Anybody having issues with radius drifting during a POI moves?

[TnDronePilot]

So this has been going on for a while but never got on the forums to ask the question. A few months ago, I did some Active Track POI moves in the parking lot of a fire station. Now it was dark outside with only the firetrucks illuminating the area with their lights but, I noticed the drone's radius was always drifting to the outside from the original imaginary radius. In fact, the drone almost hit their flagpole because I was being blinded by the red and white strobes.

Well I just did an inside drone shoot for a motorcycle museum and had the same issue where the drone would slowly drift to the outside no matter where the camera was positioned and or how slow the crab. Sometimes I had to abort the track as my drone was getting too close to other bikes around it and most of those are worth more than my life.

Now the fire trucks were 15-30 feet away from the drone versus the motorcycles that were most times just few feet away from so to me, that rules out the drone having to move away to keep the target in frame. I've attached both videos to show you the framing and moves and although they're edited videos that don't show the drift, you can at least see the difference in distance and framing. Any ideas?

Fire Trucks:

Museum:

 

[GadgetGuy]

Do the orbits manually? I have never relied on the automated orbits. They are "good enough" for those that never learned to do them manually, but I find I have far better success doing them under my skilled control.

 

[Marcass Carcass]

This a Dronelink mission controlling an MA2. The poi is the sign, although I think could "fix" it by comparing footage to the Google earth simulation and making adjustments. BUT I have been flying manually for the most part now, esp poi's........


Hopefully someone will have an answer on here. Following....I really like automation though.

 

[Marcass Carcass]

Do the orbits manually? I have never relied on the automated orbits. They are "good enough" for those that never learned to do them manually, but I find I have far better success doing them under my skilled control.

Exactly

 

[TnDronePilot]

Do the orbits manually? I have never relied on the automated orbits. They are "good enough" for those that never learned to do them manually, but I find I have far better success doing them under my skilled control.

I'm not steady enough to pull a 180 degree move and keep the camera dead set on the subject the whole time. I'm not going to stop using Active Track because of the issue as I still get exactly what I need for my edits. I was just asking if there is a way to keep it from happening.

 

[GadgetGuy]

I'm not steady enough to pull a 180 degree move and keep the camera dead set on the subject the whole time. I'm not going to stop using Active Track because of the issue as I still get exactly what I need for my edits. I was just asking if there is a way to keep it from happening.

The automation isn't "steady enough" either.
The algorithm still can't replicate a skilled pilot.
Close enough for government work, but not as good as you want.

 

[Deleted member 103366]

How does the drone compute the flight course for these POI's ?
I can think of two/three methods but don't know if they are feasible.
1) select the target in the camera image and then fly so as to keep the target a constant size (that seems fragile to me because of "but what happens if the target it obscured" and what happens if the target, as seen by the camera, changes shape as the drone orbits the target.)
2) sets a centre for the orbit in terms of Lat & Long from GPS and then flies around the target following mathmatically calculated course (but that wouldn't work for a moving target nor indoors).
3) some combination of 1 & 2.
Realistically I would be surprised if it were that 'simple'.

 

[TnDronePilot]

The automation isn't "steady enough" either.
The algorithm still can't replicate a skilled pilot.
Close enough for government work, but not as good as you want.

Well the current algorithms are making me some really good money and I don't work for the government. They're just the ones that take it.

 

[GadgetGuy]

Well the current algorithms are making me some really good money and I don't work for the government. They're just the ones that take it.

Indeed. They are good enough that the client won't complain, but pixel peepers will may find that they can do better manually, with skillful piloting, especially when the subject is difficult to track or very close. Have a look at perfection with an FPV drone:

Login • Instagram

Welcome back to Instagram. Sign in to check out what your friends, family & interests have been capturing & sharing around the world.

www.instagram.com

 

[Dave Maine]

The GPS position continuously changes as it’s base accuracy is +- 16 ft 95% of the time. It is a characteristic of the system. Nothing DJI or you can do about it. You can easily see this if you watch the drone in a hover with no stick input. It will wonder around the position several feet. Add the changing position when the drone is orbiting you can easily get some very odd orbits.

 

[Jet skier]

So this has been going on for a while but never got on the forums to ask the question. A few months ago, I did some Active Track POI moves in the parking lot of a fire station. Now it was dark outside with only the firetrucks illuminating the area with their lights but, I noticed the drone's radius was always drifting to the outside from the original imaginary radius. In fact, the drone almost hit their flagpole because I was being blinded by the red and white strobes.

Well I just did an inside drone shoot for a motorcycle museum and had the same issue where the drone would slowly drift to the outside no matter where the camera was positioned and or how slow the crab. Sometimes I had to abort the track as my drone was getting too close to other bikes around it and most of those are worth more than my life.

Now the fire trucks were 15-30 feet away from the drone versus the motorcycles that were most times just few feet away from so to me, that rules out the drone having to move away to keep the target in frame. I've attached both videos to show you the framing and moves and although they're edited videos that don't show the drift, you can at least see the difference in distance and framing. Any ideas?

Fire Trucks:

Museum:

I was going to say do the Orbitz manually and you will get a better job but I noticed everybody else said that but if you do it manually you’ll definitely need a spotter it sounds like them out environment maybe that’s why you did it with the app

 

[TnDronePilot]

Indeed. They are good enough that the client won't complain, but pixel peepers will may find that they can do better manually, with skillful piloting, especially when the subject is difficult to track or very close. Have a look at perfection with an FPV drone:

Login • Instagram

Welcome back to Instagram. Sign in to check out what your friends, family & interests have been capturing & sharing around the world.

www.instagram.com

I've been through plenty of pixel peeping lectures in my photography days but I don't care about those people. I'm in the video production business so I know all about getting the perfect shot. I was simply asking why it drifts and not how I can improve my shots.

The GPS position continuously changes as it’s base accuracy is +- 16 ft 95% of the time. It is a characteristic of the system. Nothing DJI or you can do about it. You can easily see this if you watch the drone in a hover with no stick input. It will wonder around the position several feet. Add the changing position when the drone is orbiting you can easily get some very odd orbits.

This is the kind of reply I was looking for and not a lecture on why I'm better off doing it manually. Thanks.

 

[tlswift58]

For the outside shot of the fire station, did you have GPS or not. I'm thinking "inside" the museum you did not have GPS there for sure.

With that - you were in ATTI mode and the drone will drift some. Seems nature of that beast and nothing DJI can do about it.

I have never done a POI and all the previous posts may give you some insight on that feature.

Flying at night tends to mess with all onboard controls / sensors and if in ATTI mode - that complicates it even more.

 

[GadgetGuy]

I've been through plenty of pixel peeping lectures in my photography days but I don't care about those people. I'm in the video production business so I know all about getting the perfect shot. I was simply asking why it drifts and not how I can improve my shots.

This is the kind of reply I was looking for and not a lecture on why I'm better off doing it manually. Thanks.

Your original post certainly did not make that clear.
You asked, "Any ideas?" suggesting wanting improvement, rather than just analysis.
Sounded like you were pixel peeping, wanting perfection, not yet possible with automation. Manual for the win!

 

[TnDronePilot]

Your original post certainly did not make that clear.
You asked, "Any ideas?" suggesting wanting improvement, rather than just analysis.
Sounded like you were pixel peeping, wanting perfection, not yet possible with automation. Manual for the wiI was simply asking for any ideas into why the drone drifts outwards when in a POI move and no, I'm

Your original post certainly did not make that clear.
You asked, "Any ideas?" suggesting wanting improvement, rather than just analysis.
Sounded like you were pixel peeping, wanting perfection, not yet possible with automation. Manual for the win!

You know what assumptions are like right? Questioning why the drone drifts outwards has nothing to do with pixel peeping or looking for perfection. I wasn't asking for advice on how to make my videos better but rather if anyone had any ideas why the circumference expands during a POI.

 

[GadgetGuy]

You know what assumptions are like right? Questioning why the drone drifts outwards has nothing to do with pixel peeping or looking for perfection. I wasn't asking for advice on how to make my videos better but rather if anyone had any ideas why the circumference expands during a POI.

Your questions were answered clearly. Just because you didn't like some of the answers you received, due to your own lack of clarity, is no reason to insult the responders. Try communicating better next time.