Dev Call Jan 16, 2023

Issues & Pull Requests



Attendee count (max): 18

UTC2300 - Plane: Auto Landing Refinement by rbeall · Pull Request #22378 · ArduPilot/ardupilot · GitHub

  • Auto-landing refinement
  • Headwind compensation parameter to set most headwind to assume
    • 50%
  • Back from a few weeks ago
  • Why forward and inverse operations?
    • Seemed to make it more reasonable?
    • One moves from pct to scaling factor, the other m to cm
      • They cancel, but…
  • This could make for very high energy landing?
    • Might end up landing at full throttle
    • Very few aircraft want to come in at full throttle
    • This could also be a tailwind!
      • Could come in much faster than your fbwa
    • Tridge is concerned about doing full throttle landings…
  • This might need an option bit?
    • To Use max airspeed or use cruise airspeed
  • Some users do set the arspd max past the capabilities of the vehicle
    • Don’t break existing setups….
  • We might have a bad wind estimate…
    • Different people have asked for wind speed to converge faster
      • Others want it slower…

UTC2312 - AP_Param: allow class specific defaults using pointer diff. by IamPete1 · Pull Request #22531 · ArduPilot/ardupilot · GitHub

  • Allow class-specific defaults using pointer-diff
  • Defaults don’t change
  • Costs us a little bit of RAM if you don’t have OSD
    • 1kB
    • Why?
      • This should be a win, not a loss!
      • Should be saving 20 bytes/parameter?!
      • Need to be careful in testing
        • E.g. Param ftp can allocate a thread late when you do your first ftp, and that can stuff your measurements
  • PH to work out what’s going on with memory
  • Needs to be tested
    • Flown
    • Param download defaults work?
    • Would be nice for Henry to test with OSD plane
      • Should behave normally
      • Just need to test no changes

UTC2322 - AP_Scripting:Add mission to Script_Controller and improve by Hwurzburg · Pull Request #22593 · ArduPilot/ardupilot · GitHub

  • Scripting - Add mission to script controller and improve
  • Fix mission load example in different ways
  • Merged!

UTC2324 - hwdef: no dma for unused kakuteh7 uart2 by chobitsfan · Pull Request #22624 · ArduPilot/ardupilot · GitHub

  • No dma for unused uart2
  • Merged!

UTC2324 - AP_OpenDroneID: set EMERGENCY status on crash or chute deploy by tridge · Pull Request #22626 · ArduPilot/ardupilot · GitHub

  • OpenDroneID
  • Set emergency status in uncontrolled descent
    • We don’t have a flag for this
    • We use crashed or parachute-released as a proxy
    • After discussion with Roale
  • Added option For non-GPS position
    • Otherwise No GPS we declare an emergency
  • Vertical velocity check based on GPS because of standard
  • merged!

UTC2331 - AP_Scripting:add error check to mission_load.lua by Hwurzburg · Pull Request #22627 · ArduPilot/ardupilot · GitHub

  • Add error check to mission_load.lua
  • Needs conflict resolution
  • Merge after ci pass

UTC2332 - AP_BattMoniter: don't allow equal low and critical FS thresholds by IamPete1 · Pull Request #22631 · ArduPilot/ardupilot · GitHub

  • Don’t allow low/critical failsafe values to be same values
  • ATM you don’t do anything unless your critical handler is set, which isn’t great
  • So make it < check rather than <=
  • Dropping below both thresholds might be a problem, depending on battery update rate?
  • Maybe if you have a low action you must have a critical action?
  • Merged!

UTC2334 - Plane: QRTL: use cone for QRTL climb close to home by IamPete1 · Pull Request #22636 · ArduPilot/ardupilot · GitHub

  • Use cone for qrtl climb close to home
  • This new parameter is in terrain-altitude-frame
  • SITL-tested
  • Alex might test
  • New way to get into qpos1
  • Moved home uphill or downhill needs testing
    • ReaFlight with a big hill might be a good test
  • Needs some more testing….

UTC2347 - AP_AHRS: move AP_AHRS_SIM::get_results method into AP_AHRS_SIM.cpp by peterbarker · Pull Request #22639 · ArduPilot/ardupilot · GitHub

  • Move code from one place to another in AP_AHRS
  • Merged!

UTC2350 - hwdef: skyviper: remove EK3 external nav and body-frame-odometry by peterbarker · Pull Request #22640 · ArduPilot/ardupilot · GitHub

  • Test we can compile when some things are removed
  • Merged!

UTC2350 - Add a boolean AP::ahrs().wind_estimate by peterbarker · Pull Request #22641 · ArduPilot/ardupilot · GitHub

  • Return boolean from wind estimate method from backends
  • Randy’s not sure we’d want to fall back to DCM wind estimate
  • Copter doesn’t use wind estimate except within the EKF…
  • DCM will not provide a wind estimate on Copter….
  • Privatise the method then mergeoncipass
  • Peter will create a fallback
  • Randy is concerned about taking data from this place or that place or the other place
    • Plane uses DCM airspeed estimate
      • Required
      • Plane needs an airspeed estimate!
      • Wouldn’t be able to fly a Plane without
    • Not providing a warning to the pilot that you’re using a backup solution
    • Generally we only use one backend
      • Entire switch
      • Tridge would like to be able to take attitude from EKF, rest from DCM…
      • Compass-less quadplanes in particular might benefit
        • The Xiake
        • Takes off on DCM and switches to EKF during transition
        • EKF is now used earlier with a patch fro mtridge
        • GCS convergence at low velocity is extremely unreliable and caused a crash
        • Should we add a parameterisable threshold for GSF?
    • We already have several back-doors, even in Copter
      • E.g. compass_use=0 you can take off in stab and use GSF for yaw
    • [11:18 AM]MichelleR: My concern with the Jumper is that the EKF accepted the GSF before it had fully converged, and that’s what got it a bad yaw. I think it happened to me as well some time ago when my compass wasn’t working, so I’m concerned that this may happen more frequently than we think.

UTC0018 - AC_PID: don't change pid info I in reset calls, report actual value used. by IamPete1 · Pull Request #22644 · ArduPilot/ardupilot · GitHub

  • Don’t change PID I in reset
  • By changing the I in the reset call there’s a race in logging
  • I is updated in the update call
  • We get the i that was actually used
  • Merged!

UTC0019 - SITL: add option to read Saleae Logic Async serial data into UART by peterbarker · Pull Request #22648 · ArduPilot/ardupilot · GitHub

  • Add ability to replay data from a Saleae logic capture file
  • Merged!

UTC022 - Create whitelist for undocumented SIM_ parameters by peterbarker · Pull Request #22653 · ArduPilot/ardupilot · GitHub

  • Add whitelist so new SIM_ parameters must be documented
  • Merged

UTC023 - Copter: allow for flight mode UI when FRAME_CLASS=0 by tridge · Pull Request #22658 · ArduPilot/ardupilot · GitHub

  • Allow for flight mode UI when frame class is zero
  • Some vendor was saying MissionPlanner was broken when it wasn’t, and was encouraging ancient version of MissionPlanner
    • Screenshot shows problem…
    • Chinese vendor selling Pixhawk1 clones of dubious configuration (e.g. MS5607 vs MS5611)
    • No email address
    • Facebook links don’t work
    • No ability to contact them….
    • They’re shipping 3.6.2
    • Down in the dregs here
      • But popular dregs
    • Not the only piece of misinformation
  • But not showing interface is not-so-nice
  • We do emit “must set frame class”
    • After this the flight mode interface will work
    • We say we are a quadrotor
  • Can’t set frame_class to 1
    • We do on quadplanes, that was probably a mistake
    • Might accidentally start spinning motors
  • This just fiddles the MAV_TYPE
    • No logic change
  • PH’s fix also fixes other things
    • E.g. dodecarotor stuff
  • Correct but not complete
  • Maybe a fixed wing generic?
  • Four PRs
    • This one
    • PH’s one
    • Mavlink PR to add new enum values
    • ArduPilot PR to change this stuff to return multirotor generic etc
  • merged!

UTC0023 - AP_MotorsHeli: add thrust linearization for ddfp tail rotors by bnsgeyer · Pull Request #22661 · ArduPilot/ardupilot · GitHub

  • Subgroups for parameters are a good idea
    • If you’re on the third level you’re out of luck anyway
  • Perhaps we should have a function for the thrust linearisation…
    • PH and Leonard will help out with that
  • Testing vs active throttle
    • Filter so we don’t end up with a feedback loop
      • Might be an argument for using real-time…

UTC0049 - Plane: holdoff auto takeoff till rudder released on rudder arming by tridge · Pull Request #22665 · ArduPilot/ardupilot · GitHub

  • Rudder-arming for Plane fix
  • Slewed around on arming, which is bad
  • Holds rotors to ground idle until rudder stick goes to neutral
  • Could potentially make it a timer?
    • So it could disarm again after e.g. 8 seconds
    • And perhaps allow the user to disarm over in the disarm position?
    • Would allow us to ensure control inputs are zero (even if it is a post-arm)
    • Maybe just a message?
      • Safer change?
  • Some subtleties
    • E.g. reset mission button does interesting things

UTC0059 - AP_Compass: fixed zero compass diagonals by tridge · Pull Request #22666 · ArduPilot/ardupilot · GitHub

  • Fix compass zero-diagonals bug
  • Old heavy-vehicle-cal could leave cals at zero
    • saved/restored parameter file…
  • Merged!

UTC0105 - Plane update

  • 4.3.3 beta
    • Couple of issues fixed
    • Spiral landing, the magcal thing, ftp terminate session error
      • Faster parameter download
        • MissionPlanner fix required to make it even faster
          • .05% link loss can take 5 or 10 minutes
            • MAVProxy in same takes 4 seconds
          • Issue opened
          • Qgc is fine
    • Spiral landing fix stops vehicle loitering forever
      • Out in next couple of days
    • Only one report from other people for testing
      • Thanks Tim!
      • Might be others testing, but only one reporter
      • Copter is doing better in terms of testing
        • Andy did 10 batteries on 4 quads!
        • Quadzilla tested
        • 5 or 6 testers gave it a go
          • Not 50, but still a good number
        • Tridge thinks we should give credit to all testers
          • Including Copter testers on Plane as there’s so much common code
        • Alberto and ElPara too…
      • Coders and documenters also should get credit
      • Announcements out at same time is good
        • One Facebook post is good

UTC0108 - Copter update

  • 4.3.3 final soon-ish
  • Randy is confident
    • Maybe next 2 or 3 days
  • Tridge has marked a few things for 4.3 which might just go into 4.4
  • Radar alt-hold changes are good
    • Filtered output…
    • Improved things, but position of filter is not ideal
  • Several things requiring investigation
    • Dot uses non-GPS
    • Questions on following path correctly…
      • Position moving when asking for zero velocities?
    • Takeoff alt / takeoff flip
      • Bad alt estimates can be bad
    • Maybe when we arm we should do things, rather than when switching to auto?
    • Don’t run mission commands unless armed
    • There’s a mission floating around that does allow you to arm the vehicle…
  • Duplicate log format messages?
    • External tools consuming log files
    • Peter to look at

UTC0124 - RemoteID

  • New release soon
  • Several PRs merged
    • Including the one we merged earlier today
  • Vendor adding external module
    • MSP protocol so ArduPilot knows where it is
    • Might be good for small quads

UTC0127 - Conference

  • Contact hotel and book your room now
  • Reservation expires, so you need to get your reservation in
  • There are airbnbs floating around which might an option for you
    • Tridge has been thinking about booking one on the basis someone might want it!
  • Planning on combined
    • But other organisers have said this is hard
      • Scheduling
      • Technology
      • If it doesn’t work well this year probably not next year
  • If you want to give a talk at the conference, contact tridge now
  • Pure-remote made more speaking lots available
    • Travel time locally sucks time out
    • Maybe ⅔ of the slot
    • Won’t be really late nights
  • Pre-conference is possible the best-value part of coming in person
    • Let people know if you’re coming early

UTC0132 - off-the-floor

  • OpenSolo 4.3.2 release is out
    • Two issues meaning patched version required
    • Installable via Solex!

UTC0127 - close

Regarding testing.

  1. Thanks!
  2. Suggest we have a channel for BetaTesting on Discord under VEHICLES-DEV. It might help to create a community and encourage engagement. To start perhaps just one channel, unless/until the idea gains traction (or not).

Regarding testing:

You should take into account that in the northern hemisphere we have winter right now. I have always given feedback in the corresponding thread when I have flown new beta versions. Even now several planes are equipped with the new beta, but the weather here has been so lousy for weeks that I haven’t gotten to fly yet. I guess it’s not just me in the northern hemisphere living with miserable flight weather conditions.