Dev Call Nov 6, 2023

Issues & Pull Requests

Plane

Copter

Attendee count (max): 16

UTC2300 - AP_InertialSensor: init all notch center frequencies by IamPete1 · Pull Request #25355 · ArduPilot/ardupilot · GitHub

  • Init all center notch frequencies
  • We know what’s going into 4.4.x
    • Andy and Leonard have agreed to leave the 2Hz solution (“bandaid”) solution
  • Now. short-term master vs long-term master
  • Tridge’s rework PR Tridge thinks is best in master
  • Do we need an interim solution like this one before Tridge goes in?
  • Rebase conflict needs resolution
  • Then needs testing for odd use cases
  • Tridge and Andy discussed last night for an hour
  • Octa-quad with 4-channels of bdshort?
    • I.e. only half your motors have telem!
  • There’s still a lot to understand about the ramifications of changing things
  • Logging of actual notch frequency PR from Tridge would be good
    • Current FTN logos don’t match reality
    • Notch.txt file for actual frequencies for debug on;y
    • 50-column file for all frequencies
  • Leonard to get the math right then look at Tridge’s again
  • Tridge will rebase PH’s and force-push it and then we can test that for short term
  • Other one to fix in short-term is to fix second harmonic on quadplanes on two of the motors where the quadplane has 5 ESCs with feedback
    • Already fixed by tridge’s rework
    • But a shorter fix in short-term
  • Longer term adding EMSCK parameter which would be zero meaning automatic non-zero meaning mask of frequency sources you want
    • One harmonic notch for vtol motors, separate for forward ESC
  • add INS_HNTC3….
    • So support for 3 sets of motors
    • Plot review tool is going to have to get wider!
  • Attenuation ramping stuff from Leonard will also provide challenges
  • Will see some pushes onto this PR

UTC2311 - Plane: ensure pilot throttle is valid, more set_servos_controlled rework by IamPete1 · Pull Request #25416 · ArduPilot/ardupilot · GitHub

  • Ensure pilot throttle is valid, more set_servos_controlled
  • Always check has_valid_input
  • Is_valid waits for failsafe
  • Has_valid_input vs is_valid
  • Manual vs non-manual…
    • If you are in fbwa and you’re in the intermediate stage with no RC input but not failsafe
      • We level the aircraft by squashing the inputs
        • Zeroes the throttle in non-q-mode
      • Copter keeps inputs
    • Manual vs FBWA is different in terms of failsafes?
  • Tridge needs to stare at this a chunk
  • Throttle pass-through in manual vs non-manual applies throttle min in fbwa but not in manual
    • See also direct-passthrough option
    • For starting ICE and jets
    • And idling…
      • Where some EFIs take low PWM to cut motor
      • Are we opening up more possibilities of brief RC outages causing engine to stop?
  • PH doesn’t like the set-control-in stuff

UTC2322 - AP_ADSB: avoid using GPS and RTC libraries in backend by peterbarker · Pull Request #25427 · ArduPilot/ardupilot · GitHub

  • Avoid using GPS and RTC libraries in backend
  • Merged!

UTC2330 - hwdef: add YJUAV_A6SE_H743 board support by yunjiuav · Pull Request #25431 · ArduPilot/ardupilot · GitHub

  • Add YJUAV A6SE_H743 board
  • Merged!

UTC2331 - GCS_MAVLink: handle DO_ADSB_OUT_IDENT as both long and int by peterbarker · Pull Request #25436 · ArduPilot/ardupilot · GitHub

  • ADSB_OUT_IDENT as both long and int
  • Merged!

UTC2332 - Change tense of GPS detection message by Ryanf55 · Pull Request #25445 · ArduPilot/ardupilot · GitHub

  • GSoF driver tense change
    • Ryan found the message misleading
  • Maybe the Trimble isn’t setting the auto_detected_baud
  • Change actually needs to be in _detect_isntance around GSOF

UTC2338 - AP_GPS: Expose COM port and Output Rate in header, Fix default COM port by Ryanf55 · Pull Request #25456 · ArduPilot/ardupilot · GitHub

  • Remove magic numbers
  • Checks message enablement
  • Document hardware debugging stuff
  • Merged!

UTC2341 - AP_GPS: Add configurable baud rate/port for AP_GPS_GSOF by Ryanf55 · Pull Request #25449 · ArduPilot/ardupilot · GitHub

  • Add configurable baud rate / port for GSOF
  • Want a GPS_DRIVE bit to say use the serialn_baud parameter
    • Basically a skip-auto-bauding
    • Would just instantiate the document directly
  • But also want one to say, “move to the serialn_baudrate”
  • We have a uBlox packet creation function which will be useful
  • Do we modify GPS_AUTOCONFIG?
  • Ryan will attack first bit and we’ll come back for the second

UTC2355 - GCS_MAVLink: handle MAV_CMD_DO_SPRAYER as both long and int by peterbarker · Pull Request #25454 · ArduPilot/ardupilot · GitHub

  • Sprayer commands as both long and int
  • Merged!

UTC2355 - Add expanded explanation in Fwd throttle code comments and metadata by Hwurzburg · Pull Request #25463 · ArduPilot/ardupilot · GitHub

  • Tidy forward-throttle code comments and metadata
  • Merged!

UTC2357 - Copter/Rover-4.4.0 issues list · Issue #23192 · ArduPilot/ardupilot · GitHub

  • 4.4.3beta1 is out and active for Plane
  • Copter and Rover later today
  • Couple of new issues mav-cmd-yaw using relative angles being broken
  • Searching for dummy
    • Suddenly changing yaw as going around the course
    • Edge-case?
    • Doesn’t record yaw in in picture
    • Have to point vehicle North all the time
    • If you yaw around a course it subtly changes heading?!
      • Accumulation of errors, probably
  • When should we do 4.5?
    • January, probably
    • Not really ready yet
    • 4.4 isn’t bedded in properly?
    • 1st beta in December instead?
      • Camera stuff stuck in master ATM, not available in 4.4
      • Tridge wants to fix parameters up
        • E.g. Airspeed on Plane
      • Precision landing in quadplanes?
        • A few partners have developed their own solutions for that
      • Nothing in the WIki for precision land?
        • IRLock is there
        • … doesn’t include Rishabh’s stuff
      • New page in advanced features page, probably
  • lua scripts sometimes do not start or re-start · Issue #25455 · ArduPilot/ardupilot · GitHub
    • LUA scripts not starting
    • Tridge got various debugs in place
    • Linked list just seems to be empty
    • setjmp/longjmp optimisation issue, maybe?
    • Should be able to get remote access again to the board, but probably won’t be able to reproduce
    • Probably can’t release 4.5 with scripts like this
    • Sid has seen it
    • A parameter to say how many scripts should be loaded
    • Script option to say n-scripts
    • SD-card related?
      • Tridge knows it is reading and parsing the scripts and it’s valid
      • So probably not
  • Rover pivot-turns should be fixed (SCurve-related)
    • When Randy gives up he’ll replace 4.4
    • BluRobotics BluBoat shows 4.4 works nicely

UTC0002 - Plane update

  • 4.4.3 Beta1 will go out after this call

UTC0017 - Comparing ArduPilot:master...Ryanf55:gsof-49-50 · ArduPilot/ardupilot · GitHub

  • Off the floor
  • How to handle serial link when handling serial data and handling an ack
  • Single byte returned for configuration
    • Co-mingled the rest of the stream
  • Turn off messages
  • The vendor suggestion was use ethernet
    • Two different UDP ports, one for configuration, one for data
    • Not really an option for us

UTC0022 - votes passed for the new positions

  • Contact Tridge or James if you are on the DevTeam and are interested in the positions
  • 26 responses
    • 25 for, 1 against

UTC0024 - one outstanding votes

  • See developer email list
  • Gimbal support in MissionPlanner/QGC

UTC0025 - discussion of funding

  • Josh brought up point about having a higher-level view of funding
  • James has a list he gathered earlier in the year
  • Randy will send out an email with a link to that list
  • Randy doesn’t want this to be a wish-list of things other people would do

UTC0027 - off the floor

  • ISO9001 compliance?!
    • Certificate regarding procedures of a company
    • ArduPilot foundation is
    • The dev process is separated off from that
      • In principle the foundation could get certified?
      • Craig thinks our development methods wouldn’t be compliant
  • ISO27000?!
  • If we can get these
    • We might make it easier for someone down the line to certify the autopilot for a particular use case
  • All existing code wouldn’t fall under that certification?
  • Would be a significant amount of work just to work out what what we’d need to do
  • Amilcar will do some leg-work
  • Compliance work via automation works
  • “We have a process and we follow it”
  • Related to software bill-of-materials
  • Astm certification things might also be affected
    • Python module to go off and work out what’s in a firmware build?
    • Features.txt equivalent
    • How deep does the hole go?
      • Kernel version you compiled on?

UTC0045 - feedback on funding software

  • “ArduPilot doesn’t support software development’
  • Unforeseen consequences of funding software development
    • Funding core ArduPilot code means less development happening, probably
    • If it became the norm that we paid for that dev we’d not be able to come even close to funding it
    • Then you end up with “won’t happen without funding”
      • Would change the character of AP development substantially
      • Seen this a bit with the documentation
        • It’s become more common to ask Henry to do it
        • Before that each developer tended to take responsibility
        • Would be bad if this happened in the broader code base
    • Social issue
  • MIssionPlanner and QGC pines?
    • Tridge thinks they’re worthwhile
    • Skirts closer to core code being funded
    • Separable from core cpp code
    • Josh
      • Can it be funded externally?
        • If it isn’t will there be an expectation of funding future work here?
      • QGC RID and gimbal work was funded externally already
      • Cash-in-hand is not a budget
      • Setting properties matters
    • Books need to be done
  • More people on administrative side would be nice
  • The $400,000/year mark we need to get external audit
    • Could pass this in next year or two
    • Accounting practices required at this point…

UTC0056 - close