active track worked absolutely awful for me. why?

[Sitka78]

OK I am uploading some of my active track attempts to youtube now. I will post back in the AM with links. I have to get to bed. The footage actually came out pretty disappointing. Me and the sled have way to much shadow for such a bright day with good sunlight on us. Compared to some footage I shot in the summer over that same lake this footage sucks (at least on this 4k tv im watching it on). The footage I took back then was jaw dropping it looked so good.

The sun was very high in the sky but I was heading towards it. Maybe if I was riding away from the sun with the sun behind the drone it would have helped. idk. maybe i needed filters idk. perhaps some tips once you see videos. Here is a still from the drone's video just to get you something for now. I can tell you that even if I had the drone facing 90 degrees to the left of the direction of this image, active track still failed. It failed in the video I am uploading as I was driving away from the sun as well.

View attachment 65657

My guess would be that possibly with all the brightness, you may have better success with more altitude. Sun, white background, possibly no ND filter, altitude may be your friend.

 

[Harro]

Shooting into the sun definitely does not help. Keep it back of you or too the side. Maybe (just a thought) all the tracks confused the tracking software.

 

[gerard143]

heres a clip of tons of attempts at active track. Looks like the first two tries might have been spotlight mode but when I got a little bit away it dropped out. The rest were trying trace and profile from the side. Just didn't work worth a ****. I stop many times, bring the drone closer back to myself... select myself again and fails over and over.






here's a clip.... video starts about midway thru with this link. For a little bit you can see active track actually working but then it lost me.

 

[Sitka78]

heres a clip of tons of attempts at active track. Looks like the first two tries might have been spotlight mode but when I got a little bit away it dropped out. The rest were trying trace and profile from the side. Just didn't work worth a ****. I stop many times, bring the drone closer back to myself... select myself again and fails over and over.






here's a clip.... video starts about midway thru with this link. For a little bit you can see active track actually working but then it lost me.

Were you carrying the remote, and have the antennas in the right attitude available to signal the drone? Also, how was your ground speed vs the speed of your drone? Just trying to figure out if you lost signal or not.

 

[gerard143]

Remote was mounted to sled. And yes proper orientation. Signal was not lost. Ground speed was anywhere from 0-20mph

 

[Sitka78]

Remote was mounted to sled. And yes proper orientation. Signal was not lost. Ground speed was anywhere from 0-20mph

Looks like you might consider downloading your flight log here for the experts to look at. There are several folks here that should be able to tell what was going on. Sorry for all of your problems, but, this is a great place to get answers.

 

[kanaloa6779]

I think ND Filters will help. Used it a lot for tracking people skiing-always worked fine.

 

[kanaloa6779]

If you are going with PolarPro I would suggest a custom set with ND 8 / 16 / 32.
I bought the set including 4/8/16 and never used the ND 4 up to now.

 

[Roc Rat]

I simply no longer trust Active Track. It has worked for short periods then stopped.. One time it went rogue but fortunately I was able to regain control and land it.

 

[kanaloa6779]

One more thing that worked out for me:
I mainly use 4K HQ - but if you are quite close to your subject it helps to use FOV. So if there is a quick change of direction the object does not run out of sight that easy.

 

[FormerMember07122019_2]

I wonder if it is a speed and size of motion issue. The 2 times I tried it and after seeing the same thing on a bunch of videos I noticed that there is a significant lag in the drone moving after the subject moves. It’s like it has to think (actually process) for a moment before it can react. It also seems to be a function of the percentage of the object in the frame. I assume it analyzes individual or groups of pixels to determine pattern matching of the selected object. The more pixels of the target the easier to track.

My conclusion is that if the object moves too fast or is too small in the frame, tracking is lost. Both at the same time is even worse. In addition, I’m sure lighting and contrast make it even worse.

Mostly a theory at this point. I might try to test it somewhat scientifically when I get the opportunity. I will be very interested if anyone has already done this or plans to in the future.

 

[FoxhallGH]

Post your logs for threse flights, because I'm very suspicious that this problem is caused by that big yellow globe in the sky. There have been many reports of Mavic's being stopped by the forward sensors 'seeing' the Sun as an obstacle. The logs will identify if the Mavic is going into collision-avoid!

 

[gerard143]

I posted videos. Never once does it alert of collision avoidance. Don’t think that’s it. Just think the tracking sucks and it kept losing me.

 

[FoxhallGH]

I posted videos. Never once does it alert of collision avoidance. Don’t think that’s it. Just think the tracking sucks and it kept losing me.

We need you to post the flight logs - not the videos.

 

[gerard143]

Lol. So what your saying is collision avoidance could be activating yet not showing the alert on the screen. Didn’t know that was possible.

 

[FoxhallGH]

Lol. So what your saying is collision avoidance could be activating yet not showing the alert on the screen. Didn’t know that was possible.

I know it sounds a bit crazy - and I'm not 100% on this, but in the face of nothing else being obvious, an examination of the logs might show us what's happening ...

 

[iowahill]

By posting the log file of the flight you may find that there is more certainty as to what was going on.

 

[gerard143]

Agreed. Ok. I gotta figure out how to extract them and post.

 

[FoxhallGH]

Agreed. Ok. I gotta figure out how to extract them and post.

Have a look at this site ... DJI Flight Log Viewer | Phantom Help
The files you need will be in a directory in the mobile device you are using with your RC controller. If you can find the .txt file and feed it through this process, you should get a .csv spreadsheet which shows all the parameters. When can then get @Meta4 and @sar104 to comment on your flight logs to figure out if there are any clues as to why Active Track wouldn't hold ...

 

[gerard143]

ok. I am using the smart controller not a mobile but i will try to figure it out.