Improving holding position with ground reference

[skadee]

How high can you fly and still use the downward-facing cameras for position hold?

I'm trying to make a time-lapse of the sunset, and the position of my MP varies by about 15 feet over the course of a flight, and that amount is the difference between landing in my learn and landing on my roof upon RTH. More importantly, the video shows the ground moving around a lot, as I'm flying just above the treetops.

If I place 3 lights on the ground in a triangular pattern, and flew directly above them, would it help the MP to maintain a precise position? Or is are the bottom cameras just for RTH?

 

[Simmo]

Vision System
Vision System Forward Vision System
Downward Vision System
Obstacle Sensory Range Precision measurement range: 2 ft (0.7 m) to 49 ft (15 m) Detectable range: 49 ft (15 m) to 98 ft (30 m)
Operating Environment Surface with clear pattern and adequate lighting (lux > 15)
Velocity Range ≤22.4 mph (36 kph) at 6.6 ft (2 m) above ground
Altitude Range 1 - 43 feet (0.3 - 13 m)
Operating Range 1 - 43 feet (0.3 - 13 m)

 

[Simmo]

I would say set a waypoint up to fly to each time, and yeah illuminate the area under the RPA so the vision sensors have something to look at. Becaue they wont work in low light.

 

[skadee]

Interesting. I was at 55ft, so beyond the 43ft, but might be able to launch higher to get a lower measured altitude.

I'm not sure that the waypoint would help, as the home point it recorded was never changed by me, and when RTH was initiated, the MP was about to land on my roof. If it thinks that it was over the RTH point the entire time, then it would also believe that it was holding the GPS waypoint the entire time.

 

[BossBob]

If Smart RTH is activated within 5m radius of the recorded home point the Mavic will just land where it is. It is just working as designed.

 

[skadee]

Well, it wasn't doing that!

Before initiating RTH, I positioned over its true takeoff point (clear of trees and roofs) but it swiftly diverted away from that point and started heading down toward the roof. A quick switch to Sport saved it.

 

[BossBob]

You should also know that your position (or the drones position) will wander slightly over time at the accuracy we are expecting. It can be as much as 3m, a good 50% of the accuracy you are expecting. The super accurate RTH only works if you have this requirement set as you take off and the Mavic will go through a routine which includes taking a picture of the takeoff point using the downward cameras set into the heat sink. It uses this to pinpoint the landing point.

 

[Simmo]

I meant set a waypoint for where you want to take your photo each time.
One thing puzzles me, if you flew to right over your take off point, why on earth would you hit RTH??

 

[daisy-girl]

How high can you fly and still use the downward-facing cameras for position hold?

I'm trying to make a time-lapse of the sunset, and the position of my MP varies by about 15 feet over the course of a flight, and that amount is the difference between landing in my learn and landing on my roof upon RTH. More importantly, the video shows the ground moving around a lot, as I'm flying just above the treetops.

If I place 3 lights on the ground in a triangular pattern, and flew directly above them, would it help the MP to maintain a precise position? Or is are the bottom cameras just for RTH?

The VPS of the MP uses sonar.

Trees absorb ultrasound.

 

[Kilrah]

Well, it wasn't doing that!

Before initiating RTH, I positioned over its true takeoff point (clear of trees and roofs) but it swiftly diverted away from that point and started heading down toward the roof.

You're just expecting too much precision. A 15-30ft drift is to be expected.

 

[skadee]

I meant set a waypoint for where you want to take your photo each time.
One thing puzzles me, if you flew to right over your take off point, why on earth would you hit RTH??

Mostly to see what would happen.

Since I was observing the drifting, I corrected its position to be directly overhead of the home point, wanted to see if it would go back to what it thought was the home point, or what truly was the home point. I was expecting it to return to be vertically aligned with where it had been last hovering, so one finger was at the ready on the “Sport” button, while the other tapped RTH.

 

[skadee]

You're just expecting too much precision. A 15-30ft drift is to be expected.

With some of the time-lapse videos I’ve seen, it looked as though other users were having better position hold than I experienced. I will try again when there is no wind at all, but mine floated around pretty good. Granted, I was just above the treetops, so the variation was going to be more noticeable, but I figured some folks would have done some testing by now to “dial it in”.

 

[Kilrah]

There's your problem, treetops move in the wind. You would then NOT want to have the VPS enabled as it's likely to try and follow them...