Hello,
I am having problems running autotest.py in Ubuntu. I made two small modifications to get to my current state:
- In runsim.py, I changed the JSBSim expect to wait for "Test successful" instead of "trim computation time"
- In fgout.xml, I changed "udp" to "UDP"
But when I run it, there still seems to be other problems. fly_arduplane test never seems to finish. Please see attached for the console output. Is this script still being maintained? I see a couple things that don't seem right:
- The aircraft never gets any closer to waypoint1.
- Never gets off the ground. (agl < 0)
- roll is 90 degrees
- pitch is negative (should be positive for takeoff)
- FPS is 8000 and cpu load is 100% (I would have expected this to be 1000)
Any help you could give would be greatly appreciated.
Thanks,
Ritchie
Replies
Seems your simulated flight crashed (line 5945 of the log) after some erratic flying. I'm seeing very similar behaviour running autotest using cygwin with all the default settings.
When I run fly_ArduPlane the "takeoff" phase is very unstable but usually successful. The next phase "fly_left_circuit" is done with 90+ degree rolls and knife edge descents with compensating climbs. The plane will rarely complete this test without crashing, in which case it will usually crash in the next phase, "axial_left_roll". This is all made very clear by watching the flight in Flightgear.
Although I haven't yet used ardupilot in a real system, I don't believe this instability is the intended/accepted behaviour. The PID gains being used are read from autotest/ArduPlane.parm. Is it possible these parameters are incorrect and need retuning? Any other ideas what could be wrong. Logs attached.
Thanks for any help.
fly.log
flight.tlog.raw
mav.parm
Hello Adriel,
I now have the very same problem with autotest.py and the circling after reaching he altitude.
See uploaded log.
My set-up is:
Win 7 Home Premium 64 bits
Java 7 Update 51
Framework 4 Ext.
Cygwin 1.7.28-2
Cygwin python 2.7.3-1
It seems that it is turning the wrong way after:
" Waiting for heading 270 with accuracy 10 "
Did you find a solution now?
RGDS
Staffan
skjAUTOTEST.txt
It looks like these may be symptoms that the aircraft crashed during an earlier part of the test before it hits fly_mission. Has anyone else tried running autotest.py recently? Are the parameters in ArduPlane.parm still valid?
Thanks,
Ritchie