Elevation data wrong

Geo_Drone

LV_Forestry Posted at 5-20 04:10
At least we agree on that.

For the 2cm of accuracy, can you confirm that you are talking about absolute measurement on a 3D model or a Raster?

PS1: I forgot: 2-3 cm on generated 3D model. Applies to standard distances like left-right and up-down. For calibrated distances inside 3D model, like models you have that enters in angles to inside, you need a LIDAR, cannot be precise as you cannot control them enough. (example: a window measured will have precision....a decoration that is under roof and enter in angle into house you will have errors because of planarity - sorry for english, is not my native).

PS 2: Yes, we have measured also roads and distance between different obstacles to provide the map for them, the measurement in the model map was with deviations of + - 2 cm from reality.
The +- 5 cm were achieved in angled 2 sides areas, where we compensated with 2 GCP on both plane of inclining, but at that points you need a backed RTK for cm accuracy. As we had an order for tolerance +- 10cm, was not an issue.

LV_Forestry

Labroides Posted at 5-20 20:52
If you could give a try it would be really cool. Especially since you often fly close to sea level which makes the data even easier to verify.

Excellent, thank you, at least now we are sure on our own that this is not the right way to solve the problem. At least for the Mini3Pro.

Now we have to figure out an idea for the next step to help understanding this mess.
Any idea ?

LV_Forestry

Geo_Drone Posted at 5-20 23:23
PS1: I forgot: 2-3 cm on generated 3D model. Applies to standard distances like left-right and up-down. For calibrated distances inside 3D model, like models you have that enters in angles to inside, you need a LIDAR, cannot be precise as you cannot control them enough. (example: a window measured will have precision....a decoration that is under roof and enter in angle into house you will have errors because of planarity - sorry for english, is not my native).

PS 2: Yes, we have measured also roads and distance between different obstacles to provide the map for them, the measurement in the model map was with deviations of + - 2 cm from reality.

Thanks, it's clearer that way. Conclusion: mapping with consumer drone ok but GCP are essential.

Labroides

LV_Forestry Posted at 5-21 03:38
Excellent, thank you, at least now we are sure on our own that this is not the right way to solve the problem. At least for the Mini3Pro.

Now we have to figure out an idea for the next step to help understanding this mess.

Yes ... as I've explained before, DJI has a flawed algorithm that attempts to calculate Absolute Altitude from air pressure alone, but it assumes standard temp and pressure and makes no allowance for normal changing air pressure.

There is no hint that they use GPS at all.
If GPS was used, would you expect it to change drastically from day to day?

It should be no surprise at all that unnecessarily messing with the compass made no change at all.

Labroides

LV_Forestry Posted at 5-21 03:40
Thanks, it's clearer that way. Conclusion: mapping with consumer drone ok but GCP are essential.

It shouldn't be a surprise that mapping can be satisfactorily carried out with consumer drones.
Before RTH drones were available, the whole world used the Phantom 4 pro for professional mapping and photogrammetry.
But DJI decided that using consumer drones for professional work wasn't something they liked.
The prefer users to pay double for enterprise models to do what used to be possible with consumer models.