Dev Call Feb 10, 2020 2300 UTC

Issues & Pull requests




1 Like

Attendee count (max): 24

UTC1105 - Welcome to Arash!

  • GeorgiaTech
    • Use ArduPilot

UTC1106 -

  • Add EKF uninitialised flag to mavlink packets
    • So a GCS can not say, “AHRS bad”
  • Static assert is incorrect
    • Needs to log a 32-bit number
    • get_filter_status

UTC1110 -

  • AP_HAL threadsafe object buffer
    • Played with other ways of doing it
      • Copy-and-paste was the best result
  • Needs squashing
  • Duplication is bad
  • Comments will be added and then this can be merged

UTC1113 -

  • Prevent failsafe returning to do-land-start
  • Begging MdB to have a look

UTC1114 -

  • Toshibacan / AP_Scripting
  • Usage times out of ESCs
  • Tells you how many seconds it has been spinning since it was created
  • Retrieves from ESC
  • To be exposed in LUA scripts?
    • Create AP_ESC_Telem library
      • Common interface to talking with toshibaca/kdecan/piccolocan/blheli/uavcan?
      • Minimum information in there ATM
  • Why not in AP_Vehicle?
    • Will be changed
    • KDECAN only on Plane/Copter?
    • [10:17 AM] To Weekly devcall: There’s a big comment on the KDECAN stuff IIR C.
  • Several places in the code where we check all of them individually
    • Several where we miss a type of ESC
      • Which can be a safety issue
    • So use the library to do prearm checks or whatever
    • Is_active?
    • Shouldn’t need to worry about which are being used in AP_Arming
    • Some sort of limiting code not in place?
    • 1000 to 2000 limits in blheli derived from the “digital mask” which is the mask of the motors that don’t do PWM and use 1000-2000
    • [10:22 AM] (Channel) Michael_duBreuil: AP_ESC may be misleading that its not just reporting? But maybe it shouldn’t be? Really haven’t thought enough about it/used them in this way to have real comments
    • [10:23 AM] To Weekly devcall: … like, you know, putting in 1000-2000 limit stuff isn’t necessarily “telemetry” :wink:
  • Randy will create an issue to move other ESCs into here
  • AP_Motors doesn’t cover the forward motor on a Plane.
  • AP_ESC?
  • Will be moved to AP_Vehicle
  • MdB left a review
  • Logging from this library rather than individually?
    • Randy thinks we’ll call down into them separately
    • Constant from library

UTC1130 -

UTC1134 -

  • Extends functionality that’s currently built into the Solo companion computer
  • Mavlink messages that go to the gimbal over the serial connection
  • Only thing that understands this is Solo
    • This will allow missions to toggle structure or toggle between video/photo
    • Peter’s suggestions
  • Matt will take Peter’s stuff and run with it
  • “GoPro” -> “SoloGoPro”

UTC1144 -

  • Should we log values between 1000 and 2000?
  • Onboard logs for rcout message used for diagnostics
    • “ is it between 1000 and 2000? Then digital”
      • “Outside means PWM”
  • Should we be consistent?
  • We log the actual output when the output is a PWM
    • Oneshot / oneshot125 and normal PWM are all PWM
      • Width of pulse is the data
    • All the others are not PWM
      • The output is not a pulse width
        • So we standardise the range
        • Dshot is not 0-1000
          • It has some extra bits which make things more complex
          • Zero throttle on dshot is actually something like 38
            • Lower part of range is reserved
          • High bit is also telemetry bit
          • Can’t log this - the data is crazy!
  • Log in same range as being sent to physical ESCs?
    • Can’t really as the range is wide
    • So normalize it for the digital ones?
    • Change oneshot 125 to log before we divide by 8?
    • Change servo-min and servo-max to be within the relevant range?
    • Was logged post-scaling as this is an analogue pulse width
      • Exact value can matter for some ESCs to initialise
        • Probably past these issues now
        • We used to round to an integer number of microseconds
          • We don’t any more
            • ChibiOS does 1/8th microsecond multiples
            • Full range
              • Rather than oneshot125 having 125 throttle levels
              • Now we have full 125*8 values
  • Logging 0-100 would mean logging floats?
    • Separate thing…
    • Tridge tried to push 0-1 floats throughout the code
    • This turned out to be tough
    • Linking stuff to PWM on outputs that have nothing to do with PWM is not good
  • Randy will make it log before division
    • Must check aux and main outputs

UTC1151 -

  • Fix waf --default-parameters
  • Will need to come back to this when Matt is around
  • Need to make sure traditional invocations still work
  • Note that entire hwdef file is changed so you need to make a copy of the existing hwdef
  • Tridge testing it

UTC1158 -

  • Allow do-set-servo commands to affect sprayer etc
  • Fixes regression
    • But we do allow people to manually override when needed
  • Michael’s PR made things more strict but broke things for people
  • [11:01 AM] (Channel) Michael_duBreuil: the sprayer and do set servo is a race condition IIRC which is why I don’t like it at all
  • Race condition where do-set-servo wins at the moment
    • Sprayer runs output when the vehicle is flying and is being managed
  • Correct way is to have mavlink messages

UTC0008 -

  • Use object universally
  • Rename flightmode to control_mode
  • What should we call it?
  • Autotrim_supported needed
  • Peter to review once that is done

UTC0016 -

  • Allows different instances to interact
  • Single image comes up with a singlet xml file and start many ArduPilot instances
  • Does it affect more than Scrimmage?
    • Do_build_parameters only used in SCrimmage?
    • Multicast options tested?
      • libraries/SITL/examples/morse stuff
  • Needs testing

UTC0024 -

  • Root of problem is that it doesn’t abide by the wpnav_accel parameter
    • Bill’s PR
    • Out of mission, into mission the transition back in didn’t go well
      • Initialises current wp speed
      • Uses the acceleration parameter
      • Wp speed parameter not always used
        • Do-change-speed
          • Leave mission, re-enter and you wouldn’t do the do-change-speed speed
      • [11:34 AM] To Weekly devcall: @Randy just to be clear - that behaviour is intended - you lose the DO_CHANGE_SPEED setting when transitioning out and back in again, right?
        • yep
  • Leonard to look at both

UTC0035 - Plane 4.0 release

  • Bringing Plane and Copter releases back into sync
    • Needs testing!
    • Logs please!

UTC0037 -

  • Copter
  • 4.0.4rc4
    • autotune fixes etc
  • MAVFTP causes reboot on 4.0 and master
    • Watchdog reset
    • Happens most times
    • Reset-to-defaults seems to be the key
    • Doesn’t happen in SITL
    • Randy can reproduce
      • WDOG message required
      • WDOG {TimeUS : 2998260, Task : -3, IErr : 0, IErrCnt : 0, MavMsg : 0, MavCmd : 0, SemLine : 0, FL : 100, FT : 3, FA : 1431655769, FP : 58, ICSR : 4196355}
        • In thread because of the -3
        • Thread priority 58
          • IO priority thread
    • Probably a nullptr dereference
  • CUAVv5+
    • Connection issues?
    • Can’t reproduce so far…
    • User has posted parameters from working/non-working setup
  • Issues with landing gear retract stuff
  • Blheli passthrough not working on CubeOrange
    • PR in that fixes that
    • Tridge wanted to test it a bit more
      • Iamepete said it was fine
  • Zigzag mode issues
    • spin-up/spin-down not working appropriately
      • takeoff/landing not handled correctly in zigzag mode
      • Motors may stop
    • Allow takeoff/land in zigzag
  • Eight regressions left to hammer out

UTC0046 -

  • No updates for Rover
  • Failsafe messages cleaned up

UTC0051 - vote results

  • Matt and Andy for the devteam both passed
    • Welcome!
  • Please vote on the “paid positions” PR!

UTC0052 - Close