Survey and Mapping Drones: RTK and Camera Configuration

A mapping drone is judged on one thing: does the map you produce match reality to the tolerance the job needs? Speed is nice. Accuracy is the job. And accuracy is where most of the effort goes, because it’s easy to fly a grid and surprisingly hard to fly one that holds up on the ground.

What RTK actually fixes

Every photo needs to know where it was taken. Without corrections, that position drifts by a metre or more, and you paper over it with ground control points: you go out, survey known markers, and use them to tie the model down. With RTK, corrections stream to the drone and each shot is positioned to a couple of centimetres. You still want a check point or two, but you fly fewer ground control points, or none on easy sites.

There are two ways to get corrections. A local base station sits on a known point near the site and radios corrections up. Network RTK pulls corrections from a service over mobile data. Which suits you comes down to site size, how remote you work and whether you have phone signal. We run through that choice in base vs network RTK, and the hardware sits under avionics.

Camera: global shutter, real sensor, known lens

For photogrammetry, a global shutter beats a rolling shutter. Rolling shutter scans the frame line by line, and from a moving drone that skews the image, which the software then has to undo. Global shutter captures the whole frame at once, so each photo is a clean, frozen position. That’s why serious mapping cameras cost more.

After the shutter, three things matter: sensor size, so you get clean images in less than perfect light; a lens whose focal length and distortion you can feed to the software; and enough resolution for the ground sample distance you promised, the GSD, which is simply how big one pixel is on the ground. If the client wants 2cm, you plan the flight around delivering a genuine 2cm, not a hopeful one.

Flight planning is most of the result

Overlap is the first lever. Most mapping work runs 75 to 80 percent forward overlap and 60 to 70 percent side overlap. Too little and the model has gaps; too much and you’re just burning battery. Altitude follows from GSD: fly lower, get finer detail, cover less ground per line. Speed has to be slow enough that the camera fires on the grid you planned, with no motion blur. And fly a stable altitude, because a wobbly flight path shows up as a wobbly surface.

This is where a clean, repeatable hover and cruise helps. A propulsion set that holds its altitude calmly means the camera fires where you intended. The modular propulsion systems we ship for industrial airframes are built for exactly that kind of steady flight.

Endurance for a full grid

A mapping sortie covers a rectangle, and the rectangle doesn’t care that your pack is half empty. Work out the area, the line spacing and the total distance, then add the time to climb, turn and come home. A grid that needs 25 minutes means a pack sized for 30, and the pack sizing method is in the UAV battery guide. If the area is big, the answer is usually more batteries and planned stops, not a bigger aircraft.

Data flow and the boring checkpoints

Good output needs a tidy process. Download the images, keep the RTK data with them, run the check points, and only then process. If the check points disagree with the model, something in the flight went wrong, and it’s cheaper to find that on site than after processing. Keep the flight log too. It tells you altitude, speed and battery, which is exactly what you need when a model comes out soft.

Mistakes that cost you a re-fly

Most bad maps come from a handful of avoidable things. Overflying in wind that tilts or drifts the aircraft, so the lines wander. Flying too fast for the camera to fire on the planned grid. Changing the altitude between passes when the plan said hold it level. Or forgetting that the light was changing fast, so half the images are blown out and half are dark, and the software has to fight to match them.

Then there’s the one that catches people late. Forgetting to keep the RTK corrections with the images. Without that position data the shots are just photos, and you’ve lost the accuracy you paid for. Export the correction log with the flight, every time, and you’ll never lose a day’s work to a file you didn’t copy.

A configuration that works

A stable multirotor with mid KV motors and large props for an efficient cruise, RTK on board, a global-shutter camera with known optics, and enough pack for the full grid. If you want the wider picture, the avionics guide covers the RTK and gimbal side, and the applications guide puts survey alongside the other jobs.

Frequently asked questions

Do I still need ground control points with RTK?

Fewer, and often none on easy sites. Keep a couple as a check even when the corrections look clean, as the survey configuration guide recommends.

Base station or network RTK for mapping?

Network is faster to set up and fine where you have signal. A base station wins on remote sites and large areas, where mobile coverage is unreliable, which is the comparison in our RTK guide.

Does shutter type really matter?

Yes, a lot. A global shutter stops the skew a rolling shutter causes from a moving drone, and that skew costs you accuracy.

Why did my map come out the wrong scale?

Almost always position or GSD. Check the RTK fix, the check points and the planned GSD before reprocessing.

How much overlap should I fly?

Roughly 75 to 80 percent forward and 60 to 70 percent side. More overlap is safer but eats battery, so plan it against your grid size.

Tell us the area, the GSD the client wants and how remote the site is, and we will help match the RTK, camera and propulsion to the job. The range is on our avionics page, with the configuration view in our survey and mapping guide.