It is well dark. And sadly these kinds of video shows nothing.
But most dont have equipment or any need to video theyr craft from a 3.rd perspective wile flying.
But if it is as you said, that the craft was in fact banked over, and the logs shows that this was not due to stick inpts then in fact there is a problem.
During atti the craft should not lean in any direction without stick input, even in strong wind. Well strong wind is relative, enough will toss the mavic over due to size weight.

Jason Lane Posted at 2017-1-31 18:22
Well I can share my video that I also shared in early Nov when I crashed my Mavic. I'm not saying it's a GOOD video, but it's what I have.

I was very new to quads and the Mavic, and I made the mistake of flying it in ATTI mode and at night. I know that wasn't the smartest move, and I eventually put it down to pilot error, accepted that and paid for the repairs. (No need to remind me that it was a bad move). However, I always maintained that the Mavic did not feel like it was just drifting, and it didn't seem to be following my commands. But being a novice pilot, I couldn't really say that with much authority. The video is short, and doesn't show much, but the one thing it does show is a definite bank angle as the Mavic flew sideways and up until it crashes into a tree.

In regards to unresponsiveness on the controller, there's some information on the latest User Manual (v1.2) saying that in S-Mode the responses are optimized for agility and speed making it more responsive to stick movements.

That may answer why it may not be doing what you ask immediately.

Mrimstad Posted at 2017-1-31 18:50
It is well dark. And sadly these kinds of video shows nothing.
But most dont have equipment or any need to video theyr craft from a 3.rd perspective wile flying.
But if it is as you said, that the craft was in fact banked over, and the logs shows that this was not due to stick inpts then in fact there is a problem.

Yup. Ever since that incident I now run a screen recorder on every flight, plus a chest- or head-mount camera. If I ever have an issue again, I should at least have enough data to properly diagnose what happened.

And even though my video is very dark, I can definitely make out the horizon, which looks to be tilted about 10 degrees. Maybe with your monitor's brightness turned up or something you might be able to see it.

cluc Posted at 2017-1-31 19:19
In regards to unresponsiveness on the controller, there's some information on the latest User Manual (v1.2) saying that in S-Mode the responses are optimized for agility and speed making it more responsive to stick movements.

That may answer why it may not be doing what you ask immediately.

In my case, the Mavic unfortunately flew out of my sight very quickly (it flew over my roof and over my house), and therefore I really didn't have much time to process and test the control's responsiveness. I gather than the controls might feel a little sluggish while in ATTI mode, but what concerns me more is the way that it flew sideways quite quickly without any inputs.

Mrimstad Posted at 2017-1-31 18:50
It is well dark. And sadly these kinds of video shows nothing.
But most dont have equipment or any need to video theyr craft from a 3.rd perspective wile flying.
But if it is as you said, that the craft was in fact banked over, and the logs shows that this was not due to stick inpts then in fact there is a problem.

It's true the video doesn't really offer any useful info.  The flight record/map is the go.

I've experienced the loss of GPS signal for an extended period (more than a minute) at least twice now. In both instances, there was no apparent reason for the loss of GPS signal (once I was flying over the ocean). Fortunately, I was flying fairly high with no nearby obstacles, so the MP came out unharmed in both cases. Once it happened during an auto RTH close to home - it was as if something evil possessed the MP as it was turning and flying in circles for about 15 - 20 seconds.
I can't say this for sure, but to me, it felt as if the MP is not just sluggish to respond to controller inputs when it goes into ATTI mode, but it is completely ignoring any controller input in the horizontal plane for the first 20 to 40 seconds after switching to ATTI mode. During this time, the Mavic goes crazy, turning in all directions.

After flying comfortably for several beginner and indoor flights I stuck in the 2nd charged battery and within 1 minute it began drifting up and back until it hit the ceiling. Out of nowhere it just start drifting the wrong way as if it was afraid it was going to hit the ground. I was intentionally flying indoors in a large auditorium and would like to continue doing so. Could the lower light and non-random floor trigger random climbs in elevation like this? Are the elevation/vision systems throwing this off? I echo many other in this thread and the other thread Elder started here.[/url] My flight data shows sensor activity while in this ATTI mode with no GPS. I don't believe it was confused as the conditions were very consistent. I would love to upload flight data and have it analyzed. I have also opened a support ticket with DJI(#418138).

Thanks all and thank you DJI.

If the Mavic veers off suddenly when its it switches to ATTI mode, it must mean that the IMU isn't calibrated surely? As mentioned before, the Mavic should be stabilising itself using the compass, baro and IMU.
The TBE effect is caused by a bad compass calibration but veering off should be because the IMU isn't registering the correct level.

With Mini quads the flight controller needs to be calibrated on a level surface, or modes like Horizon/Level will make it veer off in one direction.

Perhaps we should be doing an IMU calibration, despite the advice not to unless prompted ... i wish we could see the direct output from the X, Y, Z axis to confirm that you have a good calibration.

nicjames Posted at 2017-2-4 01:55
After flying comfortably for several beginner and indoor flights I stuck in the 2nd charged battery and within 1 minute it began drifting up and back until it hit the ceiling. Out of nowhere it just start drifting the wrong way as if it was afraid it was going to hit the ground. I was intentionally flying indoors in a large auditorium and would like to continue doing so. Could the lower light and non-random floor trigger random climbs in elevation like this? Are the elevation/vision systems throwing this off? I echo many other in this thread and the other thread Elder started here.[/url] My flight data shows sensor activity while in this ATTI mode with no GPS. I don't believe it was confused as the conditions were very consistent. I would love to upload flight data and have it analyzed. I have also opened a support ticket with DJI(#418138).

Thanks all and thank you DJI.

I don't think low light and/or uniform floor patterns should cause the Mavic to ascend. The Mavic doesn't rely on the visual sensors for altitude. It uses GPS, ultrasonics (when low enough) and barometric pressure when neither of the other two are available. It's possible you've experienced the VPS issue that many others are experiencing. ie. The Mavic wrongly thinks that it's basically touching the ground (0.1m or 0.3ft), and climbs to avoid crash landing.

Does your flight replay show incorrect height readings during the time the Mavic ascended? Maybe you can post a screen recording of your flight's replay.

Danny-B- Posted at 2017-2-4 05:31
If the Mavic veers off suddenly when its it switches to ATTI mode, it must mean that the IMU isn't calibrated surely? As mentioned before, the Mavic should be stabilising itself using the compass, baro and IMU.
The TBE effect is caused by a bad compass calibration but veering off should be because the IMU isn't registering the correct level.

Neither IMU BAROMETER OR COMPASS will stop you from drifting when your aircraft goes into Atti . Barometer will help with altitude, IMU will keep your AC level and Compass will keep its heading.

But if you imagine 10mph wind your aircraft looses gps then your aircraft is now traveling at 10mph, try it with your stick and see how far you get in a matter of seconds.

Gps is what holds your aircraft in place IMU keeps your aircraft level, and compass keeps your heading.
Pilot has to keep your aircraft in place when your in Atti.

nicjames Posted at 2017-2-4 01:55
After flying comfortably for several beginner and indoor flights I stuck in the 2nd charged battery and within 1 minute it began drifting up and back until it hit the ceiling. Out of nowhere it just start drifting the wrong way as if it was afraid it was going to hit the ground. I was intentionally flying indoors in a large auditorium and would like to continue doing so. Could the lower light and non-random floor trigger random climbs in elevation like this? Are the elevation/vision systems throwing this off? I echo many other in this thread and the other thread Elder started here.[/url] My flight data shows sensor activity while in this ATTI mode with no GPS. I don't believe it was confused as the conditions were very consistent. I would love to upload flight data and have it analyzed. I have also opened a support ticket with DJI(#418138).

Thanks all and thank you DJI.

Just a follow-up on something I said previously... I asked if your flight playback showed incorrect height readings, but I just checked one of my flights and realized that the playback does not appear to show the VPS height reading, only the GPS/baro reading (on my Android phone). So even if you were having the VPS issue, I'm not sure it would show up in the flight playback. I don't think the VPS issue would show any other clues in the playback. Maybe only in the detailed logs, or if you can remember seeing your VPS height icon turning red during the flight, indicating the Mavic thinks it's at 0.5m or less.

hallmark007 Posted at 2017-2-4 06:10
Neither IMU BAROMETER OR COMPASS will stop you from drifting when your aircraft goes into Atti . Barometer will help with altitude, IMU will keep your AC level and Compass will keep its heading.

But if you imagine 10mph wind your aircraft looses gps then your aircraft is now traveling at 10mph, try it with your stick and see how far you get in a matter of seconds.

I'm not talkin about it drifting though i know ATTI will not position hold, what i'm talking about (and everyone else here) is the fact that some mavic Veer off in one direction, nothing to do with wind.

Some of the accounts are when flying indoors so external forces are nill  

I have just flown my Mavic Pro for the first time. I used Beginner mode first of all and when I changed to the second battery I turned it off.

On 3 occasions the GPS signal dropped and ATTI mode was activated, all between 30m & 60m. It was pretty breezy and the Mavic drifted significantly. When using the controllers it seemed to struggle to move against the wind. One time I lost sight of the drone and it was more by luck than skill that the GPS signal came back and I was able to activate RTH.

On all occasions it was unnerving and disappointing that the GPS signal should just suddenly drop, particularly as I was flying in a country park which had not noticeable forms of interference.

After that I started to practice using the map to navigate back so that when it happens again I'll hopefully bet able to fly it back to me.

GlowRed Posted at 2017-2-4 06:57
I have just flown my Mavic Pro for the first time. I used Beginner mode first of all and when I changed to the second battery I turned it off.

On 3 occasions the GPS signal dropped and ATTI mode was activated, all between 30m & 60m. It was pretty breezy and the Mavic drifted significantly. When using the controllers it seemed to struggle to move against the wind. One time I lost sight of the drone and it was more by luck than skill that the GPS signal came back and I was able to activate RTH.

Just remember that flying by the map will only be minimally useful in ATTI mode if the Mavic's out of your sight, since it's not able to calculate a GPS fix.  It should still be able to report its compass heading and altitude, but that's about it.

Danny-B- Posted at 2017-2-4 06:56
I'm not talkin about it drifting though i know ATTI will not position hold, what i'm talking about (and everyone else here) is the fact that some mavic Veer off in one direction, nothing to do with wind.

Motors running are not an external force but a real one , take into account motors running and your in Atti mode indoors do you honestly think your aircraft will not drift or veer as some are calling it.

Some of the accounts are when flying indoors so external forces are nill

What's the difference in veer or drift , if you put an aircraft in the air with absolute no wind in Atti mode it will move try it, the same indoors , if you add to this with the merest touch on the rudder just watch and see how far it will move or veer as many seem to call it.

Your initial point was to do with compass IMU and barometer keeping aircraft in place that point was wrong .

Jason Lane Posted at 2017-2-4 07:03
Just remember that flying by the map will only be minimally useful in ATTI mode if the Mavic's out of your sight, since it's not able to calculate a GPS fix.  It should still be able to report its compass heading and altitude, but that's about it.

Your right. I have just looked at the log file. The Mavic is flying along in a straight line using the GPS. The GPS drops and suddenly jumps to a different point some distance away. After a minute, when the GPS comes back on the new point is where the Mavic is and this is when I manage to activate RTH and the drone flys back.

The drone drifted a significant distance in the wind, approximately 300m.

hallmark007 Posted at 2017-2-4 07:21
What's the difference in veer or drift , if you put an aircraft in the air with absolute no wind in Atti mode it will move try it, the same indoors , if you add to this with the merest touch on the rudder just watch and see how far it will move or veer as many seem to call it.

Your initial point was to do with compass IMU and barometer keeping aircraft in place that point was wrong .

Veering is (as i described in my first post) when the craft moves sharply to try level itself as the level isn't correct.

Do me a favour, calibrate your IMU on a non level surface ... fly it indoors and see what the effect is, it will veer off trying to maintain what it thinks is level. you'll need to fight against it to keep it level which is the point i'm trying to make.


My initial point was not insinuating that ATTI mode will keep the Mavic in one place, i'm not stupid, you are misinterpreting what i meant. I fly RC planes, Mini quads and stabilised camera platforms such as the mavic, and have for a several years now ... Trust me, i know my ATTI from GPS modes.

Danny-B- Posted at 2017-2-4 07:41
Veering is (as i described in my first post) when the craft moves sharply to try level itself as the level isn't correct.

Do me a favour, calibrate your IMU on a non level surface ... fly it indoors and see what the effect is, it will veer off trying to maintain what it thinks is level. you'll need to fight against it to keep it level which is the point i'm trying to make.

I know how important it is to correctly calibrate IMU, I also know how difficult it is to fly Atti mode indoors , but what I read about veering usually comes from inexperience pilots trying to fly indoors, for experience pilots they announced this as drifting.
There are very clear warnings through the app if IMU needs calibrating, you can also check yourself your IMU levels in the app.

Here is my input on your request.  Multirotor control systems are layered.  The most basic form has only three gyros (X Y Z axis).  With this simple stabilizer the MR is very hard to fly but with an expert at the controls its awesome.  This form is aerobatic.  Loops Rolls Flips etc.  The next layer of control is to add three accelerometers (X Y Z axis).  This adds self leveling and typically limits pitch and roll angles to 20 to 25 degrees.  Hold the stick forward the MR pitches forward, Let the stick go and it levels but continues to drift in the forward direction.  Do the same in gyro only, stick forward the MR pitches forward and flies forward, continue holding the stick forward and the MR will continue to pitch until its upside down and most likely in the dirt.  Next level of stabilization is the barometer and compass which is for altitude drift reduction and orientation.  These layers make up ATTI control.  Accessories to this is the sonar for height and downward cameras to hold position providing there is a pattern for it to look at.  A plain living room carpet is not suitable for the camera.
The last layer of control is adding GPS.  It is incorporated in this way.  The GPS control loop is always comparing positions.  If the systems detects a change in position, distance and angle, the control loop will input pitch and roll in the opposite direction to get back to the original GPS point.  It must also know which way the nose is pointing which comes from the compass.  The bigger the distance the larger the control input will be.  When hovering in the wind with GPS, the MR will have a tilt into the wind with no stick input.  Switch to ATTI mode or loose GPS signal and the control inputs are zeroed and the MR starts to drift with the wind.  A similar thing happens if the MR's mechanical alignment is off.  Say the angle of one of the motors is knocked off slightly in a crash.  GPS mode can still make the MR hold position but switch to ATTI mode and it flies off sideways because compensating inputs are zeroed.  A possible reason for MAVICs drifting off as they lose GPS is as the number of sats reduces some of the position points are way off which the control system interprets as movement and then compensates for it.  Should not happen if you NEVER get a GPS lock.

Hummingbird.UAV Posted at 2017-2-4 09:07
Here is my input on your request.  Multirotor control systems are layered.  The most basic form has only three gyros (X Y Z axis).  With this simple stabilizer the MR is very hard to fly but with an expert at the controls its awesome.  This form is aerobatic.  Loops Rolls Flips etc.  The next layer of control is to add three accelerometers (X Y Z axis).  This adds self leveling and typically limits pitch and roll angles to 20 to 25 degrees.  Hold the stick forward the MR pitches forward, Let the stick go and it levels but continues to drift in the forward direction.  Do the same in gyro only, stick forward the MR pitches forward and flies forward, continue holding the stick forward and the MR will continue to pitch until its upside down and most likely in the dirt.  Next level of stabilization is the barometer and compass which is for altitude drift reduction and orientation.  These layers make up ATTI control.  Accessories to this is the sonar for height and downward cameras to hold position providing there is a pattern for it to look at.  A plain living room carpet is not suitable for the camera.
The last layer of control is adding GPS.  It is incorporated in this way.  The GPS control loop is always comparing positions.  If the systems detects a change in position, distance and angle, the control loop will input pitch and roll in the opposite direction to get back to the original GPS point.  It must also know which way the nose is pointing which comes from the compass.  The bigger the distance the larger the control input will be.  When hovering in the wind with GPS, the MR will have a tilt into the wind with no stick input.  Switch to ATTI mode or loose GPS signal and the control inputs are zeroed and the MR starts to drift with the wind.  A similar thing happens if the MR's mechanical alignment is off.  Say the angle of one of the motors is knocked off slightly in a crash.  GPS mode can still make the MR hold position but switch to ATTI mode and it flies off sideways because compensating inputs are zeroed.  A possible reason for MAVICs drifting off as they lose GPS is as the number of sats reduces some of the position points are way off which the control system interprets as movement and then compensates for it.  Should not happen if you NEVER get a GPS lock.

Well put, gives a good understanding of balance and movements in the AC.

How long should it take between steps while calibrating the IMU ?I start th calibration but it never goes to the next step.

GlowRed,
As I understand in your case you was able to control your Mavic when GPS drop occurred.
And you said the drone drifted about 300 meters before GPS signal comeback.

Why didn't you used your phone to navigate drone back by a visual picture on the screen?  Or at least to not let it drift so far away.

Sergey_B Posted at 2017-2-4 09:33
GlowRed,
As I understand in your case you was able to control your Mavic when GPS drop occurred.
And you said the drone drifted about 300 meters before GPS signal comeback.

I was in a park and it was very difficult to get a visual reference using the Mavics camera, as everything was green / brown.

I didn't realize the drone had drifted so far, it was only when I looked at the log later on that I found out how far it had gone.

Hummingbird.UAV Posted at 2017-2-4 09:07
Say the angle of one of the motors is knocked off slightly in a crash.  GPS mode can still make the MR hold position but switch to ATTI mode and it flies off sideways because compensating inputs are zeroed.  A possible reason for MAVICs drifting off as they lose GPS is as the number of sats reduces some of the positi

Question about one thing you said:

"Say the angle of one of the motors is knocked off slightly in a crash.  GPS mode can still make the MR hold position but switch to ATTI mode and it flies off sideways because compensating inputs are zeroed."

You described earlier that the accelerometers in ATTI mode allow the MR to remain level. Wouldn't this still be done by some kind of feedback loop, not just a simple setting of "set motors speeds to xx rpm to remain level." Meaning, the MR will still be able to compensate for any external forces that cause it to tilt, right? So if one motor is mis-aligned or whatever, the MR will sense a tilt in one direction, and apply whatever compensations are necessary to bring it back to level, won't it? I feel like it shouldn't fly off sideways in the situation you've described.

Based on the video I posted above of the guy demonstrating ATTI mode with his Phantom, it certainly looked like it would be fully capable of compensating for a motor that's mis-aligned, or for any other factors that could cause an imbalance. You can see its self-leveling systems working very well when he gives the Phantom a good shove sideways.

Jason Lane Posted at 2017-2-4 18:19
Question about one thing you said:

"Say the angle of one of the motors is knocked off slightly in a crash.  GPS mode can still make the MR hold position but switch to ATTI mode and it flies off sideways because compensating inputs are zeroed."

Rewatched the guy and when he is in ATTI mode and giving the phantom a push he says it will just keep on drifting until he inputs countering control action.  With a misaligned motor in ATTI mode the craft gives itself its own push.  Oh and by the all of the gyros, accelerometers are in feedback loops

Hummingbird.UAV Posted at 2017-2-5 08:56
Rewatched the guy and when he is in ATTI mode and giving the phantom a push he says it will just keep on drifting until he inputs countering control action.  With a misaligned motor in ATTI mode the craft gives itself its own push.  Oh and by the all of the gyros, accelerometers are in feedback loops

OK, after thinking about it more, I think what I didn't take into account with your example of a mis-aligned motor is that there would then be a small amount of thrust vectored horizontally, even when the craft is level, therefore causing a bit of drift in one direction or another. So yeah, I can see how a mis-aligned motor would cause some drift even with a properly calibrated IMU. But I'm sure it would take a very severe mis-alignment to cause any significant amount of drift. Surely not the kind of sudden and quick acceleration to one side that many seem to be reporting with the Mavic in ATTI mode.

Jason Lane Posted at 2017-2-5 19:16
OK, after thinking about it more, I think what I didn't take into account with your example of a mis-aligned motor is that there would then be a small amount of thrust vectored horizontally, even when the craft is level, therefore causing a bit of drift in one direction or another. So yeah, I can see how a mis-aligned motor would cause some drift even with a properly calibrated IMU. But I'm sure it would take a very severe mis-alignment to cause any significant amount of drift. Surely not the kind of sudden and quick acceleration to one side that many seem to be reporting with the Mavic in ATTI mode.

Right, a crooked motor wouldn't do that.  Still a mystery.

I'm still pretty new to this but what I learned by forcing it to switch back and forth between weak GPS signals and ATTI/Opti is that my Mavic becomes semi-unresponsive and since I am intentionally doing this in the corner of auditorium indoors with weak GPS, it continues to bounce back and forth between modes and the unresponsiveness seems to happen for 2-3 seconds each time. While this switching is occurring it either does not seem to respond at all or it responds with only intense direction on the sticks which results in in jerking forward once it decides to respond or the 'pause' goes away. It will drift faster than usual during these transitions as well.

  I can replicate this easily without GPS being available by lowering the lights inside. I think it's VPS that stops working at the thing starts drifting consistently every time I do this. I would not have an issue with this if the controls would work normally but instead it totallys acts up. It drifts quickly up and down and does NOT respond well on the sticks and performs the same jerky motion as a result of either a delay or the fact that I have to push harder down on the sticks. I want to be able to fly in Opti mode feeling confident the drone is under my control but that's not the case yet. When both of the above these cases occur I feel a bit like I'm barely keeping a wild little beast under my control!

If I disable VPS it flies like I feel it should. It drifts slightly as I would expect but the controls are issue free and it's not acting like it has a mind of it's own. It's a bit drifty and just feels as if it is very dependant on the VPS sensors and not so much it's other sensors.

nicjames Posted at 2017-2-5 22:29
I'm still pretty new to this but what I learned by forcing it to switch back and forth between weak GPS signals and ATTI/Opti is that my Mavic becomes semi-unresponsive and since I am intentionally doing this in the corner of auditorium indoors with weak GPS, it continues to bounce back and forth between modes and the unresponsiveness seems to happen for 2-3 seconds each time. While this switching is occurring it either does not seem to respond at all or it responds with only intense direction on the sticks which results in in jerking forward once it decides to respond or the 'pause' goes away. It will drift faster than usual during these transitions as well.

  I can replicate this easily without GPS being available by lowering the lights inside. I think it's VPS that stops working at the thing starts drifting consistently every time I do this. I would not have an issue with this if the controls would work normally but instead it totallys acts up. It drifts quickly up and down and does NOT respond well on the sticks and performs the same jerky motion as a result of either a delay or the fact that I have to push harder down on the sticks. I want to be able to fly in Opti mode feeling confident the drone is under my control but that's not the case yet. When both of the above these cases occur I feel a bit like I'm barely keeping a wild little beast under my control!

If you're willing to, I think it would be very useful if you could try to capture this on video. Preferably with multiple cameras. I've been meaning to do similar experiments, but just haven't been able to, yet.

I plan to have at least 1 camera on a tripod, one head cam (a pair of camera glasses), one GoPro filming my hands on the controller (on a chest mount or clamped to my phone), and a screen recorder running on my phone as well.

Jason Lane Posted at 2017-2-6 00:17
If you're willing to, I think it would be very useful if you could try to capture this on video. Preferably with multiple cameras. I've been meaning to do similar experiments, but just haven't been able to, yet.

I plan to have at least 1 camera on a tripod, one head cam (a pair of camera glasses), one GoPro filming my hands on the controller (on a chest mount or clamped to my phone), and a screen recorder running on my phone as well.

I don't have the gear to get it on camera myself, not even a 2nd cell phone so I will see if I can recruit a buddy to film. Showing the flight-log stick movements along with the video should be interesting. We can dim the lights slowly and back up to watch it start acting up on-demand. it does sounds like they have a VPS fix in the works for release soon as well. I'm grateful for all the early flyers and folks who here willing to send in their drones so DJI had more info to enhance their software.  

GlowRed Posted at 2017-2-4 06:57
I have just flown my Mavic Pro for the first time. I used Beginner mode first of all and when I changed to the second battery I turned it off.

On 3 occasions the GPS signal dropped and ATTI mode was activated, all between 30m & 60m. It was pretty breezy and the Mavic drifted significantly. When using the controllers it seemed to struggle to move against the wind. One time I lost sight of the drone and it was more by luck than skill that the GPS signal came back and I was able to activate RTH.

DJI GO 3 or 4 ? Android ? Recently I had a lot of problems with DJI GO 4 on my android (though no ATTI-complication) and went back to DJI GO 3 which gave me hazzle-free flights from the very first beginning .............

Hummingbird.UAV Posted at 2017-2-4 09:07
Here is my input on your request.  Multirotor control systems are layered.  The most basic form has only three gyros (X Y Z axis).  With this simple stabilizer the MR is very hard to fly but with an expert at the controls its awesome.  This form is aerobatic.  Loops Rolls Flips etc.  The next layer of control is to add three accelerometers (X Y Z axis).  This adds self leveling and typically limits pitch and roll angles to 20 to 25 degrees.  Hold the stick forward the MR pitches forward, Let the stick go and it levels but continues to drift in the forward direction.  Do the same in gyro only, stick forward the MR pitches forward and flies forward, continue holding the stick forward and the MR will continue to pitch until its upside down and most likely in the dirt.  Next level of stabilization is the barometer and compass which is for altitude drift reduction and orientation.  These layers make up ATTI control.  Accessories to this is the sonar for height and downward cameras to hold position providing there is a pattern for it to look at.  A plain living room carpet is not suitable for the camera.
The last layer of control is adding GPS.  It is incorporated in this way.  The GPS control loop is always comparing positions.  If the systems detects a change in position, distance and angle, the control loop will input pitch and roll in the opposite direction to get back to the original GPS point.  It must also know which way the nose is pointing which comes from the compass.  The bigger the distance the larger the control input will be.  When hovering in the wind with GPS, the MR will have a tilt into the wind with no stick input.  Switch to ATTI mode or loose GPS signal and the control inputs are zeroed and the MR starts to drift with the wind.  A similar thing happens if the MR's mechanical alignment is off.  Say the angle of one of the motors is knocked off slightly in a crash.  GPS mode can still make the MR hold position but switch to ATTI mode and it flies off sideways because compensating inputs are zeroed.  A possible reason for MAVICs drifting off as they lose GPS is as the number of sats reduces some of the position points are way off which the control system interprets as movement and then compensates for it.  Should not happen if you NEVER get a GPS lock.

Question now is: why the heck does it switch from GPS to ATTI-mode directly in OPEN AIR and OPTI-MODE does not step in BEFORE to give at least some kind of balance and position fix navigating by landmarks and horizon ?

Questions over questions .........................

Angelo26 Posted at 2017-2-7 01:29
Question now is: why the heck does it switch from GPS to ATTI-mode directly in OPEN AIR and OPTI-MODE does not step in BEFORE to give at least some kind of balance and position fix navigating by landmarks and horizon ?

Questions over questions .........................

The OPTI mode uses the forward and downward cameras. If you are higher than 43Ft or if the surface doesn't have detectable patterns OPTI mode does not work.

Hummingbird.UAV Posted at 2017-2-7 09:10
The OPTI mode uses the forward and downward cameras. If you are higher than 43Ft or if the surface doesn't have detectable patterns OPTI mode does not work.

So that means OPTI does not detect e.g. street-patterns on the ground from - let´s say - 100 m above ? right ?

Angelo26 Posted at 2017-2-7 23:49
So that means OPTI does not detect e.g. street-patterns on the ground from - let´s say - 100 m above ? right ?

That's right, manual does say only up to 43ft.

Believe there needs to be a better, gentler handoff to just basic ATTI mode from a sudden GPS loss event.  If you are working in tight places, recovery margins can be thin.  Even adding an audible call-out warning of satellite loss would be little better than the gnawing feeling that a drift event can happen at anytime.   Was not a big issue when we only flew over fields.

-ZLV Posted at 2017-2-8 09:15
Believe there needs to be a better, gentler handoff to just basic ATTI mode from a sudden GPS loss event.  If you are working in tight places, recovery margins can be thin.  Even adding an audible call-out warning of satellite loss would be little better than the gnawing feeling that a drift event can happen at anytime.   Was not a big issue when we only flew over fields.

How might you code a gentler handoff? That's like asking the lights to dim slowly when the power goes out.

Great Question.  I wish I knew.  I am not a coder.  But if I had the chance, I would ask them if there was (in the process, somewhere) a capability during a GPS loss event, to delay ATTI long enough for manual override by the Pilot?  Otherwise, maybe an AD gets pushed out on how sudden GPS Loss occurs, how the MP responds to sudden GPS Loss, what precautions to take to recover from such events.  It occurs  randomly but pilots describe the event similarly.    Interesting issue.  Bottom-line: always carry an heir and a spare?

Hi, I observed this issue with my Mavic Pro. I noticed that in ATTI mode the drone is extremely difficult to pilot. Decided to make an indoor experiment (no GPS) in weak light (no visual positioning). The sensors do not pick up any interference, but notice how the drone speeds up backwards with no manual input. After a second or two I tried to counter the movement using the move forward control, which is shown on the video if you look closely, but the drone hardly reacts. The video could be recorded with something better than an iPhone 6 camera, but that was too terryfing for me to risk repeating this experiment (perhaps after my bumpers arrive).



Please note that ATTI mode is different from OPTI mode, where your drone doesn't have connection to GPS, but the visual positioning system keeps it stable. You probably shouldn't fly your Mavic in poor lighting conditions or wherever the visual positioning system might fail - if you end up losing GPS signal in such conditions your drone will uncontrollably fly in a random direction (and possibly hurt somebody). I would consider flying indoors a significant risk.

I started to investigate this because it caused me two crashes - in both cases Mavic switched to ATTI mode in random scenarios. I was lucky that only the propellers were damaged and my Mavic is just scratched a bit, however I am much less confident flying it now. It really looks like a serious problem - if your drone has RC connection it shouldn't be uncontrollable at any point.

atti mode is the reason i dont have my mavic now... never again