ArduPilot
fixed statbility issue in tailsitter in qrtl mode
ArduPlane : Fixing issue in tailsitter https://github.com/ArduPilot/ardupilot/issues/25771
Changed tailsitter condtion to maintain height after transition into VTOL and in QRTL mode Increased duration transition to VTOL in tailisitter to 2 seconds to provide better stability.
Before:
https://github.com/ArduPilot/ardupilot/assets/70840360/1ede6dd3-1334-4d6c-93ac-c7cb5831fa06
After:
https://www.youtube.com/watch?v=KfMJsq_d0Rk
Thanks @nishant-sg for the fix, you forgot a line, check my fork, please update the PR.
set_climb_rate_cms(0);
The right quadplane.cpp for reference.
The video you linked as solving is not correct. This is the correct link to the video showing how well this PR works.
https://youtu.be/KfMJsq_d0Rk
Another video demonstrating how well this PR works.
https://youtu.be/yc7jVmh3nIo
Another video demonstrating how well this PR works.
https://youtu.be/yc7jVmh3nIo
There is not very much we can do to take this forward as it is, there are outstanding questions that need to be resolved.
Better and better, I don't fly tailsitter without this PR.
https://youtu.be/ne_7n9bvjYg
This is the classic behavior in QRTL without this fix, from a log a user sent me. @IamPete1 if you are not conceptually okay with this fix I think something else should be done, at least similar for the tailsitter. Regardless of tuning it is still a conceptually wrong approach for a non-vector tailsitter, you can't combine a repositioning with a descent and expect the attitude to be maintained. Another VTOL does that but not a tailsitter not vectored. Repositioning after the transition dev be done with maintaining the height until the home is reached like in this fix, that way if the tuning is ok it will keep the trim. My eXplora tailsitter with this fix has never failed a QRTL, and it has failed many before this fix, and I have documented it well. Please discuss this in the next dev call, thanks (@tridge @peterbarker).
https://github.com/user-attachments/assets/47528e2d-786e-4f98-9a26-930c7934fcf8
@robustini can you submit your tailsitter model to SITL_Models, along with SITL parameters, so we can reproduce
@robustini can you submit your tailsitter model to SITL_Models, along with SITL parameters, so we can reproduce
I don't remember if the parameters are up to date, and anyway in reality the problem is bound to be more obvious. In SITL obviously some wind must be used, which is almost always present in real flight. But I am sure you understand what I mean, repositioning with lowering altitude is deleterious to tailsitters, tested countless times in real flight.