Replies

    • Developer

      Hi Mihai,

      This is strange. Your copter is going to full throttle but still descending.

      Could you turn on motor logging and repeat the problem.

      The only thing I can think of that fits the symptoms is strong oscillations in your control loops. I can't see any logging that can discount this without motor logging or IMU logging.

      • Thanks for checking Leonard, I will turn them on and repeat. The drone seems super-stable, despite some wind. I got the motor information (they went down from 1600 to 1200) from the mavlink log from Mission Planner.

        What seems to have caused the saturation in Throttle Out is a feeling of falling (CRate) which goes as low as -100cm/s around that time (could be z gyro noise?), but why was that large throttle out translated to reduce motors is beyond me. One thing that crossed my mind is that perhaps there is a bug in the hex X code? Most people run with quads, and I didn't see yet somebody with a Hex X frame.

        Either way, I"ll enable IMU and Motors and try again.

        Thanks for looking at the logs,

        Mihai

        • Hi Leonard,

          It took two more crashes, but now I know what happened - it's a firmware bug. It's repeatable. I crashed five drones (really 2 drones five times): three times from loiter and twice from RTL. The crash from RTL is immediate and deterministic as long as the return altitude is different from the current altitude, the crash from Loiter takes some time and is not deterministic.  In both cases the symptoms are similar: at some point the output to one or two motors goes to 12xy and 100-200ms later all six motors get exactly the same low value 12xy and the drone falls from the sky (as normal hover is 1600+). Since one or two of the motors shuts down slightly earlier, the drone also flips while falling down.

          I have full datalogs (with IMU and RCOU) as well as telemetry logs and videos documenting both bugs (probably one single root cause triggered both by LOITER and RTL), but I assume that here is not the right place to detail the bugs.

          BTW, the bugs are at least in APM 3.1 rc8 through the current one (3.2.1 I believe), but not in 3.0.1 which works well (RTL glitches slightly, but does not fail - the bug may be in the fix for this slight glitch). It may also only affect X Hex frames (didn't check it on any other frame).

          Thanks again for looking in the logs and suggesting to enable IMU and MOTORS (RCOU) - this confirmed that indeed was a software problem, not parameters, not hardware, not ESC, no connections, etc.

          I'm not sure where is the best place to post a detailed description of the bug - I'll look around.

          Best,

          Mihai

          rtlFromAutoCrash.log

          loiterCrash2.log

          https://storage.ning.com/topology/rest/1.0/file/get/3691109053?profile=original
          • Just for the record my now temporarily decommissioned Flamewheel Hex showed no such symptoms running autotuned 3.1 on an APM 2.6. Lots of loitering, lots of flying in stab, not so much RTL but when RTL was invoked it was usually with an altitude change. Sorry I can't try to duplicate further as I've plundered the motors and won't be replacing them for a while. 

             

          • Developer

            Thanks for your work!!!

            I will look through the logs.

            Start to post the detailed description here then copy and past it to the issues list when you are ready.

            Chat soon,

            Leonard

  • Hello,

    I re-tested the autotune on my quad:

    - qav400 with hard mount gimbal in front
    - 8x5 @ 1000kV
    - 3 Ah @ 4S 
    - 1.6 kg AUW 

    Here is what I got after a successful autotune :

    RATE_PIT_D 0,015

    RATE_PIT_I 0,2249999

    RATE_PIT_P 0,2249999

    RATE_RLL_D 0,004

    RATE_RLL_I 0,065

    RATE_RLL_P 0,065

    STB_PIT_P 8,625

    STB_RLL_P 8,625

    I'm not entirely satisfied with the results, if you pay attention to the noise of motors on this video you will see that high frequency vibrations occur during aggressive movements.

    Do you know what parameters should I change?

    I already tried :

    - Reduce RATE_RLL_P and RATE_PIT_P by 20%

    - Lower STB_RLL_P and STB_PIT_P to 7

    - Divide RATE_RLL_I and RATE_PIT_I by 2

    - Double RATE_RLL_I and RATE_PIT_I

    without significant improvement, any help will be welcome !

    Thank you,

    Pierre

    autotune.log

  • Today I did autotune in my TBS DIscover frame, landed disarmed and  turned off autotune, armed again and flew very nice, changed battery, armed and.... flipped the quad about 3 times, very agressive that it even didnt go up, I looked at the gains and in the pitch the D was very high 0.160 if I remember and also the stabilizer roll and pitch was at around 9.25 which I believe it's the problem. What I dont understand is, how I could fly after the autotune very nicely like never before and just after land and change battery I couldnt fly again? What I noticed also is if I want fly nicely in loiter I need to have stabilizer high otherwise I have a very slugissh loiter, it's all that because my frame type? I never had this kind of problem while flying with a similar setup with my ex f450. I am attaching two logs, autotune(bigger) and other from different flight. Thanx for any help, in special I appreciate if you can give me any advice to have a less slugissh loiter without having to go up with stabilizer pid too much as it  makes the copter very agressivel.

    2014-03-15 08-28-21.log

    2014-03-15 09-02-49 4.log

    • Developer

      Hi Victor,

      Could you put a couple of pictures of your quad as you were flying it on the day. Also, could you supply the log of the crash flight.

      When you were doing autotune, was it windy or perfectly calm. Also,  how much was it drifting?

      Thanks

      • it was calm with almost no wind. I marked to log everything, the only two logs I still had in my ardu was those two I posted. Thanx a lot

This reply was deleted.

Activity

DIY Robocars via Twitter
RT @breadcentric: See how the April AWS DeepRacer races have ended and a couple bits of news: https://blog.deepracing.io/2021/05/09/aws-deepracer-league-2021-update-11-end-of-april-special/ #AWSDeepRacer #Machin…
21 hours ago
DIY Robocars via Twitter
RT @sunilmallya: Representation Learning +Instance Transfer to learn new reward functions along with advantage based filtering of new exper…
yesterday
DIY Robocars via Twitter
Apr 27
DIY Robocars via Twitter
Apr 27
DIY Robocars via Twitter
RT @f1tenth: Sliding (autonomously) into the weekend like ... 🤖😎 #f1tenth #robots #AutonomousVehicles @OpenRoboticsOrg @NVIDIAEmbedded @Aut…
Apr 25
DIY Robocars via Twitter
RT @chr1sa: One of the problems with autonomous car racing is that watching software drive is not a very exciting spectator sport. To help…
Apr 25
DIY Robocars via Twitter
RT @SmallpixelCar: Replaced AGX Xavier with @NVIDIAEmbedded Jetson Xavier NX. Now both cars look fast and clean https://t.co/jlcoY2EjZf
Apr 25
DIY Robocars via Twitter
RT @chr1sa: Yesterday we had a record number of competitors (36) in our monthly @DIYRobocars virtual autonomous car race. We kept it going…
Apr 25
DIY Robocars via Twitter
Apr 16
DIY Drones via Twitter
RT @chr1sa: After more than a year of only virtual races, @DIYRobocars returns to the newly renovated @circuitlaunch on May 22 for the resu…
Apr 11
DIY Robocars via Twitter
RT @DAVGtech: And now available with LiDAR 🔥 https://twitter.com/Heavy02011/status/1381137016381964293
Apr 11
DIY Robocars via Twitter
RT @Heavy02011: #VirtualRaceLeague: @DIYRobocars Race #9 - #ParkingLotNerds #thunderhillracetrack, CA Join us for the next race April 24th,…
Apr 11
DIY Robocars via Twitter
RT @DWalmroth: Weather's finally cooperating, looking forward to racing 1:10 scale autonomous cars outdoors again! @diyrobocars, @NVIDIAEm…
Apr 7
DIY Robocars via Twitter
RT @AIDRI_: I finally succeeded in optimizing the trajectory and speed of a car on a #racetrack. Next step: implement a 2d controller and…
Apr 7
DIY Robocars via Twitter
@jetdillo @circuitlaunch Actually the second *in person* event in a year. We do virtual races every month
Apr 2
DIY Robocars via Twitter
Update: we're moving it back one day to Sunday (the 4th) at 11:00am instead
Apr 2
More…