Issues & Pull Requests
Plane
Copter
Attendee count (max): 17
UTC2301 - generator: remove written-only bufferIndex state by peterbarker · Pull Request #570 · ArduPilot/pymavlink · GitHub
- Remove written-only buffer state
- Merged!
UTC2304 - AP_HAL_ChibiOS: add hwdef for LongBowF405WING by HiRCBro · Pull Request #27009 · ArduPilot/ardupilot · GitHub
- Add hwdef for LongBowF405Wing
- Merged!
UTC2305 - AP_Mount: add topotek gimbal driver by rmackay9 · Pull Request #27053 · ArduPilot/ardupilot · GitHub
- Add Topotek gimbal driver
- Down to 6.6kB
- Merged!
UTC2309 - HAL_QURT: revived QURT HAL by tridge · Pull Request #27091 · ArduPilot/ardupilot · GitHub
- Revives QURT_HAL
- For Qualcomm Hexagon CPUs
- ModalAI
- OS is missing some pretty critical calls
- Fsync
- Chdir
- Listing of the root directory crashes?!
- Map_file_name can do allocation
- And must be freed afterwards
- Could do magically-destroying thing but would be code complexity
- Vendor-specific sub-directory for parameters?
- Need a structure as the one-level was getting congested
- Peter will have another look
- Building with clang
- Building this?
- You can download the SDK from Qualcomm (or ModalAI?)
- Very nasty agreement
- Can’t get CI to agree to license!
- We have a machine set up to build things
- Custom runner on qurt.ardupilot.org
- Push a branch, wait for github to run stuff, download the build product
- You can download the SDK from Qualcomm (or ModalAI?)
- Several bugs found as part of this…
- Drive-bys in esp32?
- Heap was enabled…. But wasn’t used
- Ifdef vs boolean
- Buzz is fine with the changes
- We should write down how to create the development environment
- Tridge will work with the ModalAI guys to make them autopilot-agnostic
- ModalAI started up shortly after we removed QURT support many years ago
- Ready-to-fly Copter with VIO is quite nice
- Tridge is holding the “PX4 Autonomy DevKit”, based on a “Starling”
- Popular among universities
- Multiple cameras
- QURT is a proprietary OS running on Qualcomm ships, including Hexagon DSPs
- PX4 treats it as a Linux port, basically
- Our port puts everything on the DSPs, except for getting mavlink data from the DSPs to the outside world
- Can keep flying on the DSPs only
- Single-system-image for the DSP
- More a DSP with multiple cores
- All threads run all the time….
- Pthread isn’t quite the same
- Peter will have another look through it
UTC2331 - AP_Math: split CRCs from checksums by tpwrules · Pull Request #27355 · ArduPilot/ardupilot · GitHub
- Split CRCs from checksums
- And rename/make more likely
- Some questionable renames
- E.g. the fport one
- Is it worth the code churn?
- 9:35 AM]rmackay9: I guess a CRC is a form of checksum so we could just put everything into checksum.cpp perhaps
- [9:35 AM]rmackay9: I don’t mind though personally…
- Straight rename of file and functions?
- If there’s no compiler output change it could
- Will postpone this for now
UTC2342 - iFlight H7 Thunder by andyp1per · Pull Request #27166 · ArduPilot/ardupilot · GitHub
- iFlight H7 Thunder hwdef
- Merged!
UTC2343 - AP_BattMonitor: Sum, ESC: all selected items must be found to not be missing by IamPete1 · Pull Request #27359 · ArduPilot/ardupilot · GitHub
- For sum and esc backends, all selected items must be found to not be missing
- Buzzer affecting compass on some vehicles
- Should the ESC telemetry going away really be a battery failsafe?
- Is it more a thrust loss or something?
- There’s no ESC failsafe, no thrust-loss on boats, props are underwater….
- Should we have ESC failsafes as well as battery failsafes?
- Yes
- Leonard says that if something has failed he wants his clients to know about it
- We’re using a side-effect of monitoring rather than raising failsafe on correct data
- Sometimes ESCs stop giving telemetry when disarmed
- This is going go very poorly if you’re using sum backends with them
- Losing a motor on an Octo?
- Leonard says this is for landing safety
- Tridge thinks users might want to continue the mission
- Should we report that we’ve lost a component of our battery monitoring?
- Tridge will check UX for this
- PH will think about ESC failsafes
UTC0006 - AP_Periph: ensure the dshot type gets set by tpwrules · Pull Request #27444 · ArduPilot/ardupilot · GitHub
- Ensure d-shot type gets set on AP_Periph
- Can be merged after the guards are put in
UTC0004 - AP_Periph: add ESC_RATE parameter to RC_OUT peripherals by tpwrules · Pull Request #27406 · ArduPilot/ardupilot · GitHub
- An rc-ouptut-speed parameter for AP_Periph
- Still a bad name, RC_SPEED
- Rename to ESC_RATE
- Or MOT_PWM_RATE?
- Rename to ESC_RATE
UTC0013 - Allow full compass calibration with location sources other than GPS by Williangalvani · Pull Request #27409 · ArduPilot/ardupilot · GitHub
- Full compass cal with location sources other than GPS
- Merged!
UTC0013 - AP_Scripting: reduce Lua state memory usage by tpwrules · Pull Request #27426 · ArduPilot/ardupilot · GitHub
- Reduce LUA state memory usage
- Saves heap memory usage using same trick we used for singletons
- Own lookup table
- Same performance
- Peter Hall to test
- Follow-up will reduce flash usage even more
UTC0015 - AP_TECS: Add a separate stall speed parameter used for minimum speed scaling by WickedShell · Pull Request #27442 · ArduPilot/ardupilot · GitHub
- Add a separate stall-speed parameter used for minimum speed scaling
- Stops TECS trying to fix your airspeed when rolling if you are flying at your minimum airspeed
- I.e. pushing throttle up
- I.e. pushing nose down!
- Uses stall speed for scaling rather than minimum-speed
- Needs to be eas not tas
- Definitely want this parameter!
- Should be AIRSPEED_STALL
UTC0019 - AP_Proximity: avoid use of int16_t-read call by peterbarker · Pull Request #27445 · ArduPilot/ardupilot · GitHub
- Remove more use of int16_t read call
- Limit int16-t call to 1024 bytes
- closed
UTC0024 - AP_Logger: remove format sent_mask, use backend formats-sent bitmask by peterbarker · Pull Request #27446 · ArduPilot/ardupilot · GitHub
- Remove duplicate checking of message-FMT-emitted
- Saves a few bytes
- Should fix the dataflash
- Merged!
UTC0040 - AP_NavEKF3: do not store prediction-enabled as state by peterbarker · Pull Request #27447 · ArduPilot/ardupilot · GitHub
- Stop storing startPredictEnabled as state in EKF3
- Pass it through as a parameter instead
- This state was not being reset when resetting the filter
- Maybe this should be renamed to something else?
- Merged!
UTC0045 - AC_WPNav: correct calculation of predict-accel when zeroing pilot desired accel by peterbarker · Pull Request #27450 · ArduPilot/ardupilot · GitHub
- Correct position controller state when we’re in loiter and in RC failsafe
- As a separate issue, what should we do in RC failsafe in Copter for control inputs?
- Make the sticks neutral?
- What do you do with throttle?
- RPY are easy
- Randy’s got concerns about logging original data
- Need a PR to discuss
- Merged this PR
- What should we do to avoid entering acro while in RC failsafe?
- Re-raise the failsafe?
- Bounce the mode change?
- PH thinks we should defer the failsafe
- So the user has actions that will change the state of the vehicle
- Let them enter loiter and the vehicle will the enter RC failsafe
- Randy doesn’t like the surprise
- He’d prefer to reject the command
- Move to BRAKE mode if we can’t enter LOITER?
- Again Randy doesn’t want to give the user something they didn’t ask for
- Bounce manual modes but allow loiter mode
- So bounce alt-hold, stabilize, acro?
- But allow loiter
- What about poshold?
- Peter will bring some PRs
- One for zoeroing control inputs in RC failsafe
- One for bouncing entry into modes when in RC failsafe
- 11:16 AM]rmackay9: @Peter Barker becareful about Auto_RTL!
UTC0117 - AP_ESC extended status packet addition by Pradeep-Carbonix · Pull Request #27248 · ArduPilot/ardupilot · GitHub
- Extended status packet addition for APD escs
- Review left
UTC0126 - AP_Mount: added sending of position data to Siyi gimbal by rmackay9 · Pull Request #27457 · ArduPilot/ardupilot · GitHub
- Add sending of position data to Siyi gimbal
- EXIF data now gets location on A8!
- Merged!
UTC0128 - AC_PID: correct error calculation to use latest target by IamPete1 · Pull Request #27460 · ArduPilot/ardupilot · GitHub
- Correct error calculation to use latest target
- #24411 introduces a bug…
- Targets updated slowly
- This fixes that
- D-updates were at right rates, which is why we got away with it
UTC0132 - AP_Logger: remove short-filename compatability code by peterbarker · Pull Request #27478 · ArduPilot/ardupilot · GitHub
- Remove compatibility code looking for log1.BIN
- Merged!
UTC0135 - AP_Common: fix and clean up Bitmask class by tpwrules · Pull Request #27481 · ArduPilot/ardupilot · GitHub
- Fix and clean up Bitmask class
- Saves a chunk of space on boards compiled for speed
- Tidying up the validation
- Good idea?
- Equality loses same-number-of-bits
- Compile-time error now
- merged!
UTC0151 - autotest: add a test for switching to auto with low speed and alt by peterbarker · Pull Request #27483 · ArduPilot/ardupilot · GitHub
- Throttle suppression logic didn’t unsuppress throttle on a flight
- An option bit to not do throttle suppression
- Maybe an option bit to MOT_SPIN_MIN for Plane
UTC0154 - AP_NavEKF: option to align extnav to optflow pos estimate by rmackay9 · Pull Request #27487 · ArduPilot/ardupilot · GitHub
- Allow extnav to align to opticalflow position
- Stops position jumping when you switch from optflow to extnav
- these techniques will help with future work coming in from Leonard
- Handle on controller side rather than estimation side
UTC0159 - hwdef: H757_EVAL: make it not-AP_Periph by peterbarker · Pull Request #27495 · ArduPilot/ardupilot · GitHub
- H757_EVAL - make it not a periph
- Can’t do dronecan bootloader as a periph
UTC0201 - Plane 4.5.x issue tracking · Issue #26202 · ArduPilot/ardupilot · GitHub
- Plane update
- Not a lot of reports from the beta
- Discovered a mode-guided problem
- Fly-to updates the position you want to go to before trying to change modes
- So if you’re in fence breach…
- Param-ftp vs networking
- Errno race condition
- Make errno thread-local?
UTC0205 - Copter update
- Things seem to be going OK