Servers by jDrones

Dev Call June 28, 2021

Issues & Pull Requests





Attendee count (max): 15

UTC2300 -

  • FF tuning items for quadplane TX tuning
  • A few things to fix yet

UTC2303 - use vehicle physics for multiple sims

UTC2317 -

  • Interpolated matrix backends
  • Useful for vehicles that can change their shape
  • Should be able to support tilting-motor quads with this

UTC2332 -

  • Send battery resistance in a separate message
  • Alternatively, use a BAT_OPTION bit to send through the resting voltage in the BATTERY_STATUS message
  • Needs to make sure information can be supplied

UTC2351 -

  • SITL -L can now take addresses

UTC2357 -

  • Randy wants to look more closely at this
  • Make setting home an arming check
  • non-GPS mode like stabilize and GPS location is on other side of planet - should you allow arming?
    • Can you then go into loiter if it gets GPS in-flight if origin is > 250km away?
  • GPS glitch puts home a long way away?
  • Does any of this stuff actually happen in reality?
  • Home and origin set before the EKF does its quality checks
  • Threshold-based distance checking - 250km or 2km doesn’t make much difference?
  • Randy will chase this up

UTC0014 -

  • Correct coordinate frame on return mavlink message for position target
  • Josh to put an autotest together showing probem being fixed

UTC0025 -

  • Double-precision EKF
  • Part of effort to get flying at very long distance
  • Accuracy of EKF over relatively short distances is also actually noticeable
  • This is post-4.1
  • Maintainability?
  • Autotests wouldn’t test single precision as-is
    • Don’t want to run both single and double tests in autotest
    • Use single-precision replay to ensure we get the same results
  • build.CopterSinglePrecision and test.CopterSinglePrecision
  • Previous PR made 64-bit PID object
    • Which one you choose may make a difference when you go to work at distance
  • One outstanding issue
    • Precision positioning of a vehicle
      • E.g. fly to a given lat/lon
        • Even with double-precision
        • @ 30k
          • With cm-accurate GPS
          • Won’t land exactly where you ask it to land
            • Maybe 0.5m at 30km
            • Single vs double using square of a distance?
      • Also affects takeoff
      • Takeoff did not complete
      • Flew to a location 2.5m from takeoff location
        • Outside the radius of takeoff controller
        • (100km origin-distance)
        • Entire mission flown with that error

UTC0039 -

  • Tiltrotot bugfix qloiter/qrt with thr_min !=0
  • Internal combustion engine with tilts?
  • Q_VFWD which controls throttle is assuming one source of forward propulsion
  • We kind of assume a single source of forward propulsion
  • Too contrived?
    • No, they actually do exist
  • Canted motors are increasingly popular
  • Don’t set throttle min?
  • This patch overrides the user’s request
    • We know better?
    • We’ve traditionally done X when the user specifies X
  • Merged as tridge doesn’t want to argue any more

UTC0057 -

  • Timing as when we get this in?
  • Post-4.1?
  • Needs Michael to look at it

UTC0128 -

  • Autotest server tidying

UTC0129 -

  • Some autotune problems have come up
  • Smax issue?
  • waiting for Copter beta5 from Randy before doing another beta

UTC0130 -

  • Pretty good shape for beta5
    • Want Leonard’s most recent guided-mode PR before we do the beta
  • Many successes over last week
    • Inc dolphin issue
    • Inc loiter bucketing
  • 3, 4 or 5 new issues
    • Multiple reports on CubePilot side that i2c ports aren’t working if uavcan gpss are enabled
      • Tfluna?
    • Build env setup on mac (Sid looking at)
    • Some reports of ground effect compensation not working as well as it did
      • Has changed a lot since 4.0.7
        • Need a Replay log
  • Beta5 this week sometime

UTC0136 -

  • Rover beta5 soon

UTC0136 - ArduPilot Configurator Demo

Servers by jDrones