Dev Call May 21, 2018 2300 UTC

Issues & Pull requests



  • F7
  • Fixed dhsot for fixed wing
  • ChibiOS port to NXP


  • Plane 3.9.0 Beta released


  • Copter 3.6 release update - want to release -rc2 in the next week or so


  • Rover-3.3 release update
  • Various bugs uncovered with acceleration limiting… being fixed
  • will push out -rc3 in the next week

Attendee count (max): 24

UTC2300 - timer registration interface

  • One timer suspension cal remaining to be dealt with
  • Tridge wants to go through it carefully
  • Pretty sure not new bugs introduced
  • Fixed on bug in the analogue interface
  • Has it exposed any other bugs?
    • E.g. missing semaphores
  • Hopefully Tridge will look at this later today or tomorrow
  • Semaphores around proc registration did lock the thing up
    • Should only be registering io processes on the main thread
    • So how do you get a lockup?
    • ChibiOS
      • Tridge will try to replicate
  • Resume_timer_procs has to die
    • It was ugly!
  • @tridge: This is the one with locking around process registration bugs in implementation are ofc possible

UTC2306 -

  • Discussed last week.
  • Need a PR to see something to discuss

UTC2306 - UAVCan message corruption

  • Default output mask changed to zero
  • Users will need to set the mask
  • Will minimise bandwidth used
    • Bandaid fix
    • Not satisfactory as longer-term solution
  • Tridge’s babel board didn’t arrive
  • Pavel is organising another one
  • Tridge is hoping to reproduce it locally
  • Francisco has been looking at this and finding other bugs
    • Bug in pyuavcan fixed
      • Even if you only used one port in AP you’d see errors - now fixed
    • Still a bug in ArduPilot
      • If you use two ports
      • Picking the cube up causes the problems?!
      • Related to when more messages in bus
      • Do we not send values if it’s the same as the previous value?
        • Relies on hal-px4 rcoutput driver not updating the PWM value
          • Need_update flag
    • Testing: remove check for need_update so it always sends outputs to can
      • So you get worst-case output
      • No need to move Cube then
      • Need to calculate total bandwidth used
      • Check bus utilisation with a scope?
  • CAN support in SITL would be great
    • Using real CAN devices

UTC2312 - deprecate old mavlink messages

  • Ongoing….
  • BATTERY_STATUS support is in QGC now, so battery2 isn’t needed there
  • MdB needs to chat with Jacob offline to figure out what is the deal with rangefinder messages for sub
  • MdB wants a kill date!

UTC2315 - review requests from Pierre

  • Would like his requests reviewed

UTC2320 - GSoC

  • Falling a bit behind in hardware
  • Ebin’s started to look at things in earnest
    • Brought simulator into master
    • Starting to look at controller
    • Found interesting stuff that jon’s been working on for simulation
  • Legal woes with University continue for one our our students

UTC2324 - Plane 3.9beta went out

  • yesterday….

UTC2326 - d-shot for fixed-wing forward-motor

  • fixed!
  • Only worked for quadplane
    • Vfr_hud and throttle absolute value
    • Because that’s what the spec says
  • Throttle percentage calculation bug was wrong yesterday
    • Fixed now in master and 3.9
    • Could have affected the attitude of the aircraft
      • Low-throttle pitch-down
      • Dual-motor-planes didn’t do this properly…

UTC2232 - NXP processor port

UTC2333 - HW updates - Phillip

UTC2343 - Plane 3.9 redux

  • Sparse release notes
    • Tridge will be doing better over the next couple of days
      • Including how-to-use-ChibiOS

UTC2345 - Randy and Copter

  • New RC out in next week or so
  • Wants to include work from tridge and Peter on Vision Position Estimate timing

UTC2345 - Randy and Rover

  • 3.3 is out there
  • We’re testing 3.3.1 (or is it 3.3.2?)
  • Zero-point-turning-tractor is showing up bugs
  • Acceleration-limiting isn’t working
  • Hitting waypoint jams velocity to zero
  • Sailboat support starting in next few months
    • Wind sensor from Philip, possibly
      • Ultrasonics on Solo were considered for wind measurement
      • AoA sensor
      • Ultrasonic wind?
        • Spinning-cups, probably
    • Boats with wings?
    • Tridge points out there’s a bit of autonomous sailboat code out there
      • Bit of a sport, apparently
    • Dev conference in China has lots of water!
      • Randy will be taking a “crazy-fast” boat, hopefully
  • Mass-flow-sensors airspeed sensors aren’t working well on Plane ATM for MdB

UTC2358 - MP and new interface issues

  • Randy will get something to MO
  • Param name discrepancies
  • Do we need a mapping for MP version to AP version?
    • Lots of work done in MP to avoid this