Autolanding & Movinghome & altitude values with GPS+Baro Setups

[julled]

Hi there,

we are using waterproof fixed wing planes with ArduPlane on the open sea by starting and landing from ships (and searching for people in distress).

problem

We want to use autolanding functionality to land next to the ship for easy useability of the drone for the shipcrew. The homeposition is updated using a movinghome-script. As of now we can only use GPS+Baro for altitude estimation.(see appendix for details)
I am aware of the different coordinate systems involved:

tests

We conducted a few tests to test the autolanding in combination with movinghome. There some questions arose regarding the relative altitudes shown in QGC / Mavproxy (mavproxy shows altitude as GPS altitude difference from Plane to GCS GPS) and the altitude used in the autolanding algorithm:

1. Autolanding without moving home
   UAV-rel.Altitude=0m in QGC before takeoff. Everything works as expected. Landing works perfect.
2. Autolanding with moving home (Use GCS GPS altitude)
   UAV-rel.Altitude=-50…50m in QGC before takeoff. This huge error in altitude might result from actualy two bad sources of altitude used here (M9N+Cheap GCS GPS). Due to high possible altitude error we didnt test the autolanding here.
3. Autolanding with moving home (Use GCS APM-Terrain altitude / 0m)
   UAV-rel.Altitude=-50…50m in QGC before takeoff. The huge error as in test 2 remains (maybe due to still bad M9N GPS altitude). Due to high possible altitude error we didnt test the autolanding here.

Questions:

1. Why is there no huge altitude error shown in test 1? I expected that there also the bad UAV GPS altitude measurements are used for the relative altitude, but the altitude remained rockstable. Does changing the homeposition via movinghome change some of the altitude estimates within ArduPlane?
2. Whats the altitude source for autolanding? Does it use the fused EKF local coordinate system estimate (see CODE), or direct altitude sensor (GPS, LiDAR,…) measurements?
3. Do you think its feasable to reliably do autolanding only using GPS+Baro? As we land on the water, precision location is not mandatory. Laser based sensors are not a option for waterlanding(shine through), Ultrasonic neither due to saltwater, but cheap RADAR could be , but should be avoided.

appendix

Our UAV sensor Setup relevant for altutide sensing:

• 1x PixRacer FCU
• 1x mRobotics GPS M9N
• 1x Baro

Our GCS sensor Setup relevant for altutide sensing:

• 1x cheap USB GPS

[Truth_T (Truth)]

I’m also programmatically updating the altitude, and found the altitude is -17m when in the ground.

thank you for replying my articles.

[VRquaeler (sascha)]

@julled

• Why is there no huge altitude error shown in test 1? I expected that there also the bad UAV GPS altitude measurements are used for the relative altitude, but the altitude remained rockstable. Does changing the homeposition via movinghome change some of the altitude estimates within ArduPlane?
• Whats the altitude source for autolanding? Does it use the fused EKF local coordinate system estimate (see CODE ), or direct altitude sensor (GPS, LiDAR,…) measurements?

Hi @julled,

Are you aware the autopilot (in default configuration) uses the barometer as main altitude source?
The relative altitude is zeroed with every arming/takeoff.