Drifting hyperlapse

[theseriouspdx]

Hi friends!

Thanks for any info in advance. Shot several hyperlapses last night on my Mavic 3 pro, without issue, however the final one of the night came out all wonky. Most of the hyperlapses I did, I used course lock and selected a subject, but this one I did a waypoint mission, and the camera was moving all over the place. Wondering if anyone knows why the drone would act this way in a waypoint mission, but not in other hyperlapse modes, and how I might avoid similar results in the future.

Other info- no wind, all the calibrations had been completed prior to flying earlier in the evening.

Video link
Flight log link

Thanks in advance!

 

[GadgetGuy]

Cool hyperlapse! I doubt the camera was pointed anywhere other than straight ahead. Looks like the drone itself might be curving through your waypoints, which is how DJI waypoints are designed. If you set two additional waypoints closely around each of your current waypoints, that should minimize the lateral movement of the drone around your original waypoints. Give it try and please report back.

 

[theseriouspdx]

Cool hyperlapse! I doubt the camera was pointed anywhere other than straight ahead. Looks like the drone itself might be curving through your waypoints, which is how DJI waypoints are designed. If you set two additional waypoints closely around each of your current waypoints, that should minimize the lateral movement of the drone around your original waypoints. Give it try and please report back.

There were three total waypoints for this flight, one at the beginning, one at the at the top of the large building and the last one at the end. what I don't care for is the drifting back and forth that that it main building (the subject) does in the frame. I'm trying to figure out why it's doing this so when I go back and reshoot it, it doesn't happen again.

 

[GadgetGuy]

There were three total waypoints for this flight, one at the beginning, one at the at the top of the large building and the last one at the end. what I don't care for is the drifting back and forth that that it main building (the subject) does in the frame. I'm trying to figure out why it's doing this so when I go back and reshoot it, it doesn't happen again.

You need more waypoints closely around the waypoint of your subject, just as I suggested above. Try it.

 

[MS Coast]

@GadgetGuy, This doesn't look like an issue with smoothing between waypoints, does it?

 

[GadgetGuy]

@GadgetGuy, This doesn't look like an issue with smoothing between waypoints, does it?

View attachment 173652

No, is that from his hyperlapse? If he only had 3 waypoints, must be something else going on. Maybe the gimbal is not holding position as it is buffeted by the wind from forward flight? Is the drone drifting back and forth sideways, or is the gimbal moving laterally? Hmmm…

Maybe one of our resident flight log analysts can chime in.
This is beyond my my grade! LOL!

 

[MS Coast]

No, is that from his hyperlapse? If he only had 3 waypoints, must be something else going on. Maybe the gimbal is not holding position as it is buffeted by the wind from forward flight? Is the drone drifting back and forth sideways, or is the gimbal moving laterally? Hmmm…

Maybe one of our resident flight log analysts can chime in.
This is beyond my my grade! LOL!

It's from the outbound leg with the drone mode listed as "Time-Lapse." That's the drone's track, not related to the gimbal. Wind buffeting doesn't seem to be the issue; on the return leg in nearly the opposite direction, the flight path is straight.

I'll look forward to more expert comments. I've reached, or exceeded, the limits of my skill with flight log analysis.

 

[GadgetGuy]

It's from the outbound leg with the drone mode listed as "Time-Lapse." That's the drone's track, not related to the gimbal. Wind buffeting doesn't seem to be the issue; on the return leg in nearly the opposite direction, the flight path is straight.

I'll look forward to more expert comments. I've reached, or exceeded, the limits of my skill with flight log analysis.

That does look very odd for the flight log path!

 

[Robert Prior]

That does look very odd for the flight log path!

Could it be related to the low light? I haven't flown at night, but I heard somewhere that the drones behave a bit differently than they do during the day because some systems aren't active (or are only intermittently active).

 

[mobilehomer]

Try that same flight with OA set to Off. It appears to be avoiding things.

 

[MS Coast]

Try that same flight with OA set to Off. It appears to be avoiding things.

The altitude during that time lapse portion increases from 168 to 382 feet. What might it be dodging?

 

[MS Coast]

Could it be related to the low light? I haven't flown at night, but I heard somewhere that the drones behave a bit differently than they do during the day because some systems aren't active (or are only intermittently active).

I've never see a flight path with a wobble like that in my night flights. Anyone else?

 

[mobilehomer]

The altitude during that time lapse portion increases from 168 to 382 feet. What might it be dodging?

The Sun.

 

[MS Coast]

The Sun.

The video clip suggests that the flight was well after sundown.

Video link

 

[MS Coast]

Any thoughts on the flight path wobble - @slup, @Meta4, @msinger, ...?

 

[Robert Prior]

The Sun.

I had my Mavic 2 stop suddenly while flying over the icefjord in Ilulissat. At the time I thought there were birds around and hurriedly returned, but I discovered later that the obstacle avoidance would get confused when flying towards bright objects in high-contrast situations and detect obstacles that aren't there.

I didn't see an error message, but I was mostly watching the drone so might have missed it. I would have expected that to show up in the log, but I don't know if it does.

 

[slup]

Any thoughts on the flight path wobble - @slup, @Meta4, @msinger, ...?

Think a couple factors contributed to this...

- The flight direction caused the drone to have a very near full tailwind
- The heading speed was very low
- That the flight mode was waypoint, that will make the drone try to hold a straight line path between the 2 waypoints
- And maybe that the flight was conducted in the dark only relying on the GPS accuracy (no VPS)

The tailwind+the low heading speed required that the drone used a very shallow tilt angle, in average approx 6 degrees. This made the drone more unstable tilt direction wise than if the tilt angle had been deeper. The tilt direction constantly moved between -49 to -15 degrees from north (meaning that the drone mostly was tilted backwards for braking to hold that low heading speed... the flight direction was nearly directly into south, approx. +175 degrees).

Imagine balancing a plate on a stick when you slowly move the stick pretty much vertical in a constant direction... & compare that with the same balancing act when you lean the stick & start to run. It's easier to keep the plate up in a more stable fashion in the later scenario.

Then the mix between a waypoint mode & only GPS accuracy could have contributed as the drone could have been pushed further off track by the tailwind before the GPS data could confirm & tell the FC to counteract... & by that the adjustments was more forceful.

In the end, the drifting & the constantly changing tilt direction affected the final hyperlapse recording because the camera couldn't be kept constantly in the same direction.

I'm pretty sure that the track had been straighter if the drone had flown into the wind.

Below a shorter bit of the flight path with wind direction & speed & further below the interesting telemetry from it.

(Click on the pics to make them larger)

 

[theseriouspdx]

So I got into experimental mode went and flew the same path again yesterday during the day, and then again at night, and both times I tried the same waypoints, and another one with two more waypoints near the start, and two more at the end. What seems to be happening is the vps senses the first building, and then goes wonky during the night flights, but during the day flights it maintains a straight path. All of these flights were conducted in nearly calm conditions. So the trick to getting the night shot to look right was to move the starting waypoint so the drone didn't fly over the first building.

 

[slup]

... What seems to be happening is the vps senses the first building, and then goes wonky during the night flights, but during the day flights it maintains a straight path.

That sounds more coincidental... if anything, the day flight would be more affected.

In order for the VPS to make the drone change flight path automatically... avoid something, the object needs to be very near, like half a meter below the drone. The other reason can be that the VPS is locking onto something below it to assist with positional hold during hovering, whilst the GPS quality is bad... & that it locked onto, moves. This can happen over flowing water for instance. Both these cases is more likely in good lightening as that is crucial to make the VPS work.

Furthermore... if the VPS have a reading in the beginning of a waypoint flight that later disappears, that initial reading doesn't affect the rest of the flight.

Pretty sure the reason is something else...