MC and FW rate controllers write integrators into one topic

[le-chat]

MC and FW rate controllers write integrators into one topic rate_ctrl_status: https://github.com/PX4/PX4-Autopilot/blob/f6eae085978d7a11483156b5d4e461c875b46729/src/modules/fw_att_control/FixedwingAttitudeControl.cpp#L622 https://github.com/PX4/PX4-Autopilot/blob/f6eae085978d7a11483156b5d4e461c875b46729/src/modules/fw_att_control/FixedwingAttitudeControl.hpp#L117 https://github.com/PX4/PX4-Autopilot/blob/f6eae085978d7a11483156b5d4e461c875b46729/src/modules/mc_rate_control/MulticopterRateControl.cpp#L228 https://github.com/PX4/PX4-Autopilot/blob/f6eae085978d7a11483156b5d4e461c875b46729/src/modules/mc_rate_control/MulticopterRateControl.hpp#L107

However, for VTOLs both controllers work simultaneously. The result is that usually MC integrator wins.

Here in the two logs pitch rate error integral is not zero (that is it is from MC -- on v1.11.3 FW integrators disabled in MC mode even for tailsitters): https://review.px4.io/plot_app?log=209cb3c3-3929-468f-83ef-7f2a9f86e2ed https://review.px4.io/plot_app?log=bdb47a46-d68d-4f06-bf3d-fbb17d260f93

Possible ideal solution:

• create topics rate_ctrl_status_fw and rate_ctrl_status_mc and let each controller write to dedicated one (probably, do so only on vtols)
• teach Flight Review to use right one depending on aircraft type, mode and roll/pitch/yaw.

[bresch]

You could do something like done here: https://github.com/PX4/PX4-Autopilot/blob/06cd3eded52297ee2277163a445efed0571ac481/src/modules/fw_att_control/FixedwingAttitudeControl.cpp#L46

_controller_status_pub(vtol ? ORB_ID(rate_ctrl_status_1) : ORB_ID(rate_ctrl_status_0))

And change the multirotor publication to rate_ctrl_status_0 And define rate_ctrl_status_0 an rate_ctrl_status_1 in the message definition, like in https://github.com/PX4/PX4-Autopilot/blob/06cd3eded52297ee2277163a445efed0571ac481/msg/actuator_controls.msg#L26

[dagar]

You could also simply publish multiple instances by using uORB::PublicationMulti<> instead. Then in each we could add an enum to identify the source.

[le-chat]

@dagar They are already uORB::PublicationMulti. However, in logs we see only MC controller's data. Can added enum change this?

[dagar]

@dagar They are already uORB::PublicationMulti. However, in logs we see only MC controller's data. Can added enum change this?

Great, then the next step is to add some identifying information to each msg (the enum).

It looks like logger is already configured to log 2 instances. https://github.com/PX4/PX4-Autopilot/blob/5f4802a239321b9f3d1845f4e7b4262030752039/src/modules/logger/logged_topics.cpp#L130

[le-chat]

Ah, yes, in master rate_ctrl_status logged with add_topic_multi, it is the stable v1.11.3 where it's logged with just add_topic. How do resulting logs look in Flight Review, interestingly. Can't browse public logs with recent master and VTOL. Probably, the simplest way to check is to build a v1.11.3 with this patch and fly with it.