RTK for precision landing

Hi everyone! I’m trying to do a Proof of Concept for a drone for precision agriculture… Point is, the drone is supposed to land automatically in a specially designed “nest”.

This requires a pretty landing precision, and we’re currently trying to find if RTK is better than VIO/SLAM.

Our drone is a Holybro X500 v2 that uses the Holybro M8P RTK rover.
After plugging in the base M8P into my laptop running QGC, i waited for the survey-in to finish (like 20 minutes) and tried doing a simple L shape mission, with the drone returning to the takeoff spot.

But out of 10 flights, during 10 of them the drone landed anywhere between 15 and 40cm away from the takeoff mat (a 40cm circle with a H on it).

Is that the expected precision of RTK? Or am I doing something wrong?
Thanks in advance!

Can you upload a log. :+1:

Sure! Here it is

https://logs.px4.io/plot_app?log=dfe0ff0b-d1fe-498d-a90e-0bdce291797f

You have pretty large vibrations and mag interference. Fix those first then see if it gets any better.

Also make sure you’re surveying-in the base RTK with a good precision (at least 0.1m).
It will take much longer but also increase landing precision.
With a decently tuned quad you can expect a deviation of 10 cm tops.
I personally wouldn’t use RTK for precision landing because it’s just to fluctuating and flakey to offer reliable precision for landings.
Cheers

I had the same experience. It got me really confused.
I made this post back then because I noticed the logs display the GPS coordinates as float: RTK GPS PX4 in float

But since it seems to work for @Christian-Born I guess I’m making a mistake somewhere. I will keep an eye on this thread and see where it goes or find my own mistake.

@ [RoboKrops_Astro]

From the Flight log analysis , I see from Thrust vs Magnetic field plot , it has a strong correlation which should not be the case at any cost .
Please ensure that you have the GPS , flight controller away from the ESC. You could achieve this by using Aluminum foil as explained here by another colleague in the forum