Lost Mavic Air 2 - Issues with RTH

[Ependeon]

Hi all,

New DJI pilot here, I decided to make the upgrade to a Mavic Air 2 after playing around with a Holy Stone for about a year and a half. It ended up being a short lived experience, lol.....

After testing it out at home I brought it up to the family cottage to stretch things out in some open areas. I flew for about 10 mins and ended up sending it out ~2.2 km (1.37 mi) at 100m (328 ft) altitude and had about 60% battery remaining before signal started to get a bit choppy at which point I activated RTH on the phone screen.

I fumbled for a few minutes to get gloves on while waiting for RTH. When I picked up the controller and looked at the display it had only come back about 100m, but had flown up over 350m (1200+ ft)! Pretty wild (and dangerous!) considering I set RTH altitude to 100m, one of the first things I adjusted as per instructions. At this point I took over control to bring it back towards me (and back down to Earth), but that only lasted about a minute and then my phone shut off because I had been outside for a while and I guess the cold had cause a quick battery drop. I pressed RTH on the controller and hoped that either that or eventually low battery would bring it back home, but it never came back. I also got the app up and running within a few minutes, which should have been fast enough that the drone would have still had some juice had it been hovering but I was never able to reestablish a connection. Bummer .

I should mention that a couple days before I had taken it out for a test run and had activated RTH from about 1km away and the drone hovered without moving for about a minute and a half, at which point I got annoyed with the beeping and brought it back manually.

Does anyone have an idea what the issue may have been here? Other than RTH altitude and camera settings every other setting was default. The weather was a bit below freezing between -5 and -7°C, within specs (14° to 104° F / -10° to 40° C), maybe pushing the lower end but the drone had been inside up until a few minutes before flying. There was a slight breeze that day, but not enough to get knocked around or have any noticeable resistance while flying. Everything seemed normal as far as I could tell other than RTH which gave me issues for the second time in as many attempts. Any insight would be much appreciated.

DJI Flight Log Viewer | Phantom Help

Upload and view flight logs from your DJI drones

www.phantomhelp.com

 

[Phantomrain.org]

So it looks like you had your RTH set to Hover and from 400 ft the wind just picked it up while you were waiting and putting your gloves on and lifted it up until it ran out of battery as there was nothing the drone could do .

Phantomrain.org
Gear to fly your Air 2 in the Rain / Snow and Land on Water

 

[boblui]

The only explanation I can think of is that during RTH, the the craft detected obstacles in front of it so it tried to ascend to avoid it. There have been reports that the sun can cause such false obstacle detection. Was the craft pointing towards the sun on it's way back ? If you can post the .DAT log file in the phone, the reason can be ascertained.

 

[Meta4]

Does anyone have an idea what the issue may have been here? Other than RTH altitude and camera settings every other setting was default. The weather was a bit below freezing between -5 and -7°C, within specs (14° to 104° F / -10° to 40° C), maybe pushing the lower end but the drone had been inside up until a few minutes before flying.

Your data confirms the event you described.
I've seen about 10 of these mystery stop-start climbing RTH events lately.
All were with Air 2 drones (and maybe Mini 2 ?) and all in snowy environments.
It seems to be related to the obstacle avoidance system and doesn't affect other, older DJI drones.

I'd suggest trying a warranty claim with DJI since there is a pattern with this happening lately.

and then my phone shut off because I had been outside for a while and I guess the cold had cause a quick battery drop. I pressed RTH on the controller and hoped that either that or eventually low battery would bring it back home, but it never came back

Losing the phone or app doesn't affect your ability to keep flying the drone.
You could have brought it home or much closer and lower.
Initiating RTH was a mistake as it just started the problem all over again (but with a 1000 ft head start).
It probably climbed to where the wind blew it far away.

 

[Ependeon]

The only explanation I can think of is that during RTH, the the craft detected obstacles in front of it so it tried to ascend to avoid it. There have been reports that the sun can cause such false obstacle detection. Was the craft pointing towards the sun on it's way back ? If you can post the .DAT log file in the phone, the reason can be ascertained.

That's interesting because I was actually flying directly towards the sun. But it was a little overcast that day I confirmed by looking at the video so it probably doesn't fit. Though at the same spot/time/direction the next day I was playing hockey with the family and my team was complaining about the brightness of the late afternoon sun, lol. Here is the .dat file.

 

[Ependeon]

Your data confirms the event you described.
I've seen about 10 of these mystery stop-start climbing RTH events lately.
All were with Air 2 drones (and maybe Mini 2 ?) and all in snowy environments.
It seems to be related to the obstacle avoidance system and doesn't affect other, older DJI drones.

I'd suggest trying a warranty claim with DJI since there is a pattern with this happening lately.

Losing the phone or app doesn't affect your ability to keep flying the drone.
You could have brought it home or much closer and lower.
Initiating RTH was a mistake as it just started the problem all over again (but with a 1000 ft head start).
It probably climbed to where the wind blew it far away.

Yeah, definitely a share of missteps on my part there, I think I panicked and thought RTH was the safe bet not weighing that it was 0 for 2 at that point. Thanks for the info, I'll definitely try a claim when I have the chance. I can see how a snowy environment and direct sun as boblui suggested might have an effect. As an electrician I've seen / heard stories about sensors being affected by backdrops and brightness. I'll update on what I hear from DJI.

 

[Deleted member 103366]

To Meta4, this is just the sort of incident I had in mind when I asked my ' will the CSC always bring the drone down?' question.

So, I will ask here.
Assuming the CSC response mode had been changed to "anytime", or what ever is the appropriate name for the mode that leaves a motor stop entirely at the pilot's command.
Would a CSC have stopped the motors in mid air if that CSC had been preformed whilst the drone was connected to the controller?

 

[slup]

...because I was actually flying directly towards the sun.

...Here is the .dat file.

The DAT log doesn't show anything else than the TXT log. This incident is pretty much what we have seen in the past earlier, especially from Mavic 1 & 2's. But now lately it's surfaced again but now involving Mavic Air 2's.

The OD sensors misjudge the bright light for an obstacle & treat it like that by following the coded process regarding obstacle avoidance during RTH.

Below is the complete RTH phase in the flight ...

The darker green background is the RTH phase
Red is height
Green is heading speed
Blue is vertical speed (positive value=descent)
Magenta is AC pitch
Grey is yaw angle

Pretty clear that as soon the AC have yawed towards HP & starts to get heading speed it pitch up (positive) for braking, there the vertical speed goes from nearly 0m/s to negative value for ascent ... then it repeats, with a slight increase in height for every loop.



Here the same but zoomed in so it's easier to follow the change in heading & vertical speed relative pitch & height increase.

So it looks like you had your RTH set to Hover and from 400 ft the wind just picked it up while you were waiting and putting your gloves on and lifted it up until it ran out of battery as there was nothing the drone could do .

Phantomrain.org
Gear to fly your Air 2 in the Rain / Snow and Land on Water

No ... nothing of that is correct, you need to read the log instead of just guessing ...

 

[Meta4]

To Meta4, this is just the sort of incident I had in mind when I asked my ' will the CSC always bring the drone down?' question.

Of course it would.
There should be no mystery.
Despite there being a problem with obstacle avoidance, the drone was connected and responding to control inputs.
You can see that with the OP cancelling the RTH and taking control at 13:44.

But if I remember correctly, you were asking about situations where the drone was not responding to control inputs.

 

[Meta4]

So it looks like you had your RTH set to Hover and from 400 ft the wind just picked it up while you were waiting and putting your gloves on and lifted it up until it ran out of battery as there was nothing the drone could do .

And you've seen an incident where the wind has lifted a drone from 400 ft to >1200 ft?
Winds in my part of the world tend to blow in a horizontal direction most of the time.

 

[Capt KO]

Mark your drone with ph #. Name and reward helpful. Usually finds its way home, eventually.
Maybe next time.

 

[BudWalker]

The DAT log doesn't show anything else than the TXT log. This incident is pretty much what we have seen in the past earlier, especially from Mavic 1 & 2's. But now lately it's surfaced again but now involving Mavic Air 2's.

The OD sensors misjudge the bright light for an obstacle & treat it like that by following the coded process regarding obstacle avoidance during RTH.

Below is the complete RTH phase in the flight ...

The darker green background is the RTH phase
Red is height
Green is heading speed
Blue is vertical speed (positive value=descent)
Magenta is AC pitch
Grey is yaw angle

Pretty clear that as soon the AC have yawed towards HP & starts to get heading speed it pitch up (positive) for braking, there the vertical speed goes from nearly 0m/s to negative value for ascent ... then it repeats, with a slight increase in height for every loop.

View attachment 124576

Here the same but zoomed in so it's easier to follow the change in heading & vertical speed relative pitch & height increase.

View attachment 124577



No ... nothing of that is correct, you need to read the log instead of just guessing ...

Just to add a bit more to this. Looking at the .DAT it can be seen that the MA2 thinks there is something in front to be avoided.

Interestingly, there is also an avoid_front_m and avoid_front_r which don't show much. I'm assuming that avoid_front_l, avoid_front_m and avoid_front_r are left, middle and right, respectively. So, the sun was in front slightly to the left.

 

[Phantomrain.org]

And you've seen an incident where the wind has lifted a drone from 400 ft to >1200 ft?
Winds in my part of the world tend to blow in a horizontal direction most of the time.

Well now that you mention it. I did have it happen where I was flying into the head wind about 20 mph and and was in hover mode and started to turn the Drone sideways slowly . I was prepared for the drone to be swept down slightly but instead climbed straight up and kept going.

Now the log does show and around 400 ft he was at 0.4 MPH which looks to me that the drone is hovering .
and than it goes straight up from there , so looked similar and it was possible he set the RTH to hover.

I have a little footage of incident on a smaller scale where I was filming
and made the turn slowly in front of the Camera and sure enough it started to rise but the winds were a little stronger.

Watch at 4:07 , apologize for the editing but you can catch a small glimpse of the drone going up on me and keep rising.

Phantomrain.org
Gear to fly your Air 2 In the Rain / Snow

 

[sar104]

Just to add a bit more to this. Looking at the .DAT it can be seen that the MA2 thinks there is something in front to be avoided.
View attachment 124581
Interestingly, there is also an avoid_front_m and avoid_front_r which don't show much. I'm assuming that avoid_front_l, avoid_front_m and avoid_front_r are left, middle and right, respectively. So, the sun was in front slightly to the left.

Very good - the sun was in fact 30° to the left of the aircraft heading. At that location, at 1505 on February 13, the solar azimuth was 222° at an elevation of 22°. The aircraft heading was 252°.

 

[slup]

Just to add a bit more to this. Looking at the .DAT it can be seen that the MA2 thinks there is something in front to be avoided.
View attachment 124581
Interestingly, there is also an avoid_front_m and avoid_front_r which don't show much. I'm assuming that avoid_front_l, avoid_front_m and avoid_front_r are left, middle and right, respectively. So, the sun was in front slightly to the left.

Yep ... trawled through all columns in the DAT searching for all with "avoid" & "brake" & saw them ... but being unsure what they actually indicated I left them unmentioned

Nice to see you again also @sar104 ?

 

[Gbower]

I had a similar incident with the Air 2 that I wrote about in these forums and actually talked to DJI about too. The outcome was that it was very bright with the snow reflections was causing the obstacle avoidance to kick in and raise the drone up and then continue home. I have since had a few flights on a cloudy day and no issues at all so I just have to be careful on the bright days and not use RTH or expect something and watch the controller extremely close and stop RTH as soon as I see either the speed slow down or the drone start to climb. I make sure to check all my RTH, max. altitude and max distance settings before every flight.

 

[tlswift58]

Hi all,

New DJI pilot here, I decided to make the upgrade to a Mavic Air 2 after playing around with a Holy Stone for about a year and a half. It ended up being a short lived experience, lol.....

After testing it out at home I brought it up to the family cottage to stretch things out in some open areas. I flew for about 10 mins and ended up sending it out ~2.2 km (1.37 mi) at 100m (328 ft) altitude and had about 60% battery remaining before signal started to get a bit choppy at which point I activated RTH on the phone screen.

I fumbled for a few minutes to get gloves on while waiting for RTH. When I picked up the controller and looked at the display it had only come back about 100m, but had flown up over 350m (1200+ ft)! Pretty wild (and dangerous!) considering I set RTH altitude to 100m, one of the first things I adjusted as per instructions. At this point I took over control to bring it back towards me (and back down to Earth), but that only lasted about a minute and then my phone shut off because I had been outside for a while and I guess the cold had cause a quick battery drop. I pressed RTH on the controller and hoped that either that or eventually low battery would bring it back home, but it never came back. I also got the app up and running within a few minutes, which should have been fast enough that the drone would have still had some juice had it been hovering but I was never able to reestablish a connection. Bummer .

I should mention that a couple days before I had taken it out for a test run and had activated RTH from about 1km away and the drone hovered without moving for about a minute and a half, at which point I got annoyed with the beeping and brought it back manually.

Does anyone have an idea what the issue may have been here? Other than RTH altitude and camera settings every other setting was default. The weather was a bit below freezing between -5 and -7°C, within specs (14° to 104° F / -10° to 40° C), maybe pushing the lower end but the drone had been inside up until a few minutes before flying. There was a slight breeze that day, but not enough to get knocked around or have any noticeable resistance while flying. Everything seemed normal as far as I could tell other than RTH which gave me issues for the second time in as many attempts. Any insight would be much appreciated.

DJI Flight Log Viewer | Phantom Help

Upload and view flight logs from your DJI drones

www.phantomhelp.com

Did you locate the drone or even try with the App?

I've got to get with you guys eye doctors as they seem to be great at giving you line of sight at over 1 mile!

 

[Meta4]

Did you locate the drone or even try with the App?

The app can only show where it last had contact with the drone.
That's not much help if the drone was still flying and going further away at the time.

 

[Meta4]

Now the log does show and around 400 ft he was at 0.4 MPH which looks to me that the drone is hovering .
and than it goes straight up from there , so looked similar and it was possible he set the RTH to hover.

His data shows that he didn't have the Failsafe Action set to hover and the drone was returning to home, but obstacle avoidance kept cutting in and climbing higher.
It had nothing to do with wind activity and his pitch/roll data indicates that winds were were not strong.

I have a little footage of incident on a smaller scale where I was filming
and made the turn slowly in front of the Camera and sure enough it started to rise but the winds were a little stronger.

Well now that you mention it. I did have it happen where I was flying into the head wind about 20 mph and and was in hover mode and started to turn the Drone sideways slowly

Flying sideways in a strong wind doesn't cause the drone to climb.
DJI drones actively work to maintain altitude.
The OA is going off and can be heard in your video.

You would have to be inside a thunderhead or flying a cliff face with a very strong wind blowing onto the cliff to produce uplift that the drone could not hold against.

 

[muchuckwagon]

I have less than a year experience flying my MA2. When I read these threads, I try to imagine what steps to take in a similar situation. Honestly, RTH would’ve been my first choice if the App crashed and I was low on battery while losing VLOS. Next, I would’ve probably tried Sport mode upon discovering the drone exceeding 400 ft once the App rebooted hoping to override whatever caused the drone to ascend.

My thought is a controlled decent in Sport mode, no matter where it might land, is almost always a better option than having the drone fall from 400+ feet from the sky.

Does that seem reasonable?