Servers by jDrones

Dev Call Aug 12, 2019 2300 UTC

(Craig Elder) #1


Issues & Pull Requests




(Craig Elder) #2

Attendee count (max): 16

UTC2304 - GSoC update - Peter Hall

  • Matt (autorotation)
    • Trying to get a PR onto master later this week
  • Trying to get motor-sailing into master
    • Motorboat with a sail or a sailboat with a motor…
    • All working quite well
  • Path planning
    • Mission would have a takeoff sort of thing
    • Path planning waypoint
    • Then 20 waypoints
    • Then a stop-path-planning waypoint
    • Visit all but will decide order itself
    • A-star?
    • No, basic algorithms
      • Greedy
      • Two-opt
        • Swap two waypoints and try again
      • Three-opt etc
    • Put algorithms in one place? E.g. Dijkstra
      • Something on the same model as the SmartRTL stuff?

UTC2308 -

  • Rangefinder subgroup renaming
  • Parameters are present twice in our documentation
  • Probably just affects our Wiki
    • And anybody using our parameters xml
  • Generator emits each group separately
    • When should it combine?
  • Michael and Peter to look at it

UTC2318 -

UTC2317 -

  • Does this take into account the RC failsafe counter?
  • [9:24 AM] (Channel) Jaime: You are discussing a higher level change. This change only parametrizes a hardcoded value. That’s all we need for now. We can change all this later.
  • Different decisions made for Plane vs Copter in terms of assumption of RC input during RC-failsafe-pending periods
  • Will need to be able to simulate and test RC
  • FS_RC_TIMEOUT rather than something to do with throttle?
  • Same timeout for mavlink overrides and normal RC?
    • They come in at different rates
  • This patch adds another layer of failsafes?
    • Another time for the rc failsafe to trigger?
    • similar parameter name we use in Copter for the failsafe value
  • [9:38 AM] (Channel) Jaime: The name can be changed, the timeout itself has been there for a long time. So we are not adding any new timer.
  • Mask for never-time-out RC values
    • Same as the thing being added?

UTC2345 -

  • Merged
  • Old compiler?
    • Some vehicles are building with PX4/NuttX builds
    • But this is the native compiler
    • We release with 4.9
    • Once we’ve no releases running on NuttX we can move forward
    • Do we keep aligning SITL compiler with release compiler?
    • Build_binaries could use different compilers for stable vs master etc?

UTC2346 -

  • Turning arming checks off in Plane - what if AHRS is not initialised?
    • Get_position still returns true when it isn’t using the GPS
      • DCM pass-through
  • Merged

UTC0002 -

  • Adds object avoidance support to Copter
    • As in navigating around things
    • Bendyruler and dijkstra
    • Tested on real vehicle many weeks ago
  • Integration is a little different to Rover
    • Wanted to be able to turn it off on fmuv2 boards
    • AC_WPNav OA library
      • Thing wrapper around AC_WPNav which calls over to OA avoidance path planner library so if there’s an intermediate point it shoves that into the underlying AC_WPNav library

UTC0004 -

  • Add support for Scrimmage
  • Adds ability to pass in a config for the model
    • May be able to get rid of (e.g.) plane-elevon
  • FF has a change requested
    • Tridge thinks this was resolved
  • Merged!

UTC0007 -

  • Matt had consistent reset on F405 wing
    • Unable to reproduce now
    • Tridge is working with him to try to reproduce it
  • Devcall should remain as this is an important thing

UTC0009 -

  • Needs to be tested under Valgrind
  • Can be merged after more tests are run

UTC0020 -

  • Need to make sure fetching parameters doesn’t interrupt ESC and servo values
  • Disable it when armed
  • Move to separate thread?
  • Need a get_semaphore method on the parameter
  • May end up corrupting parameters?
    • Declare a hal-semaphore and do a with-semaphore
    • A set should be OK
    • Pre-existing issue with set_and_save vs mavlink?
      • No thread for mavlink
    • Param_write_queue race
    • Multiple writers into queue needs locking against
    • [10:35 AM] (Channel) Michael_duBreuil: We’re proud of you, and supporting your decision to take this on at this time Peter!
    • Locking version of the buffer for writes?
    • Pushing into the buffer is done inside AP_Param
    • Save_queue_push
    • And in flush
    • ObjectBuffer -> locking object buffer

UTC0038 -

  • No work’s been done in this one
  • Maybe we’re filling up a serial buffer?

UTC0039 -

  • Probably doesn’t need to be devcall
  • Tridge can test it with type 3 or type 4
  • Shouldn’t autoprobe any more

UTC0041 - Baro-related PR

UTC0043 - Plane update

  • Anomaly
    • 120ms motor outage in quadplane

UTC0045 - Randy and copter

UTC0106 - discussion of exclusion zones

UTC0109 - stay-in-booloader not working?

  • Need more information on which bootloader etc

UTC0110 - End