Is GNSS receiver L1/L2 required on the UAV for RTK-PPK techniques?

March 6, 2019

The use of UAV to assist in the Topography activity grows every year, and the aerial photogrammetry technique is widely used to make geo-referenced maps with centimeter accuracy. But do you know the difference between accuracy and precision in the science of cartography? There are concepts that are best understood with image. You will understand the difference between precision and accuracy observed in this figure:


Figure 1 - Accuracy and Precision *

Direct Georeferencing (DG) using photogrammetry software with georeferenced photos via GNSS receiver of low precision and low accuracy (error around 3 meters) and photos superimposed in 80%, it is possible to make a georeferenced map with centimeter precision, but not it will be possible to make a georeferenced map with centimeter accuracy. The map will be offset with respect to the actual position on the planet and this error will be proportional to the error of the GNSS measurements used to geo-reference the photos, ie if the GNSS RMS error in the horizontal coordinate is 2 meters and in vertical of 4 meters the map will be with the same error. Precision will be best the lower the Ground Sample Distance (GSD) and the greater amount of pixels superimposed for map processing.

In the DG technique it is necessary to estimate the position of the UAV at the time of acquisition of the photo independent of the GNSS Rover receiver frequency and the flight speed. Current DG systems use PPK or RTK techniques and perform algorithms for more accurate position calculation through data generated by the IMU - Inertial Navigation System. In general, the IMU normally operates at the frequency of 100 Hz, so it provides data every 10 ms to calculate the estimated Drone position. These GD systems also require additional software, proprietary boards that integrate the GNSS Rover L1/L2 receiver into the IMU, and the cost can exceed the cost of the UAV solution.

To obtain a georeferenced map with high accuracy via Direct Georeferencing (DG) PPK, a high level of accuracy and precision is always required to correct the position of the camera/sensor at the moment of the photo exposure, so that the RMS error is lower at 10 cm. The PPK-RTK techniques is not enough. If you want to know more about this subject see: UAV and Direct Georeferencing (DG) solution x RTK. 

Experiments using L1 and L1/L2 GNSS receivers for Rover RTK

Now the question is, do I need GNSS on the UAV that operates on the L1 and L2 frequencies to increase the accuracy and precision of the position of each photo? Based on the result of technical experiments, in my opinion the answer is no. The L1 / L2 GNSS receiver is much more expensive than the L1 GNSS receiver and does not offer any advantages that justify the investment for the GNSS Rover RTK receiver. With new software it is possible to perform similarly on both types of receivers. Rather than explaining this by theory, which is relatively complex and massive, it is showing the practical results that compare the two technologies. The figures below show the results of the position of the different GNSS Rover RTK receivers, using the same antenna, installed on the car that runs through a city with trees and buildings. When the RTK / PPK solution exhibits centimeter accuracy, it means that it is in FIX mode:


Figure 2 - u-blox NEO M8T L1 frequency with RTK via open source software RTKLIB. FIX Rate= 94,4% of the time **



Figure 3 - GNSS Receiver ComNav using the L1 and L2 frequencies with internal RTK software, FIX Rate = 94.9% of the time **


Figure 4 - GNSS Receiver ComNav using only the L1 frequency with RTK software through RTKLIB, FIX Rate = 98.9% of the time **

In the UAV application, the GNSS Rover receiver antenna is above the trees and buildings and therefore has better reception of the satellites, so the total time with the RTK FIX system would in this case be very close to 100%.

Using appropriate RTK / PPK software, the technical differences between the L1/L2 GNSS rover receiver with a GNSS L1 u-blox Rover receiver are very small. The advantage of the L1/L2 GNSS Rover Receiver is that it reaches the first FIX faster than the GNSS L1 receiver, however, there are RTK algorithms that allow you to reach the FIX in less than 5 minutes with the UAV on the ground before taking off, this setup time is most often insignificant in UAV mapping applications.

Does using the GNSS Base receiver only with L1 frequency have any advantages for RTK system?

Another question you should be asking. The GNSS base receiver for RTK/PPK solution which is the GNSS receiver reference for precision and position accuracy of the GNSS Rover receiver, can be of the L1 only type? The answer is yes, but the application is very restricted. The GNSS L1 Base receiver achieves centimeter accuracy only after 20 hours of continuously receiving satellite data and executing the Precise Point Positioning (PPP) technique, and only then will GNSS Base L1 be ready to be used in conjunction with the Receiver GNSS Rover RTK. The GNSS Base L1 / L2 receiver needs only 4 hours on performing the PPP technique to achieve centimeter precision and accuracy. So I recommend using the base GNSS receiver that uses at least the L1 and L2 frequencies to use in conjunction with the UAV GNSS Rover RTK L1 receiver.

Is RTK / PPK solution sufficient to make georeferenced maps with centimeter accuracy?

Using RTK / PPK solution with receiver that guarantees FIX for more than 90% of flight time, with horizontal and vertical RMS error less than 10 cm, it is guaranteed that it can do Direct Georeference (DG) and obtain georeferenced maps with centimeter accuracy ? The answer is no. That's not enough. If you want to know the answer on that subject. See my other article the solution: UAV and Direct Georeferencing (DG) solution x RTK.


 Download the overview brochure of DG RTK-PPK

João Marcelo Corrêa                      Juliano Grigulo

CEO at Novarum Sky                     CTO at Novarum Sky


* Figure from article: Case Study: Accuracy vs Precision?

** Graphs from the article: Post-processing ComNav receiver data with RTKLIB – a more in-depth look