GPS accuracy decreasing without corresponding decrease in DOP or number of satellites

I have a companion computer onboard a Pixhawk drone performing a mission of:

  • store current location on companion as home location
  • takeoff to 6m altitude
  • setting position target about 30m away
  • setting position target to return to 14m above the home location
  • once reached a USB camera is turned on to assist landing
  • set position target of 6m above home location repeatedly
  • once 6m altitude reached, perform landing

Sometimes the drone flies back to a location that is over 7m horizontally away from the expected home location yet the logs do not indicate this and in the last 3 steps above, Arducopter thinks it is directly above the home location as requested (<0.2m difference).

It seems I have a GPS issue, so I examined the log of a typical flight (available here).

HDop, HAcc and NSats look good for the entire flight:

As mentioned, once landed Arducopter reports that the takeoff location is identical to the landing location. Anyone know if the information I am logging is trustworthy? I was hoping to see much more significant GPS problems in the log.


The information that you are logging comes directly from the GPS module. The arducopter firmware does to change it in any way.
So I guess you should complain to the manufacturer that the horizontal accuracy values are 7 times smaller than reality.

1 Like

Is there a parameter that describes the difference between what the accelerometers think velocity is, and what the GPS thinks? (It must be clear I have no idea how Arducopter determines velocity!)

I see NTUN includes a parameter that is “Actual accelerometer + gps velocity estimate” which is vaguely the sort of thing.