Discussions

APM Plane oscillating on upwind legs

Posted by Ondřej Procházka on January 30, 2015 at 8:30am

3691176896?profile=original

I tried to search past threads for this topic, but have not succeeded, so here is a question I humbly pose to  all APM Plane experts.

Past Wednesday, we have been flying Skywalker 1900/Pixhawk system in auto mode. There was some wind at roughly 240 degrees, about 5 m/s with gusts reaching 8 m/s.

Some relevant parameters:

NAVL1_PERIOD=14

NAVL1_DAMPING=0.65

Airspeed sensor was enabled and in use, set to 12 m/s cruise speed which the plane kept well (the frame weights about 2.7 kg and stall speed is approximately 10 m/s).

The plane navigated fairly well on downwind legs but showed strong horizontal oscillating motion when moving upwind. You can see full 3D visualisazion of flight path here (interactive - use your right mouse button). You can download the original KML from here.

I'd expect navigation performance to be affected by wind conditions but the he plane flew perfectly well in FBWA mode, even crosswind and upwind.

I tried to experiment with the NAVL1 parameters but they seemed to have little effect on the issue.

Any advice is strongly appreciated. Thanks in advance, and sorry if I am missing something obvious.

 

You need to be a member of diydrones to add comments!

Join diydrones

Tags: aerial, auto, navigation, photography, survey, waypoint
Email me when people reply –
Follow

Replies

  • Moderator
    Graham Dyer February 20, 2015 at 10:46pm

    We're running NAVL1_DAMPING at 0.8 and NAVL1_PERIOD at 19 on a similarly sized plane an our tracking is very straight. As mentioned you might want to set your airspeed up a bit to 13m/s too. I assume you've run autotune and are using the latest code?

    • Ondřej Procházka > Graham Dyer February 21, 2015 at 4:07am

      Yep, running APM:Plane 3.2.0, run autotune and the plane flies very well in FBWA mode, even in high winds. I have tried many different values for NAVL1_DAMPING and NAVL1_PERIOD, and there seemed to have little or no impact on the oscillating behavior. 

      Increasing TRIM_ARSPD_CM will likely help, though I try to keep it as low as possible too keep the plane economical. Still feel that there is some issue with L1 roll gain control at low groundspeed, as Gisela & Joe Noci suggested. 

      • Andre K. > Ondřej Procházka February 21, 2015 at 1:44pm

        Sorry, there is something going on with my ISP, I am unable to download , among others, from dropbox.

        Anyway - low airspeed is something that is likely be a problem.

        Low groundspeed, however, is not affecting the calculations. More than that; 3 weeks ago I flew a avalanche mapping mission in a longer valley, with ridiculous windspeeds at the narrower end of the valley.,

        I had as little GS as 3m/s (and AS 22m/s )  the downwind was at the minimum AS 11m/s with a 20% flaps,(8 is stall speed at that configuration) and tracking was perfect as always in both directions.

        Low GS is not the issue.  - assuming you are not using EKF. (in that case I did not test it)

        • Ondřej Procházka > Andre K. February 22, 2015 at 4:28am

          Thanks Andre. EKF is off.

          The log has about 48 megabytes, which is well bellow the diydrones limit, but I can send it to you by email if you provide me with your address.

          Interestingly, there were also some quite crazy yaw readings when the plane was oscillating upwind... the yaw readings where pointing completely off course. 

          I will run some more tests with higher airspeed when the weather gets better.

          • Andre K. > Ondřej Procházka February 22, 2015 at 5:30am

            do that. :)

            Honestly, I could have downloaded it using a VPN, but I have little time before travelling next week.

  • Andre K. February 1, 2015 at 4:46am

    post .log or .bin if you really expect help.

    I doubt this NAVL1 related at all.

  • Gisela & Joe Noci January 30, 2015 at 12:18pm

    Hi Ondrej,

    I see no one its giving any help here..Not sure I can either, since I am not familiar with APM code. We have our own design autopilot ( hardware and software) so that comments for APM would be moot.

    However, I see you were flying about 2m/s above stall speed at 12m/s and the only common issue in your flight is that the weaving occurs while your ground speed is low- around 4m/s to 7m/s it would seem? This points to a poor gain selection in the course control - normally the gains are speed scheduled and maybe the gain increase for course control increases to much at the lower ground speed ( since course is over ground, and ground speed must be used to schedule these gains).

    Can you confirm that in flights where this does not manifest, the aircraft roll angle is following the commanded roll angle with little delay, ie, the roll inner loop gains are correct? If so, the the roll outer loop - roll from course, is causing the problem, and since it seems to be only when ground speed is low, these PID gains are scheduled to to high a value. I am sorry I cannot help to tell you where to look in the APM setup - I am not familiar with that module, but I am sure many others can help with those specifics.

    Hope you come right!

    Regards

    The Nampilot..

This reply was deleted.

Activity