Dev Call June 1, 2020

Issues & Pull Requests





1 Like

Attendee count (max): 22

UTC1100 -

  • Tidy for Plane’s main include file
  • Plane.h changes
  • AP_Vehicle.cpp?
    • Moving lots and lots of code into AP_Vehicle.cpp
  • Header reorganisation?
    • Just nicely labelled
    • And reordered
    • Grouped by cpp file
    • Some dangling declarations to be removed
  • Why not PID access via parameters?
    • Make a PID and use it in scripting
    • Thread safety?
    • Will consume lots of flash space?
      • Need something to justify it
        • Transwing stuff
        • Scale factor?
          • Next
  • P and I could get out of sync without locking
  • Could grab a handle on a parameter than could be set
    • Fast access to any parameter
  • Get something to get access to a specific location in memory?
    • Typed
    • Lua object which knows its type
    • So a PID object could give you a AP_Float* equivalent
  • … but back to getting some sort of a parameter object on which you can set the value
  • Will drop the moving stuff into AP_Vehicle
    • It was supposed to be things that are inherited from the AP_Vehicle class
    • Might just end up with a random collection of functions
  • Too many bindings creating too large a flash size?

UTC1124 -

  • Force qassist defaults
  • Add switch state for qassist
  • Qassist with motors only
  • Copter gains for copter bit
  • Leave control surfaces on Plane controller
  • Removes speed scaling issues
  • Reuse q_options bit?
  • Any utility to the non-motors-only option in qassist?
    • Oscillations?
    • This isn’t for Copter tailsitters
      • Should be an improvement for all tailsitters
  • Skipping lots of stuff with the early return
  • Vector-gain no longer run?
    • Vectored tailsitters may no longer pitch correctly?
    • Vectored hover gain not run
  • Conflict with speed-scaling?

UTC1130 -

  • Merged!

UTC1137 -

  • Protect against fp error
  • On a real vehicle you’d get zero throttle output

UTC1140 -

  • Generally edge-triggered
  • The interlock switches are a problem
  • Merged!
  • Now have duplicate aux switch options for Rover and Plane

UTC1144 -

  • Rather surprised that -1 comes in
  • Serial over i2c
  • -1.00 out of range
  • If your max is 90 and you’re at 120 you will get 90 and rangefinder is good
  • Need someone to review and merge
  • Backport this, could be a fly-away

UTC1149 - hwing ESC

  • Tridge will test it next time he’s working on the hardware

UTC1151 - ublox high-baudrate

  • Tested on hardware?
    • Can be merged
  • 230kB is the limit for interrupt reasons
  • 2s lag on a GPS in flight was sub-optimal

UTC1154 -

  • Rearrange arming check order
  • Lots and lots of checks
  • This is useful for documentation

UTC1155 -

  • HC-SR04 rangefinder support
  • Merged!

UTC1159 -

  • Add example of OOP
  • There are other ways of doing OOP too
  • 9:59 AM] WickedShell: yeah I need to try the alternate version that doesn’t conflict with every manual on the internet…
  • [9:59 AM] WickedShell: Yeah, I know but it’s not PIL/main version
  • [10:00 AM] WickedShell: Yeah the main downside this way is just memory duplication, that’s all :slight_smile:
  • [10:00 AM] WickedShell: I’d prefer to poke the other one slightly first but it could if we want
  • [10:00 AM] WickedShell: But I’d hate to see us encourage this long term
  • [10:00 AM] WickedShell: (just for memory reasons)
  • MdB will investigate metatables before we go ahead and merge this one

UTC0002 -

  • More math operators for LUA
    • Named ones .dot and .cross
  • Also .micros
  • Other candidates?
    • Rotation functions

UTC0006 -

  • Add continue-after-land mission bit
  • Need to make sure motors are spooled down properly?
    • Spool-state has to be groundstate-idle by the code
  • Only affects quadplane, not fixed wing
  • If last waypoint is not nav land then the Copter will land and will just spin the motors until disarm delay happens
    • So long as next thing is not a takeoff
  • Documentation for new bit?
    • yep
  • New default is to not continue after nav land
    • Change of default behaviour for Copter
    • Will now disarm even if there are further mission items after a LAND

UTC0012 -

  • Needs to be tested on MissionPlanner
  • Needs to wait for a pymavlink release
  • Merge after it’s OK with MissionPlanner
  • Peter to write something to GCS maintainers list

UTC0020 -

  • Has to be against master
  • Tridge will cherry-pick into master

UTC0024 -

  • Plane 4.0 beta!
  • Henry and Rolfe and another have been flying it
  • Main concern around compass devid issues
    • Left-over parameters for older compasses
    • Never had more compasses
    • Unstable sort?
  • Lightware needs a patch brought back

UTC0026 - Soaring library improvements coming

  • 6.5 hour flight with no motors
  • 11,000 feet
  • Full battery the whole time
  • Stall detection and recovery
  • Auto mission
    • Flatish ground out in the desert
    • 1km square area
    • 800m alt
      • Above it the whole time
  • 32-to-1 glide rations
  • Stall recovery
    • Set Elevator and throttle
    • Doesn’t counter flat-spins
  • Stall detection
    • Bitmask specifying what is checked
      • roll/pitch deviation is one check
      • Flag from TECS control for bad descent

UTC0035 - Copter-4.0.4-rc1 released for beta testing

  • Beta test for Copter
  • Some people testing already
  • Battery testing and wikipages would be good
  • Moving baseline stuff
  • Some mission planner issues
  • Probably a long beta
  • New support issue with d-shot on Kakute-F7
    • PWM out issues
      • Could be DMA channel issue?

UTC0039 -

UTC0041 - GSoC update

  • PH is going well with MATLab
    • There were some physics issues
      • Some lucky combination of mistakes let it fly the first time
  • Harshit has opened a PR in last week
    • Core EKF work is great!
  • Rishad’s object avoidance
    • PR in for Rover avoidance
      • Would return back to original segment
        • I.e. crosstracking behaviour change
      • Now can reset the origin to where the vehicle is now
  • Object database will be moving to 3D!
  • [10:47 AM] JP: thanks to PhilK and mRo for supporting the quadraped GSoC student. Funding for Walking robot project

UTC0050 -

  • CUAV firmware should be updated
    • Tcp link! Bad!
    • Need to convince vendor to upgrade to mavesp firmware
  • Ongoing issue

UTC0057 - User alert update

UTC0108 - terrain download app

  • Server at 70% of terrain generated
    • Getting there!
  • 70% == 6.9GB (gzipped)
  • 80 or 90GB for entire world
    • Compressed files on SD card, perhaps?
    • PH: i guess continent packs might be OK with big SD cards

UTC0113 - close

1 Like