Dev Call Nov 28, 2022

Development Team
1

Issues & Pull Requests

Remote ID

Plane

Copter & Rover

2

Attendee count (max): 15

UTC2300 - message_definitions: add all_with_development.xml by julianoes · Pull Request #1924 · mavlink/mavlink · GitHub

  • Why isn’t this all about WIP tags rather than development.xml?
    • Development.xml is really just WIP messages destined for common.xml (or standard.xml) eventually?
  • Need James to weigh in
  • It’s nice that we’re moving towards QGC compiling against all.xml

UTC2311 - https://github.com/ArduPilot/ardupilot/pull/21510

  • Optical flow for Rover
  • Randy’s good with this now
  • Merged!

UTC2317 - https://github.com/ArduPilot/ardupilot/pull/21609

  • Copter weathervaning
  • Yaw rework made this a much smaller patch
  • Could do even better than this by incorporating accel data
  • Changing mode within mode is strange

UTC2323 - Plane: if RTL with VTOL active and within QRTL radius and alt then QRTL by IamPete1 · Pull Request #21883 · ArduPilot/ardupilot · GitHub

  • Actively switch to QRTL if we’re within the RTL radius
  • QRTL mode must be zero for ship landing to work

UTC2328 - Add * to the end of flight mode string in CRSF by MingchenZhang · Pull Request #22216 · ArduPilot/ardupilot · GitHub

  • Add * to end of mode string to indicate armed/disarmed
  • Merged!

UTC2332 - GCS_MAVLink: do not run all commands received on private channel by peterbarker · Pull Request #22218 · ArduPilot/ardupilot · GitHub

  • Do not run all commands received on private channels
  • Merged!

UTC2332 - AP_Scripting: dont check objects that are expected to be null by IamPete1 · Pull Request #22254 · ArduPilot/ardupilot · GitHub

  • Don’t check objects that re expected to be null
  • Merged
  • Checks are done at a higher level

UTC2333 - AP_Airspeed:add warning for large offset change by Hwurzburg · Pull Request #22256 · ArduPilot/ardupilot · GitHub

  • Add warning for large airspeed offset change
    • To catch when the calibration is done when wind is blowing through sensor
  • before/after calibration check
  • Wind blowing you will see fluctuations
    • So instead of looking at calibrated differences, look at fluctuations
  • Shorter message to fit it onto an OSD?
  • INFO level may not be seen in MissionPlanner
  • Difference matters depending on aircraft type.
  • Scale according to FBW_MIN?
  • DLVRs make everything easier

UTC2349 - Plane: Quadplane: tiltrotor: add Q_OPTION to keep motors tilted up when disarmed in FW modes by IamPete1 · Pull Request #22288 · ArduPilot/ardupilot · GitHub

  • Option to keep motors vertical even when disarmed in fw modes
  • Watchdog reset in mid-flight
    • Fixed-wings with watchdogs could be problematic
      • But this is an option
  • Merged!

UTC2351 - AP_Motors: output_motor_mask use uin16_t mask by IamPete1 · Pull Request #22290 · ArduPilot/ardupilot · GitHub

  • Expand tiltable motor mask to 16 bits for tiltable motors
    • This is motors mask, not output mask
  • Merged!

UTC2354 - AP_HAL_ChibiOS:Add moving baseline excluse to min gps by Hwurzburg · Pull Request #22294 · ArduPilot/ardupilot · GitHub

  • Add moving baseline exclude to min gps
    • Specifically re-enable it just on Pixhawk1-1M?
  • costs 12 bytes?!
  • Killing this

UTC2358 - AP_HAL_ChibiOS:free space on AtomRCF405NAVI by Hwurzburg · Pull Request #22295 · ArduPilot/ardupilot · GitHub

  • Make space on AtomRC

UTC2359 - AP_HAL_ChibiOS:remove synth curr from bds with onbd current by Hwurzburg · Pull Request #22296 · ArduPilot/ardupilot · GitHub

  • Remove synthetic current sensor from several boards
  • Merged!

UTC0001 - Move controllers and PID objects to a real time update system by lthall · Pull Request #22298 · ArduPilot/ardupilot · GitHub

  • Move PID objects to real-time-update system
  • Flow of control errors on CubeBlacks
    • Position controller sanity check
  • Just big slow-downs causing the sanity check to trigger!
  • Average loop rate is slowing down
    • Maybe 30%?!
    • Effectively increases D-term by 30% while reducing i-term by 30%
    • Also affects navigation
      • Position errors come in and out based on short-term-load of processor
  • Planning this for a long time
    • Controllers were written with an eye to doing this
  • Everything across the board seems to be better
  • Moves to using the most recent loop time delta
  • 1.5kB of flash is a lot….
    • Need to un-inline get_dt to see if it makes a size change
  • Maybe make a set_dt call on the attitude controller in the fast loop
    • Just before doing the attitude-rate updates
    • After sampling the ins
  • 11:14 AM]Peter Hall: you could also just fetch dt once per higher level function.
  • [11:14 AM]Peter Hall: eg const float _dt = get_dt();
    • Leonard was afraid of missing entry points
  • TECS now runs its own dt so we have precedent
  • Needs to use get_last_loop_time_us
    • Mode-loiter needs to use new
  • This will come back with a set_dt
  • No other warnings on the scheduler telling them roots are running slowly
    • EKF 1-in-4-loops thing
  • Run at 200Hz instead?
  • Would be nice to be able to trigger the slowdowns
  • How do we tell the user there are performance problems?
    • We really need to
  • One fixed resource we’re not tracking well
  • Would be nice to get reporting on increased CPU usage
    • SITL is so fast….
  • A failsafe if the CPU loop times go too far
  • Need to adjust the mavlink message which sends the load to the GCS to take into account extra microseconds added
  • If people have tuned on 4.3 and then they fly on 4.4 with this fix
    • Will need to retune!
  • SCHED_LOOP_RATE to 300?
    • Not optimal
  • This looks like a 4.4 change, not a 4.3 change
  • This deals with the effects of the problem but not the causes
  • Need a Wiki page and a warning
    • Turn off ekf cores?
    • Remove harmonic notches?
    • Log SD into PM message
  • Where has the CPU gone?
    • Same problem as the RAM?
      • Just losing bits and pieces over time?
  • Timestamps will always be integer multiple of sample rate
    • Two or 3 gyro samples at 400Hz
      • You won’t get a 2.5ms number here ever

UTC0048 - Issues · ArduPilot/ArduRemoteID · GitHub

  • Not a lot of activity
  • Roel and tridge have been looking at error handling on flash update

UTC0049 - Plane: stable 4.3.x issues list · Issue #15941 · ArduPilot/ardupilot · GitHub

  • Plane update
  • Second beta coming with a few more patches
  • Not much feedback on the beta
    • Not much feedback forCopter either
  • Icm26802 changes back?
  • RTL with VTOL active one?
  • uBlox M10 will probably come back
    • Pretty bad without those changed
  • MAVCAN fix too

UTC0051 - Copter/Rover-4.3 issues list · Issue #21711 · ArduPilot/ardupilot · GitHub

  • The perf issue above is the most pressing
  • Ce8 is new tiny gimbal
    • Needs work!
    • Randy will chase
  • Prearm checks are confusing users on the servo function to -1
    • 103 is an analogue pin and it can come back across to the MCU
    • Little checks are being too strict?
    • Not servo 3
    • Exception for 103?
    • There might be a conflict….
    • AM]Peter Hall: yeah 103:MainOut3
    • [11:57 AM]Peter Hall: So there is a pin conflict.
  • Boot problems if too many LEDs enabled?
  • Ongoing trouble with setting up gimbals because MissionPlanner isn’t using MNT1_ parameters
    • Servo gimbals vs brushless gimbal issue
    • camera/gimbal wiki pages need to be fixed for brushless gimbals that use PWM
  • No 4.3 for Rover yet
    • SCurve / position controller issues
      • Tracking straight lines…

UTC0100 - AP_CANManager: add an output buffer for MAVCAN by tridge · Pull Request #22306 · ArduPilot/ardupilot · GitHub

  • Add an output buffer for MAVCAN
  • Was working but things moved on
  • Sid’s best qualified to look at this
  • Will come back with DevCallEU

UTC0104 - AP_Periph: Add timeout to ESC output when CAN packets are lost by joshanne · Pull Request #22307 · ArduPilot/ardupilot · GitHub

  • Add timeout to ESC output when CAN packets are lost
  • Tested by yanking the CAN cable
  • Will remove the parameter

UTC0108 - AP_ESC_Telem: include header that adds NUM_SERVO_CHANNELS by joshanne · Pull Request #22308 · ArduPilot/ardupilot · GitHub

  • Will try to use a SRV_Channel_config.h
  • Don’t want a different length in different places

UTC0114 - SITL: fixed sign of wind with AHRS_EKF_TYPE=10 and builtin models by tridge · Pull Request #22310 · ArduPilot/ardupilot · GitHub

  • Sailboat stuff
  • SITL windvane might need fixing
    • PH will check

UTC0116 - https://github.com/ArduPilot/ardupilot/pull/22299

  • TECS fix
  • Double-flaring was broken
  • Backport?
    • Probably not
    • Not a trivial backport

UTC0118 - close