Servers by jDrones

Dev Call Apr 20, 2020

Issues & Pull Requests




Attendee count (max): 27

UTC1102 -

  • Profiled support
  • Stuff finished up from last week
  • Tridge isn’t keen on having this backported to Plane
  • Merged into Copter stable

UTC1102 -

  • Autorotate bailout stuff
  • Tell motors library we want rapid spoolup of throttle
  • Runup timer and rampup speed
  • Weren’t correctly handling timer
  • Preserve delta between rampup time and runup time
  • Should never be less than one?
    • Quadhelis have very fast runups?
  • SITL-only, being autorotate
  • End-to-end review
  • Bill needs to read and test end-to-end

UTC1109 -

  • Allow use of ramp-up when using external governors
  • Set governor up so it sees a pwm in a certain band
    • Governor knows when you return power (reengage interlock) will ramp up throttle much quicker
  • Parameter to set this window for external governor
  • New tail rotor type allowing tail rotor to be governed by external governor
  • Tested on hardware, no flying so far
  • Interesting test setup - see video in PR
  • Bill will merge this when he is happy
  • Jumps in graph?
    • minimum setting on RPM affects this?
    • This is measured from real-world-test
      • Could be the ESC feedback being wonky
  • Rob uses idle for that quick-startup point
    • That quick-spool up works OK for Rob
    • Need to be careful here as it does lead to problematic situations
  • Landed means window isn’t used; it goes to idle

UTC1121 -

  • Copter velocity control method
  • Should be non-functional-change
  • Dt handling is slightly different
    • Tridge is happy to merge anyway
  • merged

UTC1122 -

  • GPIO reorganisations for in Linux API
  • Lucas?
    • Has done a quick review
      • Hasn’t been responded to yet
  • Constexpr on some boards x= sin(3) will work if the compiler can evaluate sin, but some can’t
    • Good otherwise as it will throw error if it can’t be calculated at compile time
  • Auto keywords?
    • We do use it elsewhere
    • Main advantage is you don’t need to work out exact type
      • Used for DIR structure in autofs to allow type to be private
    • Judicious use is not a bad thing

UTC1124 -

UTC1134 -

UTC1147 -

  • Merged!
  • More documentation!

UTC1148 -

  • Henry Plane tailsitter
  • Tailsitter transition to fixedwing jerks then ramps
  • Starts transition from current pitch angle
  • Doesn’t pitch to zero at start

UTC1151 -

  • Merged!

UTC1151 -

  • Look at vehicle initialised?
  • Need to make sure we don’t lose messages

UTC1154 -

  • Race condition on storage write
  • Sentinel can be corrupted in race condition
    • Could cause current parameter loss
  • More review would be good
  • Tricky thing at low level

UTC1157 -

  • Handle external nav resets
  • Reset messages now making it into EKF
  • And now they’re being handled
  • Before picture you see a reset
    • EKF chews for a while and then swallows it
  • After picture shows EKF doing hard reset straight away to new position
  • Sign seems correct
  • Resets can occur on a regular basis
  • Delayed reading
  • Would currently need to disable GPS to use vision as primary
  • Randy and Paul will chat
  • This PR is a fairly small use-case
    • This isn’t going to be used by thousands as soon as it is merged….

UTC0005 -

  • Fix if an axis is producing unrealistic values then we fail in its entirety
  • Paul is happy to close this out as “that’s the way it works”
  • Tridge is thinking minimum-changes to EK2 as well
  • Peter still to review

UTC0012 -

  • Operation with no or bad compass data

  • Couple of time-wrap issues

  • Randy’s found a bunch of existing overruns too


  • Want this in before next devcall

  • Subtracting of timestamps must be on same type to get wrap correct

  • Paul can patch this today

  • No ekf check stuff in quadplane

  • Tridge will merge this-arvo, hopefully

UTC0018 - scripting, add bindings for servo output

UTC0019 -

  • Correction for GPS antenna offset
  • Const-ifying two functions
  • Removing scope of calls
    • Not called when not using GPS?
  • Merged!

UTC0021 -

  • Compass calibration in separate thread
  • Will need to be rebased to check stack usage
    • Could notionally force-arm and fly with the thread still running
  • Needs to be valgrinded
  • Can’t be instant-reviewed
  • Tridge and MdB will both want to review

UTC0023 -

  • Moving-baseline yaw
  • MdB was concerned about averaging time
  • 245ms if you’re a Rover gps, 215 when not
  • Rtcm corrections down both ports is bad
    • But sometimes the correct thing
      • Need to look at load on ublox
      • Tridge to prove it is a problem before we attempt to fix it
      • MdB will see if he can find an easier solution

UTC0030 -

  • Add @PARAM and @SYS virtual filesystem
  • Partial parameter downloads
  • Fixed FTP server packet timeout
    • Losing reply to open you’d have to reboot
      • Terminate was refused

UTC0035 -

  • Needs all unrelated changes split out into separate PRs

UTC0043 -

  • If anyone steps close to copter then don’t allow it to arm
  • 0.6m ATM
  • Prearm distance to allow it to be changed
  • Randy was considering removing it completely
  • We don’t back up so it doesn’t too much
  • Alt-hold is a little more interesting
  • PH thinks it is good
  • Default the value to existing default?
  • Make it a script instead?
  • Randy will just rename message

UTC0045 -

  • General apathy…

UTC0048 - Plane update

  • Chasing FRAM reset thing

UTC0048 - Copter

UTC0050 - Rover

UTC0052 - GSoC

  • A bunch of good proposals
  • Some very good students
  • Can’t discuss details of applications because of Google’s rules
  • Evaluation closes later this-evening
  • Limited by number of mentors

UTC0053 -

  • Discovery of vehicle metadata?
  • This related to our new @SYS and @PARAM filesystenms
    • Default parameter files?
      • Has to come from within vehicle
    • Parameter caching is bad
      • Fast download of parameters via FTP
    • Unset would be nice
      • Spare bit in header
      • Garbage collection
  • Will be concentrating on @PARAM / @SYS stuff for the time being
  • Once it is merged will approach Don to see about QGC support

UTC0107 - close

Servers by jDrones