Direct georeferencing (or geopositioning) is a technology that captures the GPS location and IMU orientation of individual images. This is beneficial because knowing the GPS and IMU data for each individual image greatly improves SfM processing time. Also, if the GCP and IMU information is accurate enough, it can be used to accurately scale the SfM-produced information.
The commercially available sUAS today typically incorporate a rough solution to direct georeferencing. This is usually done by using the aircraft’s internal GPS to “tag” individual images with location information. Rarely do commercially available sUAS have IMU information tagged to imagery. This lack of orientation information is resolved in the SfM processing where the software can internally interpret the orientation. This standard GPS tagging is simple and low cost, but it comes at the compromise of low accuracy, which is typically greater than a meter in X&Y axes and greater than 3 meters in the Z axis. To improve the accuracy to at least 10cm in all axes, the GPS data is corrected through a Differential GPS (DGPS) method.
Real Time Kinematic (RTK) and Post Processing Kinematic (PPK) are the two DGPS methods used to correct GPS locations. Both work under the same core concept that involves two GPS receivers. One receiver, referred to as the base, remains stationary throughout the collection period. Because this receiver is stationary, it averages its location throughout the collection period in order to create a highly accurate position. The second receiver, referred to as the rover, moves throughout the area of interest and collected the desired GPS locations. So long as the base and rover are collecting information at the same time and are co-located (within 10km of each other), a high accuracy correction can be resolved. Now, the difference between RTK and PPK is in how the corrections are implemented.
As the name details, the RTK method applies corrections in real time. This requires a wireless communications link between the base and rover receivers. This method is best for use-cases, such as vehicle navigation for precision agriculture, that demand highly accurate GPS information in real time. For sUAS data collection purposes, real time corrections are not necessary needed. Data for reconstructing a scene is typically processed after the collection has completed. This is why the PPK is ideal for sUAS direct georeferencing. Just like RTK, PPK uses a base receiver to correct the GPS information of the rover receiver. Unlike RTK, there is no real time communication between the base and rover, the GPS corrections are done in post processing. This is advantageous for two reasons. First, this eliminates the need for a data link between the base and rover receivers. This conserves precious weight on the sUAS and reduces the level of complexity for the sUAS operation. The second benefit of PPK corrections is that the GPS information can be further refined by using base station reference networks.
Base station reference networks use DGPS methods to a greater scale. They consist of multiple base stations, spread throughout a geographic area, that are constantly averaging their location and broadcast those corrections to the internet. Examples of these networks are the National Ocean and Atmospheric Administration’s publicly available Continuously Operating Reference Station (CORS) network ( https://www.ngs.noaa.gov/CORS/ ), and there are paid subscription networks such as Leica Geosystem’s SmartNet ( https://www.smartnetna.com/ ). The GPS information from your local base receiver can be further corrected by these base station reference networks, thereby improving the accuracy when compared to RTK alone.
There are multiple manufacturers of PPK systems (Micro Aerial Projects V-map http://www.microaerialprojects.com/v-map-system/ , AirGon Loki http://www.airgon.com/loki.html , EMLID Reach https://emlid.com/reach/ ) and many chose to create custom solutions for their sUAS data collection missions. The two core differences between these systems are the satellite signals they can receive and the method by which they record an image capture event. Most of these systems record both L1 and L2 GPS signals, with less expensive ones only receiving L1, and more expensive systems receiving signals from the other constellations or global navigation satellite systems (GNSS) such as GLONASS, BeiDou, and Galileo. The benefit to more signals is redundancy and a more trusted fix on the location. The other difference between PPK systems is found in how they record an image capture event. Common systems use the hot shoe port of a consumer camera to record this information. When an image is captured a pulse is simultaneously sent to the hot shoe to trigger a flash or external device. Instead of a flash, these PPK systems simply record the pulse from the hot shoe port to log the exact time of the image. The other method is to record the image capture time by tying into the storage media for the camera. These systems are particularly useful for systems that do not have a hot shoe.
Ultimately, directly georeferencing using PPK methodologies is useful because it eliminates the need for ground control points (GCP). Without direct georeferencing, an area would need to have an even spread of surveyed GCPs to have an accuracy of less than 10cm. This same accuracy is possible with PPK direct georeferencing which eliminates the need to physically enter the area of interest. Please note, that in order to “prove” the accuracy for each data collection there is still a need for check points or some sort of known reference.
- Log in to post comments