Dev Call Oct 19, 2020

Issues & Pull Requests




Attendee count (max): 22

UTC1100 -

  • Back from last weekend
  • Parameter conversion done
  • GPS_PRIMARY parameter
  • Primary to blended?
  • What’s primary in dual can?
  • Set to 2 for blended?
  • Shouldn’t change behaviour for existing users?
  • Should this be 1-or-0 based?
  • Need to test
    • Blending
    • Single gps with primary set to primary
    • Single gps with primary set to non-existant GPS
    • Dual GPS
    • Testing with first satellites than second and vice-versa
  • Allows users to specify your F9 as your primary GPS (which you’ve put on the second GPS port because it doesn’t have a plug suitable for the primary port)
  • Will be back again….

UTC1120 -

  • Wind_max disables sensor

  • Wind_warn warns about high wind

  • Disabling airspeed sensor by setting wind_max

  • No stable release with AIRSPEED_OPTIONS?

    • Might not need conversion
  • Disable on sensor failure?

    • Sets use to zero
  • Gps speed accuracy?!

    • Disabling the airspeed sensor if you lose GPS would be bad
  • Airspeed used to keep attitude in DCM….

  • Quote of the day: “Since you are going to be re-writing this code to be neater you might as well rewrite it so it works”

    • Using GPS speed accuracy looks bad in existing code
  • Probability failure work correctly 10m/s vs 100m/s?

    • Users need to configure
  • We do need to think about where we use gps-based things in the code as vision is a thing now-adays

UTC1135 -

  • Do this on the EU DevCall

UTC1136 -

  • Highlights and works around a bug in AP_FileSystem
  • Contents of file remains empty without this workaround
  • Need to close file and open it again
  • Not going to merge this - we do not want to paper over the problem
  • buffering?
  • Maybe the AP_FileSystem flush is broken?
  • PH will look at it
  • Valgrind it!

UTC1148 -

  • Discussed it on eudevcall
  • Changed float to int
  • Convention is to remove the class from the code with our #ifdef’s
    • Not just the implementations

UTC1153 -

  • gps/nongps environment transitioning
  • Alt-source and horizontal-source split
    • External-nav can be used for just-z or just-xy
  • Flown a lot
  • Part of a much more extensive PR which allows for selection of all sorts of bits and pieces for data sources
  • Useful for T265 as camera is quite sensitive to high-vibes
    • Camera can lose its position and velocity estimates
    • Loses Z-axis goes first
    • Horizontal from camera, vertical baro
  • Users will need to set their alt-source to external nav too or they’ll continue to use baro (by default)
  • Ekf3-only
  • Active-height source should drop back to baro on failure
  • Goal is to use baro with external nav
  • 50 commits in original PR
  • EKF doesn’t use rangefinder, broad look
    • Not for Plane or Copter landings, essentially
    • EKF is fundamentally flat-earth model which doesn’t get along with bumps and buildings
  • Bonus plug for ekf data source PR
    • User can switch sources using switch on transmitter
    • Also a LUA script which magically changes sources
      • Tested using three sources
      • OF, GPS, camera
      • Switching between camera and GPS
      • Gps speed accuracy was OK, but using camera innovations helped
    • Video coming!

UTC0013 -

    • Off by 3 days in timestamp from NMEA GPS
  • A chunk of code cleanup but just sharing code
  • Will also affect MTK GPSs
  • Simulated NMEA GPS showed the problem up!
  • Perhaps we did break this at some stage
    • Tridge remembers making sure he got the correct date/time

UTC0020 -

  • Some global variables in EKF are there for stack space

UTC0024 -

  • Telling users better than silently fixing it?
  • Users will get an EKF1 doesn’t exist error rather than defaulting to EK2
  • Merged!

UTC0033 -

  • Do we want to notify a user of an EKF failsafe even if we’re not going to act on it?
    • Randy likes it as it gives the user situational awareness
  • Users are getting annoyed by it?
  • Add another failsafe action which does not do reporting?
    • Rename existing one to REPORT_ONLY?
  • Change existing behaviour to not report, andd a new report-only option

UTC0040 -

  • Merged!
  • More random cleanups

UTC0041 -

  • Tridge will ping Paul on the review which was done

UTC0042 -

  • Generation of SIM_ parameters
  • Will do without HAL tag
    • Just base it on SIM_
  • Add metadata to other vehicles
  • Generate data twice, download twice on Wiki

UTC0052 -

  • More script bindings - EKF variances
  • These are the scaled
    • Values > 1 are outside the gate
  • Randy will add some comments….
  • Last entry is size of last reset
  • API concreteness issues
    • Getposresetne shouldn’t be in this call but if we remove it we’d break the bindings
    • Nobody actually uses it?
  • Make lots of calls?
    • Consistency issue

UTC0104 -

  • Additional problem:
    • Motors on heli completely stopped
    • Interlock engaged
    • Switch to stab and the rotors will start
    • This PR will be expanded to fix that
    • This PR makes this problem more likely
  • MdB tested this and it worked fine

UTC0107 -

  • Non-gps navigation stuff again!
  • Randy is really liking LUA
  • Flight mode in LUA
    • Using guided-velocity to move the vehicle around
  • No user input in guided mode normally
    • So normally we don’t let the vehicle arm
  • Extra option for allow-arming-from-scripting?
  • Why do we want to prevent people arming in guided?
    • Support calls
  • We don’t allow arming in RTL…
    • Plane uses throttle suppression
    • Auto-armed is kind of like that
    • Quadplane and throttle suppression?
  • Use a single bit for all modes?
    • FLIGHT_OPTIONS for example?
  • Whitespace in LUA scripts?
    • Spaces not tabs
    • 2-space in LUA?

UTC0124 -

  • Navigator LED
  • RPi hat that’s being worked on
  • Integrate neopixel for notify
  • Stm32 is supported via timer
  • This is a linux board and this abuses the SPI interface to drive it
  • Only thing on the bus if you do this
  • This is on the data line….
  • Slightly peculiar….
  • Needs a comment to explain what’s going on
  • Some stm32 boards do have external SPI…

UTC0132 - Plane updates

  • Plane landing issue reported by Nate
    • probably long standing
    • Full descent into ground…
    • 15608
  • Paul’s new TECS and landing improvements are great
    • Part of 4.1…

UTC0135 - Copter and Rover updates

  • 4.0.5rc2 is out there in beta
    • Stuck on one issue
      • Andreayl
      • Pitch twitches
    • Back-to-back test would be good
    • Maybe a bisect?
  • 4.1 release?
    • Want stuff done:
      • S-curve
      • Ekf-position-source-switching

UTC0139 - out of time - closed