I wanted to start this discussion believing that better knowledge on the subject will help to prevent accidents like mine (I’ll describe it later), on the other hand neither in manual nor in official tutorials is explained what is expected behavior of the aircraft in ATTI mode and how it responds to pilot commands. There are many posts in the forum about ATTI mode but I couldn’t stick them together.

  

Manual says: “...only use its barometer for positioning to control altitude.”

But it doesn’t describe the whole picture. I will try to summarize the things.

  

There are 6 degrees of freedom or 6 axes of moving:

During hovering in GPS mode all 6 axes are controlled by the aircraft.

  

During hovering in ATTI mode not controlled are only movements in horizontal direction (x and y axes) and AC is expected to move horizontally together with the surrounding air – wind drift.  All other axes are controlled: altitude -z axis, heading – yaw, level – pitch/roll i.e. AC keeps itself parallel to the horizon.

  

How the AC reacts to RC stick inputs.

Height and yaw – as usual. Pitch and roll slightly different – while in GPS mode, inputs are transferred to horizontal movement with corresponding speed; in ATTI mode inputs are transferred to corresponding flight angle - tilt (limited to a certain value – something like 15 or 30 degrees, depending on the model). When the stick is returned in neutral position, AC immediately regains level i.e. parallel to the horizon, but there is not auto braking and AC continues to move in the same direction by inertia with decreasing speed. It is pilot’s decision to brake with an appropriate opposite input.

  

Please note that I am not a pro and the explanation above is based on my amateur knowledge and experience. I will be glad if someone corrects or completes it.

  

In order to share my observations with Phantom 3 I’ve shot a video:

Recently I lost my Spark trying to fly in ATTI mode in area with no GPS signal. It was not wind cause. It went crazy, made few chaotic jumps around, became irresponsible, accelerated by tilting himself and crashed in nearby rocks.

  

Maybe I missed something and am now trying to find out the following:

  

Are there any circumstances in which after a short pitch input and stick returned in neutral position, Spark may remain tilted or leaned instead to regain level (as my P3 does) and accelerate because of this.

That was very informative.  Thank you for sharing this.

ATTI mode: I think the gyros/accelerometers will hold the inclination of the aircraft where it is until the pilot commands it otherwise.
(like my rc helicopters do).
Unfortunately I cannot  select atti mode to verify it.

Thanks for sharing this information.  Would be nice to get a tutorial video on flying in ATTI mode.

Love the video, thank you

thanks for the thorough problem description and video.  I too am curious about this behavior and will be watching this thread ;-)

Very informative, I learned a lot. Thanks for posting.

Atti is good to practice flying in in case you loose GPS while flying. It can help you be a better pilot. Thank you for making the video.

DJI-Mark Posted at 2017-9-28 11:26
Atti is good to practice flying in in case you loose GPS while flying. It can help you be a better pilot. Thank you for making the video.

How can we get your sparks to go into ATTI mode to practice

Sparky_17 Posted at 2017-9-28 11:29
How can we get your sparks to go into ATTI mode to practice

If you go to this link and click on page 9, the manual will talk about this point further: https://dl.djicdn.com/downloads/ ... ser+Manual+V1.4.pdf

You would not be able to pick flying in atti mode. The aircraft automatically goes into atti mode when GPS signal is lost. With Phantoms,  you can choose.

DJI-Mark Posted at 2017-9-28 11:41
If you go to this link and click on page 9, the manual will talk about this point further: https://dl.djicdn.com/downloads/Spark/20170928/Spark+User+Manual+V1.4.pdf

You would not be able to pick flying in atti mode. The aircraft automatically goes into atti mode when GPS signal is lost. With Phantoms,  you can choose.

ok thank you.

I've flown the Spark in Atti-mode and although tough, it kinda reminds me of flying a SYMA copter; all over the place.

Trying to fly the Mavic in Atti-mode is almost near impossible. A def don't do indoors.

Hi
I’m not that familiar with spark going into Atti mode since I’ve had it it has gone into Atti mode once due to compass problem, this happened flying close to metal roof , it went off to left in semi circle, it lasted all of 7 seconds before it recovered . It was hovering before compass problem went to Atti mode started to drift left I pulled right stick to the right and it pitched to left then I let go of stick and it recovered gps and stopped it was clear of roof at this stage.
What log said was Atti,Atti, followed by lmu exception heading speed error,speed error.
So what occurred, I think and it is somewhat of a guess that when I commanded it to go right which should send message to cpu to send Aircraft to the right it went left because of bad compass causing conflict with IMU so it reports wrong heading before it finally recovers.

But just like in your video in Atti mode and gps, Imu was continually bringing Aircraft back to correct attitude and good horizontal . So I think my command during compass problem caused aircraft to move wrong way and causing it to pitch, although not intentional.

If you didn’t touch control and it just pitched then I don’t know how this could happen.

Thanks heliman, Sportbike_Pilot, hallmark007,

It is unlikely for DJI drone to “hold inclination” as heliman suggests: “ATTI mode: I think the gyros/accelerometers will hold the inclination of the aircraft where it is until the pilot commands it otherwise. (like my rc helicopters do).”

I expected Spark to behave in ATTI mode like Phantoms:
pitch/roll stick returned in neutral position = aircraft horizontal.

We are almost guessing about this and it is important to be well known. I am waiting for a more detailed explanation from DJI expert.
Any suggestions?

djiuser_gjLUHKA Posted at 2017-9-29 08:47
Thanks heliman, Sportbike_Pilot, hallmark007,

It is unlikely for DJI drone to “hold inclination” as heliman suggests: “ATTI mode: I think the gyros/accelerometers will hold the inclination of the aircraft where it is until the pilot commands it otherwise. (like my rc helicopters do).”

I think if you have sent your logs to dji it might be better to await that outcome their engineers will have access to dat file so that should help you. When you don’t have all the information then it is merely speculation.

hallmark007 Posted at 2017-9-29 09:20
I think if you have sent your logs to dji it might be better to await that outcome their engineers will have access to dat file so that should help you. When you don’t have all the information then it is merely speculation.

Please note that I am not focusing the discussion on my case. I am using it for an example.
My goal is to clarify the things in principle.

P.S. My case was finished with conclusion – pilot error. I received some strange explanation but was not able to proof anything and gave up. I'm now thinking about buying new Spark - it best suits my mountaineering demands.

The video is great.
I think all your assumptions are right.

Here is my expirience of flying in ATTI mode. The biggest problem is that you can’t control swiching between GPS and ATTi and difference in response to pressing the RC joystick. In my caset he modes were switched constantly with an interval of a couple of seconds. You push the joystick and can’t predict AC reaction. If I could stay in ATTI  I think it will be possible to  land safely.  Definitely DJI should add possibiliti of choosing ATTI mode.

fans04bc4ba7 Posted at 2017-9-29 13:56
Here is my expirience of flying in ATTI mode. The biggest problem is that you can’t control swiching between GPS and ATTi and difference in response to pressing the RC joystick. In my caset he modes were switched constantly with an interval of a couple of seconds. You push the joystick and can’t predict AC reaction. If I could stay in ATTI  I think it will be possible to  land safely.  Definitely DJI should add possibiliti of choosing ATTI mode.
https://youtu.be/i3m3_8svCEI

I agree that inappropriate switching from atti to gps mode can cause big troubles and improvements should be discussed. In your case it’s about IMU malfunction. I hope it has ended favorably for you.

Better understanding of behavior and how inclination of AC is controlled in ATTI mode- which is nowhere clearly described- will help the user to deal with active piloting (assuming everything works properly).

It's not an issue for anyone who has taken up flying model aircraft (for that is what it is) seriously. Takes practice and perseverence.Most people can't be bothered, that's fine, as long as they don't think they're flying it.
"Drones" are very sophisticated, but today's technology enables people to just  drive them around - until atti (which is still seriously gyro stabilised ) appears.When things go wrong at sea or in the air , they go wrong very quickly indeed.
It's really dumbed down flying .

djiuser_gjLUHKA Posted at 2017-9-30 00:10
I agree that inappropriate switching from atti to gps mode can cause big troubles and improvements should be discussed. In your case it’s about IMU malfunction. I hope it has ended favorably for you.

Better understanding of behavior and how inclination of AC is controlled in ATTI mode- which is nowhere clearly described- will help the user to deal with active piloting (assuming everything works properly).

No it was compass issue. My guess - Spark try to reset GPS when detect some problems with compass.

I was lucky enough and dropping from a height of 25 meters cost me only a replacement for 4 propellers

Explanation given by DJI support about the case with my Spark (I have a permission to quote):
“the behaviour of the aircraft was the result of flying without GPS signal.
The flight in ATTI mode then results in the stick inputs being translated not to a movement, but to a tilt in a specific direction. This is not automatically controlled by the aircraft, but hat to be handled by the pilot.
The aircraft "leaned" in a specific direction and started to speed up because of this. The movement would have to be conteracted by the pilot.
depending on the "leaning angle" of the unit, a throttle up command will result in a even higher acceleration.”

I received this as an answer to my question:
“Is it normal when I push throttle up command the craft did not react – 2 seconds – no change in height?” (It was my last chance to avoid the crash.)

Screenshots related to the situation:


Flight record: http://www.phantomhelp.com/LogViewer/K1U8GU2TFCNXFKMK09BC/

Explanation above and my amateur observations are contradictory and I’m trying to understand under what circumstances Spark remains leaned and throttle up command produces acceleration in horizontal direction?

djiuser_gjLUHKA Posted at 2017-10-1 07:10
Explanation given by DJI support about the case with my Spark (I have a permission to quote):
“the behaviour of the aircraft was the result of flying without GPS signal.
The flight in ATTI mode then results in the stick inputs being translated not to a movement, but to a tilt in a specific direction. This is not automatically controlled by the aircraft, but hat to be handled by the pilot.

According to your flight log your AC raised from 1 ft to 5 feet, in that 2 second period, whether this caused the crash only you know but you did get a response from throttle.

hallmark007 Posted at 2017-10-1 10:56
According to your flight log your AC raised from 1 ft to 5 feet, in that 2 second period, whether this caused the crash only you know but you did get a response from throttle.

Yes it is, I saw this afterward. I don’t know why in app and Log viewer different values are displayed. But 4 feet for 2 sec 100% throttle is insignificant response.
I also don’t know what caused the crash. I saw several chaotic movements and sharp acceleration by pitching without command.

djiuser_gjLUHKA Posted at 2017-10-1 13:16
Yes it is, I saw this afterward. I don’t know why in app and Log viewer different values are displayed. But 4 feet for 2 sec 100% throttle is insignificant response.
I also don’t know what caused the crash. I saw several chaotic movements and sharp acceleration by pitching without command.

Looking at your log I can see one strange movement at 42 seconds, but most of the big movements came when using control sticks, was it windy?

hallmark007 Posted at 2017-10-1 13:29
Looking at your log I can see one strange movement at 42 seconds, but most of the big movements came when using control sticks, was it windy?

The last inputs were in an attempt to oppose the movements. There was weak wind - about 2 m/s.

djiuser_gjLUHKA Posted at 2017-10-1 13:43
The last inputs were in an attempt to oppose the movements. There was weak wind - about 2 m/s.

You know it seems from looking around here if you made no attempt to rescue your craft you may have got a warranty, and that is very strange.

hallmark007 Posted at 2017-10-1 13:48
You know it seems from looking around here if you made no attempt to rescue your craft you may have got a warranty, and that is very strange.

I don’t care about DJI warranty anymore. They finished their work in their own way.
Do you have an idea what exactly happened with the craft?

djiuser_gjLUHKA Posted at 2017-10-1 14:36
I don’t care about DJI warranty anymore. They finished their work in their own way.
Do you have an idea what exactly happened with the craft?

Well it’s hard to know what happened with the craft, but you were trying to fly with no gps which is always a risk and a much higher risk lower to the ground, I don’t know what area you were flying in and there are no gps coordinates to check the area, only you know this. The erratic behaviour is not really explained in your ph log. I’d put it down to experience.

hallmark007 Posted at 2017-10-1 15:31
Well it’s hard to know what happened with the craft, but you were trying to fly with no gps which is always a risk and a much higher risk lower to the ground, I don’t know what area you were flying in and there are no gps coordinates to check the area, only you know this. The erratic behaviour is not really explained in your ph log. I’d put it down to experience.

46.546083 N 12.066960 E
I definitely agree that it was a risky flight and the consequences are there. Mountaineering and shooting climbing scenes is my passion. I have made some number of flights with my P3 in similar conditions without big troubles.
There is another one strange thing. During last 2 seconds app displays GPS mode – is that means that it switched to GPS mode without enough gps lock?

djiuser_gjLUHKA Posted at 2017-10-1 22:34
46.546083 N 12.066960 E
I definitely agree that it was a risky flight and the consequences are there. Mountaineering and shooting climbing scenes is my passion. I have made some number of flights with my P3 in similar conditions without big troubles.
There is another one strange thing. During last 2 seconds app displays GPS mode – is that means that it switched to GPS mode without enough gps lock?

Sorry, my linguistic mistake above.
correction:
During the last 2 seconds app displays GPS mode – does that mean it has switched to GPS mode although there is not enough gps lock?

djiuser_gjLUHKA Posted at 2017-10-2 09:38
Sorry, my linguistic mistake above.
correction:
During the last 2 seconds app displays GPS mode – does that mean it has switched to GPS mode although there is not enough gps lock?

I would imaging it’s weak gps , no lock also if you check your flight log it shows no coordinates from beginning to end.
Usually with even 7/8 sats it will always record coordinates.

You were in a very spectacular place there, it’s a shame this happened.

When your in ATTI your flying it ,There's zero computer assist. I would only switch to ATTI in a emergency and then land immediately.

Dave Kent Posted at 2017-10-2 11:36
When your in ATTI your flying it ,There's zero computer assist. I would only switch to ATTI in a emergency and then land immediately.

You cannot switch spark to Atti , it’s part of the debate.

hallmark007 Posted at 2017-10-2 11:53
You cannot switch spark to Atti , it’s part of the debate.

We slightly deviated the topic.

In view of advice given by DJI support (quoted in #23) do you think that
in ATTI mode the pilot should care about the inclination of the AC after the stick is returned in neutral position?

I'm curious to see the comments of everyone who finds this important.

Thanks for sharing this info, I was out last night to try out some night shots with my SPARK and got scared and almost wetted my pants as i suddenly saw a message on my GO4 app saying:" AC disconnected", so the image turned black white but i still could see the camera and i still had control over my AC...don't know what it was, but i landed my SPARK ASAP to avoid any unpleasant thing to happen. Hope it does not do it again...

Viking-Pilot Posted at 2017-10-5 08:31
Thanks for sharing this info, I was out last night to try out some night shots with my SPARK and got scared and almost wetted my pants as i suddenly saw a message on my GO4 app saying:" AC disconnected", so the image turned black white but i still could see the camera and i still had control over my AC...don't know what it was, but i landed my SPARK ASAP to avoid any unpleasant thing to happen. Hope it does not do it again...

I've experienced this exact behavior a few times, where the app reported disconnect but I was still getting video and the craft was still responsive (no lag what-so-ever).  It's not something that I've been able to reproduce, and I've considered this a bug with the app.

You did right.  If there's anything unusual that occurs during a flight (be it craft behavior, software warnings, etc).

Charles Adams Posted at 2017-10-5 08:40
I've experienced this exact behavior a few times, where the app reported disconnect but I was still getting video and the craft was still responsive (no lag what-so-ever).  It's not something that I've been able to reproduce, and I've considered this a bug with the app.

You did right.  If there's anything unusual that occurs during a flight (be it craft behavior, software warnings, etc).

Scary, did you notice if you were flying near some metal structures or something? I think i was near a streetlight...don't know if that has something to do with it.

Charles Adams Posted at 2017-10-5 08:40
I've experienced this exact behavior a few times, where the app reported disconnect but I was still getting video and the craft was still responsive (no lag what-so-ever).  It's not something that I've been able to reproduce, and I've considered this a bug with the app.

You did right.  If there's anything unusual that occurs during a flight (be it craft behavior, software warnings, etc).

That exact same thing happened to me today too, several times. Luckily it was daytime and I was flying VLOS so I could see the aircraft was still responding to the RC, but the video image and flight data were frozen and I couldn't take photos or start a video - even with the RC buttons.
i guided the AC to landing with no problem, same as always. Without being able to take photos or shoot a video there was no point in remaining airborne anyway.
My phone is pretty old (LG G3 from 2014) and it sometimes overheats, I've noticed some correlation between overheating and that issue. I thought maybe when overheating some of my phone's systems begin to be erratic. Did you see that too?