Why use PPK with your drone (not RTK)
Note: This article was last updated on 10/16/2018
UAS vendors targeting markets from commercial survey to agriculture are fielding systems with real-time kinematic GNSS (RTK) capability. In principle, RTK promises accuracies at the 1-3cm level. The main purpose is to minimize or eliminate the need for ground control points, thereby reducing cost. Altavian uses GNSS receivers upgradable to RTK operation, but we favor another approach for this level of accuracy: post-processed kinematic (PPK). There are a couple of reasons why:
RTK requires a GNSS base station equipped with a transmitter with a reliable link to a fairly dynamic moving platform.
The rover (on the UAS) itself requires a dedicated receiver for the corrections.
PPK vs RTK
These primary reasons carry some further implications for the cost of deployment, especially when considered against PPK. RTK operations not only require a stationary base station, but it must be located at a previously known control point. Provided the base station is deployed for long enough periods of time, this is not too much of a problem. The base station's precise location can be determined post-mission if no control points are already present if the raw observations are logged. In this case, a global shift of the aircraft’s trajectory must be done once the position of the base station is determined, taking away some of the benefits of a ‘real-time’ solution. Additionally, the trajectory obtained from using imprecise approximate base station coordinates can lead to divergent or inconsistent RTK solutions, again defeating the goals of an RTK-enabled mapping system.
PPK requires a base station as well. In many cases in the Eastern US at least, the public CORS network and private virtual base station network services may be dense enough to provide a base station reasonably close to your project. It's still likely you will need a base station of your own. This represents slightly less investment than an over-the-air link to the rover. Both RTK and PPK are susceptible to the possibility of loss-of-lock. In both, when the rover loses lock on enough satellites, a new integer ambiguity resolution procedure must be initiated. The advantage of PPK is that the search can proceed from previous and future data relative to that instant. RTK solutions cannot use data that has not yet been recorded. Additionally, forward and reverse solutions in PPK are optimally combined and give an estimate of a solution’s consistency. If you want to eliminate ground control points and you chose an RTK system, there is no external information for basing accuracy estimates.
Finally, it is worth noting that antennas light enough to be mounted on a small UAS are not geodetic-grade and are not likely to have been calibrated for phase-center variation (PCV), let alone the actual location of the phase center, though this is beginning to change somewhat. This means that you might get a reported solution accuracy of 1-2cm, but it could easily be very misleading. With a PPK solution, at least you can see if the forward and reverse solutions agree within certain bounds (and we acknowledge this is a very limited vote of confidence for any kinematic solution, but it’s better than nothing).
Ultimately, there is no replacement for real ground truth, especially if your data product must be certified to a specific level of accuracy. Strategies to minimize the requirements on GCPs can vary widely in their effectiveness, however, depending on your needs. If positional accuracies on the order of a decimeter are acceptable, real-time direct L-band or PPP corrections, both available through subscription services such as TerraStar-D are very attractive alternatives that require no base stations at all.