Dev Call May 18, 2020

Development Team
1

Issues & Pull requests

Plane

Copter

Rover

1 Like
2

Attendee count (max): 25

UTC1100 - https://github.com/ArduPilot/ardupilot/pull/14382

  • Align GCS PID with onboard log PID values
    • Hopefully the onboard logs are correct….
      • Randy is less sure
      • Tridge is pretty sure
  • Target in degrees, actual in radians or whatnot
    • Fixed!
  • Could fix the mavlink message to remove the units
  • Note any external tool will break horribly with this
  • Rover users use it for real-time tuning
  • Passing feed-forward through?
    • Combination of cruise speed and throttle
      • Will always be zero anyway
  • Randy would like validation of the values
    • The vehicle’s actual speed should come out
    • Pete will look

UTC1104 - https://github.com/ArduPilot/ardupilot/pull/13315

  • Scripting stuff
  • Add support for dataflash logging
  • MdB thinks it’s fine
  • Tridge thinks it’s fine
  • merged!

UTC1109 - https://github.com/ArduPilot/ardupilot/pull/13155

  • Qassist switch logic change for tailsitters
  • Retested and rebased in SITL
  • Enables qassist for tailsitters
  • Copter tailsitter can use assist
  • Control-surface-less planes can work with this patch
    • Not really a good idea
  • Could this be detrimental for an elevon-style tailsitter
    • Yes, this is bad
    • Turns it off for elevator-only tailsitter
    • Don’t want to break normal tailsitters
  • Testing?
    • Normal quadplanes untested apart from sitl
    • Should test Convergence in RealFlight
  • Tridge wants no-behaviour-change for existing quadplanes
    • That is the intention of this
      • Need an airspeed reading ATM
      • Can enter it now
      • Sensor vs reading?
        • We always have a reading if you’ve got GPS lock
    • Apart from being able to force a qassist
    • More RealFlight testing required
  • [9:16 AM] (Channel) PH: Qflip ?
  • Henry’s flown this a bit

UTC1116 - https://github.com/ArduPilot/ardupilot/issues/14342

  • PR came in from MdB a while ago
  • Back-and-forth on this a bit on the PR, trying to work out what should happen
  • Outside rally-limit-km for both home and rally, do you go home or to rally?
    • MdB thinks closest
  • Few rally point users for Copter
  • Angle between them is important?
  • What would the user expect?
  • New rule: Rally_limit_km rules say that home is a rally if you’re outside that???
  • Tridge thinks we only use home if there are no rallys within rally-limit-km
    • What Randy said
    • BUT we should match what the documentation says!
  • This behaviour changed with a commit from MdB
  • Documentation
  • MdB did have a patch to change it but he lost it
  • ATM if no rally point within limit distance then we will always pick home
  • If you are 50km from home?
    • Places I planned?
  • What if home was a boat?
  • Rally-limit was just to make sure you didn’t use rally points from previous flying location
  • MdB likes the documentation description
  • [9:28 AM] To Weekly devcall: I feel a RALLY_OPTIONS parameter coming up RSN :stuck_out_tongue_winking_eye:
  • Docs will be updated to be accurate
  • Third-enum value on INCL_HOME might happen
  • Randy and MdB don’t care too much about this
    • So we will maintain existing 4.0.x behaviour

UTC1130 - https://github.com/ArduPilot/ardupilot/pull/14366

  • Servo-channel-override timeout
  • Set PWM on a channel from a script?
    • Temporary override elevator or throttle, for example
    • Counter?
    • Script to say, “I want to take over channel for certain number of loops.
    • Ms instead?
      • Main loop rate changes?
      • Will only matter if script stops functioning
      • Ask what loop rate is and convert to a counter?
      • After timeout something happens?
    • Start time / duration time
    • Wrap?
      • Subtractions over long period of time?
        • Limit it?
        • Could it be hours
  • Remap function on servo?
    • What if script dies?
      • What swaps back to normal function?
  • With small time limit
  • API in ms, implementation in counters?
  • MdB: scripts should not control servos directly
    • Script misbehaving or dying?
      • Concerned about both
      • Tricks flight mode?
  • Flip mode in Copter as a script?
  • Knife-edge in Plane as a flight mode?
  • Rolling loops
  • Stalling the aircraft
  • Spins
  • Randy thinks you’d use guided mode with an attitude controller
  • Scripts can kill an aircraft
  • Scripts for things we don’t want to put into mainline
    • E.g. boost throttle in Copter should have been a script
  • We should allow the ludicrous thing
  • Scripts can already directly control servos
    • But can only do it on functions not being used by the main flight code
  • Randy’s not against this PR, he just thinks GUIDED is the better way of doing it
  • API time will be in ms
  • MdB thinks we should keep people sandboxed
  • PH thinks we should limit it to 1s
    • Tridge doesn’t think so
  • [9:54 AM] (Channel) MdB: delibrate crash just means muck with the params lol :slight_smile:
  • Tridge doesn’t want to try to clean up from scripts
    • So no zeroing counters if the script stops running

UTC0000 - https://github.com/ArduPilot/ardupilot/pull/14390

  • Mavlink channels with overrides
  • Aion having problems with pitch/yaw sticks
  • Manual_control message
  • Channels with no overrides we ask hal for rc input
    • Badly defined if you’ve never had any RC input
      • Seeing odd -392 value
  • Maybe have a flag in rc input saying if we’ve ever seen RC input frames
  • Have SITL go via our RC protocol library?
    • In_trim_dz and get_control_in?!
      • Initial fix radio-in vs control-in
        • NFC getting rid of a crap function
  • Has-had-rc-receiver is the substantial change
  • Ap.rc_receiver_present could move up?
    • No, that’s different
  • Different paths for manual_override vs rc_override?
    • They both map into setting an rc override on a PWM override in the library
  • Looks correct to tridge
  • Needs testing
  • Zero pulses?
    • Failsafe state, maybe?

UTC0010 - https://github.com/ArduPilot/ardupilot/pull/14225

  • Debug EKF
  • “Are we in a debug build” rather than specific to EKF_NO_O2?
  • Ifndef debug_build
  • The pragma-optimize-o2 is in lots of places

UTC0012 - https://github.com/ArduPilot/ardupilot/pull/13921

  • Peter to run through Replay to ensure same results across logs

UTC0015 - Plane 4.0 release

  • Rm3100 driver problems
    • SPI works fine
    • I2c doesn’t work correctly
    • Scale factor problems
    • Maybe it’s a power supply issue?
  • Not a lot happening
  • Many patches into what will become a beta release
  • RTL_CLIMB_MIN coming
    • In metres
    • Tested by Henry

UTC0023 - https://github.com/ArduPilot/ardupilot/issues/12547

  • Copter
  • No beta release for 4.0.x
  • Need to get CI and stuff working
  • T265 non-GPS-nav is coming along well!
    • Thien’s sending vision-speed-estimate into AP now
    • Chobitsfan PR is consuming

UTC0026 - https://github.com/ArduPilot/ardupilot/issues/12509

  • Rover
  • Nothing to update here

UTC0027 - GSoC update

  • Tridge has been working with Pete on his matlab integration
    • Lockstep scheduling is problematic with MatLab
    • Research channel on discord!
  • Rishabh
    • Blog post coming
    • And some detail on how the OA algorithms work
  • New non-gps nav thing?
  • New CAN channel
    • Rewriting HAL interfaces

UTC0034 - user alerts stuff

  • James and Stephen have been working on implementation plan for user alerts
  • Got a plan up on google docs
  • https://docs.google.com/document/d/1vLX0uRuPjPImEvvol0YnI2afBFTTNbcackr2JBIw_yc/edit?usp=sharing
  • Issues we want to announce may be a mixture of hardware and software problem
    • Hardware which impacts a lot of users
  • Are HW issues precluded?
    • Didn’t want to go down the path of gating things with manufacturers
  • Processes!
  • Issues on github?
    • Instantly public, which could be a problem?
      • Private repo?
  • Safety announcements?
    • So a limited list of people who can post?
  • Release and approval processes…
  • Board-id stuff is problematic given fmuv3 covering all sorts of hardware
    • Board names rather than board ID?
  • Would channel-1-screw-on-carrier-board be covered by this process?
    • Our announcements were pre-manufacturer-forums
    • Concurrence with the manufacturers are probable required
  • Need to provide best advice ArduPilot team can to the community
    • This may not be what the manufacturer would like said
    • This is something to worry about for phase 2
  • Open vote on new team member

UTC0044 - discord testing!

  • Seems to work OK
  • Meeting next week will be on Discord
  • Paid for higher level of audio…
  • Higher level of video?
  • Jitsi?
    • Self-hosting
    • Admin and maintainer volunteer?
  • Got a bunch of people very quickly.

UTC0010 - close

1 Like