FW/VTOL Call: November 22, 2022

tstastny · November 22, 2022, 5:49pm

November 22, 2022

Tracked issues from last meeting

FW runway landing / takeoff revisions

in review @dagar
control allocation for fixed-wing issues: @ryanjAA will document and make issue

Failure detector - roll/pitch

@chris1seto will check circuit breaker and we look next meeting

Acro mode

github.com/PX4/PX4-Autopilot

Fixedwing reset rate integral in acromode

PX4:main ← PX4:pr-reset-acro-yawrate

opened 08:03AM - 26 Oct 22 UTC

Jaeyoung-Lim

+1 -0

## Describe problem solved by this pull request In acromode, the yaw integral c…an become large since the vehicle have aero dynamic forces that try to stabilize the vehicle ## Describe your solution Discussed in the FW dev call, resetting integrals in acro mode @chris1seto FYI ## Describe possible alternatives - we could also try to only reset the yaw integral and keep the integral of other axis such as roll and pitch ## Test data / coverage - Tested in SITL ``` make px4_sitl gazebo_plane ``` - Flight log: https://review.px4.io/plot_app?log=f2fd0b16-92fc-4308-b19f-a9d6a789e1a4 The integrals are reset in acro mode ![bokeh_plot (1)](https://user-images.githubusercontent.com/5248102/197974360-e5a5cf07-9150-45da-8fee-bc7eff80d229.png) ## Additional context - This topic was discussed in the fw dev call: https://discuss.px4.io/t/fw-vtol-call-october-25-2022/29505

go with the integral reset
could also disable proportional gain
@chris1seto will test all combos and report back

New issues

Quad-tailsitter SITL model flies poorly

github.com/PX4/PX4-Autopilot

Add support for quadtailsitter in SITL Gazebo

PX4:main ← PX4:pr-sitl-quadtailsitter

opened 01:12PM - 07 Nov 22 UTC

Jaeyoung-Lim

+77 -3

## Describe problem solved by this pull request The current tailsitter model ha…s control surfaces, and a control surface-less tailsitter was not available for testing in simulation ## Describe your solution This PR adds support for a quadtailsitter, which models the HEQ SWAN K1. The 3D mesh was provided by Holybro ## Test data / coverage Tested in SITL Gazebo ``` make px4_sitl gazebo_quadtailsitter ``` ## Additional context - Depends on https://github.com/PX4/PX4-Autopilot/pull/20511

CG behind CP makes it stable… but this is wrong
@sfuhrer

Unifying rate controller

could be close… @Jaeyoung-Lim checks wheel controller, @sfuhrer @tstastny support

Updates

Fixed-wing hand-launch audit

@sfuhrer did some testing and clean up!
check that launch and runway don’t interfere @tstastny
need to check what happens with commanded takeoff. maybe clean up @tstastny @sfuhrer
let’s skip to next mission item if clicking takeoff waypoint after already in air (takeoff waypoint is NOT a position setpoint)

github.com/PX4/PX4-Autopilot

FW hand-launch takeoff improvements

PX4:more-fixed-wing-takeoff-and-landing-revisions ← PX4:pr-fw-hand-launch-revisions-mfwtalr

opened 12:21PM - 07 Nov 22 UTC

sfuhrer

+692 -755

Further FW improvements based on https://github.com/PX4/PX4-Autopilot/pull/20537…, mostly centered around automatic hand-launch. ## Describe problem solved by this pull request - bad FW land detection and complicated/impossible to customize - launch detection is messed up if before launch detection the land detection triggers (to !landed) ## Describe your solution - link land detection state machine to launch detection: don't allow the land detection to go to !landed as long as no launch is detected (only when launch detection is enabled) - expose land detection triggering time as parameter (LNDFW_TRIG_TIME) - reduce default of LNDFW_VEL_Z_MAX to 1m/s (it's unlikely that a vehicle sitting on the ground has a larger v_z estimate). And remove the scale of 0.7 that was applied in the airspeed-less case to make the thresholds only tighter then. - remove LAUN_CAT_PMAX parameter that was only during the motor delay phase of a catapult start. According to the param description, it is only there to limit the max pitch up during a bungee launch, and beside that I'm doubting that many people use PX4 on bungee launches, I don't see it crucial for safe operation. - set the bearing setpoint during a handlaunch to the yaw at launch detection for NPFG While on it I couldn't hold back from refactoring the whole launch detection architecture quite a bit. The original intend (asaik) was to have different launch methods implemented, that then each have their own parameters and state machine. Implemented was though only one (catapult, which also works for hand-launch). So I removed the rather complex setup with dynamically instantiated methods and limit the launch detection to catapult, which results in having a very compact and easy to maintain state machine and structure. ## Test data / coverage SITL tested and one real flight.

chris1seto · November 26, 2022, 6:05am

I’m not sure where the very original issue was reported, but regardless of CBRK_FLIGHTTERM, the board doesn’t disarm when inverted in manual mode.

I think we should purge this item on the agenda unless someone brings out a working issue reproduction guide.