Hat Tip to DJI: A Message From An Impressed Engineer

[Kilrah]

I'm also an engineer, and have read of about a dozen reports of the Mavic stopping dead in its track due to false OA detection, staying there for minutes until the battery is empty and finally landing in position. I have not read any other report like yours, so for now it stands as the odd one out rather than normal behavior, and since it can be easily explained by the erroneous detection condition disappearing this seems much more likely than a drastic change in behavior.

Observations show it certainly didn't use to have such smarts built-in, so unless it was added in last firmware (undocumented) it still doesn't. That's of course a possibility, but should first be confirmed either by a clear message, or more widespread observation of the behavior you suggest (with reduction/disappearance of reports of it staying stuck).

An engineer's valuable trait is critical thinking, and when assessing a single observation the "proper" way to go about it is usually to try and find as many possible explanations for it, then analyze each of them and rate them by how likely they are to be the right one to explain the observation... NOT to straight jump to the only one you have in mind (there almost never is only one possible explanation, so you should never be happy with just one) or the one that pleases you, that would make a rather terrible engineer.

 

[JLane]

Observations: as soon as I hit RTH the AC will turn around, and if the sun is low on the horizon it might fly forward for a few ft, but then if fill flash obstacle on the console and began to climb, the controller will cycle between obstacle and Going Home, but basically it was climbing vertically and not moving back to home.

When the Mavic detects an obstacle during RTH (whether real or false), does it give the same on-screen warnings and radar arcs that we see during normal flight mode? I've so far never had OA triggered during RTH, but I know it's very common, and I'm just curious about the on-screen warnings when it does.

 

[znuttyone]

When the Mavic detects an obstacle during RTH (whether real or false), does it give the same on-screen warnings and radar arcs that we see during normal flight mode? I've so far never had OA triggered during RTH, but I know it's very common, and I'm just curious about the on-screen warnings when it does.

Yes , controller will rotate being going home and obstacle flashing

 

[znuttyone]

I'm also an engineer, and have read of about a dozen reports of the Mavic stopping dead in its track due to false OA detection, staying there for minutes until the battery is empty and finally landing in position. I have not read any other report like yours, so for now it stands as the odd one out rather than normal behavior, and since it can be easily explained by the erroneous detection condition disappearing this seems much more likely than a drastic change in behavior.

Observations show it certainly didn't use to have such smarts built-in, so unless it was added in last firmware (undocumented) it still doesn't. That's of course a possibility, but should first be confirmed either by a clear message, or more widespread observation of the behavior you suggest (with reduction/disappearance of reports of it staying stuck).

An engineer's valuable trait is critical thinking, and when assessing a single observation the "proper" way to go about it is usually to try and find as many possible explanations for it, then analyze each of them and rate them by how likely they are to be the right one to explain the observation... NOT to straight jump to the only one you have in mind (there almost never is only one possible explanation, so you should never be happy with just one) or the one that pleases you, that would make a rather terrible engineer.

Was not a single observations for me, I and repeat the conditions and see it every time (OA triggers increase in altitude past rth altitude when rth flight path is into the sun and sun is low on horizon with .400 firmware) try it ..

 

[znuttyone]

That is not irritating at all, since the main camera is not used for obstacle detection.
It's the stereo cameras on the front which do it and these cannot be moved.

I was seen the same, in most cases where I am above trees or obstacles , I fly with OA off, and always have the rth obstacle check setting disabiled now with an RTH altitude set above any obstacles in flight path

 

[znuttyone]

What you described has happen to me before and it makes total sense because the Sun is a big "obstacle" as are headlights of oncoming cars, high-powered street lights, or just a hand-held flashlight. I'm not sure what you expect the Mavic to do in these situations but when I had issues with RTH or just flying into the Sun, I just yaw 180 degrees and fly the orange line on the DJI app home. Pretty simple fix.

My original post was and still is a hat tip for the intelligence built into the Mavic but it may not work in all cases. Maybe my Mavic acted differently because I was already near the home point or maybe it just made the correct decision the first time and then landed. I can only say that it worked and it did it without my input.

Tesla puts a lot of intelligence into their cars but people still find a way to crash them.

Another option is just to through it in sport mode , OA gets disabled that way and you can fly it straight back with camera facing ..or you can simply jump to the settings and turn it off, it takes immediate effect in both dji go and litchi

 

[AlanTheBeast]

I'm amused at the "critical thinking" of engineers mentioned above. Yes, it is indeed a proper trait of good engineers. But the plain fact is that the OA optical sensors on the MP will both see an equally illuminated surface with no texture or dimension when lit by the sun and the stereoscopic s/w will determine it to be an obstacle at close range. I've had this happen in P-mode flight numerous times during the magic hour.

This is a problem with this sort of passive detection (all detectors have some issues). It has no ranging function other than stereoscopy - that in turn is subject to all manner of false positives due to the situation above, reflections, aliasing and more.

When it was designed, the engineers should have done a risk assessment for all flight regimes including RTH. It would appear that such was not done.

I'm not all that peeved at this issue, I've come to the conclusion that for my flying and photography that false-obstacle-detection-death-trap-by-battery-drain is a likely enough scenario and so I've mitigated it by a RTH altitude set quite high and OA off during RTH.

It would behoove the engineers at DJI to issue a safety bulletin to users so that they can make their own mitigation plans.

In the meantime DJI's engineers may come up with improvements to the situation including some smarts that recognize that an automated descent to ground of last resort is better than falling out of the sky during OA-RTH when the battery is starting to wane.

 

[znuttyone]

This.

The Mavic does not deactivate OA for RTH failure and if it does, that's incredibly stupid. Based on many prior tests and user experiences, OA needs to be turned off prior to RTH with a low angle sun or nothing but bad things happen.

Just want to clarify the settings that impact rth and OA ..
You do not need to disable OA entirely if you want to avoid the RTH OA issue where it falsely detects the sun as an obstacle when in RTH mode. You simply need the 'RTH obstacle check' setting disabled , i.e., OA can still be on for all other flying, litchi missions etc.. if you are experiencing false OA detection when attempting to fly during sunrise or sunset , then you would need to disable OA in the main visual navigation settings menu, which btw disables the RTH obstacle check setting which resides in the advance settings portion of the visual navigation menu tab, see attached screenshots View media item 423View media item 422

 

[znuttyone]

This.

The Mavic does not deactivate OA for RTH failure and if it does, that's incredibly stupid. Based on many prior tests and user experiences, OA needs to be turned off prior to RTH with a low angle sun or nothing but bad things happen.

Just to clarify. This is the setting to disable to avoid the issue View media item 422
You do not need to disable OA entirety

 

[Kilrah]

I and repeat the conditions and see it every time (OA triggers increase in altitude past rth altitude when rth flight path is into the sun and sun is low on horizon with .400 firmware) try it ..

Yes that is how it is, we know it - but that's not what OP was describing.
I've never been able to trigger a false OA detection myself so far. Has a bit to do with flying with it off most of the time anyway since that's usually the better choice for what I do.

When it was designed, the engineers should have done a risk assessment for all flight regimes including RTH. It would appear that such was not done.

They certainly did, and OA can and has helped in many scenarios. It's not perfect and will fail in some cases as can be expected from any automation that can never be perfect, but the global balance is pretty certainly positive i.e. more cases where it saves things than cases where it causes trouble, so it's still a good thing to have.

As a user if you're sure of what you're doing then OA should never help you, so you should disable it and thus will avoid the potential detrimental effects of a false detection... but for a newbie who doesn't really know what they're doing OA is more likely to save them than ditch them and thus a welcome addition.

 

[AlanTheBeast]

Yes that is how it is, we know it - but that's not what OP was describing.
I've never been able to trigger a false OA detection myself so far. Has a bit to do with flying with it off most of the time anyway since that's usually the better choice for what I do.


They certainly did, and OA can and has helped in many scenarios. It's not perfect and will fail in some cases as can be expected from any automation that can never be perfect, but the global balance is pretty certainly positive i.e. more cases where it saves things than cases where it causes trouble, so it's still a good thing to have.

As a user if you're sure of what you're doing then OA should never help you, so you should disable it and thus will avoid the potential detrimental effects of a false detection... but for a newbie who doesn't really know what they're doing OA is more likely to save them than ditch them and thus a welcome addition.

Please provide proof of the "risk assessment" that identifies the death-by-OA-on-RTH-into-the-sun issue.

Had they done so the result would have been:

- a clear, concise explanation of same in the manual along with suggested risk abatements.

Note: The word "sun" does not appear even once in the manual (ref: V1.2).​

- an automated landing straight down in all situations where the battery is nearly dead.
​

 

[Kilrah]

The manual is "toy" level, not "aviation" level, nothing more. That's unfortunate, but a completely different problem.

- an automated landing straight down in all situations where the battery is nearly dead.

That's the case...

 

[AlanTheBeast]

The manual is "toy" level, not "aviation" level, nothing more. That's unfortunate, but a completely different problem.


That's the case...

Really? Then why are people warning about battery shutdown in flight?

Where is that documented?

Still waiting on the risk assessment for OA on RTH too.

 

[Kilrah]

Manual page 14-15.

Really? Then why are people warning about battery shutdown in flight?

Uh, who is? That wouldn't happen unless there's a critical battery failure (or you do something very stupid like set the critical level as low as you can and purposefully leave the aircraft too high for that to be enough to descend safely, after already ignoring the low battery warning...)

 

[AlanTheBeast]

Manual page 14-15.

Uh, who is? That wouldn't happen unless there's a critical battery failure (or you do something very stupid like set the critical level as low as you can and purposefully leave the aircraft too high for that to be enough to descend safely, after already ignoring the low battery warning...)

What I thought others had said/confirmed was that confronted with an OA stop on RTH (out of radio contact) that the drone would sit there stopped, at RTH height, until the battery ran down and then fall to the ground.

A re-reading of p.14 seems to confirm that is wrong - colour me relieved!

But I'm still disabling OA for RTH and setting it to max legal altitude. Maybe higher for that matter.

 

[Jenee]

I am going to leave all OA on as I am now aware of the issues and can just adjust my flying to get around it. Fly sideways, fly backwards high enough to avoid obstacles where necessary.

 

[JLane]

I am going to leave all OA on as I am now aware of the issues and can just adjust my flying to get around it. Fly sideways, fly backwards high enough to avoid obstacles where necessary.

Except you won't be able to adjust anything in an emergency RTH where you've lost comms with the Mavic.

Even though I never intentionally fly my Mavic out of VLOS or where I'm likely to lose RC comms, I am aware that there is a small chance that these things could happen. eg. An RC malfunction or an unexpected loss of GPS causing a long and uncontrolled drift in ATTI mode, eventually placing my Mavic out of VLOS and comms. At this point, if the Mavic manages to regain GPS mode and initiate RTH, it should safely make its way back to me. But... If it gets hung up on a false OA warning, it might be lost forever. I'm quite sure I've read of this happening to one or more people in the forums.

So for me, the trade-off leaves me to believe that I'd rather leave OA disabled, and just make absolutely sure I set the RTH height appropriately for each flight, which I've always done, anyway.

 

[Jenee]

Except you won't be able to adjust anything in an emergency RTH where you've lost comms with the Mavic.

Even though I never intentionally fly my Mavic out of VLOS or where I'm likely to lose RC comms, I am aware that there is a small chance that these things could happen. eg. An RC malfunction or an unexpected loss of GPS causing a long and uncontrolled drift in ATTI mode, eventually placing my Mavic out of VLOS and comms. At this point, if the Mavic manages to regain GPS mode and initiate RTH, it should safely make its way back to me. But... If it gets hung up on a false OA warning, it might be lost forever. I'm quite sure I've read of this happening to one or more people in the forums.

So for me, the trade-off leaves me to believe that I'd rather leave OA disabled, and just make absolutely sure I set the RTH height appropriately for each flight, which I've always done, anyway.

I agree with you but I have a max distance set at 300mts because beyond that I just can't see it. Also if anyone loses Comms then it is a problem whether OA is enabled or not. I think some people are being a bit hysterical about OA as it really is a great feature with a current bug regarding the angle of the sun but otherwise works well.