Servers by jDrones

Dev Call Sept 6, 2021

Issues & Pull Requests




Attendee count (max): 19

UTC2302 -

  • Simplify get_wp_destination_loc
  • Back again after some testing
  • Merged

UTC2302 -

  • Relaxing quadplane fixes
  • Relax in transition to vtol
  • Check_attitude_relax doesn’t do what it says
  • Previously it says latst_attitude_control_now
  • It was last time checked, not last time run
  • Now it is last-time-run
  • Maybe we weren’t relaxing on purpose?
  • Hardcoded relax in tailsitter now gone
  • Maybe this didn’t work on tailsitter before?!
  • Under what conditions do we check but not run the attitude control?
    • There may be cases like takeoff where we’re not running the attitude control?
  • Tridge is going to have to play around with this a bit before knowing this is OK
    • Suspects it is correct
  • Henry’s been playing the tailsitter commit a bit
    • Good on hwing to avoid i-term build-up on transition
    • Split that commit out?
  • Case that matters?
    • Two that matter
      • Tailsitter
      • Tiltrotor
        • Bicopter case
    • Same use case
    • Run in main quadplane loop
      • And again when doing assistance
  • Ask attitude control library when it was last run?
  • merged!

UTC2309 -

  • Arm-with-airmode and arm-without-airmode
  • This is currently written to change behaviour
    • I.e. new switch gives you no-airmode
  • Peter and tridge are uncomfortable with changing the behaviour
  • spin-arm
  • Might interact badly with Andy’s change to stop airmode spooling down
    • At the moment we still try to detect landing
  • These vehicles could leap into air depending on your throttle mix
  • Racing quads go into airmode on the ground on betaflight and it’s not a problem – Henry
  • Disarm-delay doesn’t work when armed on switch
  • A chunk of agreeing to disagree on whether we should change existing behaviour
  • Does ArduPilot have a reputation for joggling around on the ground when using arming on a switch?
  • We’ll be going with “preserve existing behaviour”
  • Not a problem for quadplanes at this point

UTC2327 -

  • arming check for log download?
  • Boolean flag to rate-limit this
  • Good past that point

UTC2338 -

  • Fixes bit used for rc calibration
  • Stops vehicle arming when doing rc cal on the GCS
  • QGC has as “trim” button which might trigger this
  • MAVProxy doesn’t use it
  • We’ll just remove the trim radio function entirely
  • We’ll also set a flag on the rc library indicating no input
    • Need to not trigger the RC failsafe?
  • Make sure we can’t do it when armed!

UTC2356 -

  • Sprung-throttle support for Plane
  • Default here is current behaviour
  • Nice on tailsitters
  • Trim-throttle in right spot for cruise might make this quite nice for conventional fixed-wings too

UTC0003 -

  • Force disarmed logging if disarmed for nasty reason
  • Bench-testing could be problematic here?
  • Disarmed with EKF failsafe?
  • 15 seconds isn’t enough to log a descent under parachute
    • How fast am I falling, what’s my impact speed?
  • Do we get rid of the Copter log-forever-on-crash code?
  • Parameter for how long to disarm?
    • Massive logs on every disarm?
  • Massive logs are no good to anyone
    • Need to be able to log them in log analysis tools
  • Off-line debate required for this one

UTC0014 -

  • Only continue after land if next waypoint is a takeoff
  • There’s a bit you can set to continue running the state machine
  • This patch requires next command to be a takeoff
  • Get_next_nav_cmd
  • Nav_delays are skipped
  • Do_jump is handled by get_next_nav_cmd
  • At the moment it is harmless but useless as Copter won’t leave the ground without the takeoff
  • We added something to require a takeoff command at beginning of a mission
  • Users forgetting to put takeoff in second part of mission when they’ve got continue
  • Vehicles should only take off when going through a nav takeoff to ease lower-level controller work - Leonard
  • Will this stop touch-and-go for fixed wing?
    • Might stop existing workflows
      • Landing followed by a waypoint
  • Overly-prescriptive and second-guessing the user?
    • Tridge thinks we should be able to takeoff without a takeoff command if you’ve previously been flying
      • If you’ve got airspeed etc etc
    • Only used on vtol land for quadplane, so this might not be a problem?
  • Option-bit or Copter-only

UTC0033 -

  • Log entry and exit of pseudo-mode
  • One more bodgy workaround and we’ll go back and make this an actual mode

UTC00?? - hide oSD params correctly

  • Merged

UTC0035 -

  • Reverted a bad change
  • Bunch of boards broken by this one so this one reverts and we can look again later
  • Mergdf!

UTC0038 -

  • Hi-power periph on at boot so you can toggle power on sensors on e.g. Durandal
  • Merged!

UTC0041 -

  • Only issue is a watchdog reset close to the ground
    • hardfault
    • LR in scheduler but not in a function
    • In the run function?!
    • Not enough information in watchdog to assign a cause
    • Tridge has been looking at what other information we can get into our logging to help us debug these
    • There’s a few more registers we could log which might give us more information to the cause
    • 2 logs total where the LR has pointed at the scheduler run function
    • No reason to think this couldn’t happen on Rover or Copter
    • 2021-08-22 06:36:44.47: WDOG {TimeUS : 117693372, Tsk : -1, IE : 0, IEC : 0, IEL : 0, MvMsg : 0, MvCmd : 0, SmLn : 0, FL : 128, FT : 3, FA : 3758157112, FP : 180, ICSR : 4507651, LR : 134607347, TN : Ardu}

UTC0044 -

  • Copter update

  • 4.1 release is moving along

  • Several blockers removed

  • Laggy surface tracking fixed

  • Ulanding driver reported broken

    • But already fixed in master because of khancyr’s static analysis fixes
  • Dijkstras in guided mode still not working

    • Dijkstras is a a wpnav thing
    • Guided doesn’t use wpnav any more
    • If a user is pushing in slightly different targets every time Dijkstras is going to be a problem
      • We’d already have been doing that
    • We’d previously fly very slowly because of the leash
    • “Go here” vs 10Hz update
      • Two different use cases
    • Many different ways of getting targets in
      • Perhaps change behaviour based on how we get hte target?
      • Maybe just a parameter?
  • Oscillation issue still a problem;

    • Users says autopilot near centre of rotation but is it?
    • Feedback?
  • The airmode question

    • 4.1 or 4.0 issue?
  • At least one more beta

    • Within next week

UTC0058 -

  • Tradheli update
  • Autotune alpha!
    • Waiting for users to get back with results

UTC0059 - Rover update

UTC0103 -

  • Honour wpnav in-speed
  • Merged
  • Peter will do a foll0ow-up to do the vertical

UTC0105 -

  • Merged
  • Log source for mavlink commands

UTC0106 -

  • Install arm gcc compilers under cygwin
  • merged!

UTC0106 -

  • Ms5525 conversion fix
  • Libcanard isn’t 64-bit safe!
    • So we provide for building sitl as 32-bit for AP_Periph
    • Want to make it easier for sim_vehicle to run 32-bit

UTC0113 - wiki vimeo plugins broken

  • Charlie is still looking at it

UTC0116 - close

Servers by jDrones