Dev Call Jan 27, 2020 2300 UTC

Congratulations Tridge

Issues & Pull Requests E



Issues & Pull Requests A



Attendee count (max): 22

UTC1101 - Congratulations to tridge for Order of Australia entry

UTC1101 - thread-safe object buffer for param save

UTC1102 - rc-option for qassist

  • Force-always-on for quad-tailsitters where you can’t fly without it
  • Separate the assists
  • Enable for tailsitters
  • Airpseed assist to zero still means disable
  • Force-assist still makes it work
  • Qassist is kind of weird for tailsitters
  • Copter controller and flying 90-over?
    • Qassist is rates not angles
    • For a 2-rotor tailsitter:
      • At 90 degrees the Copter controller’s can’t control pitch?
      • Just yaw and a maybe a tiny bit of roll
      • Only control surfaces can help
      • Copter controller overwrites control surface outputs
        • Thrust-vectored
      • Don’t use this on non-thrust-vectored 2-rotor vehicles.
  • Stability concerns with transitions; 45-degree hack was required on initial support
  • This needs to be tested on non-thrust-vectored case
    • Could just be disabled?
      • This is a save-the-aircraft thing, we don’t want to use it if it will make things worse
        • This will turn off the airspeed scaling stuff at the very least

UTC1116 -

  • Back from last week
  • For when fixed-wing aircraft battery-failsafe once
  • On battery failsafe it calculates distance to do-land-start vs continuing
  • Edge-case central
  • Can’t handle things like loiter-unlimited
  • MdB wants do-land-end
    • Matt would prefer not
    • Assumes linear sequence between do-land-start and do-land-end
    • Do-land-end tells you if you are in the landing sequence
    • [10:25 AM] (Channel) PH: we reasoned a land was a DO_LAND_END
    • Problem is a mission may be structured so you don’t cross a DO_LAND_START but do hit a DO_LAND_END
    • OBC challenge mission used 30 or 40 DO_LAND_STARTs
    • MdB says every one of his landing sequences uses loiters to achieve correct altitude
    • Proposal, “if you cross a do-land-start and you haven’t jumped then assume you are in a landing sequence”
      • Really complicated landing pattern might have a pair of DLS?
  • Want to use scripting for this eventually
    • Stepping-stone for further improvements
  • Some confusion as to flow-of-control
    • MdB is concerned with the nested loop issue
    • Want timings on F427 on Copter on big mission
    • flag will be on default
      • Option bit for distance-based
  • Need to worry about when we ask for an RTL
    • Currently we would restart the sequence
    • Might be a sensible behaviour change
    • Lifting code out of battery failsafe
    • Might be as many people liking current behaviour
      • RTL_AUTOLAND 3?

UTC1154 - Copter-4.0.1 released!

  • Copter updates
  • Release going OK
  • Devid for baros kind of required for externals
  • One-shot-or-d-shot used is hard to tell
    • People think you can look at PWM outputs to tell. You can’t.
    • [10:57 AM] (Channel) AT: RCout: PWM:1-4,DS150:5-8
      • PWM on outputs 1 to 4, dshot on 5 to 8
  • Baro glitches have come up on Cube Orange a bit
  • S.BUS input lag?
    • 300-500ms
    • CubeBlack
  • Emlid CAN compass not found without brd boot delay
  • Mavlink gimbal stuff is reported as a regression
    • Peter to look at
  • ESC sync issues
    • IOMCU issues….
    • Upgrade failed?
  • Several MissionPlanner issues
    • Getting much better now-adays
    • Proxity viewer blowing up
  • Randy would like these little things fixed and for people to be careful about introducing them
  • [11:28 AM] (Channel) JP: I think part of it is that we don’t sync MP with firmware release cycles. We really should.
  • Funding-and-stuff
    • Lost something when FF moved on
      • Need to fund the area again

UTC0029 -

  • merged….

UTC0029 -

  • Drive-by
  • Too-close-to-read would be frozen at min distance
  • Looks correct to tridge and Peter
  • merged!

UTC0032 -

  • Initialisation moved up
  • Merged!

UTC0035 -

  • Soaring upgrades
  • Vario code going away?
    • Confusing GCS mavlink changes removed
      • Reporting vertical speed from vario was annoying
  • Logic changes in loiter?
    • Possibly reverting some of the changes MdB made?
    • MdB to review
  • Autotest for soaring!
  • Need to verify things that aren’t soaring code
  • More TECS parameters?
    • Want to change pitch rather than throttle with gliders
    • Feed-forward path to keep integrator unloaded
    • Pre-flare phase particularly a problem
    • 0 is disabled
  • Improves baseline logic for thermalling
  • UX feedback improvements
  • Three-pos switch for disabled/thermallinginloiter/thermalling
  • Auto-mode improvements
  • Max-dist-from-home stuff
  • Need to fix is_heading_lined_up
    • And commands_logic stuff cleaned up generally

UTC0041 -

  • Explanation of use case in test suite
  • [11:57 AM] (Channel) MdB: Is it really worth doing via hal instead of just #ifdef in GCS_MAVLink? Seems sorta like a HAL dumping ground
  • [11:58 AM] To Weekly devcall: No, because you could notionally trap on stm32.
  • [11:58 AM] To Weekly devcall: (my opinion)
  • Need to make sure this compiles under Cygwin
  • #ifdef of SITL in SITL is odd
  • Can be merged after cygwin is correct
  • And make sure mavlink reference is correct

UTC0102 -

  • Serialn-protocol is not set
  • This code as-is correct, just needs more comments
  • Bidirectional control of mavlink via iomcu coming
    • It’s coming….
  • RC controlled menus via HOTT?
    • Arbitrary ascii screens

UTC0110 - Plane update

1 Like