Dev Call Sept 21 2020

Issues & Pull Requests

Crash Report GCS Failsafe




1 Like

Attendee count (max): 24

UTC1100 -

  • Check motors initialises properly on quadplane….
    • Half-initialised-motors backend on tricopters…
    • left/right tilts mean servo isn’t required but we were skipping out early due to lack of servo
    • Reverse-tricopter frame type
    • Add-motor-num (which usually doesn’t matter)
    • And set-angle
    • And pitch-reversed for plus-rev
    • And initialised flag
  • Also fixes ability to sneak past initialisation with bad config on Copter too
  • Sets us up for a future change without tilt and with the tri
  • Prearms will fail if motors fail to initialise.
  • Copters could get tilt-for-yaw…
    • Not supported ATM
  • Need to avoid setting ourselves up for failure
    • Set_class_and_type
      • Change these at runtime….
      • Will initialise after these are set
    • Reboot required after setting q-enable
  • Merged!

UTC1112 -

  • Array-indexing for fields in mavlink
  • Three related PRs
  • No more _1 etc in tab-completion
  • Can graph all your battery status at the same time live and in post-analysis
  • Merged!

UTC1119 -

  • Need to use compile-time check to get rid of code
  • Still need to do nullptr check in case adsb doesn’t initialise
  • Still have double-include
  • A few messy bits highlighted by Tom
  • Mark up features in the same we mark up parameters
    • Add comment // @Feature - what you need to enable to get the feature and approximate cost
    • E.g. //@Feature ADSB
    • In ADSB.h near the #ifndef define you would have // Enable ADSB_ENABLED and // FlashSize 1.2k
    • Defaults?
      • These are per-board
  • Remove mode in Plane if ADSB not enabled?
  • merged!

UTC1134 -

  • Do we squelch using compiler pragmas or fix the problem?
  • Getting a uint32_t equal to uint32_t max could be a problem (probably get an unable-to-coerce) as it isn’t representable
  • Merged!
  • Removed a const?

UTC1136 -

  • Tridge is concerned by having the symbol twice?
  • Should this be in AP_Vehicle instead?
  • Do we need the vehicle.cpp?
  • Commit-list problem - cross-directory stuff going on
  • Reviews left

UTC1145 -

  • Rebase then merge

UTC1147 -

  • Well-tested
  • Merged
  • Allows i2c in place of uart

UTC1148 -

  • “ArduPilot ready!”
  • We could emit a message first time pre-arm checks pass?
  • GCSs should be using the prearm sys-status bit
  • Merged!
  • Message isn’t really all that good - “ArduPilot initialised”, perhaps?

UTC1152 -

  • Descriptions for bitmask entries
  • Will break MAVProxy at least
  • MdB really just didn’t want spaces in the short descriptions
  • Extend markup to allow for longer descriptions?
  • One-bit-per-line would be nice!
  • @Values[3] would also be cool
  • Henry will change this PR to go with 1-2-or-3 word shortening instead

UTC0003 -

  • Merged!

UTC0005 - what happens if you get a 2 or 4 GB file?

  • 4GB is the limit
    • Uint32_t offsets
  • An option to rotate in flight would be nice
  • Andy already had something like that….
  • Apparently it just stops as it should at the max size

UTC0008 - rotor change when upgrading from 3.6 to 4.0?

  • GCS failsafe in pilot-controlled-mode would invoke RTL
  • Bill doesn’t think the autopilot should take control from a pilot-controlled mode
  • GCS failsafe is off by default
    • But if you do enable it then it is enabled for pilot-controlled modes
  • MdB requires current behaviour - i.e. GCS failsafe in CRUISE
    • Flying beyond RC failsafe with the option but set then requires GCS failsafe
  • Tridge would like people not to fly with RC failsafe turned off
    • Don’t want a failsafe action to kick in
    • Continues to use RC inputs while they are invalid
  • Tridge wants to couple RC failsafe and GCS failsafe and change behaviour of GCS failsafe based on RC failsafe
  • General rule - there should always be some sort of failsafe enabled
    • If rc failsafe is off then gcs failsafe should change in pilot-controlled modes
      • Otherwise we’re not going to get sensible behaviour
  • Randy’s concerned that if someone is in a pilot-controlled mode
  • and has disabled rc failsafe

  • and is relying in GCS failsafe to control vehicle (e.g. joystick)

  • short timeouts are problematic ATM

  • hard-coded timeouts

  • add a timeout parameter?

  • initial change to change default value of the bit which Bill is worried about

  • changing a default might stuff people up

  • add another fs-option bit to control new behaviour?

  • change FS_OPTIONS to set it to include “4”

  • will suddenly change for people

  • Should point people to FS_OPTIONS from FS_GCS_ENABLE and FS_THR_ENABLE value 2 rather than(Deprecated in 4.0+)

UTC0024 - Plane update

  • PR in place to use AC_PID
    • Same outputs so far, comparing old and new!
    • Copter would be RAT_RLL_P
    • On the basis you tune the rate controllers first
      • Then angles
      • Then nav
  • Want to preserve rmax etc
  • Would prefer to stay away from ATC_ to avoid confusion with Q_A_ parameters
  • Q_A is equivalent to Copter’s ATC
  • Do we actually want a single object?
    • The different axes behave very differently on Plane!
  • RLL_ allows parameters to retain existing names
  • Does this undo random old PID remapping?
    • Yes
  • Plane will be tuned with FF, P and I
    • Uses old code to split into FF and P
  • Tridge is thinking of using component filtering in AC_PID
  • Axes over layer (again)
    • Prefix on front to keep them grouped together
  • There are other changes
    • Time constant vs p-gain
      • Rll2srv_t_const etc

UTC0051 - Copter update

  • 4.0.4 is out!
    • Thanks everyone.
    • Huge release
    • Smooth so far
    • MatekF405 wing error messages
    • Some question about the Hitek batteries
    • Devteam tends to fly master….
  • S-curve stuff is coming along!
    • PR at end of next week with any luck
    • For straight-line and spline-waypoints
      • 95% of our users probably just use straight-lines
      • Users of spline waypoints might not like new implementation
        • Discussions on making it simpler and splinier
        • Strings together a greater number of straight-line segments
          • Add more of these straight-line segments

UTC0055 - Rover update

  • No update
  • Want s-curve for Rover, too!

UTC0055 - Sub update

  • 4.0.2 coming!
    • Next 24 hours or so
    • Just one patch to fix prearm internal error
    • Adds expect-delay to gcs common
  • 4.1?
    • Make a static tag for beta testing the ddl integrations?
    • What should 4.1 wait on?
    • EK3 is the new default…
      • Probably not an issue for Sub
    • Branch in ArduPilot repo…
      • Sub-4.1
  • A lot of people fly Copter and Plane master
    • Autotest and a LOT of users make it pretty stable
  • S-curve stuff gets applied to Sub as well
  • Current focus is to support positioning-enable mode
  • Would be great to see the simulator that Willian did merged into master!
    • Which backend was it using?
    • Something to show at tradeshows
    • Training programs

UTC0105 - close

1 Like