Dev Call Sept 7 2020

Issues & Pull Requests




1 Like

Attendee count (max): 26

UTC1100 -

  • No time in SITL
    • Could perhaps make times match F4 hardware?
  • Memory allocation
  • Give a message back saying “set this option”?
  • Automatically set option on first fetch

UTC1106 -

  • Add pos control PID logging
  • PSCP rather than PSC1?
  • AP_Logger_Write is convenient but significantly
  • Move to normal block writing rather than Write_Log?
  • Only 10Hz so it doesn’t really matter
  • Rate pids you want high rates
  • Navigation pids can be slower
  • Can be merged after it’s using a structure to log things

UTC1114 -

  • We have alt-hold lean angle max
  • Uses throttle do determine max lean angle?
  • Why is it using current throttle rather than hover throttle?
    • Need to get Leonard to review this one

UTC115 -

  • Return uint8_t’s rather than boolean
  • Need to pass script stuff via MdB

UTC1118 -

  • Check to see if there’s any flash improvement
  • Otherwise just fix the initialisation and remove the #ifdefs

UTC1122 -

  • Lock in -Werror
  • Will prevent bugs
  • Will get CI failures where they ignored errors on their local builds
  • Moving CI to Focal
  • Developers may not get warnings on their local machine but might fail in CI
  • Some failures might be real bugs
  • Warning free on arm compilers instead?
    • Whitelist specific compilers
  • Tell user which compiler they’re using it
  • We do acknowledge the extra pain
    • But -Werror picks up real problems
  • Pierre’s fix for gebcnhmark makes related PR’s hacks unnecessary

UTC1127 -

  • Pass it by Leonard
  • Pid_info in the logger class is really not nice
  • Would be nice to pass pid_info in instead of embedded
  • Tridge has a student working on eliminating the redundant PID implementation
  • Redundant PID system for a while!
  • Merge after Leonard’s happy

UTC1140 -

  • Run configure and check hwdef.h for the board
    • Changing DMA priorities may disable DMA on other things, so you have to be a bit careful
    • See what changes in the DMA map
  • Tridge will merge once he’s happy with the DMA map

UTC1145 -

  • Set a minimum level to be transferred from ArduPilot
  • MissionPlanner shows all messages ATM regardless of severity level
  • Tridge has a preference for doing it on the ground control station
  • MAV parameter object would be done for this
  • Plane’s send text informs the GCS operator
  • Bandwidth
  • Rate-limitted status texts?
  • Rework for AP_Vehicle
  • Make sure messages still go into the log
  • Generally happy with the concept it seems

UTC0000 -

  • Merged!

UTC0002 -

  • See what SmartRTL inherits from RTL and look at removing the other inheritance
  • merged!

UTC0005 -

  • Discussion-starting PR
  • Need to let Heli rotor spool up before taking off
  • Swipe Loiter’s pre-takeoff-state?
  • Copying takeoff altitude aside
  • Remove some restrictions on Heli for auto-armed
  • Significant change to auto-armed
  • Randy will look before Friday

UTC0013 -

  • Factor a whole bunch of PWM measurements
  • Clean up debug as pointed out by tridge then merge this in

UTC0022 -

  • Specific to tilt-vectored tailsitters
  • Leonard has weighed in
    • New subclass?
  • Qloiter mode
    • Relax changes desired attitude to current attitude
  • What happens on non-level ground on the ground?
    • It will no longer build up roll integrators
    • Probably a fix
  • We don’t relax axes in stabilise in current mode
    • Qstab sets throttle-wait to false
  • We don’t want to relax the pitch axis when idle on a vectored tailsitter
    • A lot of code for just that?
    • Why touching roll/yaw, then?
  • Tridge doesn’t like duplication
    • Less work and more code?
    • Quaternions?
    • Derivative of euler is what matters at singularity
    • Distinguishing roll/yaw at 90 degrees is not possible
  • Stick it in tailsitter.cpp in quadplane?
    • Not in leonard’s bailiwick then
    • Already have other tilt compensation in tailsitter.cpp
    • Tailsitter-relax function?
  • Multirotor copters will eventually handle tilt-rotors
    • But quadplanes already can
    • Very much an in-the-future thing
  • Need to convince Leonard or shove it into tailsitter.cpp
  • Definitely need to discuss with Leonard

UTC0040 -

  • Frame Types not relevant for normal multi?

UTC0050 -

  • This one got forgotten past last November…
  • Mark’s been flying it on his convergence
  • No behaviour change if options not set

UTC0052 -

  • Removed dead code

UTC0055 -

  • Plane update
  • 4.0.6 is out!
    • We’ve learnt our lesson about shoving too much stuff into point releases
    • Lots of logs came in for beta, which was really good!
  • Throttle failsafe==2
    • Bug report from Copter user which would have liked the same for Copter
    • Land repositioning due to no throttle failsafe set….

UTC0056 -

  • rc 4 out for a week or so
  • Probably go live with it before Friday
  • There will almost certainly be a 4.0.5 but probably not too soon
  • Lots of avoidance stuff is still happening!
  • 360 lidars when leaning forward, LIDAR sees ground….
    • Might try to evade them…
    • 3D camera can improve by using different part of image depending on leaning
    • Vehicle code might need this stuff

UTC0103 - GSoC

  • Great work everyone!
1 Like