It has been pretty well known that the Mavic 2 Enterprise Advanced's RTK module is for positioning only. The sUAS does not implement Time Sync with the RTK module to take the precise and accurate coordinates and geotag your images. I purchased an RTK module for my M2EA after I found one for a good price. It will allow for repeatable inspections and similar workflows.
I was curious though on how mapping outputs would come out in Agisoft Metashape so I ran some tests.
Tests
The same flight path would be run for three different tests.
All images would be oblique at 35 degrees up from nadir.
5 Control Points were collected with Emlid RS2 for 2 minutes each in FIX
Test 1 - No RTK Module 5 Check Points
Test 2 - RTK module receivng corrections from RTK NTRIP Service with a 20 km baseline
Test 3- RTK module receiving corrections from Emlid RS2 using Emlid Caster setup on site
Unfortunately I had to conduct this in a hurry after work and the sun was going down. This had the ISO rise and the aperture fully open. M2EA does not allow for any camera settings.
RMSE
Test 1
X 7.2873 feet, Y 14.1348 feet, Z 33.1167 feet, Total 36.7371 feet
Notes: Must worse in Z than other maps I have made with this drone. With a flight speed of 11 mph and failing light the images were a little blurry. The results were not too surprising though since I have mapped with this sUAS before and had similar X and Y.
Test 2
X 7.33848 feet, Y 4.74064 feet, Z 0.398215 feet, Total 8.74561 feet
Notes: I connected to an NTRIP service and I had a baseline to the CORS/Base of about 20 km. With the RTK module installed and a FIX status, the drone collected images that again had some blur due to lighting and the 11 mph speed. I had a pleasant surprise though upon import into Pix4D. The accuracy values were populated! This gave me some hope that maybe the images were going to be geotagged with coordinates that could be of better value than the images obtained without the RTK module and the empty accuracy values. These values ranged around 0.09 ft for X Latitude and Longitude and went up to 0.22 for altitude. But I came back to reality after processing as I saw the poor RMSE values. Z though was pretty strong! Could there be something to the RTK Module adding some kind of accuracy for mapping?
Test 3
X 6.87029 feet, Y 4.95707 feet, Z 2.0558 feet, Total 8.71778 feet
Notes: I used the RS2 to collect a point for 10 minutes receiving corrections from the NTRIP service. I then used this point to setup the RS2 to send corrections the the drone. Again the M2EA with RTK module and a FIX status collected the same images and again they had some blur. I will retest this again with proper lighting. The result are again poor.
Again, while it has been stated and known that the RTK module is just for positioning, I do like to run tests just to see what happens. Overall this drone is a poor choice for mapping, and its not just the RTK module and its non-mapping integration. The images in decent to bad lighting were just plain not good. Even Test 1 images were not exactly the best with the best lighting of all tests. Next the images are large. This quad buffered camera has large images that slow down Metashape's processing. Also the drone must be set to fly slower to keep motion blur at bay.
I guess this drone will keep its day job, public safety tasks and building inspections.
I was curious though on how mapping outputs would come out in Agisoft Metashape so I ran some tests.
Tests
The same flight path would be run for three different tests.
All images would be oblique at 35 degrees up from nadir.
5 Control Points were collected with Emlid RS2 for 2 minutes each in FIX
Test 1 - No RTK Module 5 Check Points
Test 2 - RTK module receivng corrections from RTK NTRIP Service with a 20 km baseline
Test 3- RTK module receiving corrections from Emlid RS2 using Emlid Caster setup on site
Unfortunately I had to conduct this in a hurry after work and the sun was going down. This had the ISO rise and the aperture fully open. M2EA does not allow for any camera settings.
RMSE
Test 1
X 7.2873 feet, Y 14.1348 feet, Z 33.1167 feet, Total 36.7371 feet
Notes: Must worse in Z than other maps I have made with this drone. With a flight speed of 11 mph and failing light the images were a little blurry. The results were not too surprising though since I have mapped with this sUAS before and had similar X and Y.
Test 2
X 7.33848 feet, Y 4.74064 feet, Z 0.398215 feet, Total 8.74561 feet
Notes: I connected to an NTRIP service and I had a baseline to the CORS/Base of about 20 km. With the RTK module installed and a FIX status, the drone collected images that again had some blur due to lighting and the 11 mph speed. I had a pleasant surprise though upon import into Pix4D. The accuracy values were populated! This gave me some hope that maybe the images were going to be geotagged with coordinates that could be of better value than the images obtained without the RTK module and the empty accuracy values. These values ranged around 0.09 ft for X Latitude and Longitude and went up to 0.22 for altitude. But I came back to reality after processing as I saw the poor RMSE values. Z though was pretty strong! Could there be something to the RTK Module adding some kind of accuracy for mapping?
Test 3
X 6.87029 feet, Y 4.95707 feet, Z 2.0558 feet, Total 8.71778 feet
Notes: I used the RS2 to collect a point for 10 minutes receiving corrections from the NTRIP service. I then used this point to setup the RS2 to send corrections the the drone. Again the M2EA with RTK module and a FIX status collected the same images and again they had some blur. I will retest this again with proper lighting. The result are again poor.
Again, while it has been stated and known that the RTK module is just for positioning, I do like to run tests just to see what happens. Overall this drone is a poor choice for mapping, and its not just the RTK module and its non-mapping integration. The images in decent to bad lighting were just plain not good. Even Test 1 images were not exactly the best with the best lighting of all tests. Next the images are large. This quad buffered camera has large images that slow down Metashape's processing. Also the drone must be set to fly slower to keep motion blur at bay.
I guess this drone will keep its day job, public safety tasks and building inspections.