Dev Call June 26, 2023

AP_Motors: Heli: remove incorrect set range call

ArduPilot:master ← IamPete1:heli_rsc_channel

opened 12:43AM - 05 May 23 UTC

+0 -3

`SRV_Channels::set_range(SRV_Channels::get_motor_function(_aux_fn), 1000);` I…t passes `_aux_fun` heli_rsc. That is 31. It then incorrectly tries to convert "motor" 31 to a servo function. We take away 8 and then offset by `motor9` (82). 31 - 8 + 82 = 105. (Scripting 12) The function should just be: `SRV_Channels::set_range(_aux_fn, 1000);` But as I say its unneeded anyway, as its set in the servos lib. This is also proved by the fact that it has been working up to now.

AP_Compass: stop CAN auto replacement by default add option to re-enable

ArduPilot:master ← IamPete1:compass_remove_CAN_autoreplacment

opened 04:32PM - 06 May 23 UTC

+51 -46

Currently if you have multiple CAN compasses but only use and calibrate the firs…t if that compass ever does not show up at boot it will be replaced by the second. This removes that ability so compasses cannot change ordering unexpectedly. This can happen if the flight controller is powered from USB and the primary compass from battery. Because the second compass had not been calibrated its dev-id is never saved, this means the `expected_dev_id` is 0 and it is detected as a replacement compass. This can be avoided by setting all to use and calibrating all and then deselecting use for the unwanted compasses. A alternate solution would be to save all dev-id's in a calibration to "lock" the ordering, however we would then also need to force offsets to 0 for the un-calibrated compasses so they still show up as un-calibrated. TLDR: You have to got to a GCS to calibrate the new compass that has come in as a replacement anyway, I don't see that manually removing the old one and moving the new one up is significantly more effort.

Copter/Plane/Rover: Add a bitmask parameter to block mode entry from GCS

ArduPilot:master ← IamPete1:Copter_GCS_mode_block

opened 12:16PM - 08 Jun 23 UTC

+217 -0

This adds a new parameter to block mode entry from GCS. This does not block mode… entry from any other source. The idea is to prevent GCS mistakes from putting the vehicle into a mode that the pilot is not experienced with or expecting. For example if the GCS operator mean to hit alt-hold but accidentally selected acro. ![image](https://github.com/ArduPilot/ardupilot/assets/33176108/89e1f212-6f50-4e0c-b338-98ec0fc1b2d2) This does not prevent entry from any other source, the main other candidate would be RC, but in that case you can just not setup switches for modes you don't want to use. There are some modes that it is not allowed to disable, stabilize, alt-hold RTL and land. Since this is only GCS I can see the case for also allowing disabling stabilize and alt-hold, switching into them without coordination with the pilot will also cause issues. Unfortunately there are too many modes to map over to the bitmask using the existing mode numbers, so I have had to re-map using a switch case. This is just copter, but if we think this is worth while I think it makes sense to do the other vehicles too. This work is sponsored by ARACE.

Add and use AP_SBUSOUTPUT_ENABLED

ArduPilot:master ← peterbarker:pr/AP_SBUSOUTPUT_ENABLED

opened 08:54AM - 15 Jun 23 UTC

peterbarker

+33 -8

``` Board AP_Periph blimp bootloader copter heli iofirmware … plane rover sub Durandal * * * * * * * HerePro * Hitec-Airspeed * KakuteH7-bdshot * * * * * * * MatekF405 * * * * * * * Pixhawk1-1M-bdshot * * * * * * f103-QiotekPeriph * f303-Universal * iomcu * revo-mini * * * * * * * skyviper-v2450 -496 ``` I've run this through `./Tools/autotest/test_build_options.py --no-run-with-defaults --define-match-glob="*SBUSOUT*" --build-targets=copter` and it passes, showing 536 bytes for this feature on DevEBoxH2 This is probably a candidate for `save_some_flash.inc`.

Add and use AP_SCHEDULER_EXTENDED_TASKINFO_ENABLED

ArduPilot:master ← peterbarker:pr/AP_SCHEDULER_EXTENDED_DEBUG_STRINGS_ENABLED

opened 11:11PM - 15 Jun 23 UTC

peterbarker

+26 -13

``` Board AP_Periph blimp bootloader copter heli iofirmware … plane rover sub Durandal * * * * * * * HerePro * Hitec-Airspeed * KakuteH7-bdshot * * * * * * * MatekF405 * * * * * * * Pixhawk1-1M-bdshot 24 24 16 16 24 16 f103-QiotekPeriph * f303-Universal * iomcu * revo-mini * * * * * * * skyviper-v2450 * ```

ChibiOS: added force mutex release for deadlock recovery

ArduPilot:master ← tridge:pr-force-mutex-release

opened 11:59PM - 15 Jun 23 UTC

+76 -2

This allows recovery from a mutex deadlock due to lock order violation by forcin…g the release of the mutex that is holding up the main thread. Without the forced release we would watchdog. This type of deadlock should never happen, but we have had bugs like this in master, for example the bug fixed by this patch which was in master for a short time: https://github.com/ArduPilot/ardupilot/pull/23982/commits/e795dc1bf1ae2eb4bac325068c8825f7d781ddc8 When the deadlock happens the monitor thread detects it and uses chMtxForceReleaseS() to force the release of the mutex. That has the potential to cause data corruption, but it is better than the alternative of a watchdog This PR also adds test code to generate a deliberate deadlock Log message: ``` 1970-01-01 10:00:18.122 DLCK {TimeUS : 18122395, SemLine : 3187, ThdName : io, MtxP : 604463532} ``` Display to user: ![image](https://github.com/ArduPilot/ardupilot/assets/831867/f442dad7-1977-4dc0-8970-2f868f7bfec8) This was test flown without any issues on a CubeOrange quadplane

Plane: improved fwd throttle during VTOL landing

ArduPilot:master ← tridge:pr-approach-throttle

opened 12:04AM - 17 Jun 23 UTC

+28 -7

this sets up the vfwd integrator more reasonably when we are in POSITION1 stage …of VTOL landing. We need to have enough throttle to cope with a headwind, but want it lower when we are at or above our target closing speed so can minimise the amount of pitch up This also makes the landing_desired_closing_velocity() consistent with the landing speed used in approach, using average of airspeed min and cruise speed if TECS_LAND_ARSPD is not set Test flown successfully on a MakeFlyEasy Hero with CubeOrange

AP_Scripting: Only allow scripting restart if disarmed

ArduPilot:master ← IamPete1:scripting-restart-disarmed-only

opened 03:07PM - 20 Jun 23 UTC

+8 -2

This requires the vehicle to be disarmed for scripting to be re-started. Scripts… can still be stopped while armed. This is because I found another case where scripting restart causes a flight controller lockup and watchdog on real hardware. In that case I restarted the `math.lua` test. There have been several of these that we have now fixed, but there might be more we have not found. Crash dump output and watchdog decode from that one are here: https://discord.com/channels/674039678562861068/719466647710072832/1120113508877598780

ADS-B: Add a MAVLink in only type

ArduPilot:master ← WickedShell:wickedshell/adsb-mavlink-in-only

opened 05:26PM - 20 Jun 23 UTC

WickedShell

+27 -13

This does two main things: 1. Fixes a bug where we spammed "ADSB: Found transce…iver on channel" after we lost the transciever health report. Unfortunately this was incredibly confusing because we were generating this on a PingRX, which is fundamentally a Rx only device. (To make it worse we spammed this at update() rate of 10Hz, which is highly alarming on the GCS). 2. It doesn't make sense to send transciever out control messages to a device that is Rx only. This stops us from trying to get a Rx device to transmit Along these lines it changes the CubePilot boards to default to Rx in only, which is the hardware they ship with by default. ( @bugobliterator ) I chose to keep the existing `ADSB_TYPE` 1 as not changing behavior, otherwise I would break anyone who currently had setup uAvionix MAVLink out. The other option I considered was adding an options parameter, which would be reasonable, but the initial discussions with @magicrub revolved around adding a new type, which is what this presents. This was tested using a PingRx receiver, and some extra debug prints. I don't have anything that actually transmits to test with, but adding prints in the opposite case of the else's left me pretty convinced that this is working as intended if you have an actual transceiver.

AP_Mount: add Xacti camera/gimbal support

ArduPilot:master ← rmackay9:mount-dronecan4

opened 04:18AM - 22 Jun 23 UTC

rmackay9

+734 -16

This PR adds support for the Xacti camera gimbals ([Xacti web page](https://xact…i-co.com/service/drone_camera/), [video](https://www.youtube.com/watch?v=jZszQ4OmfVQ)). Enhancements include: - DroneCAN module updated to include new Xacti specific messages (see PR https://github.com/dronecan/DSDL/pull/20) - AP_Mount gets a new DEVID parameter which is set-and-saved by the Xacti backend when it first discovers a new gimbal - AP_Mount_Xacti backend allows these controls: - Gimbal attitude control using rate or angle targets (note limitations on rate control below) - Gimbal attitude feedback - Camera take picture, record video, auto focus These known issues may be resolved in a follow-up PR. - "Lock" (aka "Earth frame") mode while under RC rate control is not supported (e.g. Pilot is using RC sticks to control the yaw rate of the gimbal) because the gimbal itself simply doesn't support earth-frame control. We could pontentially add support in a follow-up PR by simulating rate control but actually use the angle controller. - Year, Month, Date are not sent to gimbal (AP's RTC library doesn't provide this) - Zoom and Manual focus are not supported (camera limitation) - Remove commented out send_camera_information() method This PR also includes these drive-by fixes: - AP_DroneCAN loses an unused semaphore - AP_Mount losses and unused "center" method - spelling fixes for AP_Airspeed and AP_Mount This PR has been extensively tested on real hardware (a Xacti CX-GB100) This PR relies on this DSDL PR: https://github.com/dronecan/DSDL/pull/20 Below is an image from desktop testing of taking pictures and recording video ![image](https://github.com/ArduPilot/ardupilot/assets/1498098/d0c0e6a9-d4b4-4786-a9ee-85ffe9b98d4a) Below is an image from MP's mavlink inspector showing the CAMERA_INFORMATION and CAMERA_SETTINGS messages being correctly sent to the GCS. ![image](https://github.com/ArduPilot/ardupilot/assets/1498098/e22f97ce-3011-4adb-bb90-e1e63decf32d) Wiki PR is here https://github.com/ArduPilot/ardupilot_wiki/pull/5201

AP_BoardConfig: fix conflicting param indexs

ArduPilot:master ← IamPete1:src_rtscts_conflict

opened 10:44PM - 22 Jun 23 UTC

+7 -3

Thanks to @andyp1per for spotting this. We have conflicting parameter indexes… between `BRD_SER3_RTSCTS` and `BRD_HEAT_LOWMGN` and between `BRD_SER4_RTSCTS` and `BRD_SD_MISSION`. This is because the higher index params were hiding higher up in the table. This is not caught by CI because we don't have RTS and CTS pins or parameters in SITL. `BRD_SER3_RTSCTS` is easy, nothing actually uses it so we can just change the index. `BRD_SER4_RTSCTS` is used by four boards: PH4-mini, PH4-mini-bdshot, Swan-K1 and TBS-Colibri-F7. But the RTS CTS setting are rarely changed, so we will just loose the config on those boards. Alternatively we could move `BRD_SD_MISSION`, it hasn't been in that long, but all "high performance" boards would loose the config. Checked what is using what by running `./Tools/scripts/configure_all.py --no-bl` and greping for `HAL_HAVE_RTSCTS_SERIAL3` and `HAL_HAVE_RTSCTS_SERIAL4` in the generated `hwdef.h` files.

AP_RCProtocol: Refactor SUMD to remove legacy cruft

ArduPilot:master ← WickedShell:wickedshell/sumd

opened 12:21AM - 23 Jun 23 UTC

WickedShell

+20 -158

Of particular note this removes the partial SUMH support. This has been broken a…s far back as 2018, as we only called add_input() if we were in the SUMD or SUMD failsafe cases, which meant SUMH didn't work. This commits to that path, and removes all the 8 bit CRC, SUMH support. This also refactors a number of things for readability (condensing white space, simplifying guards around debug code). Many of the debugs could probably be removed, but this simplifies it. This was tested against the RCProtocol example test, and everything there reports a pass. I do not have a real SUMD receiver to test against.

AP_BattMonitor: DroneCAN: allow battery reset if not using CAN SoC

ArduPilot:master ← IamPete1:DroneCAN_Battery_Reset

opened 10:44PM - 23 Jun 23 UTC

+41 -6

This allows DroneCAN battery backends to be reset correctly if were not using th…e returned SoC. This is needed for devices such as: http://www.mateksys.com/?portfolio=can-l4-bm This also changes to correctly bounce the battery reset if we are using the provided SoC. This work is sponsored by ARACE.

GPS: fixed for RTCM injection

ArduPilot:master ← tridge:pr-rtk-DroneCAN-fix

opened 05:49AM - 24 Jun 23 UTC

+72 -4

This fixes the following issues: - if the RTK base is the 2nd DroneCAN GPS the…n it couldn't receive RTMStream data from GCS - the buffer sizes on the GPS uarts were too small for RTCM data to be fully reliable - increase CAN priortity of MovingBaseline data - increase CAN buffer size when doing moving baseline - fixed a power reset issue on the HolybroG4 GPS by setting the NRST_MODE bits in the option registers This test by Foxtech shows it fixes their issue ![image](https://github.com/ArduPilot/ardupilot/assets/831867/bc800ba7-f4bf-4084-871f-38e4a47df3db)

CAN: support uCenter over DroneCAN

ArduPilot:master ← tridge:pr-uCenter-over-DroneCAN

opened 10:06AM - 26 Jun 23 UTC

+172 -21

This allows you to attach uCenter to a DroneCAN GPS, allowing for debugging sign…al issues, changing configuration etc It is enabled by setting DEBUG=4 in the GPS node Then use a new DroneCAN GUI tool with uCenter Panel and connnect on TCP port 2001 ![image](https://github.com/ArduPilot/ardupilot/assets/831867/a83b8359-cb1b-4737-847d-db7cea03c58d) ![image](https://github.com/ArduPilot/ardupilot/assets/831867/bfac979f-8b95-484c-b134-9e6968a20df5) ![image](https://github.com/ArduPilot/ardupilot/assets/831867/d595ac45-2460-402d-afbc-37a7190ba633)

Plane

Plane: stable 4.4.x issues list

opened 03:33AM - 01 Dec 20 UTC

Plane VTOL-Plane DevCallTopic ReleaseAdmin

This is an issue to gather key issues for the plane 4.2 release - [ ] display …message on pilot override active in VTOL pos controller (can prevent descent stage) - [ ] add check for RC3_MAX at least 100 above RC3_MIN - [ ] https://github.com/ArduPilot/ardupilot/pull/13760 - [ ] pre-gen lua mavlink on firmware server - [ ] we should scale baro height process noise with pressure altitude - [ ] CANFD incorrect timing, giving bus off errors - [ ] add tcp option for support.ardupilot.org - [ ] lane switch add delay till all lanes have fused, or ensure consistent for some time period (eg. 50ms) - [ ] EK3 gate mod: https://github.com/priseborough/ardupilot/commits/pr-ekf3-gps-velpos-igate-mod - [ ] make stabilize mode use proper rate control - [ ] AUTOTUNE needs to reset SMAX to 0 during twitch - [ ] possible problem with disconnected RC on H16 - [ ] resting voltage can be very bad - [ ] quadplane precland - [ ] fix i2c probing of LEDs - [ ] add CAN pkt logging to DroneCAN gui tool and ardupilot, and log loading in gui tool - [x] possible bug in ICE restart - [ ] mavcan to multiple clients at once - [ ] can gui tool panel for secure commands - [ ] add rate limited filter to airspeed to prevent large spike from corrupting airspeed - [x] default TECS_LAND_ARSPD to mid-point of cruise and min airspeed - [ ] propogate Dmod up to angle loop - [x] look at using -g -O2 for stable builds for better crash dump analysis - [ ] ATT.DesPitch and ATT.Pitch need to use the same pitch reference (ie. TRIM_PITCH_CD) - [ ] motor_lost_index never reset and not exposed in lua - [x] work out why 2048 byte reads in #23455 failed - [x] allow clear mission when disarmed in AUTO ? - [ ] add waf docs target? - [x] possible bug in RTSCTS? - [ ] Q_M_PWM_TYPE set in defaults.parm in ROMFS doesn't work for DS150 - [ ] GSF yaw errors: https://github.com/ArduPilot/ardupilot/issues/23224 - [x] mission on sdcard - [ ] more flexible M10 config: https://www.rcgroups.com/forums/showpost.php?p=50442153&postcount=8212 - [x] fix pitch limit in FBWA for transition: https://github.com/ArduPilot/ardupilot/pull/22730 - [ ] after a bounce off the ground, PIQA.I can peg high (seen in Alia model) - [ ] switching to RTL when in landing sequence to cope with VTOL motor start fail doesn't change to fw mode - [ ] ARSPD_TUBE_ORDER=3 to auto-detect order - [ ] MS4525 probe all buses not working in 4.3 after update from 4.2 - [ ] rudder arming causes yaw in AUTO takeoff #22665 - [ ] VTOL approach can turn overshoot if WP_LOITER_RAD is small - [ ] COMPASS_DIA_X of zero not treated as 1.0 #22666 - [x] level cal not clearing 3d accel cal needed msg? - [ ] CubeOrangePlus param diff BAT*PIN doesn't show changes - [ ] not using full throttle in non-airspeed takeoff: https://github.com/ArduPilot/ardupilot/issues/22572 - [ ] random voltages with voltage pin -1 - [ ] large over-tune can cause VTOL controller to climb-away, possibly lower default SMAX? - [x] with RTL autoland: Mode change to AUTO denied, in fence recovery - [x] MS5607 sub for MS5611 for Pixhawk1 Marco - [ ] annoying msg Cannot calibrate with Q_TRIM_PITCH set - [ ] not fully resetting tdrag takeoff state on new takeoff - [ ] ICE idle governor in Q modes while disarmed - [ ] Camera gimbal settings seem to be gone from QGC with Plane 4.3. ? (from fredrikfalkman) - [x] receive DO_AUX_FUNCTION in lua scripts - [ ] need better BVLOS advice and testing: https://github.com/ArduPilot/ardupilot/issues/21876 - [ ] throttle slew limit when re-starting engine on RPM outage - [ ] reduce XY gains when rate gains activate Dmod? - [ ] incorrect CRSFv2 detection: 2022-10-02 14:07:47.120 CRSFv2: requesting RX device info - [ ] remove loop rate dependency on power limit for fixed wing - [ ] limit integetrator buildup longer for catapult launch - [ ] allow idle governor to control ICE idle in Q modes when disarmed - [x] scale down rate PIDs during back-transiton based on airspeed? - [ ] auto-lower pos controller gains when have attitude error? - [ ] should use EKF height and vertical velocity estimate not DCM on loss of GPS when in quadplane mode - [ ] add package drop support - [ ] always allow at least 10 deg roll limit for fixed wing - [ ] add arming check for VTOL I gain <= P gain on roll/pitch - [ ] motors spin sometimes on soft reboot of HWing - [ ] test ublox fw update with passthru - [ ] before triggering Q_ASSIST_SPEED, check EKF fusion of airspeed, if not fusing, delay assist speed - [ ] add mission and fence extraction to MAVExplorer - [ ] add tailsitter tune param for P, I, D sequence - [ ] option for median filter between synthetic airspeed and dual-airspeed - [ ] re-load mission on arming in AUTO? See https://discuss.ardupilot.org/t/striver-mini-vtol-4-1-quadplane/71805/185?u=tridge - [ ] rapid switching between EKF and DCM on baro error: https://discuss.ardupilot.org/t/excessive-ekf3-dcm-switching-in-plane-4-1-6/82966/5 - [ ] look at methods of handling flash parity errors on H7, possibly using HW CRC - [ ] obey primary sensors with EK3_AFFINITY - [ ] motors spinning when they shouldn't? https://github.com/ArduPilot/ardupilot/issues/16630 - [ ] force arm bypass of logging start and fence auto-enable - [ ] need force_descend change for quadplane: #17824 - [ ] surface control while booting not working: #17966 Features for later 4.2.x releases: - [ ] sort out CAN baro ordering for plane, maybe option bit in BARO_OPTIONS, on plane default to can baro last - [ ] support rangefinder correction of AP_Terrain data, with configurable time constant

Copter