Dev Call May 25, 2020

Issues & Pull Requests




1 Like

ArduPilot Dev Call 2020-05-25


  • Need to modify the website to point to Discord and not Mumble

Attendee count (25):

UTC1100 -

  • Now tested for traditional quad planes
  • Merged!

UTC1101 -

  • No need to make things too glibc-like
  • Need to fix the comment

UTC1109 -

  • Merged

UTC1109 -

  • Need to move into cpp file
  • Comment to be cleaned up
  • Can be merged after that

UTC1116 -

  • Leeddarone driver has a slight change
  • Merged!
  • Cleans up discarding of bytes

UTC1121 -

  • GPS lagged by 2 seconds
  • Too much data
  • 115kbaud is too little
  • Probably filling internal buffer
  • Move to 230kB

UTC1122 -

  • Just remove the uninitialised field

UTC1129 -

  • Add unused LED PIN output as GPIO
  • 57?
  • 50 to 66 as normal range
  • So you can get to normal aux channels
  • Make it 70 or 80?

UTC1131 -

  • Can get $0 off script
  • Can be merged after that

UTC1134 -

  • Visual speed estimates!
  • Need to get chobitsfan’s stuff in first
  • Ekf3 support
  • SITL features to add velocity glitches
  • Select in real time which messages are sent
    • Tmask parameter
  • Sending both vision position and vision speed estimate
  • Remote-time-us is timestamp from remote system
  • Time-millis is when it arrives locally
  • New maths thing
    • Get vel correction for sensor offset
  • Flight test video
  • More Time wrapping issues
  • Gps horizontal velocity noise
    • Precludes using external nav and gps velocity tuning at same time
    • Need an enum class for the EKF
  • Boolean value for ready-to-use-extnav when you’ve got pos and vel?
    • Currently two booleans rather than an enum
  • Should be able to observe the scale factor by using the IMUs
  • Separate parameter table for SIM_Vicon would be nice
  • Velreset thing does seem to be an issue
    • Need to check if we’ve ever seen vel data
  • Should now get similar performance as external-gps-and-external-yaw
  • Chobitsfan’s work generally
    • Polish by Randy

UTC1156 -

  • Thread-switching can cause receiver to get out-of-sync
  • Downlink and uplink could share code

UTC0002 -

  • Runtime test GND_ options parameter
    • Allow init with no barometer
    • Runtime not required for this PR
  • Merged

UTC0005 -

  • Merged

UTC0005 -

  • Alex has been flying this a lot
    • No crashes!
    • Lucky as there’s a bug
    • Significant bug in this PR - do NOT fly it
  • Metadata is downloaded from servers
  • User then chooses which parameters to modify before going to the field
    • Empty uartdriver pointer on stack is bad
    • protected/private methods
    • Two constructors for gcs-mavlink-frsky?
    • Pointer-to-driver
  • Needs a rebase

UTC0013 - Plane 4.0 release

  • 4.0.6 beta is close
    • Wants a test flight
    • Henry will do one
    • Backport risks…
    • Randy’s been working on similar release notes for Copter
    • Major change from 4.0.5
    • Brings back compass ordering stuff
    • Rtl-climb-min
    • New boards
    • HOTT telemetry
    • Subset of LUA bindings
      • Thread-pulling didn’t extend to the LUA bindings
      • “More” is unclear in the release notes
    • Biggest behaviour change is that we compensate VTOL gains for pressure alt
      • Like Copter
      • Only matters at unreasonable altitudes
      • SITL now takes into account
        • Have alt ceiling
          • 5.4km
            • Runs out of oomph there
      • Denver, Colorado?
        • 1.6km
        • 3 or 4% change
      • Buzz Today at 10:25 AM: i did the maths once, and it worked out that adding an inch to the propsize per 1000ft alt was about right… so a copter on the top of everest would need props about 8" bigger than they started with.
      • … in theory.
  • Need to warn people they may lose their telemetry data (so if they can’t refetch at field that might be a problem)
  • In-flight reboot of IOMCU
    • Power on the IOMCU should be more resilient than the FMU
    • 4.0V?
    • Anything below 4.7 is potential risk
    • … actually 3.5/4.0V batch-to-batch
    • No brownouts enabled on stm32 so things get corrupted rather than resetting
  • Probably 4.1 for ftp
  • No configuration requests understood by our simulated uBlox
  • Flight testing to be done
  • 3 weeks of beta
  • Travis is working again on Copter stable

UTC0032 -

  • M9 from mRo is on the way
  • Not multi-band
  • F9-ey configuration stuff
  • So moving baseline
  • Faster cold-startup
  • More satellites so more constellations
  • Doesn’t need a fancy antenna
  • M9 is F9 without multiband?
  • Unit cost of GPS is similar to M8, so hopefully we’ll see lots of M9s!
  • Huge groundplane?
    • Antenna performance will drop significantly
  • M8Q / M9Q
    • Tiny, tiny ones with essentially no baseplane
  • See if Matek produces a bunch of small M9 boards?
  • S.curve stuff is basically working now
    • Corners are nice and consistent
      • Very different from leash method
      • Leash length change is the confusing thing about current code
      • Splines are not as good as they could be
    • Needs thorough review
    • Style issues
    • PR in a couple of weeks

UTC0043 -

  • No Rover update ATM

UTC0043 - GSoC update?

  • Randy’s connected with his two students
  • Blog post for avoidance is up
  • Ashvath walking robot block on the way
  • Pete’s MatLab stuff is coming along nicely
    • 2000fps
  • Physics of vehicle in matlab then fly it in ArduPilot SITL
  • Some progress on Harshitt with PaulR
    • PR in next week or so
  • Thienn’s working on the vision stuff
    • Hasn’t really stopped working since last week!
    • Passing confidence back into ArduPilot
    • 435 support

UTC0048 - close

1 Like