Dev Call Feb 3, 2020 2300 UTC

Issues & Pull Requests

Scripting REPL




1 Like

Attendee count (max): 25

UTC1100 - welcome to MikeDornisch from partner BFD Systems

UTC1101 -

  • Difference between Copter and Plane
  • Bad AHRS from the beginning
    • Randy gets too many support questions
    • Users get scared by “Bad AHRS”
  • Plane requires GPS etc
  • AHRS is not bad just because we don’t have GPS?
  • EKF could also be your secondary estimator
  • “Fit for purpose”
    • E.g. if you require position and don’t have it you would be bad, but if you flick to stabilise you go good
  • Fence enable requires a position estimate?
    • “Bad AHRS” when you have a fence but no position?
  • Add flag to status report for ekf is initialising?
  • Const-pos-mode?
  • AHRS doesn’t have to include position
    • Randy thinks it shouldn’t
    • Have AHRS only handle attitude
  • no-GPS-navigation for Plane?
    • Long-term thing to do this in Plane e.g. stabilize mode
  • [10:29 AM] (Channel) JamesPattison: As I said above: tell users what to do, not what is happening
  • PH could ask MO to act differently if the EKF is still initialising as reported by the status report
    • If in const pos mode and disarmed then lie about the EKF, say “still waiting” or whatever
  • How about leaving the vehicle in “initialising” until the EKF is using the GPS?
  • PH will work with MO to get some GCS changes in

UTC1128 -

  • Be more informative in the case of blheli tests failures
  • Tell the user via statustext
  • merged

UTC1138 - Orange Cube 4.0 and BLH passthrough

  • Blheli passthrough not working on Cube Orange?
  • Peter Hall will test it tomorrow

UTC1139 -

  • Thread-safe object buffer for param save
  • Was using a rather large amount of flash
  • Copy-and-paste job on existing function but now with a semaphore
    • Code in cpp file rather than the header?
  • Lock at bytebuffer layer?
    • Can go in the cpp file then
  • Putting code in the .cpp file might be a better solution here
  • Put semaphore into the cpp file?
  • Seen this before in matrix and vector and whatnot
  • Semaphores on f103 were a pain
    • Semaphore on RC input channels was nasty
    • 72MHz M3
  • PH will try moving stuff to cpp file

UTC1148 -

  • Doesn’t fly incredibly well
    • But doesn’t crash
  • This doesn’t make it worse
  • More testing on different vehicles on same type should show this is an improvement
  • Tridge will simulate to see if it adds anything
  • PH to simplify some code present in the PR

UTC1151 -

  • DO_LAND_START failsafe stuff
  • In_landing_sequence flag
    • Set when we cross a do_land_start
    • Whenever jump-to-landing-sequence is called it is set true
    • … and another instance in Plane
    • Any jump or set wp will zero the flag
  • Mission option bit used
  • Should be quite safe as there are few places that need to reset the flag
  • Rtl shifting to auto goes through resume which can’t reset the flag….
    • MdB points out that it’s not necessarily that case
  • Could merge this between calls

UTC0011 -

  • Bad GPS health to aggressive
  • Nate thinks it is too aggressive; MdB does not
  • 245ms interval bad == 4Hz is bad
  • Missing a single thing can spit out the message
  • [11:15 AM] (Channel) JW: @MdB how long till the message then goes away once it has been flagged?
  • Multi-constellation timings can move around a lot
    • But vendor that was sending things out configured for triple-constellation stopped doing that over a year ago
  • Need to get a log where we’re getting the problem three times/hour
  • Can simulate this - byte loss on serial line
  • Maybe add something where you need this 2 in 30 seconds?
  • Need to make sure we don’t paper over a bug with this
  • Autoconfigure forces 115200
    • Non-dma high bitrates you can lose bytes…
  • Comment adding requesting log

UTC0019 -

  • Can’t see the rc version
  • Peter will have a bit of a play with adding the extra information in the fields

UTC0030 -

UTC0033 -

  • Closed as not adding anything while destroying history

UTC0037 -

  • Can be merged after flight test

UTC0047 - change neopixel leds

UTC0049 -

  • Randy wants to test this

UTC0050 -

  • Switch to OF if GPS is bad
  • If you disable completely it will go OF
  • When does it switch?
  • Need it for EKF3 as well
  • Bitwise & for flow-use-nav
  • Needs to use ==

UTC0056 -

  • Can be merged after the typo is fixed

UTC0058 -

  • Current_loc.alt should always be above-home in Copter
  • Can cause a descent in RTL
  • API is confusing
    • Needs to be clearer on the frame
    • So a frame-in and frame-out where frame-out is what it should be stored in
  • Change should have been tested more
  • We should think about our APIs
  • … and look at where we are using magic 0 altitudes and considering altitude frame types

UTC0116 -

  • Users struggling setting up d-shot
  • So emit some more information
  • Will be merged

UTC0120 -

  • Merged!

UTC0130 - SCRIPTING repl

  • Lossy output OK?
  • Command to reload scripts enough?
  • Having REPL might allow you to remove devop
  • [12:29 PM] (Channel) RM: I wonder if developers might start using it as a replacement to mavlink commands
    • Could be rather powerful
  • REPL protocol
    • MAVFTP instead?
  • Need reliable over a telemetry radios
  • Maintains interpreter state
  • Reuse serial_control and add a reliability bit?
    • 70 byte packets are annoying
    • Currently reliability happens at a higher level
  • Need to see a PR
  • [12:36 PM] (Channel) db: … but is fixing serial_control too much work to make it out of scope for this?
  • [12:37 PM] (Channel) db: maybe once mdb’s protocol goes in as a reliable link, it might be that serial_control and it can be merged by a future fix.
  • Reliable both-ways

UTC0140 - Plane update

  • Next beta will be a resync with Copter
  • Spektrum receivers not working on Pixracer
    • Also in RC copter
    • Probably not working on other boards, too

UTC0141 - Copter update


  • Several issues coming up for Copter

    • RTL incorrect altitude thing
    • Spektrum thing
    • Mavlink gimbal limits not applied
    • Four more regressions listed there
    • Sparky2 uavcan
    • Etc
  • Jackson UAS

    • Zigzag arming issue
  • Mini-pix current sensing not working

    • 4.0.2dev works but not on 4.0.1
    • Hw def change?
  • Uavcan escs - motors spinning when disarmed

    • Probably the 1000-2000 limit assumption isn’t valid
    • Rc3 min/max thing again?
  • Rc4 is coming

UTC0150 -

  • Some nice changes coming in for failsafes
  • 4.1 Rover is a little ways off
1 Like