Dev Call Jan 20, 2020 2300 UTC

Issues & Pull Requests




1 Like

Attendee count (max): 22

UTC1105 -

  • Fixes for motor tests on tailsitters
  • Stops them overriding outputs when they shouldn’t
  • Also stops throttle stuck on on disarm
  • Outputs to motors all the time
  • This should maintain the way tilt motors work when disarmed

UTC1111 -

  • Fixes broken example
  • PH wants people to play with this
  • Might be nice to have a switch in a mission to determine if you can complete your mission without running out of battery (a condition)
  • Should probably be moved to be real code rather than a script
  • Merged because it’s a great example script
  • Logging?
    • WriteBlock
    • Varargs functions
    • Use LogWriteV instead?
    • Peter and tridge need to look through the logging
  • Need feedback on how good the prediction is food
  • Scripting on small boards….
    • Possibility without multiple EKFs in the code + RAM savings tridge and MdB are looking at.

UTC1118 -

  • Needs a review from MdB
  • Problematic pointer-comparison stuff
  • 64-bit-timestamp issues
  • Should convert to degrees
  • Emitting format for every message?
  • Would be much better to use WriteV

UTC1136 -

  • Trying to get rid of “Bad AHRS” on Plane
  • Extra fields in SYS_STATUS
  • Before AHRS initialises
  • Extensions to SYS_STATUS to indicate if sensors have initialised or not
  • Maybe set ENABLED to false until the sensor is enabled
  • Need to work out why Copter is different
  • Flying Planes without EKF being happy isn’t great, esp. If it starts to use EKF during takeoff
    • Don’t want to hide too much stuff
  • Use mav-sys-state to suppress ahrs message?
  • Use ekf flags to suppress message?
  • Step1 is to work out why the vehicles act differently

UTC1150 -

  • Tested since last week
  • Blheli fixes
  • Debug lines are a lot of flash space
  • Needs to test the client-side of the motor test
    • Set parameter, it runs the test, parameter gets reset

UTC1153 -

  • Using a suspiciously large amount of flash
  • All semaphores are now recursive
    • 4 bytes
    • About 10% more CPU vs non-recursive
    • Now using ChibiOS semaphore implementation

UTC1157 -

  • Don’t go back to DO_LAND_START if continuing would be shorter
  • Edge-case central
  • Will revert to current behaviour in the case that the normal path is complicated by conditionals and whatnot
  • If you are on final it may choose to go straight ahead
  • Tridge is worried about oscillation between different solutions
    • Once failsafe is initiated calculation function is only run once
    • Would have to have a second failsafe
  • Distance-remaining in mission vs distance to do-land-start
  • If-statement for lat,lon 0,0 incorrect
  • [11:09 AM] (Channel) HWurzburg: why not just say if past do land start and no jumps have occurred, continue landing? all this complexity seems overkill
  • [11:11 AM] To Weekly devcall: @Henry Matt did point out you might not have entered your DO_LAND_START but still be on final approach.
  • [11:12 AM] To Weekly devcall: … which IMO could be a “Don’t do that, then” kind of thing.
  • Jumps are handled, so relatively simple missions could benefit from this

UTC0019 -

  • Wing sails for Rovers
    • Needs review
  • Would like some feedback from more users of wingsails, and htis PR will let people use it more easily
  • RealFlight can’t simulate wingsails, sadly
  • Randy will look at it

UTC0021 -

  • Check return code!
  • merged

UTC0022 -

  • Check firmware size
  • Check storage is retained
  • Flight testing on Copter and Plane
  • More comments!

UTC0036 -

  • Merged!
  • Now “Initialising ArduPilot”, not “Initialising APM:

UTC0038 -

  • PX4Flow changes
  • Checks all i2c buses rather than just the external
    • No ID, just do you get a byte back
  • Hex Cube has used internal bus as external bus
    • Which is why this is required
  • This fixes a regression
  • #ifdef for CubeXXXXX?
  • Probing for i2c devices has destroyed machines
    • E.g. thinkpad
    • Randomly probing internal devices could kill boards
    • Randy will add the pragmas

UTC0045 -

  • Configure line in Waf?
  • Multiple builds
  • How does the IO firmware do it?
  • Randy would prefer disable one EKF or the other rather than provide multiple EKFs
  • merged!

UTC0053 -

  • Need to add a test for the frsky text message passthrough
  • Message to gcs maintainers list

UTC0059 -

  • Copter only?
    • Can’t know if vehicle is in GCS failsafe.
  • Should be done on all vehicles, but this works as is
  • New GCS failsafes are wonderful, too!

UTC0106 - Plane update

  • Plane 4.0 release
  • 4.0.3 out
  • Bounce-buffer allocation issue fixed
    • Log full of copies of the first bit of the bootloader…
    • Stale pointer used
  • 4.0.4 should be resynchronisation release to get closer to Copter

UTC0108 -

  • 4.0.1Rc3 is out now
  • Fixes as for Plane
  • Some support calls have come up this-morning
  • Emlid CAN compass not found with Cube Orange
    • More CAN initialisation issues
    • Can be fixed with the boot-delay parameter
  • RC failsafe and S.BUS issues
  • Windows driver updates killed a whole bunch of joystick inputs
    • Particularly in Windows
    • Right-click-rollback-driver fixes it
  • Autotune regression

UTC0107 -

  • Not much to report on Rover

UTC0117 - Close

1 Like