm not an expert. So we basically just clicked a bunch of different things and brought it in again.

AUDIENCE: [INAUDIBLE]

BRENDAN WELSH: And on the eighth or ninth try, it worked perfect. But it is a known issue. You're right. There's tons of blog posts out there that you can go and read. The forums are pretty well inundated with people having that trouble.

AUDIENCE: [INAUDIBLE] onboard when ReCap

BRENDAN WELSH: International feet?

AUDIENCE: International feet, yeah. [INAUDIBLE]

BRENDAN WELSH: Units, coordinate system, datums. There's a lot of settings. You've got your drone data. You've got your scan data. You got what ReCap wants to do and what you want it to do. It's trial and error.

AUDIENCE: [INAUDIBLE]

KEVIN FOSTER: Yes.

AUDIENCE: RTK on the drone?

BRENDAN WELSH: No. Our drone is RTK capable. I don't know how much you know about the RTK on a drone. But there's limitations with it.

AUDIENCE: Right. And then my next question was [INAUDIBLE] without RTK?

BRENDAN WELSH: Well, we're setting our ground control using RTK.

AUDIENCE: You said you had a check point? What was that?

BRENDAN WELSH: Horizontally, we can say about an inch to a tenth of a foot. Vertically, I'm typically anywhere from 1,500ths to 2/10ths. So we're like an inch horizontally, two inches vertically.

AUDIENCE: [INAUDIBLE]

BRENDAN WELSH: At 200 feet? Yeah, with a 20 megapixel camera. And if you're producing one foot contours, like we do in 90% of our surveys, national math accuracy standards say that 90% need to be within a half contour interval. So if vertically you're 2/10ths, in my opinion, you're plenty good enough.

AUDIENCE: Cool.

KEVIN FOSTER: Is there another que-- Yeah.

AUDIENCE: Yeah. This was a big area. How did you deal with [INAUDIBLE] people?

BRENDAN WELSH: So I can find my flight path just to the stream. And on the main road crossing, I can find my flight path just to the right of way And I never flew above the road. And I flew basically a small, densely-gridded mission right next to the road, but not above it, so that I had lots of side lap. And depending on the width of your right of way, if you fly high enough, with enough photos captured, you should be able to accurately map that right of way. But it's always an important consideration.

AUDIENCE: I mean, it's really crowds, right, that you worry about, not one or two individual people? [INAUDIBLE]

BRENDAN WELSH: FAA regulations say that you cannot fly above people not protected in a stationary vehicle. So I interpret that to mean people walking or people in moving cars. But it's kind of a gray area.

AUDIENCE: Yeah.

KEVIN FOSTER: Yes?

AUDIENCE: What ground control app do you guys use? Do you DJI's app?

BRENDAN WELSH: You mean to actually pilot it?

AUDIENCE: Yeah.

BRENDAN WELSH: I use Drone Deploy.

AUDIENCE: Do you? And do you find that their system-- So do you use their grid system, how it's set up?

BRENDAN WELSH: Oh. To actually pilot it and capture?

AUDIENCE: Yeah.

BRENDAN WELSH: Yeah, it's my favorite one.

AUDIENCE: Yeah. So it wasn't annual play? It was pre-setup?

BRENDAN WELSH: Yeah. You can't really do-- So the thing that you really need to rely on them for is your shutter release. Based on your altitude and the overlap that you need. There's front lap and side lap. It triggers the shutter for you every second, or every 1.4 seconds, or every two seconds. There's no way that you could-- maybe you're a better pilot than me-- but there's no way that you could accurately push the shutter in one second intervals.

AUDIENCE: I thought it interesting because this is such a oblong shape. Did you just do it overly right into this area of your capture [INAUDIBLE] on the other side and it was [INAUDIBLE]?

BRENDAN WELSH: No. I actually did six different missions that were kind of in long corridors all the way up the stretch. So that I only collected just what I needed. I don't want to spend any extra process time or disk space with extra data. I don't want to fly over people. And I really only want to get what I need. And so you wind up with a bunch of different missions.

AUDIENCE: Awesome. Thank you.

AUDIENCE: Have you ever wanted to fly lower?

BRENDAN WELSH: No. We've done a bunch of R&D in-house. If you're just doing terrain mapping, 200 to 225 feet ground control, no more than 500 feet apart, for one foot contours. If you're trying to do like a pavement assessment to look at cracks in pavement, we usually go about 100 feet. But it's a balance because the lower you are, the more photos you have to take to get the overlap. And the data set's exponentiate. The processing times exponentiate. To get that point cloud into InfraWorks or ReCap takes exponentially longer because it's that much bigger of a data set. So I always try to fly as high as I can.

AUDIENCE: Has anything [INAUDIBLE]

BRENDAN WELSH: Yeah. When we first got that Matrice, Drone Deploy wasn't compatible. So I've used the Pix4Dcapture app. I've used DJI's Ground Station app. I've used the USCG app. There's a number of them out there. Drone Deploy I like because if I'm taking less than 500 photos, it'll process them for free. If I'm taking more than 500 photos, it'll let me capture and I can process on my own. And I've been using it a long time and I think it's really user friendly.

AUDIENCE: Have you guys looked [INAUDIBLE]