ArduPilot / ardupilot

ArduPlane, ArduCopter, ArduRover, ArduSub source
http://ardupilot.org/
GNU General Public License v3.0
10.73k stars 17.18k forks source link

[VTOL Tailsitter] - Somewhat problematic transitions in automatic missions #18951

Open robustini opened 2 years ago

robustini commented 2 years ago

Unfortunately this know problem affects not only the transitions during a DO_VTOL_TRANSITION to 3, but also the transition during QRTL, which surely follows the same piece of code. It is already detectable on the SITL, and in reality it affects even more negatively. Just analyze the THo in the log to understand what happens. It is also evident that in the Plane->MC transitions the Q_TAILSIT_ANG_VT angle is not respected, just look at any other transition from Cruise to QLoiter (without transition timeout), where instead Q_TAILSIT_ANG_VT is reached. This thing I have already highlighted on Discord further detailing the problem. @IamPete1 I await your opinion, thanks!

Platform [ ] All [ ] AntennaTracker [ ] Copter [x] Plane [ ] Rover [ ] Submarine

Airframe type Non-vectored tailsitter

Hardware type SITL with Realflight

Logs

00000014.ZIP 00000013.ZIP

tailsitter_mission.zip

https://youtu.be/coN4AfPr5Sc

robustini commented 2 years ago

It affects the maximum angle a lot. In normal use Q_ANGLE_MAX is dependent on the type of tailsitter, so its value could also be for example 6000. With such a high value the transitions in VTOL during automatic mission are very bad, just bring it to 1000 for example and they are perfect. This bad result is related to the repositioning imposed by navigation, because even with a 60 degree angle during a transition in QLoiter the result is perfect, at least on the sim.