Dev Call Apr 29 2019 2300 UTC

Issues & Pull Requests


  • firmware download handling (new manifest code)






Attendee count (max): 17

UTC2304 - University of Maryland organ flight

  • Very cool and another first for ArduPilot
    *Congratulations to Ryan Henderson and the team at University of Maryland

UTC2307 -

  • Important related bugfix went in this morning

  • This one hasn’t been reviewed yet

  • Adds additional parameters for speed-scaling options

  • Bitmask for options

    • Boost for surface throws at low throttle

    • Changes default max gain goes from 5 to 2

    • Speedminspeedmax interpolates between plane and copter gains as a function of airspeed

    • Minimum value you can scale to (.2 / .4)

    • Reduces surface throws as throttle increases and tilt angle increases

    • All three options off by default

      • 3-bit bitmask
  • This seems to do other things as well

    • If statement based armed around outputs call

    • No tailsitter movement when disarmed?!

    • Will be fixed

  • Needs to be tested on real vehicles

  • Mostly tested in SITL

  • Debug code also looks a little odd

  • Style issues

  • Need to have same surface movements armed/disarmed

  • Intending for Peter Hall’s changes to affect quadplane?

    • But these changes only seem to affect tailsitter functions

UTC2309 -

  • Reinstate tailsitter roll limit

  • Bad rebase of the onion caused code loss?

  • Some sort of audit required to ensure we haven’t lost other code

    • Working out which commits may have been lost?

    • Last October?!

    • 2,500 commits since then

      • 192 in ArduPlane directory

        • Tridge will go back through the patches with any help he can muster

        • Servo jitter fixes

UTC2326 -

  • Merged!

  • Clang builds fixed!

UTC2330 -

  • Uavcan flow sensor

  • No Sid on call

  • Lots and lots and lots of patches

  • What’s the point of discussion here?

    • Randy’s keen to get this into master

    • Devcall was added 21 days ago

  • Would be nice if the patches were appropriately squashed

  • Devcalltopic removed

UTC2331 -

  • Need the signed/unsigned char thing fixes

  • Need to address Michael’s concern

UTC2335 -

  • Merged!

  • Was rewritten - delaying it was correct!

UTC2337 -

  • Merging this will let GCS authors get on with generalising their code for upload/download

  • [9:46 AM] (Channel) LV: IMHO We must try to push fence sooner rather than later due to several regulatory stuff coming to law soon

  • Hopefully get this merged this week

UTC2346 -

  • Would like to get this governor code merged

  • It’s quite mature

  • Lots of rsc changes are hinging on this

  • When RPM is unhealthy it defaults to throttle curve

    • Depends on throttle curve
  • New subgroup of parameters

  • Already being flown by a considerable number of users on 3.6 Copters

  • Droop-speed governor

    • Moves reference according to load

    • Electronic version of a Woodward mechanical governor like on constant speed Plane and Copter motors for years

  • Randy would like to review it a bit more


    • Array size assumption needs to be fixed

    • Won’t have effect if vehicle is flying

    • Some oddities commented that should be looked at

  • Initialisation of variables to zero

    • new() / static things don’t need zeroing

    • [9:57 AM] To Weekly devcall: Welll… not adding new ones, rather than removing old ones

    • [9:57 AM] To Weekly devcall: 'though removing old ones would have prevented this discussion :slight_smile:

    • Note that classic c / c++ says to zero things

      • But we’re trying to keep features on smaller CPUs so we want to save the bytes

UTC2348 -

  • Low-pass filter in Plane

  • Would want a reasonably high sink rate for Copter

  • Off by default

UTC0004 -

  • firmware download handling

  • Making changes to the way downloads are handled

  • Scraped out of Michael’s binary repo old stable releases

  • Manifest will include those

  • Michael is changing MP to use the manifest

    • Will give you a list of firmwares that match the board

    • Fmuv5 boards shipping with px4 bootloader

      • No ways to distinguish the boards in the bootloader
  • Need to coordinate with Don so QGC gets done


  • New bootloaders reboot a lot faster on watchdog reset

UTC0013 - lots of PRs being slowed down

UTC0017 - CubeBlack issue

  • Essentially resolved with new Plane and Copter releases

  • HW fault

    • Service Bulletin issued

    • Baro + 2 gyros + 2 accels on isolated board we

    • Vibration and heating over time

  • Remediations

    • If you lose an IMU then we can fly on the internal IMU if it is high enough quality

      • Clipping all the time is not good
  • CubeBlack adds boot checks that all sensors are there

    • No checks on the NuttX build
  • [10:19 AM] To Weekly devcall: Should we add equivalent checks for all boards - that all your internal sensors are actually present?

    • Sensors vary on e.g. fmuv3

    • Imagine non-CubeBlack board

      • Px4v3 doesn’t probe for external baros
  • Maybe add “required” to the hwdef?

    • Manufacturers do upgrade their sensor suite

      • Things won’t “just work”

        • But Philip has just declared he’s willing to put up with that so….
  • “Must find IMU on this chipselect line” might be a good check?

UTC0027 - GSoC update

  • Slots have been allocated

  • A couple of days until student selections are finalised

    • But we’ve done that

UTC0029 - Tracker release

  • Good feedback received

  • MissionPlanner isn’t loading correct firmware

    • Might be hard-coded Tracker stuff inside MP
  • Battery issue

UTC0031 - Plane release out

  • Remediations for Cube issue

  • Marco did a nice video

    • RC switch to lock the vehicle up
  • Septentrino updates went in

  • Retries to flash storage of parameters

  • Reduce chance of parameter reset on F4 boards

  • Microsd fixes on PixHack

  • Fixes power flags on second power input on fmuv5 so you see both supplies

UTC0034 - Copter update

  • Need to do releases of Copter and Rover with latest hardware fixes

  • CubeBlack is already out in Copter

    • But follow-on changes need to go in

    • 3.6.9rc1 later today

  • Rover later today too

  • Beta testing for a week or two

UTC0035 - navigation code

  • Randy’s planning on restructuring the navigation code

  • Navigation library at top which gives you roll/pitch

    • Attitude controller below that
  • In Rover you speed controller is at top and then your navigation controller then your steering controller

    • Will change to make it like Copter

UTC0036 - 3.7?

  • Have enough stuff to justify a 3.7 release

  • Randy wants to do more object avoidance stuff before 3.7

    • Stayout zones

    • Backing away from objects

  • Sometime in next couple of months

  • Bill and Chris want to do stuff on spool logic before 3.7 so Randy’s not the only hold-up

UTC0038 - Close