liubingzsd / gentlenav

Automatically exported from code.google.com/p/gentlenav
0 stars 0 forks source link

loss of GPS communications is not detected and leads to stabilization issues #34

Closed GoogleCodeExporter closed 9 years ago

GoogleCodeExporter commented 9 years ago
What will trigger the issue: disconnecting the GPS after initial GPS lock:

Every once in a while, someone reports the following symptom: roll and pitch 
oscillations in stabilized mode while the plane is motionless on the ground. 
Recently, "eagleeye" reported the symptom. Sid was the first to observe it in 
the earliest version of MatrixPilot. Over the course of the project, I have run 
into it myself once in a while.

A long time ago, I figured out the basic mechanism that produces the effect: 
whenever the yaw error fails to get updated, a constant yaw error feeds into 
the integrator of the PI yaw feedback controller, and ramps the yaw rate 
adjustment to infinity. This causes the plane to think it is in a terrific flat 
spin, and it causes an error in the DCM computations that I have termed 
"coning", in which the plane also thinks it has rolled and pitched as well. 
Basically, the computations get entirely wrong.

We have gradually fixed the various ways the yaw error could "get stuck", but I 
see that there is a mechanism that we have missed: if the UDB loses 
communications with the GPS entirely, there will be two problems:

1. The logic for detecting loss of GPS lock will fail.
2. The DCM computations will produce gibberish, so stabilization will fail.

Original issue reported on code.google.com by wpremerl...@gmail.com on 31 Oct 2010 at 3:58

GoogleCodeExporter commented 9 years ago
Another way I've managed to cause similar symptoms is to move/twist the plan 
while the sensors finish calibrating.

Original comment by benjie on 31 Oct 2010 at 6:04

GoogleCodeExporter commented 9 years ago

Original comment by benjie on 24 Mar 2011 at 12:52