Dev Call Sept 10, 2018 2300 UTC

InterDrone Report

Issues & Pull Requests


  • plans for plane 3.9.2, including important fixes for dead-reckoning and for VTOL takeoff


  • Release Update


  • Release Update
  • Non-GPS navigation
  • Rover simple mode
1 Like

Attendee count (max): 21

UTC2300 -

  • Merged
  • We will no longer knock back consts on pass-by-value
    • Despite extra syntactic noise
  • We do not require consts - but they’re not discouraged

UTC2303 - add cruise-trim-throttle

UTC2307 -

  • Vesca protocol
  • Some previous discussion on how this should be done
  • Developer not on call
  • Tridge added comment

UTC2311 - interdone

  • Very good attendance around booth
  • After-hours wasn’t well attended
    • Not sure what the problem was
    • Overlap?
    • We get a big room because it’s usually packed…
    • Was someone else offering free food?
    • Maybe next year we offer free pizza?
  • Jeff also attended ID and astm meetings and utm meetings
    • Tom and Jeff spent Thursday running through all of these meetings
    • Lots of protocol details
    • Bouncing back and forwards between technical and high-level items
    • Intel, amazon, google whitefox ardupilot, faa, research orgs, …
    • Specific formatting of data items
      • Uncorrected pressure or density altitude etc etc
      • Faa guys want integration with TCAS (uncorrected alt)
    • Long, brutal days
    • Intel’s POC implementation seems to be doing its job highlighting issues
      • Bluetooth was 400m, bluetooth 5 should be >1km
      • Spec is in flux
      • Protocol should be medium-agnostic
      • Trying to future-proof it
    • Signing support
      • public/private key setup?
      • Can-o-worms
      • If you do this badly people will go out of their way to break your system
        • Better off not doing it if you’re not doing it properly/securely
      • ADSB isn’t signed
        • The penalties for faking it would be extreme
      • “License plates” for drones
        • It already has an ID field
      • Where does the private key go?
      • Key-per-vehicle?
      • Use unique-id on processor?
      • Bolt-on-support?
      • Not much concern about hobby/diy crowd
        • Or thinking they’d get a module from an approved vendor
      • Some sort of booking system for people to say where they’ll be flying if they don’t have a device?!

UTC2325 - Plane update with tridge

  • 3.9.2 betas starting soon
    • Fix to dead reckoning in case of loss of GPS
      • Tested on real aircraft, working very well
    • Quadplane takeoff
      • Bug in Copter’s attitude controller
      • Caused CanberraUAV some grief
        • Battery fade
        • High fuel level
        • Yaw issues
          • MdB - Looking at Leonard’s patches?
            • Including a transition issue
          • We are looking at others of Leonard’s patches
            • Probably should base it on the spin status of the props
              • E.g. shutdown and spin-armed
            • “Wing twist and oscillation”
            • “inverting”
      • Probably affects ordinary multicopters
        • I.e. Copter 3.6
        • But only affects underpowered vehicles
      • If aircraft doesn’t make progress on takeoff then it may start drifting horizontally, and if it drifts for 15 seconds it will go crazy and roll/pitch corrections get applied too fast
    • Tracking issue related to crash detection code
      • Want Tom’s input on this one

UTC2336 - Randy and Copter

  • 3.6rc9 went out on Saturday
  • Probably an rc10 before release
  • Rob: Failsafe events / parsing the data?
    • entering/exitting s.bus frsky receiver s.bus outputs 874 bug
    • Debouncing stuff has gone in
    • On rob’s 3.5 branch where Rob’s fix wasn’t inplemented
    • Channel-8 motor interlock
    • Nate also saw this yesterday
    • One-off glitches are hanled with debouncing (one or two bad frames)
      • Have to debounce all switch activities
        • Including interlock
        • Should be fixed in 3.6
    • When frsky receivers go out of range and they’re set to no-pulses they usually send no pulses
      • Sometimes they send some garbage instead
      • Nate saw most channels go out of range
        • Except channel 5
      • The NuttX implementation doesn’t check parity bits on s.bus
        • Signature byte helps
        • Every byte has a parity bit
          • NuttX doesn’t check this
        • NuttX doesn’t understand parity bits on uarts
      • ChibiOS does check parity bits on the aux ports
        • But not on the IO firmware
        • Sid’s IO firmware will add checking for parity-checking serial streams
    • Set-range, set-angle - maybe a set-switch?
      • Which adds debouncing
      • Would like to handle flaps as part of same thing
  • 3.6 needs to fix the 874 bug
    • Not all channels exit this failsafe state at same time
    • increase debounce time?
    • 8 data log messages
    • 2 or 4 frames make sense to tridge - but 8?
    • Spektrum send higher channels less frequently
      • To fit more stuff into bandwidth
  • Need to get this sorted
  • S.bus decoder gets magic values for 874?
    • This probably translates to an in-air value of 0xff or something
      • There’s multipliers/shifts/additions to translate on-air to ranges
  • Exact value could be dependant on s.bus decoder implementation
  • Test it rapidly by turning internal transmission circuit on/off rapidly in transmitter
  • (Rob’s patch)
  • Tridge is on the OBC prep ATM
  • Started to push some things back for point releases
  • [10:10 AM] (Channel) AndrewTridgell: 874 == SBUS 0
  • [10:11 AM] To Weekly devcall: Nice!
  • Rob to raise an issue

UTC0006 - Randy and Rover

  • Planning on getting sailboat support in before 3.5 betas
  • non-GPS navigation is coming along
    • Working on Cartographer
    • Passing 360-Lidar through ROS down into AP
  • Simple mode has been merged