Waypoint

Ask The Operators — A Survey Drone Q&A (Part 2)

During senseFly’s recent Drone Surveying 101 webinar, we dedicated over 20 minutes to an open Q&A session on survey drone technology, with answers supplied in the main by our two client presenters: Marc Cañas of Jacobs and Aidan O’Connor of ASM Ireland. In part two of this extended article (read part one), we explore more of what Marc and Aidan had to say on topics such as: absolute accuracy, BVLOS benefits, turning a DSM into a DTM, data management and more.

Specifically, our presenters answer the following questions (click a question to go straight to its answer):

survey_drone_qa_speakers
Aidan O’Connor (ASM Ireland), Marc Cañas (Jacobs), Brock Ryder (senseFly)

Marc, what’s the highest accuracy you’ve been able to achieve with the drone in your projects?

Marc Cañas, Jacobs: We fly the eBee RTK and we’ve found that the highest accuracy is about two inches. Actually one and a half inches in confined areas, but once you get into longer areas the average we’ve found is about two inches vertical and horizontal. That’s actually with greater than 95% confidence. Those of you who are surveyors will understand that term.

Survey_Drone_QA_Contour
A section of eBee-derived contour map from Marc Cañas’ Californian high-speed rail route survey (read the case study).

Aidan, somebody asked you previously about the 2-3 cm ground sampling distance accuracy. Is this sigma max? Is that what you meant on that?

Aidan O’Connor, ASM Ireland: No, the ground sampling distance is basically the resolution at which the camera takes the photo. The lower you fly, the higher the resolution, but also the less ground you cover. So basically what we cover generally—what we find is a nice fit—is three centimeters per pixel for us.

The lower you fly, the higher the resolution, but also the less ground you cover

You can go higher, you can cover further, but again you’re limited. In Ireland, we’re limited to flying at 120 metres, so we find, roughly, that around 3.2 centimeters per pixel is a height of about 110, 115 metres, so that’s kind of the area we fly at. We find that the resolution is perfect for what we do. So it’s a ground sampling distance.

OK, and one of you brought up BVLOS [Beyond Visual Line of Sight]. Here’s a question we received: were you flying with an exemption for BVLOS? If not, do you think it would that have saved time and money if you had?

Marc: That was me. Absolutely, if we were allowed to fly in BVLOS it would have saved time and money because we could have extended the limits of our initial flights instead of sticking to line of sight only.

Conceivably, with the automation and the autopilot and what we’ve seen on the eBee and the RTK model specifically, we would feel very comfortable having it go out of line of sight. So yes, it would have reduced our costs. No, we did not fly BVLOS; that was not allowed for us. I think there were only three companies at the time that had an exemption for that and we were not one of them.

Conceivably, with the automation and the autopilot and what we’ve seen on the eBee and the RTK model specifically, we would feel very comfortable having it go out of line of sight

Marc, what software did you use to go from DSM to DTM?

Marc: We ended up using Correlator [Correlator3D], I believe, and our mapping arm also employed some conventional—believe it or not—LIDAR tools to go to the DTM. So they cleared out that dataset. I don’t know if Aidan has anything additional to add to that?

Aidan: Yeah, generally for smaller projects what we use is a software called Global Mapper. That’s our main tool. You can view and edit what you want out of the terrain. That’s been quite a useful software for us.

Survey_Drone_QA_UAV_Deliverable
An example survey drone deliverable: a digital elevation model of a quarry produced by ASM Ireland.

Here’s an interesting question: how do you determine at what height you should fly a drone when you want to create an orthomosaic?

Marc: Quite honestly, the orthomosaic is almost a byproduct, but if the orthomosaic is the only thing you’re looking at, the only thing that’s going to limit you is our restriction, at least in the US, that we can’t fly higher than 400 feet, essentially 120 metres. So, that’s really going to determine your biggest [coverage] area.

The other thing you really want to consider is: how much resolution do you want on the imaging? If one and a half inches, or if you need more resolution than that, then you would fly lower, but for us, we’re always just about 390, 375 feet. That seems to be about the average for us.

The files are so huge on the deliverables; how do you deliver them? How do you get them to your end user in a size and a format that they can actually use?

Marc: That was one of my takeaways. Managing the file data and data sizes is something that you’re going to have to take into consideration. Most of our clients are connected pretty well via cloud services so we’re able to deliver most of our data, whether it’s the raw data or processed data, via cloud service.

Managing the file data and data sizes is something that you’re going to have to take into consideration

For those clients who cannot handle that, we’ll actually put them on portable hard drives and send hard drives out. That’s really how we’re transferring data between us. And again, I said that’s one of the challenges, managing that data transfer. Not only the data transfer but the storage, and in the US, privacy is a big deal as well, so having a secure server where you host that data is important.

Aidan: It’s important to bear in mind that we get our surveys generally processed at a grid spacing of approximately ten centimetres. I have yet to come across a consultancy that can use that level of detail, so generally what we’re having to do is to strip it back to maybe half a meter or one meter grid spacing. It’s just that the data is absolutely enormous. Sometimes you can have a surface model that’s well over a gigabyte for a single project. That’s a single small quarry or a single small area that’s maybe 150-200 hectares. So really, what we are generally doing is stripping back the data because people can’t use it.

That’s going to change and we’re constantly evolving different types of point cloud format, stuff like this. Every consultant in Ireland uses different software and they all have their preferred software formats. Some are better than others and we’re constantly tweaking to find out what is the best approach to take.

quarry_survey_output_drone
An example image from an ASM Ireland quarry survey.

Aidan, what was the margin of error you had on your elevations?

Aidan: In terms of absolute accuracy, as I stated in the presentation, what we do, as well as surveying ground control points, we also survey independent checkpoints. These are points that we don’t give to the software, so the software doesn’t know these points. We check these independently, so that can be any other sort of point on the ground.

… generally what we get is a margin of error somewhere in the region of 50 millimetres. That’s across the board

We’ll check five or ten of these per project and generally what we get is a margin of error somewhere in the region of 50 millimetres. That’s across the board. We’ve often got much better and sometimes we get a little bit worse, but generally it doesn’t average much more than 50 millimetres for absolute accuracy. Again, it depends on what you’re using to take your ground control points, but that’s the sort of error that you’re getting.

And for you Brock, we have a lot of questions that have come in about the eBee Plus. How does one go about actually acquiring one?

Brock Ryder, senseFly: Well you know, that’s never an issue, to be able to buy a product from us! senseFly has over 200 points of sale around the world now through our distribution network. The best thing to do is to look at our website, in the Where To Buy section, and that will direct you to your local distributor. Or you can simply pop us an email via info@sensefly.com and we’ll speak to you, direct you and help you as quickly as we can.

OK great. Here’s the last question for you Marc: which drone and camera were you using on your rail project in California?

Marc: On that project we used the eBee RTK and the camera that we were using was the Sony WX camera.

Alright, very good. We’ll wrap up the questions and I apologise that we couldn’t get to all of them. Thanks!

Watch senseFly’s full Drone Surveying 101 webinar on demand:

drone_surveying_webinar_watch

2 Comments

  1. Sheldon Simon

    13.03.2017 - 05:35
    Reply

    Dear Team,
    Want to make a proposal to my manager.
    Question:
    1. can my company purchase a drone (ebee) without processing software and then engage your company for data processing and map preparation? The idea is to reduce the cost to the company yet getting the same quality information without having a member of your team visiting our operations regularly.

    Sheldon

    • senseFly

      21.03.2017 - 11:31
      Reply

      Hello Sheldon. Yes, senseFly drones are currently available with or without image processing software (such as Pix4D or now, also, MicaSense Atlas). As for processing and map preparation, senseFly does not provide these services direct. You would need to identify a provider of such services in your region. Your nearest senseFly distributor can probably help with this – you can locate yours here: https://www.sensefly.com/about/where-to-buy.html Hope that helps!

Leave a Reply

Your email address will not be published. Required fields are marked *