ArduPilot EU Dev Call 2025-10-22

Georacer · October 22, 2025, 8:15am

Attendees (unique): 11

UTC0702

github.com/ArduPilot/ardupilot

Tools: autotest: param parse: check values and ranges don't contradict one another

master ← IamPete1:paramParseRangeValuesCheck

opened 11:58PM - 20 Oct 25 UTC

IamPete1

+21 -2

This currently fails, but not as badly as I would have expected. ``` Error i…n ../AP_DroneCAN/AP_DroneCAN.cpp At param CAN_D1_UC_S1_IDX Range of 0.000000 to 100.000000 and value of: -1.000000 Error in ../AP_DroneCAN/AP_DroneCAN.cpp At param CAN_D1_UC_S2_IDX Range of 0.000000 to 100.000000 and value of: -1.000000 Error in ../AP_DroneCAN/AP_DroneCAN.cpp At param CAN_D1_UC_S3_IDX Range of 0.000000 to 100.000000 and value of: -1.000000 Error in ../AP_DroneCAN/AP_DroneCAN.cpp At param CAN_D2_UC_S1_IDX Range of 0.000000 to 100.000000 and value of: -1.000000 Error in ../AP_DroneCAN/AP_DroneCAN.cpp At param CAN_D2_UC_S2_IDX Range of 0.000000 to 100.000000 and value of: -1.000000 Error in ../AP_DroneCAN/AP_DroneCAN.cpp At param CAN_D2_UC_S3_IDX Range of 0.000000 to 100.000000 and value of: -1.000000 Error in ../AP_DroneCAN/AP_DroneCAN.cpp At param CAN_D3_UC_S1_IDX Range of 0.000000 to 100.000000 and value of: -1.000000 Error in ../AP_DroneCAN/AP_DroneCAN.cpp At param CAN_D3_UC_S2_IDX Range of 0.000000 to 100.000000 and value of: -1.000000 Error in ../AP_DroneCAN/AP_DroneCAN.cpp At param CAN_D3_UC_S3_IDX Range of 0.000000 to 100.000000 and value of: -1.000000 ``` ~Would be easy to fix those and lock it in~ I have added a commit to make this new check pass.

Merged!

UTC0704

github.com/ArduPilot/ardupilot

Tools: autotest: param_metadata: check values and increments are numbers

master ← IamPete1:paramParseNumeric

opened 11:22PM - 20 Oct 25 UTC

IamPete1

+13 -0

This would have caught the original error that https://github.com/ArduPilot/ardu…pilot/pull/31334 fixes. That PR will need to be merged first as master currently fails. This is the output: ``` Error in ../libraries/AP_Scripting/drivers/LTE_modem.lua At param LTE_MCCMNC Value not number: "AU-Telstra" Error in ../libraries/AP_Scripting/drivers/LTE_modem.lua At param LTE_MCCMNC Value not number: "AU-Optus" Error in ../libraries/AP_Scripting/drivers/LTE_modem.lua At param LTE_MCCMNC Value not number: "AU-Vodaphone" ```

It looks good but it now has conflicts.
Peter rebased it.

MergeOnCIPass.

UTC0708

This needs a Copter maintainer to merge it.

UTC0709

This needs a Copter maintainer to merge it.

Would be nice to have a separate error rate getter.
It’s a bit unclear what the error rate refers to. Not that this is PR is to blame for this.

UTC0716

github.com/ArduPilot/ardupilot

Plane: prevent bad climbs in auto mode

master ← peterbarker:pr/bad-plane-climb-fix

opened 10:32PM - 21 Sep 25 UTC

peterbarker

+54 -4

if we entered auto mode before we get our home location then we can end up subtr…acting a 0m AMSL from a supplied absolute altitude, making the relative-takeoff-altitude a Very Big Number Supplied autotest fails before and passes after. From before the fix: <img width="1933" height="919" alt="image" src="https://github.com/user-attachments/assets/c8cfc940-b621-48e1-85e0-6bf78d29a5f4" /> (I'm not even going to speculate on what's going on with that altitude at the end of the test, but you can see we climb to ~1000m AMSL rather than the mission-desired 650m) After the fix: <img width="1933" height="919" alt="image" src="https://github.com/user-attachments/assets/1c998fb6-1a3a-4644-bee8-ce6102b6da50" /> I'm labeling this for backport, but I'm dubious as this is probably a very long-standing bug. The pre-arm check is probably overkill as Plane currently requires position for arming anyway and home will get set rapidly from that. Might close a thin race.

Needs a Plane maintainer to merge it.

UTC0716

Would be nice for Randy to have a look.

UTC0718

github.com/ArduPilot/ardupilot

handle PLND_XY_DIST_MAX = 0

master ← AlexisMGL:zero_PLND_XY_DIST_MAX

opened 09:35AM - 18 Jul 25 UTC

AlexisMGL

+1 -1

I believe there's a bug in plane_precland.lua related to _PLND_XY_DIST_MAX._ …While testing Precision Landing in SITL/RealFlight, I noticed that setting _PLND_XY_DIST_MAX_ to 0 caused the drone to hover indefinitely — it never started descending. According to the parameter description (“Set 0 to always descend”), the expected behavior is that distance should not be a limiting factor when the value is 0. The issue was resolved after changing the value back to 10. It looks like plane_precland.lua does not currently handle the 0 case properly. I suggest we add a condition to explicitly allow descent when PLND_XY_DIST_MAX = 0. This would make the behavior match the documented intent.

Peter: This is a discontinuity in value behaviour, but it’s according to documentation.
George: I find this behaviour actually intuitive.

Merged!

UTC0721

github.com/ArduPilot/ardupilot

AP_SmartRTL: Adding a simple logic for checking circles

master ← ahmedmsalah99:smartrtl_fix_buffer_full

opened 05:30AM - 18 Jul 25 UTC

ahmedmsalah99

+5 -3

Addresses: #20716 Added a for loop for checking if a path point is close to c…urrent position. Number of points is not growing up anymore now if we go in circles. <img width="1046" height="661" alt="Screenshot from 2025-07-18 06-49-58" src="https://github.com/user-attachments/assets/44e4e34a-12c2-4e48-bdcf-8593ab1f5565" />

We’d need Randy to pitch in.

UTC0722

github.com/ArduPilot/ardupilot

Rename HAL_PROBE_EXTERNAL_I2C_COMPASSES to AP_COMPASS_PROBING_ENABLED

master ← peterbarker:pr/HAL_PROBE_EXTERNAL_I2C_COMPASSES-to-AP_COMPASS_PROBING_ENABLED

opened 10:53PM - 11 Jun 25 UTC

peterbarker

+239 -233

This is a no-compiler-output change. This PR also force it to have a value, n…ot just be defined. ``` pbarker@threads:~/rc/ardupilot$ ~/rc/ardupilot/Tools/scripts/filter_size_compare_branches_csv.py --hide-binary-identical /tmp/some.csv ----- --------- ----- ---------- ------ ---- ---------- ----- ----- --- Board AP_Periph blimp bootloader copter heli iofirmware plane rover sub ----- --------- ----- ---------- ------ ---- ---------- ----- ----- --- pbarker@threads:~/rc/ardupilot$ ```

Andy: Let’s make sure that we remember this change, when we back-port boards.
Peter: Left a note in the 4.6 Issues List.

Bumped for the next Dev Call.

UTC0732

Andy: I2C is quite slow, so when sharing DMA with the SD card, there’s the danger of failure to write in time.

But I’m okay with merging it, it’s specific to that vendor.

Merged!

UTC0734

We’ll need Sid to weigh in on this.

UTC0742

Andy: This is tested working.
Peter: Is the scale factor correct?
Andy: Haven’t checked.

I’ll make the changes then split this PR into two, one for the driver and one for the new board.

Peter: The PX4-2 connector description doesn’t match the silkscreen. Same for UART4.

The board_type ID is alrady taken.
Boards with 1M flash or less might not want this driver baked-in.

UTC0757

Amilcar: I can no longer access the parameter server, so I can’t test this anymore.

Perhaps my IP address changed and I can’t rsync in anymore.
This PR will allow GCSs (i.e. Mission Planner) to get the stable parameter .xml list, not the one from master.
I would like that we upload parameter descriptions for each patch release.

Peter: It would be nice if we uploaded them next to the board binaries.
Amilcar: I had suggested the same, but tridge wanted them in a separate, dedicated directory.
Peter: We could also be pointing the GCS to the correct parameter file in the manifest.json.

In any case, if you can set it up to push the parameter metadata next to the binaries, I’ll advocate for it.
I’ll go and refactor param_parse.py into a importable module.