Dev Call Feb 21, 2022

Issues & Pull Requests





Attendee count (max): 19

UTC1100 -

  • Max altitude for FBWB
  • Works in same way as minimum altitude
  • Parameter docs need some work
    • cm rather than metres in the name, remove _CM
  • Arming check?
  • Cruise mode and you move home alt with alt
    • Cruise will will dive or climb
  • _ALT_FRAME parameters are good in Copter - might be nice in Plane too

UTC1106 -

  • High latency protocol link support
  • Mark a link as high latency via serial protocol
  • Runtime mavlink message enables this on all links?
    • Probably shouldn’t
  • Need to not automatically send HIGH_LATENCY2 unless requested / high latency link
  • Some of our drivers are getting SERIAL2_PROTOCOL so spewing random mavlink messages at them

UTC1128 -

  • Prevent switching to a dead GPS
  • We will switch to a dead GPS if blending fails
    • For a few seconds
    • Timeout on blending is shorter than the timeout for a dead GPS
  • We don’t currently tell the user which GPS is currently being used
    • What does “used” mean?
      • EKF affinity…
    • We’ll use a Named_value_int to send through which GPS is being used… somehow…
      • From LUA?
  • Tridge will tidy the logic and merge

UTC1138 -

  • Print axis gains on axis completion
  • Bill to test
  • Leonard’s good with it

UTC1140 -

  • Combines is_flying and is_lfying_vtol
    • Quadplane is_flying tells you if you are on the wing
    • Quadplane us_flying_vtol tells you if you are vtol’ing
  • Methods were very similar….
  • Slight change
    • Lower-limit is now we-are-flying
    • High throttle / lower limit
  • Safety things
    • On ground, manual and you arm it
    • While you walk to pilot box your radio loses link
    • Throttle suppression may fail
    • Suppress_throttle function in servos.cpp
    • Is_flying_vtol vs is_flying
  • The two methods seem to be used randomly in different places
  • High barrier to entry as this doesn’t seem to fix any bug
    • Landing detection
      • Is_lfying isn’t matching with is_flying_probablility
  • Lots of careful testing required…

UTC1150 -

  • Alarm pwm function
  • Need Andy, DevCallEU

UTC1154 -

  • If pulse width was above 2200 it would trigger high
  • Now it will trigger low
  • Two use cases
    • High signal which would interrupt the PWM
    • Wire cut or button pushed
    • 2000 → 0
    • Driving from other system
  • Invalid means trigger low
  • Option bit to give behavior
  • Need to document this

UTC0004 -

  • Add new target revo-mini-sd
  • Can we BRD_ALT_CONFIG instead?
    • Probably right now
  • Commit list needs fixing

UTC0010 -

  • Terrain check sends back zeroes for terrain/current alt if it doesn’t know terrain height
  • Just check the spacing field…

UTC0014 -

  • Clock multipliers need more work

UTC0014 -

  • Get_location in scripting is broken
  • Alias → rename
  • Or make alias create an alias
  • Both functions work in master right now
    • Fine for now
  • Longer-term we should fix aliasing to create an actual alias
  • Can we deprecate bindings?
    • Singleton method deprecated
  • Deprecated functions and deprecated aliases would be cool

UTC0021 -

  • Run LUA scripts as part of missions on Copter and Rover
  • You can do anything…
  • Can access attitude and velocity control and position control
    • Anything you can do in guided mode
  • Rate on three axes + throttle
  • Rate and attitude inputs….
  • Path uses Lua function which gives a path through space
    • Lua code which tracks that path
    • Derivate of path to get velocity vector
    • Copter can do better; you pass in path and differentiated path and you don’t get integration errors building up
    • Be at this angle at this rate is a thing
    • Rolling loops coming up….
  • Tiktok.lua… rolling loops… helis are going to be fun…
  • Rolling loops with helis are hard without forward thrust
  • Andy’s keen on doing rolling circles
  • Leonard will put up a Callisto to demonstrate this behavior at the next Dev Conference
  • Some debates around things actually terminating.
  • Two NAV_SCRIPTs in a row might be a problem
    • Nav_scripting_done init needed

UTC0028 -

  • Need to send mavlink enable command for a parachute before being able to trigger parachute via mavlink
    • Can’t change this behaviour
      • Might cause dangerous situation
    • Might be able to add another function which does trigger, ‘though

[11:38 AM] Josh Henderson:


[11:39 AM] Peter Hall: Some won’t work. They need the low PWM before they see the high.

[11:40 AM] Peter Hall: * assuming disabled parachute does not ouput PWM.

[11:43 AM] Peter Hall: Basicaly the problem is that the enable switch sets the enable param.

[11:47 AM] Peter Hall: You could run the aux function over mavlink of course…

UTC0052 -

  • Aux switch to start turbine
  • Couple of minor things

UTC0052 - Plane update

UTC0057 -

  • Copter update
  • Andy’s chasing an omnibus vs d-shot thing
  • There’s always another patch…
  • Need to get 4.2 out there…
  • Landing vs position controller vs loiter nav stuff is being problematic
    • Maybe drop it for 4.2?
    • Current code works?
    • Slight wrong altitude ATM
  • Start the beta so other things can be tested
  • Beta before Friday
  • Mustafa has an SCurve pause thing to go in before the beta
  • SCurves probably not going to make it in
    • Getting many testers now, ‘though

UTC0058 -

  • Build efficiency
  • Ccache efficiency plummeted over the last little while
  • Matekf405 and -bdshot were completely different builds
    • 75% reduction in build time after this
  • Durandal vs CubeOrange only gets 25%
    • Few small changes in boardconfig and serialmanager can get that really high
    • Uart count
    • Serialmanager header file different because number of uarts
  • Tags dirty changes?
  • Autotest for 3 board types
  • guard= allows you tp specify the outlying guards to improve hit rates
    • Ifndef AP_CUBEORANGE_HWDEF was bad

UTC0113 -

  • Timer vs decimation problem with stack log frequency
  • Merged

UTC0117 -

  • Tridge is working on tuning a quadplane quickly in vtol mode
  • This RMS stuff has been replaced with SMAX stuff
  • These messages clogged up MissionPlanner’s display
  • CTRL message is really useful in log messages
    • Better than RATE in some cases
    • [12:21 PM] Peter Hall: I mean if we can default PID logging on in copter…
    • [12:26 PM] Peter Hall: I have never looked at it…

UTC0030 - close