I might ask @tridge and/or @priseborough to chime in here but I think it must be best to get the stable position during pivots and if this leads to oscillations in navigation then it should be possible to re-tune either NAVL1 param sor the turn rate controller and restore performance to what it was (or better).
So the key point I think is whether the GPS_POSn_X/Y/Z positions are relative to the autopilot’s IMU or relative to the middle of the rear-axle (aka COG in the docs). The docs say this:
In practice the distance to the sensor can be measured from the center of the autopilot unless the autopilot itself is placed a significant distance from the vehicle’s center of gravity in which case the IMU position offsets can be specified and then the other sensor’s position offsets can be specified from the vehicle’s center of gravity.
If we have an issue in the code I think it could be here in NavEKF3_core::CorrectGPSForAntennaOffset and this code does seem to use both the GPS offset and the IMU offset. TBH I would have thought that we shouldn’t have the IMU offset in the calc but I think I better let Tridge and Paul weigh in.
Txs for the report.