Mavic Air drifting after take-off

[ManuCH]

Today I've had an interesting experience. After waiting for the Mavic Air to lock enough GPS satellites, I took off manually and went up to about 10 meter and let it hovering there.

For some reason, the drone started slowly drifting to the left and backwards for a few meters. I noticed, gave it some control input, flew forward and everything was normal from that moment on.

Interestingly enough, this drift can be seen in the logs, so the Mavic was fully "aware" of it. It looked like it was getting some slight control input from my side (and I'm pretty sure I wasn't touching the sticks).

Can some of you experts have a look at the logs to figure out what happened? It was all during the first 30 seconds.

DJI Flight Log Viewer - PhantomHelp.com

Thanks!

 

[InvisibleName]

Maybe try an RC calibration while you wait for a response?

 

[Simmo]

I'm gonna call GPS Health had not achieved a sufficient level, therefore Atti mode flight, due to a magnetically distorted takeoff location.

 

[hko]

I had also few similar problems today..
1. Really bad GPS reception (9-10) sats on ground. When ascending to about 10m GPS was ok but after some minutes when i hover bofore landing it starts slowly drifting.
2. What scares me more there was not home point set and even i did it manually there was not H point on map, no drawing of flight path and no aircraft arrow on map. Map was only black square. When i checked log there was home point coordinates but no any message like "Your home point has been updated".
3. Also after some time of flying video stream start dorp out/tear and there was about 4 seconds lag, RC signal was ok, picture not became like where there is weak signal gray.
4. And last thing when take off and ascending it after some time or in some height (about 10m) suddenly turn (yaw) about 20-30 degeeres to left and then it became again stable.

I had no any of those problems before. Still same firmware and latest DJI Go app version since update. I using old Galaxy S5 with Android 7.1 just for Mavic.

So today i have lot of questions from my flights instead of enjoying it...

 

[Simmo]

I believe that you should spool up, then wait for an announcement, then check it on the map! Is the map correct??
If not, your future flight is unsafe... IMO

 

[InvisibleName]

You may have had 9 or 10 satellites but what was your GPS health like? If you didn’t have 5 bars, you shouldn’t have taken off

 

[ManuCH]

You may have had 9 or 10 satellites but what was your GPS health like? If you didn’t have 5 bars, you shouldn’t have taken off

It's indeed possible that the GPS health wasn't good enough. I will pay more attention to that next time. What I usually do is waiting for the DJI Go 4 app to say "GPS mode" (instead of "Opti mode"), then I consider it safe for take-off. Probably not the best procedure.

Maybe try an RC calibration while you wait for a response?

I did that in the meantime. I also did an IMU and compass calibration as well as a gimbal auto-adjustment, as I never did those. I will check how it behaves during the next flight.

One other issue I am encountering is that spheric panorama pictures aren't stitched together properly, meaning that the stitches are visible and slightly mis-aligned.

I had also few similar problems today..
1. Really bad GPS reception (9-10) sats on ground. When ascending to about 10m GPS was ok but after some minutes when i hover bofore landing it starts slowly drifting.
2. What scares me more there was not home point set and even i did it manually there was not H point on map, no drawing of flight path and no aircraft arrow on map. Map was only black square. When i checked log there was home point coordinates but no any message like "Your home point has been updated".
3. Also after some time of flying video stream start dorp out/tear and there was about 4 seconds lag, RC signal was ok, picture not became like where there is weak signal gray.
4. And last thing when take off and ascending it after some time or in some height (about 10m) suddenly turn (yaw) about 20-30 degeeres to left and then it became again stable.

I had issues 1 and 2 as well. I guess GPS reception was just too weak to properly fix a home point.

Issue 3 sounds like interference, try another Wifi channel. And I've never personally witnessed issue 4 before, although I've read of other people having this in the forums.

 

[CanadaDrone]

How many satellites did you have? You need at least 12 or ideally 13 for a proper lock. I wouldn't leave the ground without at least that many and max bars. The app will clear you for take off with only 8. That was most likely the issue if you just took off when the app said it was OK.

 

[ManuCH]

How many satellites did you have? You need at least 12 or ideally 13 for a proper lock. I wouldn't leave the ground without at least that many and max bars. The app will clear you for take off with only 8. That was most likely the issue if you just took off when the app said it was OK.

I had 9 at take-off. Then this was probably it. Thank you!

I have done 5 other take-off attempts after waiting for 12 satellites and it was rock solid.

 

[Mossiback]

Make sure the Mavic and the pointer in Go 4 are both pointing in the same direction. If not, there could be some magnetic distortion at your takeoff point that will not be there when aloft. This can cause some control issues.

 

[CanadaDrone]

I had 9 at take-off. Then this was probably it. Thank you!

I have done 5 other take-off attempts after waiting for 12 satellites and it was rock solid.

Yes that was most likely it. You certainly can't be blamed for taking off when DJI says it's OK to do so, but without at least 12 satellites (and good signal), you don't have a perfect lock and that is the most common reason for drifting.

 

[sar104]

The left and back drift was seen by the GPS, but not by the IMU, assuming that the green track below is approximately what you say, so it was not the result of poor GPS accuracy:



The IMU did not really see the drift to the SW, or at least only a meter or so.

GPS was not great, but acceptable:



The initial motion forwards at 14 s was due to elevator input:



The drift back and left does not have any significant associated stick inputs, so it's not clear what caused that.

 

[ManuCH]

The left and back drift was seen by the GPS, but not by the IMU, assuming that the green track below is approximately what you say, so it was not the result of poor GPS accuracy:

The IMU did not really see the drift to the SW, or at least only a meter or so.

GPS was not great, but acceptable:

The initial motion forwards at 14 s was due to elevator input:

The drift back and left does not have any significant associated stick inputs, so it's not clear what caused that.

Thanks a lot for this, I really appreciate it! Yes, the green track is exactly what I meant. It's really weird that the IMU wouldn't pick up that movement.

In the meantime I recalibrated the IMU, so let's see if this happens again. Considering that I also get these error messages from time to time, I'm starting to think I might have a defective unit. What do you think?
IMU redundancy switch

One last question for you: I would like to create graphs like you do myself. I could do that manually via CSV file and Excel graphs - I tried, it works, but it's a bit tedious. Do you use a dedicated software for that? If so, would you mind sharing which one? Thank you!

 

[sar104]

Thanks a lot for this, I really appreciate it! Yes, the green track is exactly what I meant. It's really weird that the IMU wouldn't pick up that movement.

In the meantime I recalibrated the IMU, so let's see if this happens again. Considering that I also get these error messages from time to time, I'm starting to think I might have a defective unit. What do you think?
IMU redundancy switch

One last question for you: I would like to create graphs like you do myself. I could do that manually via CSV file and Excel graphs - I tried, it works, but it's a bit tedious. Do you use a dedicated software for that? If so, would you mind sharing which one? Thank you!

I would probably calibrate the IMU and compass. It would also be instructive to look at the DAT file and compare the magnetometer data to the IMU rate gyro data.

I use Igor Pro to analyze and display the data.

 

[VLL]

I know this is an older thread, and is in the Mavic Air forum, but I thought I would add my experience.

I got a new Mavic 2 Pro yesterday as a replacement for one that had an IMU problem. I charged up batteries, updated software, and took it to the backyard for a maiden flight. The compass needed calibrating so I did that in the same spot in my backyard where I try to do all compass calibrations.

I took off manually and stopped to hover at about 4 ft. The AC instantly drifted back two feet or so. Slowly, but a distinct drift. I thought it was just the forward obstacle sensing at first, since the trees have leafed out in the 3 weeks I've been without a drone. Or maybe this AC is more sensitive? I wasn't sure. And wasn't sure if it was partly my imagination. But it seemed like a possible problem.

On a second flight early this am, I experienced the same thing after turning off the front/rear obstacle avoidance. Throughout the flight, I felt I was fighting constant drift backwards every time I stopped it to hover. Now I was sure it wasn't my imagination and that it was excessive.

I searched the forum and found a few threads suggesting things like poor GPS (I had 14-15 satellites today), need for stick calibration (A-ok on mine), wind factors (light for my flight). I was getting discouraged when I found this. I probably should have considered an IMU calibration, but I hadn't. So I recalibrated the IMU. For good measure, I did another compass calibration outside.

I then put the AC up in the air and, yay, the problem appears to be solved. The AC seemed appropriately rock solid at a low hover. I flew it around, stopping frequently to hover. Each time, it held position well. I won't have time to fly more until later today or tomorrow, but I feel reasonably confident all is well.

Hope my positive experience with the solution suggested by SAR104 proves helpful to others.