Accuracy of UAV Photogrammetry compared with network RTK GPS

I hope this document can help clarify on the accuracy that can be achieved using drones.


Oh man that’s cool. I’ll dive in deeper next week. If you feel the urge, a summary added to would be fantastic. Repository here: GitHub - OpenDroneMap/docs: 🎉 Contribute to OpenDroneMap's documentation! Read how below! 🎉


ok, I’ll try

Awesome. One place it might fit well is a discussion of accuracy at the end of / docs/gcp.rst at publish · OpenDroneMap/docs · GitHub

But a top level page would be ok too.


1 Like

I’m thinking on extending the GCP section, but I will consider to make it a top level page.

Thank you.

1 Like


I hope this helps put some light on the question of accuracy.

To start using this software for large scale photogrammetry, suitable for a fit-for-purpose engineering grade the image convergence really needs to conform to surveyed ground control. Basically creating a plan instead of a map.

UTM datums are basically generated to make calculating areas & distances easier than using Lat/Long.

Ideal survey control stations to utilise the bundle adjustment performed by ODM.

AeroTriangulation Bundle Adjustment

AerTringulation Exposures

Notice the survey control all on the outside and heavily braced at the end of the image runs.

Unfortunately, survey control is time consuming and expensive but for quantifiable georeferenced accuracies, it’s needed.

The other way to quantify the photogrammetry results is to utilise LiDAR on the drone, essentially measuring one nearly accurate reduction against another for redundancy.

Bramor seems right in saying that the photogrammetry does tie itself to the RTK system between 0 & 0.047 Horizontally and 0 & 0.07 Vertically.

Yet this could mean that one point relative to another could be -0.07 and +0.07. a difference of 0.14m between control points.

I hope this sheds some more light on accracy.




I definitevely agree with you.
Thank you for this information, I really appreciate it.