Dev Call Feb 12, 2024

Issues & Pull Requests


Copter & Rover

Attendee count (max): 15

UTC1100 - Copter: improved follow at close distances by tridge · Pull Request #26138 · ArduPilot/ardupilot · GitHub

  • Follow at close distances fix
  • Review by Georacer
  • Need to fix z-limit of velocity
  • Will close tridge’s and use Georacer’s PR
  • GeoRacer will ping us in #code-review to have a look at it

UTC1106 - AP_Scripting: raise max FLTE for yaw to 8 by tridge · Pull Request #24933 · ArduPilot/ardupilot · GitHub

  • Raise max FLTE for yaw to 8
  • Scripted Quiktune limit
    • Unlimited bad
    • TVBS dolphin didn’t go well with this sort of thing
  • Perhaps a function which indicates whether we should use higher gains or not based on many other parameters
    • Tridge and Henry will work on that
  • Filtering on vectored thrust shouldn’t have values other than 0 says Leonard
  • Quick-tune might not be appropriate for yaw on these vehicles
  • Change default filter on yaw axis for tilt-vectored vehicle from 2.5?
    • Might break people who have set a P-gain manually
      • Leonard’s not sure
  • Recommending quiktune over autotune?
    • Warning of ESC de-sync issue on autotune
  • Would be nice to have autotune testing
    • Tridge has seen some bad results
    • Leonard thinks relaxing the prereqs is the problem
    • Needs a base reasonable tune
      • Quik-tune should be good for this
    • Autotune is not takeoff-and-fly-autotune-to-get-working-vehicle
      • Needs base-tune
      • Standards were dropped and it stopped working well and people lost confidence
    • Filtering failures?
  • Quik-tune is based around Leonard’s tuning method
    • MUST be able to detect oscillations
    • Can’t touch accels
    • People are generally happy with the way it works out…

UTC1135 - Mode takeoff fixes for fence autoenable by Hwurzburg · Pull Request #25804 · ArduPilot/ardupilot · GitHub

  • Auto-enable fence on takeoff
  • Uses correct nav roll/pitch and throttle calculations afterwards
  • This changes the takeoff profile a lot
    • Was causing Pitot tune test to fail because the pressures were moving too rapidly
  • Merged!

UTC1145 - Plane: rework `isHeadingLinedUp` function for loiter breakout by IamPete1 · Pull Request #26102 · ArduPilot/ardupilot · GitHub

  • isHeadlingLinedUp fixes
  • Some weirdnesses fixed up
  • JR opened up a related issue
  • Curve-into-other-curve with normal ArduPilot missions
    • Currently breakout logic is a bit weird within 5% of radius
  • This PR is much less sharp and more intuitive
    • Clean curve exit
  • Probably some follow-up PRs
    • JR is most of the way there with his stuff
    • But some around mavlink compatability
  • Merged!

UTC1148 - Plane: output nav scripting throttle with rest of nav scripting stuff by IamPete1 · Pull Request #26148 · ArduPilot/ardupilot · GitHub

  • Output nav-scripting-throttle differently
  • From stabilize function which is why it works for tricks-on-a-switch
  • Tricks-on-switch definitely not going to work in training or manual!
  • merged!

UTC1149 - Plane: output nav scripting throttle with rest of nav scripting stuff by IamPete1 · Pull Request #26148 · ArduPilot/ardupilot · GitHub

  • Move logging up to AP_Vehicle class
  • Merged!

UTC1153 - Move parameter storage formatting up to AP_Vehicle by peterbarker · Pull Request #26181 · ArduPilot/ardupilot · GitHub

  • Move formatting of parameter storage up to AP_Vehicle
  • Conflicting
  • MergeOnCIPass

UTC1156 - Decode high duty-cycle CRSF frames using frame marker rather than timeouts by andyp1per · Pull Request #26183 · ArduPilot/ardupilot · GitHub

  • High-duty CRSF using frame markers
  • We use gap-framing in many of our backends for reset
    • Cheap synchronisation
    • Works will with low-duty-cycle protocols
    • Crsf is not low-duty-cycle!
    • Pushing uart to limit
    • Multiple output uarts
    • Normally no DMA on sending devices
      • Gaps in middle of frames!
  • This is more complicated than it needs to be?
    • Tridge has requested changes to simplify
    • Will discuss at DevCallEU

UTC1204 - Copter: run rate loop at full filter rate in its own thread by tridge · Pull Request #26189 · ArduPilot/ardupilot · GitHub

  • Run rate loops at IMU sample rate
  • Very much a WIP
  • 2kHz 4kHz 8kHz rate loops in Copter!
  • Need a bdshot vehicle, really
  • Need to show it provides a benefit on some vehicles
    • Flight options=8 to enable this on this PR
    • Why not always do it?
      • More combinations == more changes for failure
      • Not much point doing this on large vehicles
      • Only on tiny vehicles does it really matter
      • This should remove the PID aliasing
      • Can’t do it with just PWM outputs?
        • Definite advantages in running filtering at the higher rate
  • Should be able to update notch centre frequencies much faster
    • And have narrower notches
  • Lots of testing on Saturday
  • Logging to see effects?
    • How do we evaluate without better logging?
    • Can see amount of noise power in sub-50Hz range
  • Things don’t fall out of the sky with this
  • Probably of most use to TinyWhoops
  • Only Copter for now, maybe Plane later
  • Dt is a real problem when passing stuff into attitude controller
  • Implications for autotune
  • Motor slew issues too?
  • Vertical controller (accel-z-pid)?
    • Probably
  • Direct-measurement accel loop…
    • Attitude controller into faster thread
    • Forward-propogate attitude based on IMU buffer
    • Computation cost of PIDs is very small
    • The attitude controller is doing a huge amount of maths in contrast
    • Wait to see if rate-controllers work out first
  • Testers please!

UTC1227 - HAL_Linux: reduce delay(ms) jitter by rsaxvc · Pull Request #26195 · ArduPilot/ardupilot · GitHub

  • Fix weird delay problems with Linux
  • TOCTTOU issue
    • May skip millisecond delay
  • Delay_millis should multiply by 1000 and call delay_micros
  • Testing?
    • Willian to test?
    • Will merge if that goes OK

UTC1233 - treewide: fix shebangs - /bin/bash -> /usr/bin/env bash by sorki · Pull Request #26196 · ArduPilot/ardupilot · GitHub

  • Uyse /usr/bin/env for bash
  • For nyxos and openbsd
  • Merged!

UTC1236 - AP_Vehicle: correct compilation when AHRS not available by peterbarker · Pull Request #26198 · ArduPilot/ardupilot · GitHub

  • Fix compilation of heavy periphs
  • Merged!

UTC1237 - Plane update

UTC1249 - Copter/Rover-4.5.0 issues list · Issue #26103 · ArduPilot/ardupilot · GitHub

  • Copter update
  • Some feedback on 4.5 received - thanks!
    • Siyi old camera firmware
      • Reporting 0.0.0 firmware version for a while at startup
    • Watchdogs on MatekHl43
      • Link to discussion
    • Some logging problems too?

UTC1256 - Conference

  • Randy posted about the timing
  • Currently scheduled end-of-October
    • Randy was a bit worried about weather
    • Kaga high 17 low 13
    • Aug too hot
    • Could come back to July?
    • October means no chance of typhoon
  • Randy’s going to go out on Friday
  • Tourism ideas will get a document!

UTC0103 - GSoC

  • If you want to be a mentor ping Randy

UTC0103 - close