-
Notifications
You must be signed in to change notification settings - Fork 13.3k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. Weβll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[v1.14 Release Candidate] - Flight testing & Flight Issues (logs) #21358
Comments
All on 1.14b on FW. :) |
Btw, I would like to note that since we agreed to wait for the rebasing of main branch into release/1.14 until we resolve the outstanding issues in our project board: https://github.com/orgs/PX4/projects/45, I'm not sure if we can consider current beta releases as a valid one or not π€· |
Hey everyone, did some testing today in a couple of different modes and wrote up some stuff I thought might be of interest. Manual Mode Altitude Mode Stablized Mode Position Mode Attached below are flight logs for the altitude mode and position mode behavior described above. Let me know if I can provide any more clarifying information. Altitude Mode: https://logs.px4.io/plot_app?log=c597875e-feab-40d4-a633-a709a72dbe99 |
@junwoo091400 anything tagged v1.14 (beta/rc... etc) or coming from the main branch (before the official release or branching out) should be considered 1.14, and as such the logs are definitely helpful. |
https://logs.px4.io/plot_app?log=7324a947-33f9-4c63-a5a1-2ce1c7208560 Flew my hex for a quick few minutes in position mode. No problems noted. |
When checking out the beta and updating submodules. |
This is a 250 quad with ARKV6X, ARK GPS, and ARK Flow. I think my rates tuning was way off to start. The vehicle always wanted to pitch backwards on takeoff. After some rates tuning it seems okay. @dagar I was able to arm in position mode with optical flow for some reason this time. https://review.px4.io/plot_app?log=96d5db15-dbcd-4b5e-aa64-f85d6ca41da4 |
This issue has been mentioned on Discussion Forum for PX4, Pixhawk, QGroundControl, MAVSDK, MAVLink. There might be relevant details there: https://discuss.px4.io/t/px4-maintainers-call-april-04-2023/31415/1 |
@MDecarabas Thanks for the tests. I inspected the altitude issue and the drone is constantly climbing because the GNSS altitude tells that the drone is actually descending (compared to the baro that seems to be more reasonable). You could change the height reference to baro (`EKF2_HGT_REF) to lower the effect of the GNSS height drift or even disable the GNSS height fusion to let the baro do the job alone (EKF2_GPS_CTRL). See https://docs.px4.io/main/en/advanced_config/tuning_the_ecl_ekf.html#height for more details. |
This issue has been mentioned on Discussion Forum for PX4, Pixhawk, QGroundControl, MAVSDK, MAVLink. There might be relevant details there: https://discuss.px4.io/t/px4-community-q-a-april-05-2023/31435/1 |
This issue has been mentioned on Discussion Forum for PX4, Pixhawk, QGroundControl, MAVSDK, MAVLink. There might be relevant details there: https://discuss.px4.io/t/v1-14-beta-call-for-flight-testing/31551/1 |
This issue has been mentioned on Discussion Forum for PX4, Pixhawk, QGroundControl, MAVSDK, MAVLink. There might be relevant details there: https://discuss.px4.io/t/px4-maintainers-call-april-11-2023/31524/1 |
This is a 250 quad with ARKV6X, ARK GPS, and ARK Flow. Very windy day. https://review.px4.io/plot_app?log=1263490b-aecf-420e-af44-b1013802bd0a 650 quad with ARKV6X, Dual ARK RTK GPS, ARK Flow, and a dronecan smart battery. Moving baseline for yaw. Very windy. https://review.px4.io/plot_app?log=95943f96-4039-4b80-ad57-92311a29e532 Different 250 quad with Pixhawk 4 Mini, ARK GPS, and ARK Flow Rev 3 with PAA3905 and AFBR_S50LX85D. Very windy day. https://review.px4.io/plot_app?log=0b1a730d-147b-47e3-b25e-c56fe27dacf8 The optical flow bug is still an issue when the sensor is touching the ground. On the two vehicles with longer legs, I am able to arm. |
From sturner in discord: https://discord.com/channels/1022170275984457759/1022185721450213396/1092830779429625977 Just tested with latest 1.14 and an ARK Flow. Possible that we have other issues with our configuration. This is with external vision inputs enabled where we are able to arm: With external vision inputs disabled, we are unable to arm: https://review.px4.io/plot_app?log=c2b658cc-72ea-4b9f-9153-c13852865406 |
We are preparing for tests with 1.14.0-beta2 and we experienced the same issues as the ones described by @MDecarabas #21358 (comment) We are working with fixedwings, so I cannot fully reproduce everything, but, in particular, we were also experiencing a similar problem when trying to arm in manual mode: "Unable to arm because throttle is high when using sbus rc.". When debugging, we've realized that manual_control_setpoint orb topic contained throttle from 0 to 1 and I understand that in 1.14 it should be from -1 to 1. After some extensive debugging and trying random things, we've managed to make it work; we think the problem was related to the fact that throttle output of RC was inverted. @MDecarabas can you maybe confirm if your RC was also set to invert throttle? |
Tested with the main branch commit on ATL Mantis Edu (Quadcopter) today. Yaw estimate error bug test
Random 100% thrust issue test I couldn't reproduce the random 100% thrust scenario. https://logs.px4.io/plot_app?log=60f24c0a-d422-472e-b5ff-2cea91ec4978 |
A few of our flights with a fixed-wing model. All done with 1.14.0-beta2. During the flights, everything went smoothly. The only problem we had was with calibrating RC with the inverted throttle as mentioned in one of the comments above. Is it a known problem, or should we raise it as an issue? |
I've raised the PR #21576 which solves this problem. |
Flight before was completely fine and from takeoff to landing in mission but it doesnβt have PR 21988 in it. I should have been more clear, same pr as above meaning same as the previous flights I posted. |
Thanks a lot for doing that. Appreciate the work! Too bad there is nothing point to something clear. Not sure what to think/do at this point. |
hello guys I wanna share my logs from my quad copter's successful and unsuccessful fligths. It has a body made by ourselves, around 35kg. Px4 version=v1.14.0 (RC) ([5359fe5] my first look from good flight. it is fly very good in mission mode and returned with pilot. after 4 5 hours. we try to fly without change anything. after we plugging. we start the mission and we see something wrong we landed with manuel. and I reupload mission again. nothing change. frame and all electronics looking okey. Our drone video while flying: |
Exactly, I wonder why. The GPS itself looks fine. Any idea @bresch? Now when it comes to the crash, it looks to me like it run out of battery. The voltage suddenly plummets, doesn't it? |
Could something be loose? Vibration is so high first vs last flight. |
Is there anything else that can be done past this? |
@julianoes @ryanjAA today I will fly with my pixhawk 4 + 2 GPS and 1.14 version and changed RC transmitter. |
@dagar @KonradRudin while SITL testing this fix #21988 I found another corner case where takeoff command engaged trim throttle somehow with no valid position setpoint and led to fly-away -- im assuming because i hadnt re-uploaded the mission on the first go after building, and tecs was likely initialized at some point with trim throttle. Then it stayed in PX4-Autopilot/src/modules/fw_pos_control/FixedwingPositionControl.cpp Lines 2365 to 2366 in 0f260fc
Flags on arm indicating this would fall into PX4-Autopilot/src/modules/fw_pos_control/FixedwingPositionControl.cpp Lines 714 to 812 in 0f260fc
log: https://review.px4.io/plot_app?log=98c21633-7be5-4ce4-a77f-45baf6b19990 Brain is fried for tonight. Will need to look at this tomorrow. If you have any ideas for a patch that isnt a complete refactor... very open to recommendations. We REALLY need to do a broad refactoring of FixedWingPositionControl for 1.15 to clean up state machine and APIs ... and unit/integration test the hell out of it so it cant devolve again. |
@dagar - another tangent.. (scope is creeping quickly here) - we totally let this one slip some time ago #21581 -- i didnt want to hack in the bool as it's done there.. but it would be the safest way to get the proper airspeed septoints for takeoff into this release. i can take this one back up again tomorrow as well. |
Hi, I have been trying to configure a quad tailsitter on RC-1, using a cubeblack and last QGC V4.2.8. I have bench tested it and found this: 1-Can't calibrate ESCs in the new range, since cubeorange and black do not work in QGC in windows via USB. I managed to calibrate 3 ESCs using the actuator testing but 3 motors spin at idle an one is stopped. Finally I calibrated the ESCs using a servo tester. All other calibrations done via telemetry. 2- The message Critical: Preflight: CPU load too high: 95.2% appears several times 3- I was not able to configure properly the elevons gain. Actuators/ Pich & Yaw Scale does not seem to work. Elevons move with random gains with different values of the parameters set. Both elevons same movement in pitch with different gains: Movement not proportional to gain: CA_SV_CS0_TRQ_P=0.5 4- It can't do the transition on stabilize. 5- It does not perfom the transition properly in alt hold. 5- Preflight fail on flight on FW mode: |
I made autolanding Tests with the fix #21988 Observations in all three flights: WARN [health_and_arming_checks] Preflight Fail: height estimate not stable Flight 1 Handlauch (Manual, stabilized, Mission) Flight 2 Handlauch (Manual, stabilized, Mission) Flight 3 Launch direct into Mission |
It was always slowly yawing to the north. Did a mag cal and this is slightly better. Still has the bug where it slowly descends until it lands. |
@VTOLDavid thanks for the test, here some feedback for your raised points:
Could it have been triggered by parameter changes? Then it's not concerning. The average load in your shared logs seems okay (80% max).
What did you try to achieve with the tuning of the elevon gains? I would recommend to leave them as default and then tune the responsiveness of the controller through the PID gains of the fixed-wing rate controller (which is the controller controlling the elevons).
In Stabilized mode you have to manually pitch forward until the transition pitch is achieved and the airspeed is reached. Were you aware of that and is the issue something else?
That sounds severe. I also noticed that in your logs the attitude and rate setpoints are not logged, which could be related. I will check this further.
It's due to pitch angle errors. The first time in FW (I guess that can be expected with the issues noted in 5), and then the second right after the backtransition is over. |
It looks like the issue is that you don't have a valid global position, and thus no Home position resp. local_position.ref_alt is not set. We don't handle that in the FW position controller cleanly (the controller needs absolute altitude, and thus in your case gets a NAN as input).
|
I do not remember changing any parameters when that message was shown, but that error has not being shown again in later tests
I want to have the following gains:
We need full deflection of elevons for Yaw in MC and small deflection in pitch, but use the FW controler gains FW flight. For example, how can I get this configuration in the new system?
I had no idea about this. What parameter defines transition pitch? How is the procedure? Switch to FW mode an then pitch forward to accelerate?
But how is possible that it gives a preflight fail if it is armed? |
@sfuhrer We usually have GPS, but we fly stabilize and alt hold modes without GPS to be sure we can still fly if GPS fails. |
Test hop of an old quad that I just converted to an ARKV6X. 6s, Here3, gimbal. https://review.px4.io/plot_app?log=c6db6c7d-255c-46f9-b49d-5c8f83dd6e6a |
If you want more deflection for yaw then for pitch you can reduce the yaw effectiveness and increase the pitch one. E.g. 0.3 and 0.7.
It is currently hard-coded to 60Β°. Would you want to be able to tune it and thus would rather want a parameter for it?
Yes exactly. It's been like that for a while now but I also don't like it that much, I'm thinking about also doing the automatic pitch ramp in Stabilized.
The "preflight" shouldn't be in the message, agree. |
I will test it. Have you tested if the deflection of the surfaces for a tailsitter is correct? I saw a very random behaviour in my tests. You have the logs in one of the previous posts.
As I understand for transition to FW you need:
Once the conditions are meet direct change to FW Is that correct? Back transition to MC is as usual, using the transition switch? I see 2 problems:
I like very much the v1.12 pitch ramp approach in Stabilized where you can control thrust. Is very helpful for tuning the transition of large or high speed stall speed UAVs, where you have very narrow range between stall and airframe damaging G forces. Specially the transition back, where if it is done too slow the plane can stall and if it is too fast the airframe can be damaged. In these tests if pilot can control thrust can avoid a crash. |
I think we might have a big release blocker in that camera trigger and capture do not appear to be working in 1.14. |
Ugh - ya thats an important one. |
Thank you to everyone who contributed in the v1.14 beta testing! @mrpollo shall we close this issue? |
Thanks @junwoo091400 and everyone else who helped get this out the door the release is out and we can finally close this issue π |
This issue has been mentioned on Discussion Forum for PX4, Pixhawk, QGroundControl, MAVSDK, MAVLink. There might be relevant details there: https://discuss.px4.io/t/cube-o-px4-v1-14-0-erroneous-pitch-roll-setpoint-in-autonomous-mode/35557/4 |
β About
π‘ For the upcoming v1.14 release, we are asking the Community for flight tests. This issue will be used for keeping track of all the tests from the Community.
If you're unsure about upgrading, please read the preliminary release notes on the PX4 docs. Pay close attention to the Upgrade Guide before deciding if v1.14 is right for you.
Pending Issues
β Why we need flight testing
π Thousands of people around the world depend on the stable release of PX4.
ποΈ Before every major release, as much testing as possible is needed to ensure that our release actually works!
π― Simply said, more testing = better release
π€ We would like to encourage every community member to join us in improving PX4!
π€· How you can help with flight testing
π± Even if you have never contributed a flight testing report, fear not! Please follow the steps below for the details of how you can contribute flight testing.
π Flight testing matrix
π Here are the table of things we need tests for.
This will constantly get updated, and your contribution to testing the missing slots (tests needed) will be greatly appreciated!
Random 100% ThrustHow to flash v1.14 beta firmware
This should automatically flash & reboot your flight controller with the latest v1.14 beta release!
How to share your log
First, retreive the log from the flight controller in QGC Q icon > Analyze Tools > Log Download, and find the log you want to download.
Then:
When posting on this thread, please mention what kind of test it was (especially if it is one of the testing matrix options) & general info behind it!
Questions?
Feel free to ask any questions regarding flight testing in the #dev-team channel in the discord server!
The text was updated successfully, but these errors were encountered: