Dev Call Aug 9, 2021

*Issues & pull Requests





Logfile analysis Seminar

Attendee count (max): 17

UTC2257 - looking at the SkyDweller

UTC2300 -

  • Add options to pymavlink magfit_wmm to save outputs
  • Batch processing
  • merged

UTC2301 -

  • Add mode 27 into pymavlink
    • RTL_AUTO
  • Merged

UTC2302 -

  • Don’t report remaining battery capacity if remaining capacity is zero
  • Peter will look at this after the call and merge it if he’s happy

UTC2304 -

  • Straight into AP_Vehicle?
  • 5kB
  • Having the vehicle type in the header is problematic but probably worth it here
  • This is documented and tested
  • Will merge it now but move it into vehicle

UTC2311 -

  • Add a bit to force ublox to 115200
  • Some GPSs don’t work at 230400
  • Need to fix the regression for 4.1
    • Not the best solution
  • MergeOnCIPass

UTC2322 -

  • Fence alt-frames
  • Update 7 hours ago
  • Peter and Randy will have another look
  • Randy doesn’t want to hold this up
  • Peter will look at it later today and merge it if it good

UTC2331 -

  • Parameter renaming
  • Makes everything nice and consistent
  • Backport to 4.1?
    • yes
  • Should retune SkyViper with all the new stuff

UTC2334 -

  • Flywoo fix
  • Also for 4.1

UTC2335 -

  • Save 3kB of flash by removing checks on floats
  • Maybe also remove it on ints at some stage
  • Write’nocheck
    • Maybe have ‘norangecheck instead
  • New compiler saved 13kB on -Os!
  • ‘no_range_check perhaps

UTC2339 -

  • Moves code into frontend from backend
  • Merged

UTC2341 -

  • Allow any number of active gsf switches
    • Coming afoul of the maximum number of times we allow the gsf to be initialised and used
  • Compass learn functionality value 3 might be reworked to use gsf
    • Currently uses a related algorithm
    • Takeoff in althold, fly around fora while, land and you’re all done
  • Paul’s already looked
  • Merged!

UTC2345 -

  • Brings Plane up-to-speed with Copter
  • Never get sad noise at the moment as mode entry never fails
  • No beep when moving into initialising or into your initial mode
  • Beep when first polls mode switch
  • Merged!

UTC2347 -

  • Remove pointless override in Plane
  • Merged!

UTC2348 -

  • Tested and doesn’t work
  • Pixhawk4 LEDs don’t suit the way we do LEDs
  • PixRacer LED we assume the LEDs are colocated
  • Not colocated on the PixHawk4
  • Not an RGB LED
  • Three separately coloured LEDs
    • Code probably still around from APM days?!
  • Expand SITL LED setup?
  • Board LEDs
    • Act b/e and power leds
  • Just like the original PixHawk?
  • If they’re simply reversed then we’ll reverse them, but this isn’t a full-on toshibaled thing

UTC2353 -

  • Move singleton method bindings to C++
  • Saves 12.5kB of RAM
    • 1.5kB of flash
  • Need make sure performance hasn’t dropped
  • External flash might be an issue? QSPI
  • Slow deterioration in RAM was concerning
  • We can get more back by moving more stuff

UTC2356 -

  • Does this look familiar enough that it should be factoring?
  • Should try to do this as a single one if we can
  • Two different backends?

UTC0004 -

  • Variant of fmuv5
  • Ships by FoxTech with PX4
    • Allows ArduPilot to run on it
      • [10:05 AM] Peter Hall: bugfix!
  • Sbus timing is a bit out on this board
  • Flown on Sunday
    • Flies very nicely
    • Surface-less tailsitter works well
    • Yaw-d term made the yaw tuning vastly better
  • Probably want a tailsitter fiddle for mix/max
    • Single-axis only
    • Should be multi-axis
  • Roll axes should have a low mix/max
  • More tweaking this Sunday…
  • Wiki and Blog post coming up

UTC0010 -

  • Oscillations at 22Hz weren’t really being caught when logging at 25Hz
  • So add a bitmask to log attitude at full rate
  • Merged

UTC0013 -

  • Servo_gpio_mask
  • Try defaulting SERVOn_FUNCTION to -1 based on BRD_PWM_COUNT old value or define
    • Tridge will try it
  • Getting rid of BRD_PWM_COUNT meant we broke relays
    • Had a motor fire up on a Rover

UTC0017 -

  • Plane update
  • Looking good
  • Clarify gps imu cog
  • Ekf3 had a bug related to IMU position
  • Didn’t calculate lat/long correctly taking into account IMU position
  • Bit of oscillation on one Plane after an autotune
    • Log didn’t show the oscillation
    • Losing a bunch of log messages so it might have been hidden
  • Rcgroups are actively testing Plane
    • Thanks!
    • Henry helps out a lot over there

UTC0019 -

  • Beta7 coming up this week
    • Fixes from Leonard
    • Removing the commanded limit on guided mode
    • Control system changes that added flexibility
  • Couple of things to fix before then
  • Beta6 was a big step forward
  • Tradheli guided stuff still needs to go in
  • Ekf3 sources require rangefinder issue
  • If a user sets up rangefinder2 then it won’t find a rangefinder

UTC0024 -

  • Not much in the way of new issues
  • IMU offsets
    • INS_POS_XYZ not honoured for lat/lng
    • Very Noticeable with rtk gps and lawn mowers!

UTC0026 - GSoC update

  • Custom build server

  • Might have a new server up at soon

  • Not pretty……

  • Python and WSGI

  • Will run as a separate user

  • Everything should be nice’d

  • Could potentially run it as a separate apache process

  • No dependency tracking

  • No feature sizes

  • Will grow a lot more features into the future


    • Bit rough….
    • Still WIP
  • Willian and Buzz will have a squiz

  • DDS

    • Tridge is going to work with Arsh some more

UTC0034 - upcoming log analysis seminar

  • Monday 16th 9am Canberra time
  • Maybe do one with MissionPlanner at some stage
  • 3 logs on 2 hour call
  • Will put them up as videos later
  • Seminars might become a bit of a thing
    • PH will do one for scripting
    • Maybe one for adding a flight mode etc
  • [10:38 AM] Charlie(TunaLobster): ArduPilot video series by the devs? I’ll subscribe!
  • Started as an anonymizer
  • Declination gives you ambiguous results but can tightly give you the location
  • Whitelist has to be used

UTC0051 -

  • Xml parameter format issues?
  • Note all of the value tables are being caught by the regex