Hi

In DJI's web I can read:
Accuracy of the L1 system at 50m flight height (RMS) 10 cm horizontal and 5 cm vertical.
I undestand that is the L1 IMU's precision  (It's L1's topographical and cartographic precison)


But I also can read:
Range Accuracy (RMS 1) 3 cm to 100m

Somebody who can explain me what it means:.. ???
What is the diference?


And in another place say:
Relative accuracy:  Horizontal R.P. 2 x GSD P.R. Vertical     3 x GSD
And this? OMG?
What we are talking about

I'm going crazy




Thank you

Your subject line is misleading, as your post is specific to the L1 on the M300 RTK.

The first value is overall accuracy, the total propagated uncertainty of the system as a whole.

The second is the range accuracy of the laser itself. That's why it is smaller, it a a component of the overall accuracy.

I have no idea about the 3rd and do not see it in either the L1 or M300 RTK specs. Got a link?

Here's my experience after almost 2 years and 20+ surveys with the L1, in different situation (from 300+ha moutain survey, from corridor road mapping).

The overall absolute accuracy without GCP is a bit unsufficient, specially if you make the "point optimization" in DJI Terra and then adding GCP in Terrascan. I measured good accuracy in XY (even in worse condition never more than 10cm), but on Z axis i measured differences that can reach easily +-30cm. The worse results was from a 110m AGL flight, i had 53cm differences in height from GCP. That's why i bought TerraMatch for alignment, but i haven't tried it yet on new works, but i see it works way better for stip alignment.
With low level scanning the accuracy is way better, but we're far from the 5cm declared. I see it also depends on scan pattern, with Repetitive scan way more precise than Non-Repetitive.
In non-repetitive scan, the points far from the center of the scanner area are so imprecise, in terramatch there's an option to not use that points for alignment (you can choose the angle from the center of the frame the software use for alignment, i use +-20° in front-backward).

About the laser precision, even here the 2cm (or 3 at 100m) is the declared but not the real accuracy. On flat surface (for example roads), the "spread" of points (in a 10x10cm sample, the difference between the highest and the lowest points) can reach +15cm with non-repetitive scan at 100m, about 6-8cm at 50m. With repetitive pattern we had better results, with about 8cm at 100m, 3-4cm at 50m. After this raw data, with terrascan you can lower the "spread", but losing data accuracy on edges (smoothing and removing noise).

For my experience, L1 system are great for forestry mapping or survey don't need cm reliability.
But it's compulsory, in order to increase Z accuracy, to have GCP in every project. For "small" area, i use to map with L1 and P1, and use P1 for orthomosaic and for GCP. P1 are super great in accuracy, i've never experience errors more than 5cm in project without GCP, usually with 80m AGL flight 2-3cm XYZ.

patiam Posted at 8-17 06:26
Your subject line is misleading, as your post is specific to the L1 on the M300 RTK.

The first value is overall accuracy, the total propagated uncertainty of the system as a whole.

HI, sorry for title post, but I don't know how to change it.

Thank you for you reply

And I have to look for the link of the 3rd cuestion, i don't  see it...

Panph Posted at 8-17 23:24
Here's my experience after almost 2 years and 20+ surveys with the L1, in different situation (from 300+ha moutain survey, from corridor road mapping).

The overall absolute accuracy without GCP is a bit unsufficient, specially if you make the "point optimization" in DJI Terra and then adding GCP in Terrascan. I measured good accuracy in XY (even in worse condition never more than 10cm), but on Z axis i measured differences that can reach easily +-30cm. The worse results was from a 110m AGL flight, i had 53cm differences in height from GCP. That's why i bought TerraMatch for alignment, but i haven't tried it yet on new works, but i see it works way better for stip alignment.

Thank you very much for you reply

I understand you always use Terrascan to reprocess the terra's clouds.  An always use GPC to get a better accuary.
Even when do you work in forest proyects ??

One more thing , (although I think I'll start a new post)   Do you have problems with the overlapping pf the diferents passes of th L1?   Are they at diferente heights?

Thank you again

To @Panph's points, I was merely defining what the values represent, not giving any sort of confirmation of the veracity of the values DJI publishes.