Practical Autel Evo II RTK accuracy test without GCPs

[mauricioranzan]

I did a mapping with my Evo II RTK to serve as a base map for a forensic investigation.
I flew in RTK mode with my Emlid RS02 base on site.
I collected some various points to use as GCPs.
But, I wanted to process the photos without GCPs to see how the result would be.
Then I loaded the orthomosaic in QGis, loaded the points and checked the discrepancy of the points on the targets in the orthomosaic.
I got a very good result (in my eyes).
The mean positional error was 5cm, with points from 1cm to 10cm.
I flew at 18km/h.

I accepted a very fair result.

What do colleagues have to say to me?

Attached is the Metashape report.

 

[hewell]

I never use Metashape, only pix4d but in 2D in my firts flight i got 2 cm precision with few point from ortophoto to GPS survey data.
One year using the drone I can't say than it's a 1 to 10 cm (rarely 15-20 cm) precision but is not the same in heigth, I had seen 5 to 50 cm precision, regular 20cm, what i correct with GCPs, so I used it just to link Drone GPS height to Survey GPS height data.

 

[hewell]

One more thing, I had noticed a precision degraded fliying at 6 m/s or more, not at 5m/s or less.

 

[mauricioranzan]

OK,
Então, entendemos que 5m/s ou 18km/h seria o máximo ideal.

Mais uma coisa, notei uma precisão degradada voando a 6 m/s ou mais, não a 5 m/s ou menos.

OK,
Então, entendemos que 5m/s ou 18km/h seria o máximo ideal.

 

[jaja6009]

First thing, although the report is in another language, it looks like you did not Optimize. If you did not Optimize you are losing a lot of accuracy right there.

Second thing, import your collected control points into Agisoft. You are causing a lot more work by trying to use QGIS to "check" your accuracy.
After importing your collected points (Agisoft calls them markers) assign them as Check Points.
You could also use just one as a GCP/Control Point and the rest as Check Points. This could help to further better your accuracy.

Start to learn how to use Gradual Selection. It can help to get rid of lower quality points, and can also help get better accuracy.

Overall you are off to a good start, but you have to optimize at least one time for best results.
With RTK receiving corrections you should have sub 6cm RMSE.