On December 1, 2017, Aerotas co-founder Daniel Katz presented a webinar for the New York State Association of Professional Land Surveyors. In the webinar, Daniel covers key lessons Aerotas has learned from training over 200 land surveyors on high-accuracy and profitable drone mapping.
Watch the webinar below, and when you are ready to start using a drone as a profitable high-accuracy tool, contact us.
Drone Surveying Webinar Transcript
Amber: Good morning! Thank you for joining us for the NYSAPLS first Friday Webinar Series. Today we’re joined by Daniel Katz, who will be presenting a program on Drone Surveying: The Complete Story. Before we get started I wanted to go over a couple items to make sure you all understand our webinar process. First, on the right side of your screen you should see a box we call the dashboard. On the dashboard you’ll see a” Raise your hand" symbol with the picture of a hand. If you can hear me, please click that button and raise your hand now. Wonderful. And while we’re waiting for some hands to go up, I’d like to remind you that we cannot hear you. Therefore, if you have a question, please use the questions box on your dashboard to type your questions and we’ll address them as needed or at the end of the presentation. Okay, I see many of your hands up. It looks like we’re good to go. I’m going to turn everything over to our speaker. It’s my pleasure to introduce Mr. Daniel Katz. Good morning, Daniel!
Daniel: Good morning, Amber! Thank you so much for arranging this and thank you to the NYSAPLS members for starting off your Friday with me. So, as Amber mentioned, I’m going to take the group through, a presentation a little bit of a demo on what we've learned about our best implementing drone surveying for organizations like your own. Before I dive into that, a little bit of quick background about me and my team. I’m one of the co-owners of the firm called Aerotas. We provide complete drone mapping solutions for land surveyors. So we work all over the country. At this point we've gotten close to 70 survey teams up and running with drone survey operations. We have completed with our clients, we completed about 600 drone surveys and in the neighborhood or 450 to 500 hours of field trainings and testing and R&D.
What I want to share with you all today is a lot of the best practices we’ve learned and the hard-won lessons we have learned about how surveyors can benefit from drones in their organizations. Another word of quick background: I myself, I am not a surveyor by background, so what I will do my best to do is to share with you the principles that I’ve learned from my clients, that I get back from my clients. Forgive me if I invariably, as I do, say something dumb. I have to ask for a little bit of forgiveness there in advance. And certainly, as Amber mentioned, if you have questions, please do not be shy about sending those over. I’d be more than happy to clarify and dig in where there are specific questions. And if you have questions that are very specific to your own organization and you want to have that conversation directly one on one, you can see that down at the bottom of the slides there’s a phone number and e-mail address, that comes directly to me. So if you want to shoot over questions there that are specific to your organization, I’m happy to address those after the webinar.
Alright, with that there are four main topics that I want to cover with you all this morning. It can go on some tangents along these but the big things I want you all to be able to take away from this morning: One are some key best practices and the lessons learned about how to build a professional drone program for land surveyors. The second is what the current status is of drone survey technology. Thinking of this in terms of ... For your businesses or organizations – what’s the place to make an investment? Next is some of the actual field and office workflow for using a drone as a survey tool. I mean, I’m going to actually walk you through a quick example of some of that – data processing, getting to find a line work from a drone 3D model. And then last are some bigger picture implications for surveyors looking to use a drone to grow their survey operations. Before diving into the nitty gritty and I think it’s really important to understand what you should expect from this technology. This technology is still new enough that we frequently find people with really-really divergent expectations about how a drone should be benefiting them. Everything from people who just expect it to be a nice working tool, to the people that expect the drone to do their complete job for them. And the truth is, well surprisingly which you should be expecting is something in the middle. So the key thing is that the key kind of metrics or check boxes you should be using when evaluating either a drone program you currently have in place yourself or whether you’re determining what it’s... Starting a drone program is a good investment for your team are as follows: First, probably most importantly, is the drone program should be getting you to your final line-work. So what I mean by this is you should have as part of your drone program a streamline and efficient way to get to those final contours, break-lines, feature lines, and surface as a CAD file. I’m going to show you an example of the workflow we’ve identified to do this effectively, but the important takeaway here is it should not just be getting you a pretty picture. If all you’re getting from your drone is a video or some aerial imagery, you’re leaving money on the table. You’re leaving some benefit on the table. With that, your total investment, a good benchmark to have is about $10,000. So you should be expecting if you’re ramping up a drone operation, your investment shouldn’t be more than about $10,000 total. I’ll get in a moment a little bit of what should draw into that. Accuracy wise you should be expecting a 0.1' vertical or better and that’s when tested to ASPRS Positional Accuracy Standards, so not one of individual test shots, but actually the total accuracy of all in error of a project should be at about a tenth of a foot or better. In terms of the actual benefits to your company - the big thing is time savings. So you should be with the right filed operations and the right equipment, you should be saving 60% to 90% of man hours in the field. So a project that would’ve taken a two-men crew a whole day should now be taking one person just a couple of hours in the field. Obviously, there are some variations in terms of types of projects and some factors in the field, but it still should, you know, really should never be taking you, or saving you less than half the time. And then your office time shouldn’t be that much different. We hear a lot of horror stories about folks using drones saving much time in the field and then it doubles, triples, quadruples the time in the office and so the net gain is pretty minimal. But the right data processing workflow in the office, your office time should be about a wash with what it is if you’re doing the same project conventionally. The operation should be designed such as you don’t need the highest specialized drone operator. You shouldn't have to go out and post job postings for drone pilot or drone surveyor -- it should be a tool like a GPS or a total station that most of your survey staff can, existing survey staff, can use with training. Then also obviously critically important is you should not be introducing any additional risk to your people or to your business by using this tool. And that comes with having the right insurance, having the right compliance strategy and most importantly, having the right field operation. So this is kind of… I see these as mistakes. This is what you should be… you should be demanding a lot of this technology and this is what it should look like. So in terms of digging into the actual benefit to your organization, there are a couple key things we see from our clients, as the main ways they’re benefiting from this tool in their operation. The big banner headline, the most important benefit is time savings. That should always be what you’re focused on first, that should be your first priority in implementing a drone program. In what that time savings should look like, I mentioned before about 60% to 90% time saving in the field. In terms of the man-hours it takes to complete a survey. When you’re using a drone, you should have a saving of 40% to 60% - that includes field time and office time. So obviously, that means if you’re producing your final deliverable so much faster, that means you're turning projects around faster, that means you can bid them a bit lower and still profit even more substantially from them. It also means that your people, your field crew, is going to be doing a lot more work in the same amount of time. So it allows you to punch above your weight with the smaller field operations. Beyond that time savings benefit, you have a handful of other benefits as well. One I think it is preemption – meaning that it’s really easy to collect more data than you necessarily need. Great quick anecdote from one of our clients - He went out to do a project at a landowner site and the landowner said: No, I just want you guys to survey the front half of this project, I think it's about a 10 acres total. So just the front 5 acres. And they check and double check and say: Are you sure? We’re going to be out there, are you sure you don’t want us to survey the back half of the property as well? No, no, no, just the front half. They said: okay. They had a hunch. They knew, they worked with this client before so they just, you know, they took their drone up, they flew the entire property. You know, they were already out there, it took an additional maybe 10 minutes to do. Go back to the office, they produce the survey for the front half of the property, hand it over to the landowner. Sure enough, he says: Hey, this is great. Can I actually get the back half of the property also? They had already collected that data using the drone when they were in the field with marginal additional work in the field. They already had that data in the office, all they had to do was pull it up and complete the survey. So they executed that second survey with nearly, with 0 additional field work and marginal additional office work. So it allows you to kind of collect more than you need. With that, on a similar note, it’s easy to collect this data, it functions effectively as a record that you maintain. So every time you use the drone on a site, you’re collecting this reach, accurate, 3D data, comprehensive 3D data of that project site. Meaning you’re not just relying on what specific points your field crew collects in the field, but if you come back to the office and realize that: Oh, we actually needed to collect some more data on the site, you can just pull out a 3D model and actually extract your survey shots from that record. Never mind the fact that in the unfortunate incident of disputes with clients or landowners, you have this record, this very high-resolution image record of exactly what the site looked like on the day you were there. That really high-resolution imagery also means that you can provide really high quality, high resolution base-maps behind your surveys so create more valuable final deliverable to your clients. And then, of course, you have the benefit of cost savings from aerial contractors if you want you can start doing more of those projects yourself in house rather than having to shop them out to aerial photogrammetrists. And then last and very importantly, a drone allows you to reduce the amount of time your crew has to spend in dangerous situations. Standing in roadways, you know that a surveyor has to stand and take shots in a busy roadway, you can just fly a drone over the road or steeper, unstable slopes or other sorts of hazardous or difficult to access areas. You just fly over without having to walk through it. So these are the varied ways you can expect a drone to be benefiting your business. Again, I really encourage you to focus primarily on that time savings benefit and then all the rest of this is kind of the icing on the cake. So what it actually takes to capture these benefits is to think of a drone as part of a comprehensive drone program. So it’s not just a matter of to find a drone on Amazon or at your local distributor, throw it in the air and expecting magic to happen. You really do need to take a systematic approach to this. And what we have found is that an effective drone program, a professional drone program has five key components, five key categories of things that you have to put in place. The first, obviously, the one that everybody thinks about it’s the equipment itself. The drone, but also all the other gear you want to have in place in order to operate that drone effectively – spare batteries, controller with the auto-pilot software installed, and even the rugged carrying case, we found that having the right kind of case can actually be really important when thrown into the truck and not worry about damaging the drone. Yeah, the equipment, you also need to be compliant. Anybody who is going fly the drone needs to have a special commercial drone pilot license, it needs to be very familiar with the rules so they make sure every time they go out they’re using the drone in a way that is compliant with regulation. Third, insurance. You have to make sure that, as I mentioned before, you shouldn’t be exposing your business to any additional risk. Most business liability insurance policies have an aviation exclusion – they say that if it flies it’s not covered so you do need to get insurance that is going to plug back the aviation exclusion gap. And then, the last two are the two that folks seem to think about the least but also the most important. First is having clear, well-defined filed operations or standard operation procedures and training for your staff on those operations. And then last, is having a clear workflow in the office for getting to your final line-work deliverable. Now, I’m going to get up on my soap box for just a moment because we have the benefit of seeing the patterns in how businesses, you know, the struggles they have in implementing drone programs. And we see a lot of folks running into very similar problems. And this one really consistent theme we see in companies when they struggle with this and that is that they over emphasize the drone itself. They just think about that first bullet point, just the equipment. That’s what they put all their energy into, they go shopping, they compare tax, they get completely sucked into that and they think that it’s all just about getting the right drone. Honestly, at this point, the drone is the easy part. The drone is not the part to focus your energy on. This technology has actually matured really-really quickly to the point that you can have a really effective drone program with a pretty inexpensive, easy to use drone. The place where companies make money or lose money, the place where they are effective or ineffective, that results in having an accurate, efficient drone program or a wasteful one comes out of these last two bullet points. It’s about having good, clear operating procedures so that every time your crews go out, they are collecting the right data, they’re doing it efficiently, they’re doing it in a way that’s repeatable and having really clear training on that. This goes back to when I said you should not be needing a high specialized drone pilot for your drone program, you should have a good training that allows anybody, any one of your field crew members to be able to become a proficient drone operator. And then, that last bullet point of deliverable production – if you just have the complete, the field operation, the equipment, everything dialed in but what you get back is this overly complex 3D point – if you don’t have an efficient way to get to that final line work deliverable, you can’t use it to actually produce surveys. It’s a great marketing tool, but it’s not actually getting you to the value that it could. So this is really what we are focused on – unsurprisingly, what my company provides is one solution that encapsulates all five of those components. And honestly, most of our time and energy goes into these strategy points. This is really where we’ve put a lot of focus on, on optimizing the force. So thinking about, now that I’ve talked about how unimportant the drone is, I'm going to talk about the drone. A lot of folks want to know what is the best technology, where if I’m thinking about purchasing a drone, what types of features should I be focused on. From our experience, having done a ton of testing, having had the opportunity to test pretty much everything under the sun at this point, we keep coming back, and we’re continuing testing this stuff, we keep coming back to the fact that right now the smart investment is actually on a pretty inexpensive camera drone. So I can get into exactly the drone use in a moment, but I want to talk about the three technologies that people think about when they call us asking: Hey, should I buy a drone that has this feature? And so I want to kind of talk about each of those in a moment. And those are fixed-winged drones, so airplane style as opposed to the helicopter style which is called the multi-rotor, drones carrying lidar sensors or laser scanners and third is drones carrying basically survey grade GPS - RTK or PPK or Referencing integration. So right now none of the three of these are really a good investment and I’ll talk to you quickly why. A fixed-wing aircraft, the primary benefit of the airplane style drone over the helicopter style drone is range. It can cover a lot more ground, a lot more quickly because it flies more efficiently. However, one of the primary regulations in place right now when using a drone as a tool is the operator has to be able to maintain sight. They need to be actually watching the drone the entire time it’s flying. And then sure, a multi-rotor helicopter style drone it already can cover enough area that you won’t lose… it can fly beyond your sight line as is. So the benefit of a fixed-wing is kind of moot because if a multi-rotor can already cover that much ground then you can’t take advantage of that extra benefit in the first place. Accuracy is pretty comparable and so is what comes out of it but your field operation and the operating part of it is considerably more complex because you have to start getting to think, thinking about things like take-off and landing. It takes off at an airplane rather than vertically and it lands like an airplane, rather than vertically. So it actually becomes considerably more complex in the field figuring out which direction do I take it off, what, you know, making sure it angles right planning for your landing and they also, they belly land, which they don't end up landing like an airplane, they just come in and they land on the ground, which is great as long as you only every work in very soft grass. But if you're working in areas that have any sort of dirt or pebbles or rocks, these things get shoot up really-really quickly. And they're pretty expensive. So the type of small multi-rotor drone we use costs about $15,000 out of the box. Fixed-wing aircraft is about $30,000, usually. Next up, lidar sensor - so putting a laser scanner on a drone. I personally think that, like, I'm very excited about this, I think this is going to be a great opportunity once the technology matures. But right now, it's just not there yet. After the testing we've seen, accuracy it's still pretty low. I should say - the primary benefit of using laser scan on a drone is it allows you to shoot through ground cover a bit better. So you can get down to bare earth when you’ve got some cover. Accuracy, from the testing we have done and we have seen, is not there yet. Even if it has half million-dollar lidar drone in from field testing barely comes in at three times vertical. Again, you got really-really complex operation, you’re just setting the drone up for the first time, much less setting it up and operating it at each time your field, it requires a lot of pretty regular calibration work to be done, whereas an inexpensive camera drone doesn't. And the data workflow is pretty cumbersome as well, as anybody who works with laser scanner data has. And they're really expensive so bare, bare, bare bond minimum is $15,000 more likely about $100,000 - $250,000. So given the ground cover benefits I do think this is one worth keeping really close eye on but it's going to be a little while before it's maturing up to be really a good investment. Particularly if you're starting out your drone program. Next up, survey GPS integration so RTK, PPK, or direct geo-referencing on a drone. The benefit of this is that these systems they basically tell you the location of the drone to extremely high accuracy. The problem is knowing the location of the drone doesn't tell you the location of the ground. So we actually have on staff one of the firsts ASPRS which is the American Society for Photogrammetry and Remote Sensing has now started credentialing Mapping scientists so we have one of the first certified scientists on our staff. And it explained to us after having, you know, reveal all the research and done a lot of testing that using RTK or PPK on a drone, the primary benefit is that it allows you to use a little bit less ground control, you can set fewer ground control panels or targets on a project site, should mean it saves a little bit of time on the site. But you still need to set some ground control and it's about getting too much into the weeds on most projects sites that most surveyors are doing this size requires you to set pretty much the same number of ground control, the benefit really only starts being significant when you get up about at 100 acres and if you're already on site setting some ground control, it's not that hard to set a couple more ground control points. And ultimately, you're producing data that is of comparable accuracy at the end of the day but you still do have, again, you have pretty complex integration - the setup of the equipment is pretty difficult, every time you're using the drone in the field there's a lot of calibration to be done, you're adding a lot more complexity to the system so it's a lot more things that can go wrong. And, again, they're quite expensive. So you're looking at minimum is about $30,000 to $40,000 for a drone that has one of these integrations. And then you're, another thing to be aware of is you're adding another step to your data processing. You actually have to post process that as well. Again, I think this is a technology, all three of these are technologies that we're constantly testing and checking in with our partners on the equipment site, we're excited about them, we think there is going to be some potential in the future but right now the investment just, it just doesn't pencil out. So, again, looping back to those expectations I said at the beginning, the place to... Kind of the benchmark to have in mind is your total investment for your drone program, including the equipment, including training, including insurance, should be around $10,000, or even a bit less than $10,000. That's going to get you kind of the sweet spot of maximum cost-benefit right now and obviously, given the prices of these other technologies, right now they're not a great investment. In terms of the type of drones that we use and recommend, the model that we use, that our whole system is built around, is a DJI Phantom 4 Pro. As you can see in the picture, it's a pretty small multi-rotor drone, easy manageable, few years in the field. The things to prioritize when you use this drone or you use a different one, the things to prioritize honestly are reliability and ease of use, those should be the things that you're most focused on. What can my guys in the field, what can they use with the fewest number of headaches or what I'm going to have to deal with the least amount of maintenance, the fewest things who are wrong. Basic To have in mind, I mentioned before, you want to go with a multi-rotor helicopter style drone so it takes off vertical, takes off and lands going straight up and straight down. Camera, 20 Megapixels is a good baseline to have. We tested better cameras, we've actually only gotten, you know, not really gotten better accuracy results. And you want that camera to have a mechanical shutter, as opposed to a digital or a linear rolling shutter. And kind of a watch out is - it can be really tempting to get sucked into the marketing about a professional grade drone and half the time you find one that's, you know, the company will have their1 consumer model and their professional grade model. The professional grade one is just the same thing That's not necessary, there isn't really a considerable difference, there's not something you need to look for that is Professional grade. Focus more on something that is easy to use, reliable and it's going to meet these basic text facts. Okay. Obviously, for surveyors, this technology is only as good as the accuracy can expect from it so I'm going to dig into that a little bit because probably you all won't be surprised by this, but the drone industry has not done a great job of being clear about what to expect with accuracy. Oh, good question from Tom about American made drones saying that the U.S. military has DJI, the drone that we actually use is Chinese on 99,9% of projects, that's not an issue, but every once in a while, we do have clients who have projects for the U.S. Government or for the military which requires an U.S. made drone. Right now, unfortunately, there isn't a really good U.S. made drone that we’d recommend. The one that we were... There is a company called 3DR that was making drones in the U.S. but that has stopped. It's something worth keeping eyes out for. Right now there's just not a great solution, which is disappointing. But we're certainly keeping our eyes out on that. Jeremy asks for me to repeat the brand and model of the drone, probably a bit more slowly and clearly. So the brand is DJI and the model number is, and the model is Phantom 4 Professional. Alright, back to accuracy. So, you find that you're going around and talking to drone companies that there's often some confusion about how they refer to accuracy. And it's not out of some sort of duplicity, it's just, you know, not everyone thinks about accuracy as much as clearly as surveyors do. So the first thing to be aware of is often times what a drone company quotes as the accuracy, will turn out to be their ground sampling business. Basically, the effective pixel size of photos from the drone. So you want to make sure, you know, a lot of times you'll hear things like: Oh Accuracy, or accuracy of 100 or something like that and you want to clarify like, you know, we're talking about all in error, we're being square error on it, you know, with all the projects, not on the pixel size, the ground sampling distance. That's the first thing just to have in mind. In the, taking a bit of a step back, obviously accuracy it's a deep rabbit hole and I'm not a surveyor by background, so I need to be careful about how far I read into it, . But the important thing to understand is that error or inaccuracy when using a drone can come from a lot of sources and most importantly, it can come from a lot of unpredictable sources. A certain, you know, there are a handful of sources of error that are expectable or that are predictable - things like, of course, your camera resolution or the field that you and kind of, lens distortion of the camera, your flight altitude, whether you're using that digital rolling shutter versus a global or mechanical shutter, camera focal length and things like that, photo overlap. But there are also a lot of factors that can be unpredictable, that can change on a project by project basis in a way that makes it impossible to predict - things like vibration in the air frame or temperature, humidity changes, how lightning changes, even subtlety over the course of the project. What's your exact distribution is ground control or areal target, subtle terrain fluctuations and even just small human errors on the post processing. Even, I love this example somebody from my team uses having one of the propellers hit a particularly large bug that leaves a particularly large amount of bug gut on the propeller which causes a slight amount of vibration in the air frame. So the implication of this is you can't just, you can't really say the drone gets a certain level of accuracy because it's impossible predict. So the way to assess accuracy of a system is the calculating your all in error, calculating that root-mean-square error. And this is where ASPRS again has these really helpful standards - I call them positional accuracy standards. So basically amounts to some guidelines for how to assess accuracy on a projects using checkpoints. So I have appeared just a screenshot from a recent project we did that we, that we had verified by a third party surveyor using ASPRS standards and you can see, basically these blue markers identify where we had ground control, where we had aerial targets and the green markers are all the places they had additional checkpoints. So they went, they followed ASPRS positional accuracy standards. The letter here had a pretty large number of checkpoints and they use that to do a statistical analysis of the all-in error on the project. On this project was 500 vertical. But the important takeaway here is when you're assessing the accuracy system, first of a talk to a drone company, ask the specific hard questions, make sure that you're speaking about accuracy in the same way and second of all, frankly, never trust the drone, always set checkpoints, always make sure that you are determining for yourself what the accuracy is of that project. Looks like I have another question here. Noise, the background my mic, sorry about that, I think that is actually it hitting the zipper of my jacket, apologies for that. Another question from Joseph about questions on camera calibration and how to handle that. It gets pretty technical pretty quickly - it even goes over my head pretty quickly. The important thing to keep in mind - one are the way that drone photogrammetry works is a bit different from how conventional photogrammetry works. In conventional photogrammetry where you've got a big metric camera on an airplane, the camera is, they're using highly calibrated cameras, metric cameras, that they're so highly calibrated that you can just take two photos and you overlay them on top of each other and have a stereo pair and just extract your data from there. So in order to do that, you need to make sure that those photos are perfectly, they're corrected to represent the ground. With a drone, it kind of flips that model So rather than calibrating the camera, you take, and only require two photos, you take a lot of photos that really-really-really high overlap such that any single point on the ground is going to be captured in as many as 16 photos. And with that much overlapping imagery, what that allows you to do is actually to effectively calibrate the imagery in the data processing portion rather than needing to calibrate the camera itself. So there's really no need to calibrate and recalibrate the camera on the drone over time because you're making those corrections post facto in that image session software. Basically, there are techniques to identify where your operations are in any one image or consistently images and use the overlapping image to culminate format. Another question from Paul about accuracy if differences on sunny versus cloudy days. We've seen no issues. Time of day does not make a considerable issue. It's okay if you've got shadows, it's okay if you've got a cloudy day versus sunny day. The only real issue, or the only impact of that is going to be what the photo looks like. You know, if it's a cloudy day it's going to make prettier picture because you're not going to have shadows, but in terms of accuracy, picking shots off the model we've actually found there's no considerable difference. Okay. Getting back into accuracy, I mentioned this at the beginning, what we've generally seen in this industry, industry standard or industry benchmark is 0.1' vertical. This is, I just mentioned of course, this one project where we came in a 500, it was pretty ideal conditions. I'm a big believer in you make these decisions and you purchase decisions assuming the worst case so what we've seen is over a lot of projects and confirming with and constantly checking with our clients were using the drone in the field, about how the checkpoints are coming in. You should assume 0.1' vertical and you're generally going to be pleasantly surprised. And again, this has been a work using a pretty inexpensive drone producing this sort of accuracy and we are, folks that are using really expensive like our Are getting the same accuracy, nobody's really getting considerably better than this on a consistent basis. Having said that there's a little bit of new ones to be aware of. There's a lot of new ones to be aware of in that accuracy, but one thing to keep in mind is that you do have a trade off in the accuracy you can produce and the amount of acreage you can cover. And that's because if you want to produce the highest accuracy you need to fly the drone a bit lower. What we found is at about 100 feet is kind of a sweet spot, but you fly lower than 100 feet you actually don't really get that much better accuracy because you start introducing more noise and more variables and so flying at about 100 feet is where you can get that vertical accuracy. But if you want to cover more ground, if you want to, you got a bigger project site, you can fly the drone higher, but that is going to come at a cost of accuracy because your effective photo resolution is being reduced. One photo covers more ground, but each pixel gets bigger. So this is what, you know, this table summarizes a bunch of testing we've done. Basically, what you can expect to go on from maximum accuracy if you want to get to that 0.1' vertical you can cover about 25 acres in an hour with this type of drone we're using. At the other end of the spectrum, if you want to cover more ground in that same hours of flight time, you can cover about 250 acres flying at your , 400 ft and you're still going to be at about 0.4' or better, which means can be good enough if you just need to produce one foot contours. So this is part of the learning curve, this is something you should kind of expect in training or in your testing, is to dial in these best practices about how you're choosing making the choice about this tradeoff between acreage and accuracy. Okay, some questions about processing, I'm going to hold off on that for just a moment so I'm going to get into that in a second. In terms of workflow, a big picture, your workflow has three primary steps - you've got your field operation, actually operating the drone that you've got to take the photos, fly the drone, stitch them together and then you need to extract the actual Cad line work from that stitched together 3D model. When you're flying the drone, the field operation itself, what that actually requires is setting your ground control, setting up the autopilot on the drone and then collecting any other critical field shots and hitting, and the drone should be, the operation is completely autopilot controlled. You take those images into an image post processing solution, that's when you stitch those together into your ortho-photo and your 3D model, and as I had done in the beginning, you need to have a way of getting from that 3D model actually taking your survey shots, your break-lines directly from that 3D model in a way that is going to be able to come over to Cad or to whatever your finish software is. In terms of field operations, again, I'm going to do it again. Having really clear dialed in field operation is one of the most important - it's probably the most important thing to getting high accuracy, to be safe and to be efficient and profitable. And what we found is the single most important tool in a drone program, honestly, is actually not the drone itself, it's having a really good checklist. It seems trivial, but a well-developed field operations checklist makes sure that when your crew goes out, they're going to be efficient, they're going to be consistent, they're going to, you know, having things on that that make sure they don't forget to charge the batteries and go out in the field and burn the day because they forgot to charge the batteries or forgot to grab the memory card or something like that. It's going to make sure that you're checking all the boxes in terms of legality and not violating any rules. Most importantly, it's going to make sure that your people are safe and that they are reliable so that the results they produce are repeatable. We like paper checklists, it's not the most high-tech thing, but we find that paper is still the most reliable technology so we'll operate with paper checklists and get a lot of our clients to see back to the office part of their workload is just to scan that into a digital format now they got it as part of their record. Field operations. So very similar principles to working with conventional aerial contractors. The first thing you need to do is to set ground targets. The reason for this is the drone has on board GPS - it's not, we're not talking RTK or PPK gpss, it's standard Receiver and that allows, so the photos that the drone takes are geotagged but there's a decent amount of inaccuracy, obviously, in just relying on the drone's GPS and targets why to do it is to effectively nail that model down to that 0.1' vertical accuracy we were talking about earlier. Basic rules in terms of ground control is you want to have 5 targets per battery that you're using. So, earlier I mentioned if you're flying for high accuracy you can cover about 25 acres in one hour of flight time. In real world terms, one hour of flight time is going through four batteries. Many factors date and flight time on the drone uses is thirty minutes, but for a lot of operational reasons, you want to assume 15. So for each 15 minutes flight time, which is about 7 acres as you're flying for high accuracy, you want to have 5 targets -one near each corner still a little inside, one near the middle. If you have any considerably high points or low points you want to put a target there, if you have any out property, you want to put a target there as well. When you're out there collecting, your ground control is where you're also going to be setting or collecting any critical high accuracy points. So building corners, Compliant ramps as well collecting any shots that are hidden below tree cover, or other sorts of ground cover. I see other questions coming in, I'm going to stay or keep going through the presentation for a couple minutes here, because I know we want to make sure I get to that workload demo so I'm going to hold off on questions for just a moment and I’ll get back towards the end. I'm happy to stay on longer to answer questions. So once that filed operation or setting ground control is completed, setting up the drone, again, mostly is about following that checklist and making sure that you're covering everything you need to. In the full, the operation should be auto pilot controlled so the setup we use, the auto pilot software lives on a tablet that's attached to the controller, the interface looks like Google earth and all you do is basic trace and outline the area you need the drone to survey. It sets your flight altitude, it creates the drone flight pass, you hit go, it takes off by itself, flies overhead by itself in a lawnmower pattern comes back and lands by itself. The surveyor is holding the controller, they're watching they're maintaining on the site, keep an eye on the And status on the controller. The technology is smart enough to recognize if you have an area that is larger that can't be covered in one battery, it will recognize that it's running low on battery, bring itself home, swapping the battery takes about 30 seconds and then can pick up where it left off. Now the last step, obviously, once the drone operation is being completed is getting to that actual survey. So, first part of that is stitching these photos together and then actually extracting your survey from that. In terms or that actual image stitching, this is one of the most complex parts of this entire process. This is the most technically challenging part. So there are few options you have in terms of picking a solution for doing that image stitching. There are three kind of big categories you'll find when you're looking into this. One is using these kind of third party automated image stitching services such as DroneDeploy or Propeller or or you can purchase that software yourself and run it in house using Pix4D, Agisoft Photo Scan, or you can use specialized outsourced provider such as ourselves that are survey focused and actually do that image stitching specifically for survey operations. There are tradeoffs with all of these. The automated, these purely automated processing solutions where you're basically just, you know, you just simply upload your photos to a server - the main benefit there is they're going to be really-really quick and they're going to be really easy to use but your accuracy is going to be the lowest. They’re not focused specifically on survey deliverables, they don't have a human in the loop so they're not catching how to make the small modifications in that image stitching process that are going to get you to the highest possible accuracy. The local, purchasing the software yourself is going to give you the maximum flexibility - obviously, you can do it when you want, how you want, where you want. But it is very, the learning curve is steep and long. We're among the most experienced user of this software in the world and we're still learning on a pretty regular basis and it's expensive. The software itself is expensive and your using your people's man-hours and your people's time in the office and your employees' time is expensive. Expert out sourcing options like what we provide is going to get you, it’s going to be the best for getting to the highest accuracy since we've done this a lot of times, you can figure out that QA/QC procedures to get to the most consistent high accuracy but it's going to be a slower turnaround time. So our usual turnaround time is about 24 hours but every once in a while we get slammed with work and it can be a couple days and that's because we always have a human in the loop doing QA and QC. So in terms of what each of these options is best for, if the types of work you're doing is more about progress monitoring, you want to generate a visual record of a site you just want to have something for planning and discussion, that cloud-based automated processing is the best. If your office is in a remote location, you have poor internet connection at that local, doing the processing yourself locally could be the best. Or if you're dealing with really sensitive data - you have a government client or something where part of your contract is that you have to keep all the information in house, it makes sense to do the processing yourself in house. Or if you're for standard land survey on a high accuracy mapping without this critical data sensitivity issues, using a provider like ourselves could be the best option. So what anyone of these options is going to produce for you is a 3D model, a digital surface model, and an ortho-photo. So let me quickly flip over and show you then a solution we have found for actually getting from that 3D model to an actual survey. So this is software that we partner with called Virtual Surveyor - it's a company based out of Belgium I believe - we've partnered with very closely and helped all with the software. We're looking at here is a stitched together 3D model from a drone project. So you can see kind of fly around in here very similarly to how I would in google earth. A couple quick questions I'm going to address. The software we use for processing - we use Pix4D ourselves in house, the autopilot software we use is called map pilot app and Asks: Do we outsource data processing outside the US - we do not, that all happens in our office, in California. So as the name of this software suggest, Virtual Surveyor, basically what allows you to do is to survey in a virtual environment. So if I take my point tool here and I click on the ground on this model, what I'm effectively doing is taking a, creating a Cad smart point right here. So it's just like I was out on this project site with my GPS And taking field shots. And anything I do in this software, I can ultimately just export as a ..dxf or as a ..csv, an File and it can come over into whatever finish software I'm using - 3D, Carlson So these, each one of these individual shots is just recording my Location at that point. You can see over here on the site I'm creating these elevation shots. I can even do some layers if I want, I want to create my edge of pavement layer. I can do that, I can change this, give it a free fix so then when I zoom in here, on my edge of pavement, I can run a series of shots along the edge of pavement and if I export this as a .csv and import it in Or Carlson and apply my templates, you know, I can basically set this Convention to match my standard Convention and will automatically apply my line work. Or I can even just do my line work directly in here. So I can actually create 3D polylines here. And that's, oh, that's white, that's hard to see. Sorry about that. And that 3D polyline, I can set that to function as a break-line or I can just use it as a feature. So if I show you what a project site like this would look like, what my ultimate product might look like Turn on my completed survey here. And you can see that what I’ve done is I gridded up these areas - you can drop pretty easily drop an automated grid and then go and clear up any shots that drop on top of cars and trucks, it's pretty quick and easy to do. In my vegetated areas, in order to drop shots in here, I can use a couple tools. I have this automatic terrain map here I can use to guide me to what my low points are using these automatic contours and make sure I drop survey shots in those lowest points. Understanding that ground cover is a limitation of this technology right now. As an example, if I come over here under these trees, there's no way I'm going to see through those trees but I can take reference shots from the field, bring those over and Modified model. So I did my survey shots in that vegetated area, then I went in and started doing my line work so you can see I’ve basically marked out this kind of marked out this kind of dirt lot area, drew it like Line here, my back of curve and my front of curve and my edge of pavement, marked out all of my paint striping. Now, once I’ve finished the survey shots of line work, I can run a tin. SO this is now taking this really-really rich 3D model and - to clarify this: the reason you need this software is if I just try to open this full 3D model into any sort of Cad software so the 3D autocad Carlson, it would break the software. Those softwares are just not designed to handle the richness and the size of this 3D data so effectively what I'm using is Virtual Surveyor software to do is to reduce down that 3D model. So you can see this Tin I’ve created is now just connecting my spot elevations, breaking across my break-lines and circulating through any lines that I have in there as features, like my paint striping and then once I’ve got that done, I can run these contours over the top. Turn off the Tin so it's easier to see. Now I have these contours in white that are just running over the top. And once i've completed this, I can export it as a .dxf if I want my full line work or my individual points with labels, I export it as a ..csv and if I open that up in a simple autocad viewer, this is just my export from Virtual Surveyor. See, all of my layers match the layering I created in Virtual Surveyor. I can even turn on that Tin, don't want to use that. And you can see if I turn off my contours, it gives us a View. We actually have full 3D elevation data around here as well. So now you can see this data, you know, we have now left the drone behind. This is getting you to what you actually need in order to complete your survey. And again, this was using the .dxf. If instead I want to just apply my naming conventions and work in points, or just get points, I would export that as a .csv file. Alright, we're running short on time here so a couple more quick words. Obviously, you know, you can see from that workflow, this is where the real efficiency gains come in. My field operation, in order to do that, would have been a matter of a couple of hours. Setting ground control and operating the drone rather than having to do all that - walk the grid, mark out all those , mark out my edge of pavement, do all that in the field so my field time on a project of that size is going to go from a full day with a two-men crew down to a couple of hours in the field, including collecting ground shots, collecting any of my critical high accuracy shots. In the office work, using that Virtual Surveyor software as you saw, it's pretty quick and easy to use, it's really intuitive so that total` office work it's usually about a wash with what it would have looked like to do that office work convention, if I had done the survey conventionally. A couple of comments on the state of the survey industry and the reason that a drone is a very valuable tool particularly in this moment. I'm sure all of you are very familiar with this but the survey industry is in kind of a critical moment right now. 2017 was one of the biggest construction years in a long time so we hear this story all across the country that surveyors had one of the busiest and most hectic summers they've had in a long time and a lot of survey firms are looking to grow. They have more demand, more work than they can keep up with but folks are having a really-really-really hard time hiring. After 2008 there is a shortage in the country of hirable surveyors and the average age of a land surveyor in the US is in the neighborhood of 55-65 years old. So it's not only a shortage of surveyors to hire, there's also a shortage of young early career surveyors. So the best option for this, you know, long term, is to be attracting more people into the industry, but short term is to find ways to be able to run survey operations more efficiently. Obviously, a drone, the reason why we're focusing on using a drone as a survey tool is because it's a really effective way to get that sort of proficiency, but there are a couple business implications that you need to be aware of if you're going to incorporate this too. One of these is that the focus on efficiency needs to be a core focus of the business. You know, and what that means is this is not to say that it's more important, it's not more important that accuracy, it's not more important than client service, but it needs to be put on That. This needs to be a primary focus you're thinking about when you're planning projects, when you're teaching new surveyors or in the culture of your business. And it can be tempting to think like Well, the way I’ve always done it has worked for me and for surveyors and for me so I'm going to keep doing it that way. But the survey industry is changing right now so there's need to be, to move away from the trap: I’ve always done it this way. There is, well, we see a lot of our most effective clients doing is shipping to smaller survey crews, doing a lot of projects with one-man crews and focusing putting a really heavy emphasis on avoiding return trips. And when planning for projects, thinking about opportunity costs, thinking about if I'm putting this surveyor over here to do this, what else can he not do? So it's not just about doing the job in fewer hours for the sake of being more profitably, it's also doing it in a few hours so that then you can take another job as well. Your surveyor can finish the job in a couple hours rather than in a couple days so he can do and do another job in that same time. There also needs to be a shift in how you think about how So a lot of surveyors function on a kind of time and materials pricing so it's like a cost basis for pricing. They say: Well, this is going to cost me X dollars to do it so I'm going to charge you, you know, some margin on top of that. If you're running a much, much, much faster field operation, you're completing projects in way fewer hours, that's not useful if the impact of it is that you have to charge time and materials and you make far less money for it. So what a lot of our clients do is they basically, they price the project based on what it would have cost them to do traditionally, they, we are all very experienced, you know how you can estimate those and then they will provide a discount on it. They'll say: normally, I'd have to charge you extra for doing this project, I’ll charge you a little bit less so that More competitive and then they are running that project more profitably. So it's a shift in thinking about how you plan for projects. And then like we talked about with that the image stitching options thinking strategically about when you use outsourced tools, right? When you use one of those automated image stitching solutions or a service like ours rather than having to do everything in house and there, by being a focus the staff you have on whether to be most impactable. And, of course, there's a benefit with drones that, you know, it's what a lot of younger people want for Christmas this year and so a drone is, you know, land surveying is in a really advantaged position right now where you come across a young person who has fallen in love with drones and wants to find a career that allows him to use drones - I'm going to be a land surveyor. I can tell you from, we work in other industries as well, we For the drone industry, land surveyors are using drones more than anybody else right now, so it's a great way to attract the next generation. Alright, I'm going to, we're at time right now so I want to wrap up the content there. Amber, pipe in and let me know if you need to jump in and wrap anything up on terms of housekeeping, otherwise I can stick around for a little bit and address, I see that they've been a handful more questions that have come in so I can stick around for a little bit and address those. And, again, remember, if you want to, if you have any questions that you want to ask one on one, I believe in the handouts slides, on the last page, there are included my direct phone number and e-mail address. The phone number and the e-mail address in the bottom of the slides here also, get to me pretty quickly As well. Alright, so I'm going to look back to some of the questions now. I was trying to skim them as I was going. There was one about photo overlap. Peter had asked: 80-90% overlap it goes long processing times, what processing times you're experiencing on the job with say 300 to 400 images. So, first of all, our benchmark for photo overlap is 75%. We've found that is the sweet spot for getting to accuracy without getting totally killed on processing. Also, actually, we found if you can significantly increase processing, overlap, you can actually introduce more noise and result in having lower accuracy, so I would encourage you to dial it down a little bit to about 75%. In terms of processing times, 300 to 400 images, so in terms of computer time granted we're using some pretty high-power computers on our site. We're using about $3,000 to $4,000 really high-power CPU GPU processors and computers. It’d probably be about 4 to 5 hours of processing time as well as 1 to 2 hours of human time doing the correction work and the QA/QC work. So, total, you're looking at most of a day for a 300 to 400 photo project but probably about 2/3 of that, 3/4 its going to be computer processing time. Okay. Paul asks about the size of the data files. So, it's a really good question because the size of that 3D model can be several Gigabytes, depending of the size of the project site and the number of photos that went into it, which is why it's important to have that step in-between, the Virtual Surveyor cause if you just try to plug that really-really large 3D model files directly into Silver3D or Carlson, it can't handle that. The .dxf files that come out of Virtual Surveyor are a couple of Kilobytes. So those Cad softwares work totally fine with that. Okay. Peter asks: Are you relying on a third party to process the data and would you rely on a third party to translate the data that your client requires? Not 100% sure I understand exactly what you're asking. We are a third party to our clients, so we do process the data for them and we translate into whatever data they require. So we can operate on any We can operate on arbitrary coordinates. That is one of the things that separates us from your kind of automated third party processing solution as well as most of them don't have that. That's a very kind of surveyor specific requirement that a lot of them don't have and that's something that we're pretty comfortable with. Okay, more questions. Keith asked: Thanks for doing this program. Thanks, Keith. Do you guys offer training? Particularly around that data reduction processing part. We do provide training. A lot of our training is folks from field operation so you The most important piece of the puzzle. In terms of the data processing, we do not do training on the Pix4D processing because data, that image stitching processing because it's such a steep learning curve that even Pix4D provides 3 day workshops that only really teach you the basics but we do include it as part of our package training on that Virtual Surveyor software so we get our clients proficient on how to operate that software. Joe asked about getting to bare earth. I gave a really-really quick demonstration of some more grounds for that. There are actually a lot of work around. Oh, first I should say, like I mentioned, ground cover is one of the main Of this technology so the most important takeaway is to think of a drone is a tool in a toolbox. It's not going to do all of every project for you. So on a lot of projects, you're going to use You know, if you've got to be really-really tight and accurate on an area that's got a lot of Or grass or trees, you're going to want to take out your rover, you're going to want to survey that conventionally. But if that same project has areas that are bare earth or a hardscape or where you don't need to be quite as high in terms of accuracy, you can rely on a drone for those parts of that project and you can merge all that data together in Virtual Surveyor or in your Cad finish software. Having said that, there are some workarounds in that Virtual Surveyor software for getting pretty close to bare earth using those automated contours I showed Contour tool, there's a couple of it Allow you to get pretty tight down. You can also modify the terrain so you can cut just a couple reference shots in` the field or take reference shots from another part of the project and you have to drop to basically set your elevation in those areas that has ground cover. So there are some tools you can use there to . Tom asks: How do you test your data to ASPRS? Software or Manually? Okay, I'm going to do my best on this. I am not the ASPRS certified mapping scientist on the team. Basically, what it involves doing, I'm going to bring that slide back up where I had that image. Basically, what it involves is, you know, the surveyor on the ground surveyed in all these points that are marked in green and then, when we handed over the 3D model to them, they went and they took, you know, using Virtual Surveyor, they effectively took the survey shot at each one of those points. And they did a simple comparison. I should see if I could pull up a report from this, it will take me just a moment. They did a quick comparison and they looked at what was the difference in elevation between those two points - the point that they shot in the field and the point they took off the 3D model. And if I pull up the report they produced, I can find it quickly. Here we go. Basically, this is what Looked like and these are kind of the important numbers to pay attention to. Surveys is going to be the shot they took in the field of what their vertical was there and then the shot they took off of the 3D model and the difference between the two. So you can see, you know, there, I think about 48 survey shot, individual shots they took here, some were high, some were low, somewhere even off by a 0.1 those individual points, some even a little bit more. But what's important, basically, the ASPRS standards, what they instruct is to do a... To calculate across all of these shots what your root-mean-square error is, which is basically just the standard deviation of these vertical differences. So across all of those shots, the deviation, which will be root-mean-square error which basically what you can say you're all in error was on that project was 500 vertical. And they publish standards for how to do this, it's actually fairly straightforward. Emmett asks: Does that $10,000 investment that I suggested include a software that has been demonstrated to produce Profile. Short answer, yes, it's a little bit more, that software, that company actually doesn't offer a perpetual license, you can't just buy the software, there offer monthly or an annual subscription for 125$ a month or I want to say $12,000 a year. So at $10,000 you should still, that should cover, probably you could lump a year of that software into that investment, so it'd be pretty close. Peter asks: Could this office processing be handled by say Carlson survey? What is showed in Virtual Surveyor, taking the 3D model and actually reducing it down to that Tin in the contours and break-lines, Carlson cannot do that but what I exported out of Virtual Surveyor, that .dxf file or even just points in a .csv, yes, Carlson can interpret those just fine, a lot of our clients use Carlson. Peter asks: What FAA waivers do you hold? We apply for waivers for our clients where necessary. Right now it's really just focused on getting airspace permissions. The process of getting other sorts of waivers is we found that's frankly not worth it right now - the Time is so long that it went to not be worth it to our project basis. We have pretty good lines communication with the FAA and they're moving pretty quickly to relax some of the more high priority rules so we don't have to focus too much on applying specific for waivers for . Michael asked: What altitude was the demo capture from? I believe that 120 ft so close to what you'd expect of highest accuracy, highest resolution. Brandon asks: Do we sell the Virtual Surveyor software? WE include it as part of our package so, yes we do. Joe asks: Have any of the firms you worked with expressed license issues? No, we have that conversation a lot of times, you know, we work with clients on pretty much every state and every state has different license requirements. So far we've never ran into an issue with it. Basically, this is still happening under your license that's why we equip you with that Virtual Surveyor software and why we always recommend you doing those, taking those In the filed to determine, to assess the accuracy. So, no, that has not been an issue to this point. Mike asks: What brand was your rugged drone carrying box? We use the Go Professional cases one Alright, I think that is all the questions.
Amber: Oh, a lot of questions. Great!
Amber: I actually have five or six more, but just looking at the time, what I’ll do, Daniel, is I’ll send those over to you and if you could maybe correspond by e-mail and we'll also share the attendees your contact information in an e-mail with our CE certificate so they'll have it and we'll share the attendee list with you so everybody can correspond and continue the conversation as well.
Amber: And thank you so much. Before we sign off, I want to thank everyone once again for joining us today and a big thank you to Daniel for your service to our organization. Daniel is actually joining us from California this morning so he had quite an early morning getting up at 5:30 this morning. As a reminder to our attendees, your CE certificate will be posted on our website next week. We will send you an e-mail to let you know once they've been posted and you'll be able to log in and view and/or print your certificate directly from our website. So that is the close of today's webinar. Thanks again for attending and have a wonderful weekend!