Dev Call June 8, 2020

Issues & Pull Requests

Student updates



1 Like

Attendee count (max): 29

UTC1100 -

  • Calling calc_pwm twice
  • Not needed now
  • Overriding channels from scripts was busted
    • Couldn’t override AP_Motors, for example
  • E-stop wasn’t overriding LUA scripts

UTC1103 -

  • Restart scripts automatically
  • Avoids reboots
  • Only does this when disarmed
  • Tridge thinks it should be restarted when armed
    • Add something into the script so it can say it’s good to be restarted
      • For scripts that are critical
        • Rare paths / race conditions / interpreted language
  • All scripts would have to fail for the restart to happen
  • Per-script on binding?
  • If code doesn’t compile it wouldn’t be able to set the restart itself
  • Handling everything in-script?
    • Eval’ing
  • Requires support?
    • Little bit of scope creep….
  • Opt-in at bind time
    • Suggest when debugging scripts then adding debug
  • [9:14 AM] Peter Barker: Another bit for armed/disarmed reload?
  • [9:14 AM] WickedShell: peter nah, if you need armed use it via the script one
  • [9:14 AM] WickedShell: the disarmed gets you around compilation error
  • [9:14 AM] WickedShell: then you can have the lua call
  • [9:14 AM] Peter Barker: So two features.
  • [9:14 AM] WickedShell: yeah
  • … and another command_long to restart scripts
  • Delay script until someone connects?
    • Would be good to have an example of that

UTC1115 -

  • Add minimum-megabyte-free parameter
  • Merged!

UTC1124 -

  • Would be nice to get rid of BRD_PWM_COUNT
  • Which servo functions need a BRD_PWM_COUNT change?
    • Say it is 4
      • 4 aux available as servo functions
  • Say you have a relay pin set to one of those
  • Would be nice to define functions of a pin solely on servo function values
  • We defaulted relay pin and relay pin to 56,57 which means a lot of people may be relying on defaults values
    • They’ll start to get warnings
  • Changing the relay defaults
    • Arming check

UTC1135 -

  • Accept external position estimates e.g. T265 at 50Hz
  • More storage requirements
  • Would be nice to only allocate when needed
    • E.g. when data is first pushed in
  • Nothing to tell if the buffer is too small
    • Samples get thrown away without processing
    • SITL warning?
    • EKF time Delay horizon means you stop fusing anything as no data ages sufficiently
    • Look at the delta-t and the buffer length?
    • Dynamically allocate buffers might fail if you’re not feeding data in e.g. rangefinder-min-cm
  • Lots of code in the .h file
    • Repeated template class in a header file

UTC1137 -

  • Add glitch filter to Maxbotix
  • Same as in the echo-trigger driver
  • Cleans things up a bit
    • Tridge thinks this should be moved up
      • Default glitch filter to off
    • Glitches on radars and lidars
      • More glitches on these though
  • Henry will have a shot at moving it up

UTC1146 -

  • Hard to remember which switch positions are which
  • Some positions don’t have an observable effect
  • Several items need to be removed
  • Text strings needs to be fixed in a few cases
  • SW:LOW -> LOW
  • Henry will change things
  • So long as toggle a switch indicates you’ve done something this feature is a success

UTC1152 -

  • Rotors on a convergence will tilt forward
  • Works on all quadplanes
  • Not active in position-controlled-modes
    • In qloiter the forward throttle is handled automatically
  • Tilt on ground works
    • So you know where the props will be when you start them spinning
  • Plane-specific arming check that it’s zero when you arm
  • Parameter needs a better description as it’s misleading
  • Slider on side of throttle for forward throttle….

UTC0005 -

  • Vicon
  • Couple of parameters for visualodom
    • Use covariance fields
    • There is a minimum error
      • Some external systems are over-optimistic about their errors
  • Timing errors need to be compounded onto the source errors
  • cbrtf … never called it before
    • For ellipsoid fitting
  • Is_nan on first element?
  • Have to be careful with passing arrays to functions
  • The size is just a hint, sizeof() doesn’t return the array size but the pointer size
  • Needs units

UTC0016 -

  • Adds telemetry radio RSSI as a source
  • Merged!

UTC0020 -

  • Linux HAL uses std vector which still uses empty()
  • Avoids trap where empty() is ambiguous
  • Merged

UTC0023 -

  • Needs to pass CI then can be merged
  • Alignment issues

UTC0030 -

  • Strings of boards in output
  • Merged

UTC0031 -

  • Make is_active and is_private orthogonal
  • Slightly difficult one to test
  • Merged!
  • Gcs-private option coming?
    • Uavionix ping is spamming message interval confirmations
      • Don’t want these going to GCS
      • We don’t forward ADSB vehicles, we consume and transmit
  • Serial options bit
  • Would be nice to dig ourselves out of the SR0_*

UTC0037 -

  • Needs more testing
  • Put it on a CubeBlack

UTC0048 -

  • Log fence enable/disable always
  • merged!

UTC0052 -

  • Automated decimation
    • So you don’t need several lines to do decimation
  • Just a developer tool
  • Should not increase flash size unless used

UTC0056 -

  • Shouldn’t have the bits at the end
  • Need to make sure we don’t get stuck in missions
    • If you try to do a command which doesn’t exist we shouldn’t block

UTC0110 -

  • Merged!
  • Rc-channel honour reversals

UTC0113 -

  • Limit forward throttle during auto transition
  • Simplify by saying in-takeoff or in-quadplane-transition
    • Do it in one spot
      • Set-servos-control
  • Battery sag in transition is a problem
    • This can reduce the problem
  • Needs to be configurable

UTC0119 -

  • Noop if you don’t enable it
  • Yaw motors tilt can be tested

UTC0124 -

  • No answers…

UTC0128 -

  • Compass issues still to be worked on

UTC0129 -

  • Several new items
  • Cubepurple has an extra defunct parameter
  • HOTT telem fix is coming in rc2
  • Ap-proximity no-zones patch coming in
  • 4.0.4rc2 coming soon

UTC0130 - Nothing for Rover

UTC0132 - terrain generation

  • Tile generation was incorrect
  • Regeneration is happening now
  • Correct format. Probably.
  • Audrey says he’s still seeing oddities

UTC0132 - User alerts

UTC0139 - close

1 Like