https://drive.google.com/file/d/1Qr-EsF0pznWxyK7LzpKu3VjxTEQzHAjA/view?usp=sharing
Greetings all
We are trying to upgrade our configuration to a firmware version beyond 4.3.8. Our crafts are GPS-for-Yaw style, with a stack of Ublox boards where the moving-baseline rover feeds the Autopilot with the position and rtk vector, and the moving-baseline base is physically attached to UART2 of the rover, feeding the rover RTCM messages. Due to our design, we cannot rely on the Cube’s internal compass. Due to the GSF not being available yet in this version, we set COMPASS_ENABLE=1 and COMPASS_USE=0. We get a yaw solution and are off to the races.
Every time we have tested 4.4.0 and above, some part of the ArduPilot core code is unhappy with the gps solution, but there is no obvious culprit. In the above log, you’ll see that the message delta is mostly 200ms, the solution is RTK-Fixed, and the GPS antenna locations are reasonably accurate. In this craft, they are on opposite sides of the cube laterally, 53cm apart. The EKF reports that it has no horizontal position so hence we fail prearm checks that require a position, as well as being unable to go into modes requiring this. EKF innovations for POSD are below 1. I think the GSF is configured correctly, but I don’t know for certain, regardless, the heading doesn’t seem to reflect the compass even if it were turned on.
In the above log, I attempt to use GPS autoconfig and mess up the GPS type, it is supposed to be ublox movingbase-rover. I will fix this tomorrow, but before the majority of the log this was set to type=ublox This configuration works for us in versions 4.3.8 but not after, and I don’t know why.
I am at a loss