Feedback on APM Plane test - possible bug?

We planned to do a 40Km test flight this weekend.

We used APM 2.78

We setup a circular mission of aprox 2km and set the mission to do it 20 times.

Everything looked fine until the 12th lap. Plane started to clime and headed in wrong direction. Eventually I took it off auto and landed the plane.

We saw a few funny readings, Alt was constant on 98.33 even through the plane had landed.

Attached is the tlog file.
Failsafe was not set.

In the kmz file we saw that in the first and 12th lap apm went into failsafe for a short while then back to auto. On the 12th lap when apm when back into auto mode, some information was not updated there after.
If you graph alt_vfr_t_r v/s alt_gps_raw_int_t
alt_vfr show alt of around 100m until 14:15 from there it is a fixed line. Compare to the alt_gps one can see that plane increase altitude to about 300m before I changed modes to stabilize. The same time the plane was going 90 degrees off course.

Is the a possible bug when the apm goes in and out of auto mode because of failsafe?



Why did you not have a failsafe set?


At the time I wanted APM to continue with the mission and not RTL when a possible failsafe situation occur. I couldn’t get information to setup failsafe and continue with mission in auto so I just left it out. Last night I found out how to setup failsafe and have auto to continue with the mission, and tested it in the bench.

The furthest my plane should have been away from me was 800m on this mission. The reason it did lose 2.4Ghz was probably interference with other 2.4ghz signal and not RC transmitter as I was the only person flying.