Dev Call July 15 2019 2300 UTC



1 Like

Attendee count (max): 23

UTC2303 - GSoC with Matt kear

  • Get entry and glide working

  • Work on additional phases afterwards

  • Headspeed controller has been developed for entry and guide

    • PD controller with feed-forward / follow-up-trim
    • P-term gives direction feed-forward needs to move towards
    • Works somewhat like an i-term
    • Works well for the filters
    • In steady-state-glide frequency is decreased to provide smoother flight
  • Has been using one nav speed/height controller for forward airspeed

  • Getting ready to test on real airframe

    • Need a reliable bail-out of the autorotation
    • Pitching up aggressively on bail-out because of switching back to previous mode
    • Setting altitude and then aggressively trying to attain it
    • Working on code to ease into replacement mode
      • Maintaining airspeed through the rotors
      • Currently using a simple timer
      • Measuring rotor RPM at the moment which will help determining a bail-out time
    • Seems to be some time constant involved in motor RPM
      • Are we ramping the motor?
        • Bill has written code to avoid this
  • [9:11 AM] To Weekly devcall: Most critical thing for testing on real hardware: make sure you get video!

  • [9:13 AM] (Channel) MdB: Things go most wrong when logs or video aren’t present. Enforcing both reduces aircraft crashes. Proven fact :slight_smile:

UTC2313 - Peter Hall and sailing

  • KF PR is in
  • Working on a mapping mode
  • “Active” mapping mode concept
    • Deals with failures
    • Off the line
    • Missing the waypoint
    • Reaches a fence and realises it can’t go past the fence so gives up

UTC2322 - Rajat AirSim PRs are in

  • PR against airsim

UTC2322 - MAVPproxy stuff

  • Hopefully catching up with Akshath in next couple of days
  • Something in place to catch the GTK issues

UTC2323 - move rc channel option to reset mission

  • Merged!

UTC2324 -

  • Serialmanager cleanups
  • Why not make fmuv3 -O3?
    • Scripting means we’re over 1MB now
  • merged!

UTC2330 -

  • Play tune on no-SD
  • We can boot without microsd on ChibiOS
    • Should we play a tune in this case?
  • File stuff from the boot thread?
  • Merged!

UTC2332 -

  • Mtime not updated on write
  • 1st January 1980
  • Clock update and state update only on ChibiOS build?

UTC2335 - move polyfence code?

  • We should try to share the read/write of the messages
  • Would be nice to get OSD-style scrolling of statustext messages in MiussionPlanner HUD
  • Dealing with old fence type
  • Working from EKF origin
    • No EKF origin in DCM!
  • AP_Fence?
    • Plane is fancy when it comes to fences
    • Lower-fence
      • Major source of embarrassment if you fail to disable the lower fence when demonstrating a landing

UTC2345 -

  • Merged!

UTC2353 -

  • Can be merged when it passes CI

UTC2357 -

  • Probably already merged
  • Can be merged otherwise

UTC2357 -

  • Move in right direction
  • Randy will have a look at it

UTC0000 -

  • Behaviour change
  • Bad rebase comment present
  • Should be two PRs, a separate one for the behaviour change
  • Peter and ntridge to help get this one in

UTC0008 -

  • Has somebody tested this using SITL NMEA GPS?
  • Marked as NeedTesting

UTC0009 -

  • %d vs %u

UTC0010 -

  • AP declination update / magnetic tables

  • Difficult to make package work said somebody

  • Python package isn’t getting updated?

  • Online server may be closer to truth?

  • Continue with existing setup

  • Compass calibration didn’t converge in very low magnetic field environment

    • Somewhere in Africa
  • Compass calibration with no GPS?

    • Drop back to old methods
    • Most people will have a fix?
    • Doing stuff indoors?!
      • Very bad

UTC0017 -

  • Which firmwares are coming up?
  • Missing Rover 3.4.2
  • Missing Copter-3.5
  • Directory name incorrect?
  • Manual work done by tridge
    • Manually fixed by tridge
  • Will rebuild the Copter 3.5 stuff
  • Rover 3.4.2 Pixhawk1 build?
  • Randy and tridge will work through the issues

UTC0031 - Copter update

  • 3.6.10rc1 reports say motor test doesn’t work on Pixracer
    • Safety switch handling?
    • PK will test
  • Pixhawk mini not working with SBus?
    • Randy doesn’t have a pixhawk4 mini
    • Says it also doesn’t work with latest
    • Separate SBus input connector is NOT used on this
      • May be inconvenient to move cable… enhance to use this one?
      • Dual pin inputs?
      • RC input port can be used as a uart instead!
  • Use serial protocol manager to allow selection of which uart is used for RC input?
    • Only recover a full uart on some boards where the inversion can be done in software

UTC0038 -

  • Nothing new

UTC0039 - Leonard’s PID changes

  • Two extra filters on AC_PID
  • Copter-3.7?
  • Want to finish off the high-vibration failsafe
  • Calculates accel from altitude delta when in failsafe
  • Hopefully beta-test in 4 weeks

UTC0042 - Plane and tridge

    • Chinook flying in Loiter
  • Need to get video from other people’s
  • Beta2 is out
  • Mro zero f7 support
  • Pass-through-control on loss of fmu was not enabled prior to beta2
  • Needs a lot of testing
    • Please test and report!

UTC0045 - USB dual-function

  • T265 cameras
  • Need a companion computer ATM
  • What would it take to use USB port?
    • No exposed uart
  • Substantial amount of work to act as a USB host
  • Camera allows for a lot of configuration
  • Consume directly in ArduPilot?
    • Could develop in SITL
    • Could pass through the USB data stream over a normal UART from a CC
  • Probably a couple of weeks
  • Depth messages would be cool
  • Could be cheaper than a 360 lidar
    • And they’re doing SLAM
  • Tridge’s indoor-vs-outdoor channel mode option
  • Existing switch for disabling GPS

UTC0049 - stack usage

  • Reduce memory usage by moving allocation from heap to stack
  • MdB had a bug which went away when increasing stack
  • Tridge is working on statically determining stack usage

UTC0101 - setting bad value in rcmap will cause vehicle to fail to boot

  • Peter’s die-rcmap-die branch does fix that
1 Like