We had a flyaway incident that was primarily due to power loss to the pixhawk cube orange. The pixhawk is powered with POWER1, POWER2, as well as USB side power. The rail is also powered for peripherals. After power loss, both GPSs rebooted and the drone entered EKF failsafe and switched to LAND mode. Unfortunately, the drone did not actually land. It continued to hold it’s attitude (as far as I can tell) and flew away until it ran into a tree.
Could someone help us investigate the log and determine why the FC didn’t actually land the drone?
Any insight is appreciated, thank you.
I don’t see that the GPS’s are rebooted before the EKF Failsafe happened. Both gps provided updates in every 200mS till your copter hit the trees. So the main reason for the EKF failsafe was not this.
Based on the Throttle/Compass check both of your compasses are affected by the motor throttle. Which easily can lead to EKF failsafe on high position innovations.
Why it did not landed (apparently)
LAND initiated by failsafe has a four second delay, during this delay your copter kept the altitude, then started slowly descend. But your LAND_SPEED was 60 cm/s ! So the descend speed was almost undetectable. Then the EKF failsafe cleared, (lanes were switched) and you switched to RTL which cancels the LAND. After three seconds the now active lane failsafed as well, EKF failsafe kicked in and switched to LAND, but the four second delay started again, and then the 60cm/s descend started to wind up again. until copter hit trees.
Thanks for the analysis. I had a few followup questions:
I saw a lot of repeated “GPS detected as UAVCAN” messages in the logs. We usually only see those on bootup which is why I thought they were rebooting. Is there some other reason those might show up like that?
I was not aware that 60 cm/s was a difficult to detect land speed. It’s worked pretty well for us the last month or so. What’s the minimum speed you recommend?
We’re looking into the compass issue as well