Dev Call July 4, 2022

Issues & Pull Requests




Attendee count (max): 13

UTC1059 - ArduPilot: SENSOR_OFFSETS add deprecated note by hendjoshsr71 · Pull Request #277 · ArduPilot/mavlink · GitHub

  • Add deprecation message to mavlink

UTC1100 - Remove experimental warning by Hwurzburg · Pull Request #36 · ArduPilot/CustomBuild · GitHub

  • Remove warning aon build server about being experimental

UTC1105 - Support VTOL_LAND options by WickedShell · Pull Request #14874 · ArduPilot/ardupilot · GitHub

  • Support VTOL landing options
  • Mavlink update
  • Needs double-shuffle mavlink update
  • Mavlink commit needs some PRs squished
  • Backport isn’t going to happen…

UTC1110 - AP_Vehicle: add AIS by IamPete1 · Pull Request #18332 · ArduPilot/ardupilot · GitHub

  • Move AIS up to vehicle
  • Weird dummy methods thing
    • Costs 300 bytes when feature disabled
  • Rover will no longer get this
  • Dummy methods on everything else 300 bytes
  • Separate AIS message
  • How would users know they can get AIS from the custom build server?
    • Something in the AIS section of the Wiki?
    • Need to make that bit common….
  • Use multiple values for AP_AIS_ENABLED to work out whether to use dummy methods or real methods (i.e. 1 gets you full support, 2 gets you the dummy methods, with the default of 2)
  • Ship-landing people will like this

UTC1125 - GCS_MAVLink: improved battery cell monitoring by tridge · Pull Request #20840 · ArduPilot/ardupilot · GitHub

  • Resting voltage not working with smart batteries
    • This fixes that
  • Backend limit
  • Distributes extra voltages across all cells
  • Individual backends are packing cells separately which isn’t bad
  • If you don’t have too many cells then this won’t change things EXCEPT fix sag corrected voltage
    • Can only support up to 14 cells in current mavlink, so this allows batteries with more cells
    • 16, 18 and more
    • LH: we’ll get to 400V
    • Some people have 100 cell battery
  • Scale-factor hack doesn’t work on smart batteries

UTC1141 - AP_Baro: auto-set field elevation from origin by tridge · Pull Request #20867 · ArduPilot/ardupilot · GitHub

  • Automatically set field elevation from origin
  • Microcontrollers on sensors can heat the air causing bad readings…
    • Keep a history of outside air temperatures
    • Look for gaps in times, find most recent temperature
    • Use a separate temperature sensor?
    • Use atmospheric model to calibrate temperature sensors?

UTC1154 - AP_InertialSensor: remove voiding of parameter by peterbarker · Pull Request #20889 · ArduPilot/ardupilot · GitHub

  • Merged
  • Removed pointless (void) of parameter

UTC1155 - AP_Param: add `@SYS/defaults.parm` and log default values by IamPete1 · Pull Request #20972 · ArduPilot/ardupilot · GitHub

  • Parameter defaults
  • What’s the default value?
    • How about ones we set every boot not the default?
    • E.g. Mot_pwm_min is set to 1000 for digital
      • Tridge thinks 1000 is the default, not whatever is in MOT_PWM_MIN default value
      • A tickbox for “show non-default parameters” shouldn’t be cluttered by this?
      • Is it useful to know that the default has been set through a mechanism other than the defaults table?
      • “The default value of a parameter is the value a parameter would have if all other parameters were as they are now and the user has never touched this parameter”
      • Mechanics first, getting everything right later?
        • Tridge thinks we should merge the ones we’ve found
        • GCS compare button will make things very obvious
  • This is close
  • Just need to work on all the corner cases
  • Maybe get this in DevCallEU
  • Unrelated: FTP on F4 boards?
    • If it fails to create FTP thread then it uses the IO thread
    • Make IO thread default on F4 boards?

UTC0017 - hwdef: remove redundant 'define HAL_MINIMIZE_FEATURES 0' by peterbarker · Pull Request #21051 · ArduPilot/ardupilot · GitHub

  • Remove verbiage from hwdef files
  • MergeOnCIPass

UTC0021 - Plane: fixed check for fixed wing approach QRTL start by tridge · Pull Request #21064 · ArduPilot/ardupilot · GitHub

  • fixed check for fixed wing approach QRTL start
  • 4.3 use full position control for entire landing approach?
  • We use main velocity controller in 4.2
    • But not position controller
    • Asymmetries in constraints make this stuff tricky
      • Some sort of override to the way limits work?
      • May need a more complex controller all over to do it
  • Merged!

UTC0021 - Dev: backing out a stable release is difficult · Issue #21066 · ArduPilot/ardupilot · GitHub

  • Backing out a stable release is difficult
  • Discuss off-line or after call?
    • 4.2.2 early report from partner of difficulties
      • Probably GPS-related, not the software at all
    • Beta is 4.2.2rc2 but official is
    • Need to update both stable and beta tags
    • Then need to delete new version from build server

UTC0031 - AP_Scripting: Generator: fix docs generator aliasing by IamPete1 · Pull Request #21072 · ArduPilot/ardupilot · GitHub

  • Fix docs generator aliasing
  • LUA docs only
  • Merged!

UTC0032 - hwdef: added telem3 on Pixhawk6X by tridge · Pull Request #21112 · ArduPilot/ardupilot · GitHub

  • Adds usart2 to pixhawk6x
  • Another hardware flow control
    • Benefit of coming to the modular party late was they knew to add more pins

UTC0034 - HAL_ChibiOS: hwdef: add new hardware MFE-PixSurveyA1 by mfe-xiao · Pull Request #21113 · ArduPilot/ardupilot · GitHub

  • New board
  • Fmuv3 variant
  • Lsm - boo!
  • Heater
  • MFE-PixSurveyA1
  • No bootloader
  • Might be nice to have a document in the dev wiki outlining what we like to have in the hwdef directories
    • E.g. pinouts

UTC0036 - Plane update

UTC0045 - Copter/Rover: 4.2.0 issues list · Issue #20192 · ArduPilot/ardupilot · GitHub

  • We’ll restart beta testing for 4.2.2 soon
    • Allow partner to re-test to make sure problem was unrelated to 4.2.2
    • No changes will be made to 4.2.2
    • Randy didn’t have a lot of confidence in the release
      • Only left it out for a week
      • Andy said he was confident of it
        • But after the claw-back was done
    • Circling issues when circle rate exceeds required yaw rate
      • 90 degree/second fast turn and yaw couldn’t keep up
        • Document it?
      • DO_CHANGE_SPEED doesn’t work directly after takeoff command
      • Because takeoff doesn’t use wpnav
      • Because pnav gets initiated on first actual waypoint
  • Solo gimbal not working?
    • It does move but Solex app can’t see it
    • Maybe looking for heartbeat?
  • Not far from releasing 4.3
    • Early August we’d start beta testing 4.3

UTC0052 - GSoC update

  • Autonomous docking update
    • Videos of testing in Gazebo with new DOCK mode
    • Precision landing library can be used horizontally
    • Virtual docking port with April tag on it and the vehicle can get there
      • Comes at it at an angle ATM
      • Wants to come at it square-on, ‘though
  • Sankeet is working on easing ROS integration
    • Allow AP to offload navigation to ROS
      • Cleaning up Randy’s work…
  • Luxonis camera over water is being tested
    • Would like Henry to get camera so he can help with testing
    • New version of camera coming…
  • Emre’s on call, PH’s student
    • PR in next couple of days
    • Easier for researchers to work new attitude controllers into ArduPIlot
      • E.g. matlab-based controllers
      • Runtime-selectable with any lucky
        • RC switch to change attitude controller

UTC0059 - Here3 GPS no fix in Mission Planner, but UAVCAN inspector shows Coordinates - Here GPS - Cubepilot

  • CAN GPSs we don’t auto-config except for CAN moving baseline and the like
  • We don’t configure message rates and whatnot
  • Moving base only applies to F9s
  • Documentation could be improved….
  • AP_Periph-based devices do the moving baseline stuff appropriately

UTC0102 - Callisto model

  • Yaw a bit funny and wobbled around
  • SITL model has no dependency on size of aircraft
  • Torque is fixed
    • So larger aircraft, large mass, yaw control is same as every other aircraft
    • Small fix that scales yaw control with thrust….
      • Unsure if that’s correct….
  • Make change in such a way that is doesn’t affect default model
    • Unless that model is broken
      • In which case might have a bit of pain….
  • Maybe a nicer simulation - closer physical model?
    • More work on Callisto?
  • Current values in JSON file are generally easy to work with
  • Default parameters stuff PH is working on would be useful here
  • PH has a few bits floating around

UTC0012 - triple-notch-filter

  • Same attenuation number for single/double/triple
  • Overlap makes for massive attenuation
  • Goal is someone can go single ->triple and retain existing tuning performance
  • Two approaches
    • To reap benefits for phase lang you must lower the attenuation
      • Improve-your-tuning perspective
      • Would need to retune to get benefirt
    • From improve noise-attenuation perspective
      • Can go single → triple and not change any tuning
      • Just get reduced noise levels
  • Leonard’s happy with the way stuff is currently done with the three-notch
  • Triple the CPU cost
    • But we stopped doing notch on every IMU
  • Could we create a triple-notch filter class that does all the maths in one place and save CPU?
    • Each notch has poles and zeroes
      • Two forms of implementations of discrete-time filters to represent those poles and zeroes
        • Can’t do equivalent three-notch filter with fewer polls/zeroes
          • Maybe something that’s close?
          • There are band-stop filters which are combinations of lp/hf
            • In order to get sharp edges you need higher orders (poles/zeroes)
    • Might be able to get marginal improvement but it’s going to be hardcore digital filtering
    • Band-stop filters Cheby-chef style filters
      • Different ripple characteristics
    • If you have really precise filter requirements
      • But when you’re just doing what we’re doing the notch is a very efficient filter
      • Fire
  • 18Hz ICE engines are a pain
  • Very low frequency vibrations
    • Several Hz wide notch
    • Significant phase lag
    • Uncertainty in RPM
  • Can now log pre and post filter at same time
  • Want real-time fft to be able to do cleanup of the throttle/esc rpm filters
  • Want to be able to log each filter separately
  • Wrapping in FilterTool?
    • Many other cool changes too (inc log-scale)
    • Full PID models
      • Maybe later
    • Want to be able to upload logs and see what difference it would make / where the filters would apply
  • 4.2.2?
    • No, probably 4.3
    • About 8 week’s time

UTC0139 - close