Dev Call July 27, 2020

Issues & Pull Requests




Possible alternate time zone dev call proposal 0700 UTC

1 Like

Attendee count (max): 20

UTC1100 -

  • Static casts for access to rc binding and returns
  • No MdB again
  • Needs a rebase
  • Can be merged after it is rebased

UTC1104 -

  • Configurable rates on gryo backends
  • Parameter for selecting gyro rates
  • Use an absolute value rather than using an enumeration?
    • Round down?
      • Banner will tell the user the actual rates
  • MPU6000 will be able to do 8kHz gyros
    • Smoother flights on all of the older boards?!
    • Apparently betaflight users think the MPU6000 is good at gyros…
      • and don’t care about accels so much
  • Some crappy stray debug
  • Can be merged after debug is removed
  • Needs wiki label added

UTC1107 - server mute

  • Persistent setting on Discord
  • Careful about using it as users will stay muted until it is removed

UTC1108 -

  • Where does the applets directory go?
  • Separate repo or subdirectory
  • Autotest would be easier with subdirectory
  • Smartaudio would be first entry
    • Makes use of both params and rc channels
  • Can be merged after CI passes
  • Threshold for getting applets in?
    • Working, tested, documented sufficient?
    • There’s a
  • Need to work out how people might include them….
    • Downloading from github can be problematic
    • Firmware server?
  • Versioning is going to be interesting
    • Expose version to script and have it check?

UTC1114 -

  • Do we really need the two constrains (MIN and constrain)?
    • Do we need “MIN”?
  • What are the units of slew rate?
    • percentage/second
  • Integer maths?
  • Original purpose of the MAX was to ensure throttle was a little bit
    • Throttle wasn’t quite getting to zero
  • What’s the range of the return value?!
    • -1000 to 1000?
    • -1 t o1?
    • -100 to 100?
  • Low slewrate you end up rounding down to nearest integer -> 0.5 it never changes
  • This is so throttle can move by at least 1% per loop
    • If you run at 400Hz you will get a different answer
  • We should be able to remove the MIN(…)
  • We’re doing floats in here should be OK, not needing the “have to be able to change in the face of integer changes)

UTC1129 -

  • Addressed 2 weeks ago
  • Mark made required changes
  • Not a quick one to review
  • Tridge will need to chat with Mark on this one to discuss testing

UTC1134 -

  • Included in stable release already
  • Henry has tested it
  • Peter will read through it today and merge it if he doesn’t see anything wrong with it

UTC1136 -

  • Can be merged after CI passes

UTC1139 -

  • New Plane beta
  • Uavacan compass swapping went in
  • Sensor fixes
  • Fixed vtol missions in quadplane
    • Only specially shaped missions would suffer
    • Auto mission in pure vtol mode, for example
  • GSF coming in 4.1
  • GSF in planes with no compass works surprisingly well
    • Initial yaw from groundstation…
      • How is that conveyed?
      • MAV_CMD?
      • Maybe one of the visual position estimate messages?
      • Takeoff platform can tell vehicle what vehicle orientation is
      • Providing a full pose can help with setting accel offsets
    • A little movement tells vehicle which direction the vehicle is going in
      • So ships do well

UTC1147 -

  • Another scripting example
  • Merged!

UTC1148 -

  • New Beta will come out today
  • Tridge couldn’t fault the new compass stuff from Sid
  • New compass stuff is complicated
    • Would be nice if we could find some simplifications
  • UI is elaborate in MP
    • What about QGC?
    • What about MAVProxy?
    • Big numbers, inconvenient to set
    • Tridge has managed it using MAVProxy
    • Not intuitive without UI
    • HereLink patches for it?
      • Are patches being fed back up?
    • Randy will ping Don to see if there’s a possibility for a GUI

UTC1151 -

  • Nothing to add here

UTC1152 - GSoC

  • Rishabh is doing really well with avoidance
    • Copter backing up video
  • Walking robots using LUA scripts, very cool
  • Oscillation detection from MATLab
    • Cuts gains down automatically
    • Works well in SITL…
    • Pseudo-manual-tuner-quicktune
      • Push gains up until oscillates, then back them down
        • Then move to proper autotune
    • Could be good for quadplanes
      • Normal autotune is too slow
        • No battery for it, even for single axis
    • Based on PID Info object so general applicable
    • If anybody has ideas on tuning FF, please poke PH
      • Get into a constant rate, look at steady state then you can calculate the ff state from that and what you’re feeding into maintain that steady state
        • Like the way autotrim works?
        • Tradhelis are rate-command, multis are acceleration-based
        • Feed-forward d-term
        • Constant acceleration => steady state
          • Driver Rover in circle
    • Need to make FF an axis
      • Don’t tie to vehicle type
      • Constant offset in multicopter gives constant acceleration
      • Tailsitters some axes need FF, some don’t
      • “Does this axis need FF?”
        • If FF starts at zero then don’t use FF
        • PH is currently using a bitmask
        • Mixer really ought to know
        • TVBS quadplane roll axis is like quadrotor
          • So wants FF
        • Pitch axis is more like a heli
          • It’s a vane
          • Same with yaw
        • Method on the mixer to ask if an axis is FF?
        • Or set things up at init
          • Or like tricopter with the servo
    • Trigger this with a switch?
      • Per axis?
        • Cycle through them?
      • Three-pos
        • Middle starts
        • Top saves
      • Quadplane transmitter tune-like
    • Have this active all the time?
      • Oscillations can start in flight
      • Octaquad with motor failure might lead to oscillation?
      • Send this out as “makes no changes” initially?
      • Could do wonders for quadplanes in transition
      • Currently this works on gross oscillations, might not work for finer stuff
  • Second evaluation due RSN!

UTC0005 -

  • Obviously unused

UTC0009 -

  • Expose and log the synthetic airspeed
    • In CTUN
  • Rename the variable _last_airspeed -> _last_synthetic_airspeed

UTC0018 - European devcall

  • 0700 UTC
  • 5pm Canberra time
  • 30-ish hours

UTC0022 - close

1 Like