Dev Call Oct 11, 2021

Issues & Pull Requests

add ability to trigger other calibration options via RC Channel options
Networking options




Attendee count (max): 16

UTC2301 -

  • modalAI port
  • New partner
  • Interesting Copters
  • 4 cameras
  • Depth camera
  • Vio camera
  • Vision copter
  • Used in a bunch of commercial vehicles
  • Adding support for their boards
  • Couple of new battery monitor chips i2c
  • New feature to use uart pin as gpio save/restore for gpio mode
    • No separate pin for spektrum binding
    • Use receive pin on uart
  • Better support for px4 bootloader upload script
  • AP_Analogue In maybe later
  • Vio already runs on SnapDragon
    • Uses different mavlink message fro common.xml
    • Will add support for this message into ArduPilot
  • Would be nice to have depth cameras work, but VIO first

UTC2307 - AP_Scripting: allow to mark arguments for no range check, optimize for size by IamPete1 · Pull Request #18221 · ArduPilot/ardupilot · GitHub

  • Scripting mark arguments for range checks
    • Saves bytes
  • Merged!

UTC2309 - Copter: Support parachute battery failsafe action by tridge · Pull Request #18490 · ArduPilot/ardupilot · GitHub

  • Support parachute battery failsafe action
  • We don’t have any prearms that your parachute is actually enabled
    • E.g. compiled out
  • Too low, no failsafe?
    • Need a backup action
      • Land

UTC2322 - AP_Motors: always use PWM min and max, don't take from throttle. by IamPete1 · Pull Request #18576 · ArduPilot/ardupilot · GitHub

  • Always use pwm min/max don’t take from throttle
  • Parameter descriptions need to stay for a while after we remove the parameter as MissionPlanner uses latest
  • MergeOnCIPass

UTC2324 - Copter: move simple transform into mode class stop using set_control_in by IamPete1 · Pull Request #18581 · ArduPilot/ardupilot · GitHub

  • Tidies up simple mode rotations
  • Randy will look at it

UTC2329 - triggering of accelcal via RC switch

  • 6-orientation less import as IMUs get better
  • EK3 does bias estimation
  • Scale factor becomes unimportant in modern IMUs
  • Our advice might be come “place on level surface and press a button”
  • Scale factor !=1 might be a calibration error on the user’s behalf
  • Anything after the “2” series would probably work fine with single-axis
  • We’d accept single-axis calibration
    • Wouldn’t catch mis-aligned flight controllers
  • Three different things for accelscals
    • Trims
    • Another that sets scale factor to 1.0 and offsets to zero and zeroes trim
      • “Single-axis calibration”
      • This is the one to trigger on a switch
    • Six-axis calibration
  • Icm2 maybe 8% off, icm3 0.1% off?
  • Need GPS to allow for bias calibration

UTC2338 - Added logging of stack free per thread by tridge · Pull Request #18587 · ArduPilot/ardupilot · GitHub

  • Add logging of stack free per thread
  • Tridge will merge later since it has an approval

UTC2340 - ArduCopter: Guided pos_control_run() add yaw_rate timeout by hendjoshsr71 · Pull Request #18616 · ArduPilot/ardupilot · GitHub

  • Add timeout for yaw rate directive?
    • This is a velocity control should reset the rate to zero like this
    • Position vs rate control
      • Position is set once and vehicle will complete that command
      • Velocity has to be sent all the time or vehicle will stop
  • Can’t revert yaw mode in guided?
  • Position part of message won’t time out
  • Yaw rate will time out
  • The go-to-position message doesn’t time out why?
  • Should we fall back to looking along track or something when the yaw rate times out?
  • Maybe this message really tries to do too much
  • Randy: as a side note I have an item on my to-do list to improve our developer wiki to clarify how to get the vehicle to do what you want in Guided mode

UTC2349 - Copter: GCSMavlink Guided inputs check force_set and reject by hendjoshsr71 · Pull Request #18720 · ArduPilot/ardupilot · GitHub

  • Forces vs m/s/s
  • Reject if force bit is not set
  • Acceleration in wrong units would be bad
  • Merged!

UTC2352 - HAL_ChibiOS: fix G4 I2c bus speeds by tridge · Pull Request #18856 · ArduPilot/ardupilot · GitHub

  • Fix g4 i2c bus speeds
  • Merged!

UTC2354 - HAL_ChibiOS: add missing return to Dual CDC get_usb_baud() by sypaq-nexton · Pull Request #18865 · ArduPilot/ardupilot · GitHub

  • Missing return statement to return rate
  • Merged

UTC2356 -

  • Produce elf_diff output in CI
  • Merged

UTC2358 -

  • Merged already
  • Emit telemetry while doing compassmot

UTC2358 -

  • Rc input using f-port on a uart
  • Serialn_protocol to 23
  • A lot of people are doing this
  • When rc failsafe happened it started to RTL
  • Receiver was set up for no-pulses
    • The ppm-in pin was floating
    • Noise on the vehicle appears to have caused enough noise to be detected as dsm
    • Interpreted random input as being manual mode
    • Random controls
    • Destroyed plane
  • Two PRs to fix this
  • This one changes everything to pulldown
  • This might cause problems on some boards
  • Can we get away with just pulldown rather than just pullup
  • Long lead issues?
    • Need higher drive strengths
  • Copy-and-paste errors might have selected PULLUP
    • Consistency more important than preserving existing pullup/pulldown
  • Once we detect protocol we could change pullup/pulldown based on the detected protocol
  • We were using float because we were worried about current draw
  • Tridge will talk with hardware vendor on this one

UTC0011 -

  • Don’t allow redetect of RC protocols in-flight
    • Option to get back old behaviour
  • Armed check?
  • Failover should fail on ground
  • Allow switching based ons trong crc?
    • Most protocols don’t
    • sbus/spektrum are worst protocols but also most popular
  • Would be nice to log where the RC data is coming from
  • Could we use jitter / some sort of quality to decide if dsm is good?
    • We already do some of that with the valid-frame-count
  • Peter will have a look at it
  • We’ll discuss at DevCallEU
  • Leonard thinks we need a positive action to give/take control
    • He has some notes on that
  • RC_PRIMARY option?
  • Peter to look at this

UTC0024 - networking discussion

  • PR in place to add LWIP to ChibiOS
    • Brings ChibiOS to latest version
    • Needs attention
  • Very few parameters
  • Do advanced networking stuff in scripting
  • Most people will use private network addresses?
  • Hide in HAL or do like AP_FileSystem?
    • More like AP_FileSystem
  • Need ground control stations to use 32-bit-clean implementation
  • AP_Networking with a POSIX / LWIP backends
  • LWIP wrappers for BSD sockets?
  • Portscans…
  • SerialManager becomes aware of these networking endpoints
  • Accessor on stream device class (subclass of betterstream)
    • Baudrate_capable?
  • Autobauding?
    • Would probably just work at this point
  • Link-type-aware?
    • “Do you have a baudrate?”
  • Bandwidth?
    • Blindly reporting baudrate bad?
      • No
      • Artificially capped at e.g. 1Mb
  • QOS
  • Tcp has flow control, udp does not?!
  • The key change is that SerialManager returns a StreamThingy not a UARTDriver
    • Needs a rename?

UTC0051 - Plane update

  • 4.1.2 beta is out
    • Serious bug found and fixed in autotune
    • New release soon
    • Tridge wants the RC fix in at same time
    • Wants this out in next day or two

UTC0052 - Copter update

  • 4.1.0 official is now out there!
  • New report out this morning
  • Perhaps a change in behaviour in extended_sys_status flags
  • Existing problem with helis that allows for landing detector to trigger in-flight
    • Need to release a 4.1.1 beta to get a fix
    • Age-old problem back in 4.0
      • Won’t do a beta period of 4.0.8 beta, rely on Bill to round up testers
  • Rover 4.1.1 will come out at same time
  • 4.1.0 weren’t completely synced
  • 4.1.2 will be a full-sync
  • Double-shuffle rebase stuff is a good way to do things
    • Git diff should be no differences

UTC0055 - Sub update

  • 4.1 plans!
  • Position controller issues need addressing
  • Will sync with Plane-4.1 and Copter-4.1 branches

UTC0056 - Tracker?

  • Peter Hall and Peter Barker will make an effort to get a 4.1.2 out, matching up with Copter and Plane

UTC0100 - using new compiler on old firmware

  • Might need to for size reasons
  • We build with 10.2 and 6.x in github actions
  • We really should advertise the fact we changed our release compiler a bit more!

UTC0101 - Rover update

  • Will release 4.1.1 alongside Copter
  • Uri has found a dangerous condition to do with GPIOs
    • Maybe try master?

UTC0106 - close

Post-Meeting Networking discussion