So, you have just bought a drone with obstacle avoidance. Do you .........

[Dronecation]

I never use OA. Just don't need it for the type of flying I do. So I have never tried to test it out.

 

[Felix le Chat]

I have driven a car that has automatic braking to avoid collisions. I do not test it by driving at a wall full speed to see if it stops in time. That's stupid.

Same with my drone. I do not test APAS/OA by flying it like an idiot to see what it will stop for and what it won't.

I fly with due care and caution, and if it gets too close to an object and stops or avoids it, great.

Worse, my Air 2S doesn't have side OA, which would be more helpful for most autonomous shooting modes. But that just means I must fly it more carefully.

OA is not for testing. It's in case you completely bullocksed things up and got too close to something, then it MAY help avoid a costly collision.

I hardly think that testing if something works instead of taking someone else's word for it is the mark of an 'idiot'. If you took DJI's word for it and believed the shullbit they crank out in their marketing videos, you WOULD be flying like an idiot because you'd believe the things were all but impossible to crash.

 

[hansenn]

I enjoy watching drone fail movies on youtube and this keeps me away from pushing my expensive mavic 3 classic to the limits.

 

[Dave Maine]

There are all kinds of risks inherent in drone piloting. For me, the important thing is to be aware of them and fly consistently with them. There are error bands with the GPS positioning and repeatability, the barometric altimiter is notably erratic, the control signals between the drone and the controller are subject to interference from physical obstacles and competing signals on their operating frequencies.

The issue with performing tests is that they may pass once, but not a second time. GPS for example is specified for positional accuracy of +- 16 ft, 95%of the time. Most of the time it is far better, but 5% of the time it will be greater than 16 ft. from where you expect it.

A good pilot needs to be aware of all those uncertainties.

 

[Yorkshire_Pud]

A good pilot needs to be aware of all those uncertainties.

But how many are "aware" and how many use OA in "blind faith"?

 

[James in the Desert]

Immediately try to emulate the you tubers flying along forest paths whilst tracking themselves moving at speed or do you test it first by, perhaps, slowly flying towards the like of a solid wall and then gradually make faster runs at trickier objects etc.?
Just curious.

EDIT. Since some people seem to question the wisdom of 'testing' I will add the following

I.e. do you experiment with OA so that you 'get a feel ' for its limitations before you use it in a real situation, or, do you use it with 'no feel' for its limitations.

I never rely on obstacle avoidance, however, when I started flying drones at the beginning, I pushed the return home repeatedly just to get that confidence level

 

[JoelP]

I tested my Mavic 3 by flying toward a tree with sparsely populated branches to see how the avoidance behaved and where I knew I couldn't get into much trouble. This built my confidence in how well it worked and also showed me that there were circumstances where it could get trapped in a pocket between branches. It was instructive for me to practice how to back out of that spot. There is a blind spot on the Mavic 3 and I remain aware of not testing in that direction. Another test was inadvertent as I was flying along an abandoned railroad track and failed to see a fine branch that was extended over the track. It did admirably. I have yet to test the latest SW that shows me the images from the avoidance cameras, but am feeling that this will be helpful since I like to fly close to terrain.

 

[lensimages]

So, does that mean you have never used OA?

No it means that I never fly any of my drones into areas that I consider to be dangerous, cheers Len

 

[rjwmorrell]

I never use OA. Just don't need it for the type of flying I do. So I have never tried to test it out.

This is why I went with the Mini 3, not the Mini 3 Pro. Same camera, but cheaper. I'm only missing automated flight modes I don't use and OA I don't need.

 

[JimSteadman]

I use RTH all the time, but have had OA kick in only once - flying up and backward to get the subject in the perfect position for a photo. It stopped just before flying into a tree. I have however lost a drone to trees when flying sideways, with the drone focused on a point of interest, but the drone had no sideways sensors and my point of view prevented me from seeing the danger (you can think you have a clear view, only to find that the offending branch was not where you thought it was.) For me, the omnidirectional OA is a bit of an insurance policy. And like insurance, I hope I do not have to rely upon it but like the security of knowing it is there.

 

[Zbip57]

I tested my Mavic 3 by [...]. This built my confidence in how well it worked and also showed me that there were circumstances where it could get trapped in a pocket between branches. It was instructive for me to practice how to back out of that spot.

The more complex and "idiot-proof" these drones become, the more important it becomes to test all the various features to familiarize yourself with how they actually work.

All too often we see posts from people claiming their drone malfunctioned or "went out of control" or did something completely unexpected resulting in a crash. Only very rare cases can be attributed to actual hardware or software failures. The vast majority of cases by far are due to operator error caused by ignorance of how their drone actually is programmed to behave.

You don't need to deliberately run your drone at full speed toward a wall to test Object Avoidance. Surely there are sensible ways to test those systems to see what they can/can't do and what sort of limitations exist, so you won't be surprised when they do/don't kick in when unexpected/expected.

 

[Torque]

If only our drones had some sort of braking system so we could stop them on a dime. It would make testing the OA so easy.
Hmmm....

 

[LoudThunder]

Just to remind everyone…

The obstacle avoidance will not work in Sport Mode.

In Normal Mode, the maximum horizontal speed is 10m/s.

Faster than that, your Care Refresh had better be current…

 

[4 inch pistons]

It will not detect power lines in any mode.

 

[rjwmorrell]

It will not detect power lines in any mode.

True.

I have been looking at enterprise level drones for work, and even the latest greatest Matrice 350 RTK requires an install of a CSM radar unit to detect wires.

The fact is our hobby/prosumer level drones are not meant to be flown into fine objects like branches and wires. The tests I've seen of the latest Mini 4 Pro OA on YouTube are very deceptive. Flying a drone, even in ActiveTrack mode, into a bunch of tall trees with wide bare trunks, where the foliage is higher up the trunk than the drone, that's not that difficult. My Air 2S can avoid (mostly -- I wouldn't trust it in bypass) those potential hazards.

 

[Cooner]

Lol ... excellent responses. All I can say is "now ya tell me". Had MA2 in active track with OA enabled and wanted it to follow me on bike down off road path in the trees. Worked well through first few obstacles then decided it had had enough and tried to eat its way through tree branches. Had to send it in for repairs. Still want to get video of that path but this time it will be walking with me driving drone .... slowly. I guess the tech is just not quite there yet.

 

[LoudThunder]

All I can say is "now ya tell me".

Some folk have an over estimation of the capability of DJI's active track. They see a lesser drone that can follow the subject through trees, behind cars, around building, etc… (like there is a "tracking device" on the subject)

Well they do have a tracking device… The lesser drones are tracking the WiFi signal from the device the subject is carrying and as long as the drone receives the signal, it stays on target. When the subject is out of sight, the signal is still there and the drone soon picks up the subject again once the signal is direct…

DJI active track locks on the subject visually using a special Vision Recognition technique and if the subject travels behind a tree, bush, car, building, the track is broken and deactivated…

You have read and all need to take heed that Obstacle avoidance cannot necessary "see" thin branches of trees and bushes nor wires. It's all on you, the Pilot…

 

[Cooner]

I think what threw me is that the drone had navigated fairly easily around lesser obstacles successfully. It had tracked me fairly well. It slowed down to work its way through and around previous branches with no problem and then would speed up to catch up. Then it just ran straight into the branch that toasted it when there was plenty of room to either side and below the obstacle. I had tested OA with the drone before and it seemed to work amazingly well. As the general consensus of this thread seems to indicate .. if it works count yourself lucky but don't depend on it.

 

[rjwmorrell]

I think what threw me is that the drone had navigated fairly easily around lesser obstacles successfully. It had tracked me fairly well. It slowed down to work its way through and around previous branches with no problem and then would speed up to catch up. Then it just ran straight into the branch that toasted it when there was plenty of room to either side and below the obstacle. I had tested OA with the drone before and it seemed to work amazingly well. As the general consensus of this thread seems to indicate .. if it works count yourself lucky but don't depend on it.

That's definitely my approach. if it works and stops the drone from colliding, great. But I won't depend on it. Just like the auto-braking feature in cars now. I won't rely on it, I'm still going to follow cars at a safe distance and speed and use the normal foot brake.

 

[Ty Pilot]

Gonna take some flak for this but . . . .

Obstacle Avoidance on consumer drones is a sales gimmick at best, and at worst; a painful lesson for unsuspecting pilots. I have never used OA on any drone I've owned, and have never felt the need to 'test' it out. Why test something that is known to fail? All my drones have had the OA turned OFF on day one.

The irony of OA is that; the very type flying that would prompt someone to use it (OA), is precisely the type environment or flight; most likely to induce the failure. If the OA in 50,000 automobiles can fail, do we really believe a 1,000 dollar drone has a chance?

In my opinion, OA on drones is the antithesis of one of the bedrock fundamentals of sound piloting - risk mitigation. It is a known risk, it can (and will) fail under very well documented scenarios and conditions. If I need to put a camera into those conditions, then I'll find a way to do it without relying a gimmick that has a bad gambling habit piloting my drone.

Okay, off rant.