Dev Call Apr 08 2019 2300 UTC

** GSoc & GDoc**

  • Student submission Deadline

Issues & Pull Requests


  • I love onions!



Attendee count (max): 16

UTC2305 - GSoC

  • Submissions due today!

  • Many, many vision-related projects

    • Just Jaime for mentoring on these, really?

    • Dan Pollock?

    • Patrick

  • Some high-scored submissions haven’t been finalised

UTC2308 -

  • Removing stale binaries

  • Sub beta is now based on something that doesn’t have

    • New Sub is on ChibiOS!
  • Tridge removed the beta sub directory

UTC2313 -

  • Conflicts

  • Our conflicts might be due to repository setting change

  • Discussed the “Rebasing the commits of this branch on top of the base branch cannot be performed automatically as this would create a different result than a regular merge. “ problem

  • Peter’s rebase scripts to go into repo

  • Should this be get_offset_ned?

    • Make sure operations are correct way around
  • Make sure arguments are correct way around?

    • Tridge thinks it is
  • Can be merged as soon as it is deconflicted

UTC2322 -

  • No throttle min any more in fbwa?

    • Contrary to docs
  • Non-manual throttle modes are affected

  • thr_pass_stab will make throttle pass-through

  • Straight-forward fix says Michael

  • Pete will put something together

UTC2326 -

  • Argument made to move logging into libraries generally accepted

    • allows more stuff to be kept private

UTC2326 -

  • AP_RSSI: make type enum class, remove default clause in type switch

  • merged!

UTC2326 -

  • Use singleton for RSSI logging

  • merged!

UTC2334 -

  • Rover shouldn’t be doing special RangeFinder logging

  • Will need to log avoidance values in a separate log message

  • Should stop using backends directly in logging

  • Good example of why logging should be in Rangefinder

  • Can be merged once it is ready

UTC2336 -

  • Spammed messages

  • Added a timer to limit message rate

    • Should be a 1 second timeout, not 5
  • Certainly an improvement

  • Refactoring, however?

    • Needs testing on hardware, has this been done?

    • Randy might pluck in the rate-limiting first

UTC2338 -

  • Scan past end of eeprom

    • Initialiser ordering

    • Constructors are the wrong place to be doing parameters

  • Going to make scan past end of eeprom fatal in SITL

    • Removing eeprom.bin will probably emit this, ‘though
  • Could add a check in the public interfaces on AP_Param to initialise itself

UTC2345 -

  • Problem with None for firmware file to upload

  • F4 vs H7 vs F7 issues

  • Just a tool to help people know which board they are running

  • MP does it implicitly as part of firmware upload?

  • Should we really just print the flash space?

  • Extend meminfo to include amount of flash space available

    • Has the drawback we’ll send it forever more
  • Include it in ardupilot_version

  • Can go in once the bugs are fixed

UTC2355 -

  • Accessors should be floats

    • Updated at runtime based on what the sensor can do
  • Tridge is reluctant to add the floating point designators

  • Tom wants the designators

  • [10:01 AM] (Channel) MdB: I like them, but I’m demented like that :smiley:

  • This PR can go in

  • Randy likes the f-designators

  • We should add the compiler flags for floating-point constants everywhere if we are going to use it

UTC0000 -

  • We could use a static-assert

    • But too much work for not much gain
  • Fixes a bug if you were to allocate way too much space for rally points?

UTC0007 -

UTC0010 -

  • Shows a basic problem with C++/C

    • Checking arguments or checking results

    • Is_positive etc don’t work

      • Or at least not sufficient
  • Should we move to check-the-outputs rather than check-the-outputs?

  • MdB is in favour

    • Constrain_value could take optional arguments to determine what happens if things are out of bounds….
  • Can go in

UTC0018 -

  • Merged!

  • Another one from the new compiler warning

UTC0021 -

  • Fixed throttle curve for qstabilize mode

  • Good for tailsitters and regular quadplanes

  • Qacro affected by new parameter

  • Rudder is currently pass-through on qacro

  • Problem with trim and min for throttle found

    • Should use range rather than 100 for division (so we can fix the 1% resolution issue in the future)
  • Tridge would like to have expos for manual etc

  • Heli spline code could be instructive

UTC0034 -

UTC0035 - SITL add Copter tailsitter


  • Moving towards getting these into autotune

  • Arming checks 0?

    • Ekf warnings otherwise
  • Spinning around without the ground behaviour?!

    • That’s not good

    • Was put down to euler angle funny stuff

    • Probably should be tracked down!

  • Why dcm-to-euler when not using results?

  • Non-copter tailsitter did work at one stage

UTC0042 -

  • Arming check failure found and fixed

  • Any more changes needed?

  • Rover and Copter now print messages slightly differently

  • Possible we might not get updated strings from AHRS?

  • Merged

UTC0053 -

  • Add heap for scripting

  • Lucas doesn’t like constrained heap at all for Linux

    • Just use the system realloc
  • Original idea was to limit the memory available to scripting

  • Some Linux boards have very little memory

    • Would want to constrain here

    • Tridge thinks we should constrain

  • Scripting has been tested on Linux

  • Need response from Lucas

UTC0053 -

  • Tridge is happy for this to come in

  • Off by default

  • Would have been good to get some review…

UTC0057 -

  • Can be merged after it passes CI

UTC0103 -

  • Fix condition-yaw in auto mode

  • Does it cope with not having a set-yaw?

  • Randy test

UTC0107 -

  • Randy and peter to discuss

UTC0113 -

  • Fixes bug in TECS which tridge found recently

  • Conflicts with in-tree fixes

  • Merging this on time would have saved at least one aircraft

  • Low-pass filter on tecs dem

  • This would need some attention from Paul as well

  • Tridge will try to rebase this

UTC0115 - onion update

  • It’s merged, great!

  • Fixed on fmuv2

    • And fmuv2 added to CI

UTC0118 - Copter

  • Has notch filter been tested and known to work

  • Yes

  • Fft data presented in MP is pre-any-filtering

  • Some people’s vehicles are flipping when using notch filter

  • Don’t put your notch filter near your control frequencies!

  • Lots of notch filters in the future?

  • More boards coming

  • CXOF support coming on lower-end boards

  • Buzzer-climb problem is being looked at again

UTC0125 - ESP32

  • Alex got it to the point of booting + HAL

    • SPI working but i2c not
  • Now we have rcin and rcout and i2c

  • … and uarts (gps etc)

  • We’re getting a mavlink stream over tcp over wifi

  • Ready to fly in next day or two

  • A week or two for a pull request to bring it into master