Drone Stuttering/shuttering/hesitating during mission

joystix2

So far creating and and flying missions are working but starting from my first mission when the drone starts the mission it starts what I can only describe as shuttering/stuttering/hesitating while in flight. It happens throughout the whole mission and stops when the mission is completed. Is anyone else having this issue? The video comes out fine but I think its because of the stabilization of the gimbal

0

Comments

23 comments

  • Comment author
    Steve Shackelford

    I see this a lot on YouTube Videos. Anyone know the reason for this? If I can ever get a test flight in, I'm curious to see if it happens to me.

    0
  • Comment author
    Jim McAndrew Dronelink Staff

    https://support.dronelink.com/hc/en-us/articles/360052533153-Why-does-the-drone-fly-off-course-or-make-jerky-movements-during-certain-mission-plans-

    0
  • Comment author
    Patrick Bernhart
    • Edited

    Hi, i fly two missions one with Mavic Mini and second with Mavic 2 Pro, on both missions i hear a sputtering sound from the drones? What that sound ? i never heard with Go or Fly App? It happened during straight line map capturing, during curve i do not hear this

    0
  • Comment author
    Jim McAndrew Dronelink Staff

    This is the sound the motors make when Dronelink is sending micro-corrections to the course and velocity through the virtual stick commands. DJI has assured us that it is not a problem and it only happens on these models (for example, the same exact code does not make these sounds on the Phantom or Inspire series).

    0
  • Comment author
    Patrick Bernhart

    Is it possible smooth this micro-corrections inside plans ? this stuttering sound produce a lot of attentiveness during flight in contrast a smooth sound.

    0
  • Comment author
    Jim McAndrew Dronelink Staff

    Our testing has shown that any change in course or velocity causes this to happen, so it isn't clear how to solve this problem at the software layer - it appears to be something that DJI needs to correct in their firmware (since it is specific to these models only). Furthermore, even if we did try to "smooth it out," it would result in the drone flying off course because the whole point of sending changes in course and velocity is to keep the drone on the path in the first place.

    0
  • Comment author
    Patrick Bernhart

    Ok , i can not evaluate how often you send joystick commands , but for me seems to be the regulator gets to many or spiky commands. Visual there is not course deviation. Do you ask DJI correct the SDK?

    0
  • Comment author
    Jim McAndrew Dronelink Staff

    We told them about the issue several years ago.

    0
  • Comment author
    Bob Scheffler

    I also have this same stuttering with my mini even when flying with almost no wind, but it does get worse when the wind picks up so I am confident it is as Jim says, it happens when software is making micro connections to stay on course.  I have noticed something that does not seem to be related.  When flying straight with no wind the drone sometimes, not always, looks like it is popping mini wheelies.  The nose of the drone bounces up and down, just the nose, looks very strange.

    0
  • Comment author
    Patrick Bernhart

    i fly map missions with DJI GS Pro and Dronedeploy , never see this behavior before. I will not using Dronelink with my Mavic 2 Pro again , i will not risk a expensive crash.

    0
  • Comment author
    Bob Scheffler

    Patrick I am relatively new to flying drones (7 months) and less for using a program like dronelink so I am not sure what to expect.  I think the program is safe, I have flown about 2 dozen missions and I have lost connection, came low on battery power and had telemetry problems but none have caused a crash. I am not sure how great the video would be if you were using it professionally especially since the drone is jumping around like a Mexican jumping bean.  It does draw attention though when you are flying a mission and it flies overhead sounding like a sputtering plane ready to crash, I've had a few people look like they were trying to get out of it's path, even though I was 50' up.

    0
  • Comment author
    Jim McAndrew Dronelink Staff

    We are continually trying to make Dronelink better, and there may be ways to mitigate / lessen the effect that this DJI SDK bug has on certain drones, but we have many priorities and technically this issue isn't causing any problems. As for your perceived risk, you have to make the best decision you can given what you know, and if you don't want to use Dronelink that is understandable. You may be interested to know that there has not been a single crashed reported by any of our users where the software was at fault (and there have only been a few crashes reported total).

    0
  • Comment author
    Patrick Bernhart

    Jim, all high frequency vibrations create mechanical stress and can cause malfunctions. Also during image capture vibrations create blurry images. This is not my goal creating aerial imaging. But not a big problem , with M2P i can use different apps, only MM fails. I hope you can fix it because your solution is very promising!

    0
  • Comment author
    Patrick Bernhart

    Jim, do you work on this behavior with Beta 2.0.0 App ?

    0
  • Comment author
    Jim McAndrew Dronelink Staff
    • Edited

    The short answer is yes: we are always trying to make things better, and v2.0.0 has many improvements.

    The long answer:

    We introduced Kalman Filters into the kernel to smooth out sensor data (mainly sporadic GPS updates) for use in the upcoming modes, particularly Follow.

    We then applied these filters to the Path and Orbit components and were able to see a marked improvement in the issue presented in this thread (the graphs were smoothed out). In the real world, we still observed the "motor stuttering", but subjectively it seemed to not be "as bad". It is hard to quantify the improvement, however, because the only way to judge it is by how it sounds, which seems to vary from mission to mission, day to day, and drone to drone (i.e. it is subjective as long as some noise remains).

    Having said that, we also introduced a code path in the kernel that uses the other mode that DJI allows for virtual stick: percentage based input (vs the current course + velocity mode). You can think of percentage based input as a mirror image to what a human would be doing on the sticks.

    We introduced this for the purpose of Focus mode in order to provide the user with a more "direct stick input" experience, and we noticed that when virtual stick is using this mode, the motor stuttering sound is not present.

    This evidence, combined with the fact that it only happens on certain drones (the Mavic line), adds further support to the theory that there is an issue in the DJI SDK with virtual stick in course+velocity mode that is only present in certain firmwares (Mavics). While I think that introducing the Kalman filters is a win in any case and may improve the situation (even if only marginally), I still suspect that we are nearing the limit of what we can do from a software standpoint, outside of a fix to the firmware from DJI.

    The only other option at this point would be to abandon course+velocity virtual stick mode, and use percentage mode (what they call angle), and maybe do it only for certain drones (Mavics). Unfortunately, not only will this complicate the implementation and double the QA effort when doing regression tests, but it also comes with certain drawbacks, such as not being able to take advantage of DJI's onboard wind-correction algorithms. It is possible to perform device-based wind correction, but it will never be as good as onboard correction. Furthermore, the PID has to dynamically approximate percentages with real world courses and velocities. Again, not an insurmountable task (indeed, I wrote this type of code for the first version of Autopilot because course+velocity virtual stick mode didn't exist in the DJI SDK at the time), but it adds further complexity to an already complex implementation. It also results in real-world performance trade-offs, such as a reduction in the quality of path adherence as the PID attempts to converge to the approximate values.

    You are welcome to opt-in to the beta and try the improvements for yourself and we welcome your feedback.

    0
  • Comment author
    Patrick Bernhart

    Jim, thank you for your detailed answer! i am not familiar with the SDK. For my understanding , you send speed and absolute course data to the drone. If this data do not have „holes“ and are smooth from one to the other there is no explanation for the noise. I am not sure if you have the possibility logging all data the software send to the drone. If yes you normaly see what happend. I can understand that percentage mode create much more calculation for the control. but for now i sell the Mini 1 and buy Mini 2. So now i have to wait until DJI release a SDK Update. Hope this will fix some Issues.

    0
  • Comment author
    Glenn J

    Flew my mavic 1 today for the first time with this app, and I can confirm the Stuttering/shuttering/hesitating is still there when executing any mission. 

    Reading the above I would be concerned with mechanical stress and things becoming loose. 

    Jim any work arounds from the last time this topic was touched? 

    I am loving the app otherwise. Fantastic work, keep it up and much appreciated.

    0
  • Comment author
    Jim McAndrew Dronelink Staff

    There are no known work-arounds at this time. You can always subscribe to updates on the release notes to keep up.

    0
  • Comment author
    Glenn J

    ok thanks. 

    0
  • Comment author
    Daniel

    Today i Was flying my first mapping Mission with Mi Mini 2 and was wondering about the Sound when drone is stopping. So one year later the Problem with stuttering Motors are still real?

    0
  • Comment author
    Patrick Bernhart

    @ Daniel, do you encounter problems with tilt down camera 90° and stopping ? during normal flight without dronelink , camera slightly goes up ~70° during stop. I think stuttering Motors can not prevent, also happend with Air 2 but not so noissy

    0
  • Comment author
    Jim McAndrew Dronelink Staff

    So one year later the Problem with stuttering Motors are still real

    DJI has not addressed the underlying issue in the drone's firmware...

    0
  • Comment author
    Daniel

    @ Patrick, no, the prop-motor sound when the drone is breaking.

    @Jim
    ☹️Ok

    0

Please sign in to leave a comment.