YDQA: Ep 56- "How To Achieve Optimal Accuracy With A Drone?"

John Dickow: [00:00:00] Hello and welcome back to another episode of Your Drone Questions Answered. I'm John Duca with the Drone Launch Academy, here to find the answers to your drone questions. These are questions that you submit yourself. Today's question is, How do you get high accuracy with a drone, or optimal accuracy? And so with me, I have Dylan Gorman.

He's the founder of pilot bite. He's been with us before. Thanks for coming back, Dylan.

Dylan Gorman: Yeah, I appreciate you guys bringing me back on for some fun questions.

John Dickow: Well, first, I mean, just as a refresher, do you mind just kind of introducing yourself? Tell us what you do, especially at pilot bite.

Dylan Gorman: Yeah. So, I actually have an extensive background in doing mapping and photogrammetry.

I've been doing it professionally for six years. but I started pilot bite as an education platform to build on top of the, free YouTube material that I put out there. so ultimately pilot bite is just an extension of what I already do, publicly online, but I have some online courses working with other industry experts, just providing a really in depth, Technical training that is really hard to find out there, which is sounds like we're gonna be answering some of those questions today.

John Dickow: Perfect. Well, let's get into the question. Our [00:01:00] question being, how do you achieve high accuracy with a drone or if you call it optimal accuracy?

Dylan Gorman: Yeah. So there's really three main components that you have to look at. not three components, but there's three different systems that you can use or methods of achieving high accuracy.

and actually there's two different kinds of accuracy. So if somebody is looking for a high fidelity on a, on a map, then you're going to be looking at the visual accuracy side of it. And that's going to come down to how high you are off the ground, the camera resolution and the focal length.

That has to do with your G. S. D. Known as your ground sampling distance. and so those are kind of the three main factors that go into creating high visual accuracy and more high fidelity on the maps and models that you create. Now, when it comes to the physical accuracy side of things, you have ground control points, which is kind of the tried and true, method of guaranteeing, a high precision.

and then you have P. P. K. And R. T. K. Which are more of the two newer technologies. R. T. K. Being the one that is Essentially taking the industry by storm. but really what all three of these [00:02:00] systems have in common is you're referencing some known position in the real, in the real world with your digital data that you capture with your drone.

So with ground control points, those are typically set by surveyors. and then you, as the drone pilot are going to set out these aerial targets. and then you're going to go out and map the area. And then you're going to take basically these aerial targets, sit on ground control points, or known points, and those known points are defined by three data points.

You have the latitude, longitude, and elevation. You can also see it as an X, Y, and Z or, a northing, easting, and elevation. And that basically defines that exact position in the real world. And then that aerial target is going to sit on top of that as the visual representation of it. As you go and map your mission out, you're going to then take that data into whatever photogrammetry and mapping.

Processor that you like. I often use Pex 40 or drone deploy. and then you're going to take the control book. All control book is is a list of every single one of those points that you put your aerial target on with the coordinates on it, and then you're going to match it in this in the software, and that's gonna get [00:03:00] you your the highest precision, the highest accuracy on your maps and models because you're tying a plethora of images to known positions in the real world.

Now, if you don't have access to a surveyor or you're doing a mapping project that you can tie to ground control points, that's when you're going to shift into doing the P. P. K. Method. P. P. K. Stands for post process kinematic. the process to that is your drone is going to be capturing its trajectory.

Basically it's positioned throughout the entire duration of the flight. that's called a static survey of basically the position as it's flying out. You're also going to have a base station. Now this base station, again, very similar to a ground control point is a single reference point, a single known point in the real world.

And you're going to be within. 25 to 50 kilometers depends on the region. If you're out in Florida, you can be 50 kilometers away from Ron. If you're out in the Rockies, more up in the northeast where there's a ton of mountains, you want to be a little bit closer from that 15 to 20 kilometer range from that base station.

And essentially what's going to be happening is throughout the duration of the flight that you're capturing the data, that [00:04:00] base station is also going to be recording a static position of it, of its exact location. And then after the fact, Because all of those photos that were captured on your drone has GPS data that's tied to the trajectory of the drone.

You're going to take both of those files, put it into a PPK processing software. I like to use Emlid studio. It's a free application that you can use out there. and then what it's going to do is triangulate and basically figure out where that drone was. Cause the drone can be off anywhere from 10, 15 to, I mean, see a hundred feet off, in its positional accuracy, which.

If you know this, this is going to get tied down to your actual accuracy on the map, which is going to make it inaccurate. So what PPK helps it do is filter out any of the bad data and actually help position where that drone actually was in the real world, but after the fact. And then this leads us into RTK.

So that entire equation that I just explained to you with you. PPK and doing the comparison that's actually happening in real time, which real time kinematics, that's what our TK stands for. And [00:05:00] again, same thing with ground control points and PPK. That base station is going to be sitting on a known position somewhere in the real world.

Again, you want to be within 15 of it. This is actually completely different from PPK. You actually do want to be close to the RTK base station just so you can Calculate less inaccuracies in terms of what's going on in the area with you. and then on top of that, as you're flying the mission, you're getting those corrections in real time.

So all of that GPS data is being stamped or added onto every single image that gets captured. And that's really like the main method that a lot of people are doing. nowadays, especially with drones like the Mavic 3 Enterprise, that is really easy to get access to, you know, a sub 5, 000 drone to be able to do RTK mapping is like a godsend nowadays, where a few years ago, because I've been doing this for six years, it used to cost you 10, 15, 20, 000 to do RTK mapping.

Now you can do it for sub 5, 000 and maintain a high level accuracy all the way down to almost a 10th of a foot of accuracy on most [00:06:00] projects. And I've seen that time and time again. and it's, a big shift that's happening in the industry. A lot more people are shifting over to RTK cause it's just a faster, more streamlined way of.

Doing that high precision and high accuracy on your maps and models. Just to kind of recap with your visual accuracy, you're going to be focusing on your ground sampling distance. That has to do more with your camera settings, not necessarily the GPS systems. And if you want good physical accuracy, that's where you're going to look at your ground control points.

Your PPK and your RTK.

John Dickow: The person who asked the question also asked if there is just an ideal altitude or percentage of overlap that would optimize accuracy, but from what I'm hearing, it sounds like there's so many other factors.

Yeah, there's

Dylan Gorman: definitely are, in terms of being able to not guarantee accuracy, but, give your software the best chance of producing a higher accurate mapper model, you do want to stick within, I would say, No less than 65 percent frontal and side overlap. and these are just standard settings that you'll see in really any mapping application.

so your frontal overlap is as the drone is flying [00:07:00] forward, let's say 70 percent overlap with it. Every photo is going to see 70 percent of, itself. And then for the side overlap, Think a lawnmower as that drone comes back the other direction, every photo to the left and to the right is also going to have that 70 percent overlap.

And that is honestly really kind of industry standard. like I said, no less than 65 percent just depends on the size and the scale of the project. If you're doing a massive mapping project, then yeah, you can drop it down to 65%. Um, but if you're doing, you know, really small areas, something that's like five, 10 minute flight, then, you know, You can fly 70, 75%, even 80%.

Now, if you're trying to achieve high accuracy with models doing a lot of three d work, then you're going to want to have Your overlap be at like 75 80%. Now, in terms of altitude, my preferred altitude, I like to stick around 160 ft. and that's with a 20 megapixel camera drone.

I've seen great accuracies anywhere from 0. 4 to 0. 6, inch per pixel GSD. and basically just to kind of explain what that means is, the measurement between each pixel in [00:08:00] an image, is the equation of what that real measurement is in the real world. So, you know, for some projects I have to do pavement analysis and they need to see cracks that are bigger than half of an inch.

Well, in order to see that crack that's bigger than half of an inch, you have to shoot less than half of an inch to do that. So that's where that physical, that visual accuracy comes into play. But for most mapping projects, because I did, I've done a lot of construction progress updates, just A lot of construction and engineering projects, I typically stick at about 160 feet.

If it's for anything that's four, five, 600 acres, then I'll probably bump it up to 200 or 220. really just depends on what your optimal output is and what your client is looking for, but taking into consideration the visual and the physical accuracy, that's how you can kind of get your equation of, okay, how high do I need to fly?

What's my overlap? You know, what is your desired output? and then you kind of work backwards from there.

John Dickow: Excellent. Dylan, I, I really appreciate your explanation behind this And we get so many questions about mapping and our, drone launch community we're a community of people who are learning about drones and want to apply drones to our everyday work.

And it sounds like so [00:09:00] much of it comes, plays into mapping and people, a lot of people are interested in mapping. And so, do you have any advice for, for, Kind of people starting off wanting to kind of break through or start off with mapping. How do you start?

Dylan Gorman: I would say starting in 2024, moving forward is completely than starting six years ago.

there's a lot of resources that exist out there now. There's a lot of misinformation that is out there as well. so do be careful, you know, which sources you pull from and, who you listen to. But the one, thing that I advise most people to do is get educated first, get technical training on this information find your own resources to get technically educated because you're going to spend a whole lot less money in the long run doing that as opposed to, okay, I see an M three 50 and the L two LIDAR sensor. Right. Okay. 35, 40 grand. Let me just spend that because you know, the more money that I put in, the more money that I can get out of it.

That's kind of a backwards way of thinking. And that's actually the thinking that I had very early on where I've spent well over a hundred thousand dollars on all kinds of pieces of equipment that I probably shouldn't [00:10:00] have bought because I was in over my head. And I should have actually taken a few steps back, gotten technical education first and technical education can be even be working at some of these drone provider networks that exist out there, getting your feet wet, getting, just some, basic experience to just really learn what this industry is about because mapping as a whole mapping is a very base level of what you can actually do.

Every single day, there's new industries leveraging just the core concepts of photogrammetry and mapping all the way from inspecting power lines and solar panels to helping engineers and architects build the next, you know, biggest burst cliff or whatever. Right. So, you know, drones are being used in so many different areas specifically for mapping and photogrammetry.

And you'd be surprised at. You know what? People are willing to pay for this stuff. If you have the right technical experience and being able to explain it to your client, because at the end of the day, they're paying you as an expert to do this. I'm not saying you have to be 100 percent [00:11:00] expert. You can get started within a few weeks and get good paying jobs.

Trust me, there's quite a few of my students that have made some really good money very early on, but that's because they've gotten the technical education first.

John Dickow: Perfect. Dylan Gorman, founder at Pilot Byte. Dylan, thank you so much for coming back onto this podcast. Your explanations are, wonderful.

They're, they're understandable even for those who have very little experience in, in drone mapping. So thank you. And we hope to have you back on this podcast again. Hey, and if you have a drone question yourself, please send it our way. We are happy to find the answer to it. You can submit your questions at ydqa.

io or if you're part of the Drone Launch Connect community, go ahead and type it in there. We'll see you. And we'll find somebody, like Dylan, to answer the question. Until then, we'll see you in the sky.