Log vehicle local position setpoints for fixedwing position control

[Jaeyoung-Lim]

Describe problem solved by this pull request
Previously it was hard to interpret what the fw position controller was doing in term of tracking performance since the position setpoints were not logged in the local frame.

• It is hard to associate waypoints switching with the vehicle behavior, given that the fw pos control overrides the setpoints internally sometimes
• This is especially a problem when passing high bandwidth setpoints from offboard or for path tracking.
• For VTOLs the position setpoints just stops logging when the vehicle transitions to FW mode, since the local_position_setpoint is only published from the MC position controller module

Describe your solution
This adds publication of the vehicle_local_position_setpoint so that it gets logged as a local setpoint.

Describe possible alternatives

• I think a bigger problem creeping is that all our position setpoints are in the global frame due to L1 only supporting lat/lon as a input. For offboard mode, we are converting local->global and global->local two times for this reason. I would like to move towards doing fw control in local coordinates since this would allow us to fly even in GPS denied situations.

Test data / coverage
Tested in SITL Gazebo

make px4_sitl gazebo_plane

and the log correctly shows the position setpoints and how the waypoints are switching.

• Before PR: FlightLog
  

• After PR: Flightlog
  

Additional context

[dagar]

populate the unused fields with NANs?

local_position_setpoint.vx = NAN, etc

[dagar]

Alternatively does populating any other existing data make sense (for casual interest)?

• acceleration from L1?
• velocities from TECS (remove wind?)

[Jaeyoung-Lim]

I think for now IMO there is not much more to add - e.g. velocity setpoints are not used byt the controller / accelerations are not used by the controller. I think we can add these by cleaning up the controller one by one (e.g. change L1 controller to command accelerations, rather than roll angles)

[sfuhrer]

Some stuff - e.g. lateral acceleration- will be published with the new L1 controller (#18810) in npfg_status. I would also find it nicer if we publish whatever we can in local_position_setpoint. That though also requires some controller interface alignment - using local coordinates prob is one thing of that.

[dagar]

Looks good, although doesn't this need to be skipped during VTOL transitions?

[Jaeyoung-Lim]

Looks good, although doesn't this need to be skipped during VTOL transitions?

@dagar Done! Happy new year! 🎉

[sfuhrer]

Thanks, that makes sense.

Let's figure out if it makes sense to pull some more data out of tecs_status and npfg_status into vehicle_local_position setpoint.

[sfuhrer]

Some stuff - e.g. lateral acceleration- will be published with the new L1 controller (#18810) in npfg_status. I would also find it nicer if we publish whatever we can in local_position_setpoint. That though also requires some controller interface alignment - using local coordinates prob is one thing of that.

[Jaeyoung-Lim]

Some stuff - e.g. lateral acceleration- will be published with the new L1 controller (#18810) in npfg_status. I would also find it nicer if we publish whatever we can in local_position_setpoint. That though also requires some controller interface alignment - using local coordinates prob is one thing of that.

Fully agreed! I think now that we have two controllers we can use this as an opportunity to clean up the controller interface