Lost my second DJI drone....this time a Mavic Air 2

[fGene]

I'm not that familiar with the 6-axis integration algorithms. But, after that step the results then have to fused with GPS, ultrasonic height, barometer and vision system data. It's this part that I'm supposing is done in different ways.

There have been several incidents where Yaw has been abruptly changed in one IMU and not the other. The gyro data (excuse me, I mean the angular velocity data ω ) clearly shows the rotation didn't actually happen. The FC was adjusting Yaw to what it thought it should be. Since this happened with just one of the IMUs I'm speculating that there must be some type of secondary non-conventional fusing that's different for each IMU.

Please, I am not picky about terms as long as I can understand the intended meaning. I know my English isn't perfect, but I am trying my best here as this discussion is very interesting. I used to write FC software myself, and apparently there has been significant progress in this field in recent years which I am trying to catch up to.

You have found an important clue which might explain everything. Here is my hypothesis. Yaw axis is different from the other 2 axii in that its gyro-derived angular position sometimes cannot be reliably corrected by accelerometers. At zero inclination (rest or hovering) the accelerometer influence equals zero, and in such case the yaw axis can only be corrected by fusion with magnetometer data, which is slow and error-prone but provides the only means for correction. On the other hand, once drone starts moving/changes attitude, the accelerometers provide enough meaningful data that can be used for yaw correction, and magnetometer data becomes less relevant. What if what you saw in the graph was the result of fusion of the same gyro data with magnetometer data for one virtual IMU and fusion with accelerometer data for another virtual IMU? If true, the observed variations between IMUs in the graph could be explained. In most cases they would match each other relatively closely, but in some cases deviation could be significant -- which is what you probably observed.

If this is indeed the case, DJI might be right about MA2 having a single hardware/physical IMU (that is, one having a single gyro, accel ang mag triad) .

 

[BudWalker]

In my example, I am assuming that the compass is correctly calibrated, only the external magnetic field is distorted. If all sensors are bad then the fused result will not be any better.



I have some doubt on this because without GPS, the drone will be in ATTI mode and we have all seen the resulting magnitude of drift in position. With GPS avaiable, it holds position a lot better. GPS can detect change in position very accurately by sensing the doppler shift. Every GPS satellite seen by the drone acts like a laser gun pointing to it. That's how GPS speedometers work.

I think that the FC turns off any attempt at position hold when the FC thinks the GPS or compass data isn't valid. It's not that position holding is degraded, it's not even attempted. The navHealth (aka gpsLevel or gpsHealth) signal is the confidence that the FC has in the location computed from the GPS and compass data. NavHealth has the range [0, 5] with a value >= 4 being required for GPS_ATTI mode.

Your supposition that VelN, VelE and VelZ (the doppler shift derived speeds) are also used in the location fusing algorithm makes sense. The signal nonGPSCause can take on the value SPEED_ERROR_LARGE. I had always thought that occurred as a result of a computation involving time and positions. But, nonGPSCause can also take on the value GPS_POSITION_NONMATCH.

I've always thought the gyro and accelerometer data was also used to compute location. But, I've not seen any direct evidence of this.

Are VelZ, VelE and VelZ accurate enough at the low speeds encountered in hovering? If not that implies the gyro and accelerometer data are also used in the location fusing algorithm.

 

[moonlitnite]

I think that as operators, it's at least as important to know the "why" as well as the "what." For instance cameras focus based on contrast ratio detection, phase detection, etc. So if you are taking a picture of a white wall, a dark scene like at nighttime, try as you might the camera will not focus at all. And even if you think the scene in the field of view has light/white and dark/black contrast variations, if the metering mode is set too tight (like spot mode) trying to shoot a photo of something like a white linen wedding dress will also fail to focus, or the camera lens will "hunt" and go in and out of focus during a video.

So this is not a hardware or software issue for a camera, or a problem with the UAV - it's a failure of the Operator to RTFM...Read The "Fine" Manual

 

[Capt KO]

Just add a float when over water. Simple, and with Care get a new one. Even my Phantoms slowly descend when hovering low over water. The MMini is much worse.

 

[G Randy Brown]

I had difficulty with the learning curve needed for the video and audio post production software I use...it looks like I have another challenge ahead understanding logs...glad I found this forum ?

 

[Capt KO]

I had difficulty with the learning curve needed for the video and audio post production software I use...it looks like I have another challenge ahead understanding logs...glad I found this forum ?

Just learn how to retrieve and post the logs. Leave the understanding to the experts here. Hurts my old head just lookin, much less, reading them. ? God bless sar104 and his understudies for all the help.?

 

[HWCM]

So after losing my Mavic mini in the water, I decided to upgrade to the Mavic Air 2. I purchased a Sealander kit for it but it did not arrive before my fishing trip that I wanted to use the drone on. The primary reason I bought the drone was to film fishing and boating activities so flying over the water is what I do mostly. I was flying the drone around the boat videoing us offshore and everything was going fine. I recorded a circle quickshot video and everything was fine. Then I had the drone hovering about 5 feet above the water off the side of the boat and I selected the dronie quickshot. When I selected it, the drone lowered itself right into the water and sank. I watched the flight log on my phone to verify that I was not touching down on the left joystick and I in fact was not. Any idea what happened?

Once again, I purchased the DJI refresh on this drone but cannot use it because I could not recover the drone. If this was a malfunction of the drone, do you think DJI will replace it?

Operator error.
Mini, Air 2, now get a Pro 2 and you will have a complete set in the water.
The owner's manual is a wonderful tool.

 

[crs06e]

Operator error.
Mini, Air 2, now get a Pro 2 and you will have a complete set in the water.
The owner's manual is a wonderful tool.

What was the operator error? Drone hovering and select dronie quickshot and it descends into the water.

 

[Alex 007]

The drone uses several tools to keep it in place and stable. The tool that controls stable altitude is the built in barometer, which works pretty well as I have another drone that has altitude hold only and has no other aids. However if atmospheric pressure changes while you are flying, which is does very often, the drone will often suddenly ascend or decend a couple of meters. Obviously if you are flying high up, this is not a problem. The GPS is mainly for controlling the latitude and longitude stability and is not reliable at all for helping with altitude. DJI does warn that the downward sensors do not work when flying over water, so this cannot be considered a malfunction.

 

[Dakarenduro]

What was the operator error? Drone hovering and select dronie quickshot and it descends into the water.

My opinion, flying 5 ft over the water and attempting an automated flight mode.

If you are going to ignore the manufacturer's warnings, and your own experience with already losing a device to the water gods, you should - my opinion - maintain manual control so you can correct any issues. in this instance when you witness the drone start to descend you should be manually ascending.

i know there is a debate on whether sensors should be on or off over water. my personal experience with Mavic 2 flying low over water....as I was about 2m above the water and trying to fly forward at speed to capture that "skimming across the water" footage, I noticed the drone to start to descend. Scared the crap out of me. I quickly adjusted the flight and realized that I was too low for my own risk/reward comfort zone. Had I been in an automated flight mode (circle, waypoint, courselock, etc) at that height and it began to descend I would probably not be quick enough to pause or override the descent.

I fly probably 90% over the water, but at much higher altitudes. One benefit to my Zoom... I can get a lot closer via optics over physical distance.

 

[Dakarenduro]

And per DJI's own advice, flight within 2meters of water may have erratic behavior and should be manually operated-
How to Fly Safely Over Water - DJI Guides

"All DJI drones come equipped with VPS, a technology that maps the surface below to help position the drone. When flying over water, however, light reflecting off of the surface can cause issues with this technology. Because of this, it’s best to maintain a height of at least two meters above the water. If you have to fly below this height for a special shot, we recommend you turn off the VPS to avoid any unstable actions by the drone. Caution: When you disable the VPS of your aircraft and the distance between water surface and drone is smaller than 2 meters, you should fly the drone with a low speed in your FOV (field of view). Also, the height needs to be adjusted frequently since the aircraft could possibly drift downwards. Be aware that the minimum height has to be above 1 meter when flying over water."

 

[SkyeHigh]

The drone uses several tools to keep it in place and stable. The tool that controls stable altitude is the built in barometer, which works pretty well as I have another drone that has altitude hold only and has no other aids. However if atmospheric pressure changes while you are flying, which is does very often, the drone will often suddenly ascend or decend a couple of meters. Obviously if you are flying high up, this is not a problem. The GPS is mainly for controlling the latitude and longitude stability and is not reliable at all for helping with altitude. DJI does warn that the downward sensors do not work when flying over water, so this cannot be considered a malfunction.

Atmospheric pressure doesn’t change that quickly to cause sudden changes in height. The typical barometric pressure sensors in drones are usually only accurate to +- 1-metre equivalent measurements. They also have inherent noise to deal with especially at low altitude, which typically is the equivalent of 25cm. Depending on how accurately the IMU uses this data depends on how level a hover is maintained.