The new RTK drone from DJI was eagerly awaited in the UAV scene. We were given a test device and put it through its paces. We wanted to know exactly what advantages and accuracies we can expect from this technology, ultimately this device costs only a fraction of conventional RTK drones. As practitioners with high demands, we have repeatedly questioned our attitude towards high-speed UAVs with RTK accuracy and put them to the test. RTK in drones has not really convinced us to this day, as the problems of eccentricity when the aircraft tilts sharply were glaring.
Our test scenario
We used our test area directly in front of our office in Munich am Schatzbogen. Here we have distributed both cadastral fixed points of the survey management, height fixed points and a large field of highly accurate control points over 3 levels that are sufficiently overdetermined in multiple measurements, so have an accuracy window of less than 1 cm.
We flew this control point field with RTK equipment, as well as with the standard Phantom4pro (for comparison). Here we used the new flight modes of the new DJI-RTK App with 2D evaluation with photogrammetry mode, as well as the 3D mode. The normal mode photogrammetry mode is better for orthophotos, while the 3D mode is better suited for 3D models of buildings or excavations and mining sites. Unfortunately, there are still some modes missing in the app that you can find in flight planning apps today, which has no influence on the test result here.
We determined 50, 60 and 70 m as altitudes and as overlap we chose 80% in longitudinal and transversal. We chose 60 and 80 grad for the slanted images. We used 3Dsurvey as the evaluation software, because here we find higher standards in terms of surveying.
The control points were as mentioned repeatedly overdetermined and are with an accuracy of +/- 1 cm in the position and height before. They were measured with a Stonex S900 calibrated at Sapos control points and converted to metric coordinate systems and official NTv2 grids and the current height network 2016 using WebCADdy. The comparison of the accuracies, however, was done in order to avoid conversion losses in the ETRS89 system, ie the NativGPS coordinate system. For better clarity, we have converted the lists into the UTM coordinate system.
The basic accuracy of the PH4pro + RTK
For the RTK capability, the remote control is equipped with an LTE stick, which allows to communicate on the Internet with the so-called NTRIP services of surveying management. Every surveyor knows that from his GNSS receiver. A prerequisite is, of course, a Sapos or AXIO account to retrieve correction data in RTCM 3.0-3.2 format. The configuration is to be made according to a normal GNSS from other providers.
If the device is configured correctly, the PH4 can of course also be used as a measuring device. We did that to test the basic accuracy. The result is amazing. If we place the camera exactly over the control point (instrument height 4,5 cm), position the camera horizontally, we are able to approx. 1,5 cm and in height at approx. 2 cm deviation. Mind you, we have only exemplified on individual points. This is slightly more than the manufacturer indicates (1 cm position and 1,5 cm height). But for a real practice test and an 1 frequency device that is very good in our opinion. Also, the initialization time of about 10 seconds on average is excellent for a fixedsolution for this technology standard.
The accuracy in a stationary position has nothing to do with the accuracy at high speed and the associated inclinations and the resulting eccentricity of course. On the first day we have little wind and good weather conditions, but on the 2. Day strong wind had to announce. So strong slopes during the flight and different airspeeds. The drone has a fixed solution continuously and therefore a high level of accuracy. Altitude was 50 and 70 m for the photogrammetry solution and 60 m for the 3D solution. There were no warnings (except for strong wind at high altitudes (about 45 km / H), but this does not detract from the suitability of the PH4.) For the flight arrangement 2 options of the basic equipment of the DJIGo AppRTK were selected.One is the photogrammetry mode for a The best possible orthophoto and the 3D mode for the best possible 3D model: No special flight arrangements from the Flying Surveyorfundus were chosen because they are not available on the DJI-RTK app (and we did not want to install special apps on the demo device) and because of us wanted to have a fair comparison to other test.
Use with reference station
The problem still exists in areas with poor network reception, which can only be measured to a limited extent. Our preferred solution with MultiSIM cards doesn't work here, since you can call up the correction data with it, but you don't have free Internet access. And this is what the DJIpro+RTK needs. As DJI has another option on offer. A dedicated reference station with a 2-frequency as a base station. The price alone is a treat. With only 2100 € surcharge you get a technically perfect high-quality reference station that you can set up on a known point or on a point with internet reception. Thus, for a price of €7800*, you are also equipped for measuring tasks with RTK drones in “black holes” (areas without correction data reception).
We had gotten the base by the editorial deadline, but had no opportunity to test it. We again expect a significantly higher basic accuracy with respect to the reference point. We will inform you immediately when the test measurements are completed.
