Split up control methods for backtransition #23893
Conversation
…sition Signed-off-by: RomanBapst <bapstroman@gmail.com>
| _npfg.setAirspeedNom(_performance_model.getCalibratedTrimAirspeed() * _eas2tas); | ||
| _npfg.setAirspeedMax(_performance_model.getMaximumCalibratedAirspeed() * _eas2tas); | ||
|
|
||
| Vector2f virtual_target_point = Vector2f(cosf(_backtrans_heading), sinf(_backtrans_heading)) * 100.0f; |
There was a problem hiding this comment.
Can you make a comment for the "100" magic number? Is it enough for all types of VTOLs?
@sfuhrer Yes, the reason I implemented this was actually because the test vehicle flew a turn with the previous implementation, which was just to set roll to 0. I guess it makes a big difference it you have crosswind or not. |
| } | ||
|
|
||
| Vector2f airspeed_vector = Vector2f(cosf(_local_pos.heading), sinf(_local_pos.heading)) * true_airspeed; | ||
| Vector2f ground_speed = airspeed_vector - _wind_vel; |
There was a problem hiding this comment.
@sfuhrer Sign error
Signed-off-by: RomanBapst <bapstroman@gmail.com>
| true_airspeed = _performance_model.getCalibratedTrimAirspeed() * _eas2tas; | ||
| } | ||
|
|
||
| Vector2f airspeed_vector = Vector2f(cosf(_local_pos.heading), sinf(_local_pos.heading)) * true_airspeed; |
There was a problem hiding this comment.
@RomanBapst have you checked in simulation how well this works with a high crosswind backtransition? As there the course is not equal tot he heading, so this assumption here gets stretched...
There was a problem hiding this comment.
@sfuhrer I will run a simulation with high wind and post results here.
Solved Problem
Previously, the method to control a transition to hover was handling both cases when position was available, and when it was not available.
It's cleaner to have the methods separates for once. Other than that, the method which controls the transition for the case when position is not available, now does heading hold, which is better than just demanded 0 roll, which was done previously.
Changelog Entry
Control heading during backtransition when local position is not available.
Test coverage
Currently tested in SITL, following up with plots from real tests.
Context
Related links, screenshot before/after, video