Servers by jDrones

Dev Call Oct 1, 2018 2300 UTC


(Craig Elder) #1

OBC Report

Issues & Pull Requests












Plane

3.9.2 release plans

Copter

Release progress

Rover

Release progress
Sailboat support - should we support this sensor as an airspeed sensor? https://moderndevice.com/product/wind-sensor-rev-p/


(Craig Elder) #2

Attendee count (max): 21

UTC2300 - https://github.com/ArduPilot/ardupilot/pull/9274

  • Make sure clang builds don’t break
  • Add waf-configure-args instead of use-clang (pb)

UTC2303 - https://github.com/ArduPilot/ardupilot/pull/9470

  • Solo fix
  • Needs flight testing
  • Related to https://github.com/ArduPilot/ardupilot/pull/9381
  • Counter seems to be inverted?
    • If someone’s flying around with a high lean angle they need a higher thrust
    • Could descend under full throttle in this case

UTC2313 - the battery one

  • https://github.com/ArduPilot/ardupilot/pull/9414
  • 9 battery instances
  • 10 cells per instance
  • Luis wants 12
    • Mavlink spec can’t handle more
    • Pushing on mavlink side for 14 cells per instance
    • Waste of memory until we can support it
      • Need a driver
  • 7 extra enable parameters
  • Arming checks move into battery monitor
    • Gets a no-arm-with-battery-failsafe
    • Arming_min_volt won’t be translated
      • Tridge is OK with that
  • For 3.7
  • 8+1
    • One-per-motor and forward thrust
    • Not necessarily a per-battery thing, just used for monitoring
  • Need to do tool fixes because doing all of the battery things consumes vast numbers of logging messages

UTC2325 - transition airspeed

  • https://github.com/ArduPilot/ardupilot/pull/9444
  • During transition airspeed wait we continue to use quadplane controller
    • Makes transition slower
    • Should be a quick part of the transition
    • Using vertical thrust would seem to be aerodynamically unsound
  • Low transitions - really need lift if you’re transitioning from low altitude
    • Will likely crash
    • Climb rate you’re achieving should be same as climb rate trying to achieve from vertical lift or you will not climb as hard as you want
    • For some aircraft not adding the vertical power the vehicle will descend
  • “We shouldn’t be attempting to climb when transitioning”
  • Assist component
    • Downside
      • As soon as quad motors kick in for assist in fbwa
        • If you nose-down to increase speed the quadmotors go to maximum to counter negative list of wings
        • Can enter nasty situations
          • Can we get this to behave better
          • Can get stuck in quadplane-assist mode
            • Too much drag to move fast
          • MdB: I mostly hold that if you needed quad lift to be safe you transitioned to low :slight_smile:
          • [9:31 AM] (Channel) MdB: Just my opinion
    • Climb rate is linked to pitch stick in fbwa/assist regime
      • Might not be balanced correctly?
      • Full-down-elevator (full-nose-down)
        • Will ask quadmotors to descend at q_z_max
  • Merged the level-transition PR
  • https://github.com/ArduPilot/ardupilot/blob/master/ArduPlane/quadplane.cpp#L1127
    • Uses nav_pitch_cd
    • Separate parameter for descent vs climb?
      • Use angle-of-attack instead?
        • Will this work with quadmotors involved?!

UTC2325 - https://github.com/ArduPilot/ardupilot/pull/9274

  • Not allow anything that breaks clang++
  • MdB is happy with the changes
  • FF had reservations on bits and pieces
  • PB to help get this in this week

UTC2344 - updates to servo.cpp

UTC2345 - https://github.com/ArduPilot/ardupilot/pull/9369

  • Peter to add an autotest to make sure the limit is applied

UTC2351 - https://github.com/ArduPilot/ardupilot/pull/9167

  • Tailsitter / airspeed / qassist
  • Fly well in over and forward flight
  • This is supposed to help in the middle bits
  • Welcome iampete!
  • Sounds like a good idea
  • Too much to go through on call
  • Tridge will discuss with Paul
    • Concept of still in transition or not
  • See also https://github.com/ArduPilot/ardupilot/pull/9066

UTC2352 - https://github.com/ArduPilot/ardupilot/pull/8175

  • Yaw trim?!
  • Ahrs trims in radians was a mistake
    • Should have been degrees
  • Having a preflight “this is upright” calibration would be nice
  • What’s the CPU cost?
  • Will this update the sin_pitch cos_pitch etc etc
  • Maybe just add one for pitch rather than all three?
    • Q_TRIM_PITCH?

UTC0002 - https://github.com/ArduPilot/ardupilot/pull/9066

  • Fly forwards quite nicely but if you tail-sitter-backwards it doesn’t work quite so well
  • Non-symmetric-aerofoils
  • Centidegrees and floats are contradictory
    • Better to be floats in degrees
  • Could add to Copter roll/pitch max for up and down

UTC0009 - Plane and release plans

  • 3.9.2 very shortly
    • Based on current patches
    • Only a few extra patches over previous beta
  • Then looking at PRs for 3.9.3
  • A few bug reports from the 3.6 beta testing to look at before new release
    • D-shot report in particular

UTC0010 - OBC update

  • Full writeup will come along soon
  • Good news is that AP come 1st, 2nd, 3rd, 4th and 5th
  • 8/11 started with AP
  • 9/11 ended with AP
    • The changing team successfully flew!
  • CUAV came third
    • 30 points on time
    • 15 points on hatch
    • Serial cable went bad
    • No connection between Pi and STM32
      • Routed via Canberra instead
        • “The long way round”
        • Latency issues
        • Didn’t know it was working
        • Georeferenced correctly
        • No updates on the GCS to indicate it was working
        • Overrode landing position
          • Had we not it would have landed perfectly.
  • Delft almost got their avoidance working
  • Avoidance worked perfectly
    • http://uav.tridgell.net/tmp/DNFZ_testing.png
    • Throttle-based avoidance to be added
    • http://uav.tridgell.net/tmp/DFNZ_lots.png
    • Static no-fly
    • Dynamic no-fly
    • All on autopilot
    • Mission includes static no-fly zones
    • GCS talks to relay and retrieval vehicle
    • Both vehicles avoid
      • One “antenna vehicle” was constantly dodging
    • Asterix radar data to GCS
    • Converted to ADSB packets and filtered
      • Projected location, 10 seconds
    • DNFZ in image; number at end is number of objects within 1km of vehicle + statics
      • Closest is 114m
    • Max allowed was 32 statics (they gave 22)
    • Max allowed was 32 dynamics
      • They tried on 200 for size but got bounced
  • Copter did really well
    • Hybrid hexacopter
    • 2 of them
      • High antenna
    • 0.85 behind lead time
    • 2 points per minute late
    • No onboard starter
  • Low-vertical-thrust quadplanes were an issue
  • Tilting-rotor quadplane
  • Forward Robotics - great stuff as usual!

UTC0039 - Randy and Copter

  • 3.6rc11 went out yesterday
    • Hopefully last rc before soft-launch of 3.6
    • Few issues
      • PixHawk4 user can’t save params
      • D-shot on hexacopter freezes up after 10 seconds
      • NaNs in logs
        • False-positives in loganalyzer
      • Compass doesn’t work for one user
      • Fast sampling doesn’t work
      • Safety button not working on PixRcaer
      • Lidarlite not working under ChibiOS?
      • Solo gimbal under ChiibiOS not working?
    • Now’s the time to look at it Copter 3.6!
    • ChibiOS issues will go into ChibiOS issue
    • Others will go… somewhere….
    • Heli’s loiter-mode-when-disarmed thing should be fixed in this rc series

UTC0042 - Randy and Rover

UTC0044 - Hex conference in less than two weeks

  • Hopefully have sailboat in before then
  • Jani will not be at Hex conference
  • Nor will tridge

UTC0048 - Boeing and RFDesign are doing some great stuff together with ArduPilot