Dev Call Nov 18 2019 2300 UTC

Issues & Pull Requests




1 Like

Attendee count (max): 22


  • Matt has the test heli for the autorotation
    • Has also been arranging stuff with Ryan for a heli in Canberra
  • General oo-ing and aah-ing about the tiny little helicopter

UTC1102 - Welcome to Braedon

  • Arizona company doing aircraft inspections

UTC1103 -

  • Merged!

UTC1104 -

  • Adding dynamic notch filter to Plane
  • Was Copter only
  • One parameter was exposed as a uint8_t, now exposed as a boolean
  • Using one filter on the vertical lift motors, the other on the forward motor
  • Sheep-shearing facility did not smell of sheep
  • Randy’s keen to merge it.
  • merged!

UTC1107 -

  • Two F9 GPS with Rover and Base and you end up with yaw input
    • not need for compass
  • Being able to fall back to compass in case of GPS loss would be good
  • Need to wait until MdB is on the call as it
  • EKF3_MAG_CAL to 5 means yaw comes in externally
  • No option to fall back to normal compass in the EK3 code
  • No UART required between the GPS units
  • Just plug in the GPSs to the flight controller

UTC1110 -

  • Don’t show parameters for serial ports that don’t exist
  • A whole bunch of #ifdefs in serial manager
  • For a board that only has 3 uarts we stop showing 3+ sets of inappropriate parameters
  • CAN/Serial bridges - don’t have them yet, so not relevant
  • Serial8 is coming on a new flight controller
  • Merged!

UTC1113 -

  • AP_Arming prearm check for terrain allocation failure
  • We can allocate memory in many, many places now
    • So catch some places where allocation has failed.
      • This is the first example
      • Mark’s EKF failure PR is another
  • Our past pattern of setting parameters to zero when initialisation fails is now considered an anti-pattern
    • Stop doing that
    • And consider trying to eliminate current instances
      • AP_Avoidance, navek2 and navekf3, quadplane all need fixing
        • Two places in each of navekf2 and navekf3
  • Merged!
  • Need to get this sort of thing into backports

UTC1120 -

  • Circle mode changes
  • Use roll stick to control circle direction and speed
  • Use pitch stick to increase/decrease the radius
  • Rate is limited to 90 degrees per second
  • Any limit required on radius?
  • Bitmask option to enable this?
  • Confusing with reposition?
    • Radio tuning parameter
    • Rc6 can increase/decrease the circle rate
      • Having this set eliminates the stick input
    • Use an RC channel option to switch between adjusting circle rates vs position?
  • Very rough control when working with very large circles?
    • degrees/second can mean a lot of tangential velocity out there
    • Currently a very gentle increase/decrease in speed
  • Increase in radius and not increasing speed?
    • Could lose control
      • Recalculation doesn’t work very well when changing the radius/speed
        • Can cause short-cutting across the circle
      • Leonard’s feed-forward circle changes will conflict with this
  • Not well-used
  • Randy will review
  • Need to make sure it is stable as speed is increased
    • Small radius, large radius
    • Stick inputs scaled by radius?
  • To parameter or not to parameter?
  • Don’t scope-creep this PR to add the circle-centre-moving!
  • Need an upper limit on circle radius
    • Something arbitrary
      • 50km?
  • Options bitmask for enable-speed-change and enable-radius-change-on-stick
    • Three for both features
    • Would need another bit to enable repositioning the circle

UTC1137 -

  • If you have a GPS which does its own logging and you’ve asked it to log then treat that as we do our current onboard logging
  • uBlox stuff still coming
  • Confusing bits?
    • There are already different statustext bits
    • If one goes wrong while you’re flying then there’s no disam
  • Can be pushed in once methods are renamed to match the ones on Logger

UTC1146 -

  • Back to uBlox moving baseline configuration
  • MdB has a pending review
    • A few tweaks and errors to fix
    • So no pushing/merging just yet!

UTC1147 -

  • Remove the is-zeros
  • An optimisation
  • Merged!

UTC1149 -

  • Support wider range of frequencies
  • Clock-multiple issues meant NeoPixels didn’t work
  • These only work if you run them faster than they are spec’d
    • Round-up vs round-down clocks
    • Ws2812b strips
    • Doesn’t change roundup/rounddown for d-shot
  • Calculations done at startup, so the loops aren’t a huge concern
  • merged!

UTC1151 - LED question (off-script!)

  • Some variant takes 32-bits
    • RGB + white?!
  • Can take a parameter….
  • 63 LEDs on a pin from some company
    • So a larger mask is coming…
      • Or maybe an arbitrary-number API

UTC1155 -

  • Merging the simple fix
  • Peter will continue to work on the more complicated fix, still with an eye on a backport to Plane stable

UTC0000 -

  • Tridge had to run off

UTC0002 -

  • Tridge had to run off

UTC0002 -

  • Tridge had to run off
  • PB to rebase

UTC0004 -

  • Peter Hall still needs to look at it
    • Actually, currently flying it in RealFlight
  • Addiction-x flew really well
    • Option2 for gain-scaling
  • Needs more testing!
    • Merging would lead to more testing….
    • Mark has some builds on Google Drive for people to test….
  • Peter Hall will continue to look at it

UTC0006 - Mark and q-thr changes

UTC0010 - Copter update

  • No huge progress
  • Rc3 coming soon
    • 5 resolved issues
  • TMODE gps fix to come back in
    • No RC before this comes back in
  • Somebody checked and d-shot is working
  • Still time to get minor features n
    • E.g. Rob’s servo-testing stuff

UTC0011 -

  • Rover update
  • Rc3 is still the current one
  • GPS issue means an rc4 is coming

UTC0015 -

  • Calibration bug
  • Mag cal on compasses aren’t working sometimes
  • Off-diags and diags aren’t set
  • It’s in 3.6 and 4.0 Copter
    • Unsure when this first came in
      • Unrelated to the auto-orientation code?
  • People need to know, announcement to be made
  • Solution is to do the calibration again and then check the results

UTC0021 - two votes for core developer team to address

UTC0040 - end

1 Like