Dev Call Feb 8, 2021

Issues & Pull Requests




1 Like

Attendee count (max): 20

UTC1100 - Developer conference dates

  • 9,10,11 April for the ArduPilot Developer Conference (Australian dates)
  • Virtual (duh)
    • SPI will not be reimbursing internet expenses for zoom-only conference….
  • Any objections, please let tridge know
  • 2022 dates also locked in (hotel dates deferred again)
  • Speakers will speak at a time sensible to their timezone
  • Call for speakers!

UTC1105 -

  • Merged
  • Constrain roll on takeoff

UTC1109 -

  • Get logs (artifacts) of failed tests from CI now
  • Can now see details of flapping tests
  • merged

UTC1110 -

  • Layering violation
  • Special case for CubeOrange’s heater to add countering-offset for electromagnetic field
  • Different integration times on the compass didn’t modify the noise sufficiently to make changes worthwhile
  • If we removed the dithering we could eliminate the noise
    • But that screws the baro
  • Will also be in AP_Periph

UTC1114 -

  • Durandal with moving baseline yaw with GPS
  • 460800 not needed when data is flowing between GPSs
    • Lack of DMA on both UARTS can cause instability at 460800
    • Ergo, stop using 460800 if the data is flowing directly between the uarts
  • Will wait for results

UTC1120 -

  • Scripting example for toggling relay at 50Hz
  • Merged

UTC1123 -

  • Discovering significant things when working at very low temperatures
  • 14km discrepancy at -40C
    • Starts going very bad at -36
  • Datasheet gives corrections for below -15 and -20
    • We did have a comment for below -15
      • Reduces amount of error
  • Still stuffed at -35
    • “Well don’t do that then”
    • Operating in Arctic or Antarctic?
      • Reboot after you’re warmed up!
      • Prearm check your baro hasn’t moved too much?
        • We don’t know if all Baros suffer like this
        • Look for a height change instead?
        • People might deliberately have a change….
  • Large changes in magnetometer based on temperature
    • Calibrate in Europe, ship to Africa and get terrible results?
      • Could be temperature?
    • External GPS/cmpasses are particularly vulnerable
  • 10:26 AM] Jeff Wurzbach: i have a little print out in the checklist screen of MP for this
  • Disarmed Plane in graphs was continuously recalibrating baro
  • Jeff has a calibration chamber going down to -68C
    • Amilcar also has one…
  • AK09916 is relatively linear
    • Single slope and caltemp
    • No need to third order polynomial like we do for IMU

UTC1133 -

  • Chris Olson has been running a governor on a separate branch
  • Auto-spool-up
  • Look at rotor speed and judge how much power to add
  • At a certain RPM it goes into governor
  • Also has cool-down for ICE engines
  • Singleton-returning-pointer == check for nullptr
  • Use #if RPM_ENABLED
  • Need to watch out for parameter changes

UTC1142 -

  • Peter will try to reproduce later today
  • Potential Rtl-climb-alt-min bug

UTC1148 -

  • Funding to work on SITL stuff
  • This issue is to track improvements we’d like to do in SITL
  • Feel free to add in ideas on what you’d like to see
  • Tridge has asked KnifeEdge about fixing RealFlight
    • Lockstep
    • Tcp-connection-per-frame

UTC1150 -

  • Test with byteloss
    • Including on hardware
  • Fix sanity check
  • Ensure debug is off
  • Can be merged after that

UTC0003 -

  • Logic-affecting changes make developers nervous
  • Servo_reversed?
  • Need to test on older compilers before we merge this

UTC0012 -

  • Needs testing
  • Don’t have an autotest for VISION_POSITION_DELTA yet

UTC0014 -

UTC0021 -

  • Randy’s not keen on the backport

UTC0024 -

  • Peter to to quick test including sub after call
  • Can be merged if it is all good

UTC0030 -

  • Fix external yaw when using external nav
  • Arming checks…
    • Vehicle’s AP_Arming - mandatory
  • What does “using external yaw” mean?
    • Configured or actually using it?
    • Means if you’re fusing yaw from gps or yaw from external nav
  • Just a downright bug in the function
    • Because of the order andthe use_compass call it was never executing the check below
    • So e’d always so false even if you’re using external nav
  • Need to check users are using it in correct sense
    • Bool configured_to_use
    • Bool actually_using() ?

UTC0034 -

  • Magic undocumented registers to avoid issue
  • Just follow manufacturer’s advice
  • Doesn’t seem to destroy anybody’s sensor
  • Merged

UTC0037 -

  • Review issues fixed, hopefully
  • Tridge Wanted config errors rather than panics
    • Done
  • Remove unused method
  • Many more comments and some renaming
  • This is flying and tested….
    • Works with existing position controller because its scaled according to throttle
    • Leaves some performance on the table, ‘though
  • Leonard’s been thinking of this for a long time
    • The vehicle is exactly what Leonard was thinking of
  • The s-curve stuff starts moving us towards natively supporting this sort of vehicle
    • The position and attitude controllers need a 6-axis variant which might be a direct feedthrough…
  • Want this sort of vehicle and subs to more naturally supported in pos and att controllers
  • Leonard doesn’t have a strong opinion on what we do now
    • Reservation is that when you do a quick half-arsed approach to bring something in it can be a support burden in the future
    • Done properly can be an asset for the future
  • This flies great as-is
  • Leonard doesn’t have any big problems with this
  • Do we merge now?
    • Will it make it more difficult later?
      • Should just be a bit to turn on or off the scaling
  • Multirotor attitude controllers might be able to evolve
    • Bit might then break this aircraft
  • Removed function is actually critical for upcoming mavlink method
    • Function’s been there for three years and nobody’s used it
    • The function is good, modifying it is complicated
      • Not required for the 6dof vehicle

UTC0106 -

  • Tridge isn’t convinced this is a good thing

UTC0106 - plane update

  • Both betas flown on the weekend with no issue
  • Hopefully they go out RSN

UTC0106 - Copter update

  • Leaving stable release to Amilcar and Bill
  • Main fix is the code to recover from parameter resets, seen on F7
  • S-curves is making good progress
    • Improvements in last week
    • Can change speeds between waypoints
    • Do-commands when vehicle is closest to corner
      • Currently they trigger when leash hit
    • Still has to pass autotest
  • Rishabh has improvements to object avoidance to go in
    • Ground-plane-detection
    • Rangefinder-based enablement
  • No Rover update

UTC0110 - close - out of time