Drive-by DTM. and Navigation at our university in cooperation

Transcription

1 Drive-by DTM GPS and GSM/GPRS Power Cost-Effective Terrain Modeling A data teletransmission system for quick and efficient creation of digital terrain models (DTMs) forms the backbone of experimental work in the transmission of differential GPS and real- kinematic (RTK) corrections. A roving vehicle carries the system collects data for a precise DTM. Adam Ciecko, Bartlomiej Oszczak, and Stanislaw Oszczak, University of Warmia and Mazury, Poland THE ATV WITH assembled equipment; backup reference station in background Precise navigation, land surveying, and fleet-vehicle monitoring all require reliable, stable, and inexpensive wireless connections. Our previous experience with conventional UHF radio modems clearly showed that this type of data teletransmission is inadequate in modern positioning and monitoring systems. Especially in urban environments, the limitations of radio modem distribution of DGPS and RTK corrections became obvious. Our research team performed experiments and practical tests that showed the maximum range reached by the GPS reference station, transmitting at 10 watts of power, was 10 kilometers in the urban environment. In hilly and wooded urban areas the range was much shorter. We hypothesized that these problems could be overcome by the use of GPRS transmission. We constructed our system of GSM/GPRS teletransmission of data for monitoring, satellite positioning, and navigation at the Departent of Satellite Geodesy and Navigation at our university in cooperation with Polish companies and institutions including Biatel S.A., Polkomtel S.A., the Mari Office in Gdynia, and the Naval Academy in Gdynia. In the first phase of the project (P2P connection between the GPS reference station and the roving GPS receiver) we simply replaced a standard UHF radio modem wireless connection with the GPRS connection (FIGURE 1). We used this connection for the initial tests of GPRS teletransmission. BI- ATEL designed and built a special GPRS modem was designed and built for our purposes. It is compact, light-weight, and operated by just one switch (FIGURE 2). In addition, the modem s software can be upgraded, an essential feature for our project. After many trials of various software and configurations of the GPRS modem, we made the system work. First we successfully sent the DGPS corrections in Radio Technical Commission for Mari Services (RTCM) format. Next, modifications to the GPRS modem s software contributed to the first fixed solution in RTK surveying mode. This success revealed an obvious limitation of the solution we achieved in phase one of our project. In P2P solution, only one user could receive the corrections from the reference station, which of course is impractical in a modern teletransmission system. In the second phase, we introduced a corrections server (FIGURE 3). The server gathers information from multiple reference stations, manages data, and redistributes corrections to multiple users of the network. This allows a practically unlimited number 2 GPS World September

2 Survey/Mapping SURVEY & CONSTRUCTION FIGURE 1 First phase of the GPRS project FIGURE 3 Second phase of GPRS project FIGURE 4 Third (existing) phase of GPRS project FIGURE 2 GPRS modem designed by Biatel of GPS reference stations as well as users. The GSM/GPRS network enabled the distribution of corrections from the reference station to the server and from the server to the user. This solution provided promising results, greater development possibilities, and introduction of new features to the system. The limitation was no longer the number of the users. We did encounter in this phase of the project the problem of transmission delay. The delay of GPRS transmission caused problems with continuous real- solution, and was especially problematic with the RTK method. In the third phase, we connected the corrections server, also called the replicator, to the GPS reference stations via a secure Internet connection (FIGURE 4). This system, built on existing infrastructure, significantly improved the transmission speed, making it more reliable and accessible for different users. Our system, which is still in the testing phase (initial operational capability status), consists of a network of GPS reference stations (three at the moment) connected to the system's main server using Internet Protocol Security tunnels (FIGURE 5). The server collects data from all existing GPS reference stations, manages data, and distributes it to mobile users in real. The distribution of corrections is possible via a network of GSM operators in Poland, which makes the system fully independent. Each mobile receiver is connected to the main system server via the GSM network and has a predefined primary GPS reference station. In case of failure of the primary station (for example, Internet breakdown), the server detects the emergency situation and automatically switches the user to another nearby station. September 2006 GPS World 3

3 FIGURE 5 System design and theoretical range of the DGPS/RTK services in northeast Poland FIGURE 6 The results of the nine-hour RTK test Each GPS reference station can be remotely controlled from anywhere in the world. The system can significantly increase GPS positioning accuracy even with the most inexpensive receivers and is fully compatible with all GPS receivers that have an RTCM option. The only user requirement is the GPRS modem with an activated SIM card dedicated to server application. DGPS/GPRS positioning and navigation may be an interesting alternative for the European Geostationary Navigation Overlay System (EGNOS) as well as the U.S. Wide Area Augmentation System (WAAS). RTK positioning and navigation with GPRS is a good solution in places where availability of UHF radio signals is problematic or limited by the urban environment or natural terrain obstructions. FIGURE 7 Measurement profiles (in blue) and actual track of the vehicle (in black) during real- measurement Static Tests Before executing field tests with the moving vehicle, we performed several static experiments to confirm the continuity, reliability, and integrity of our system of data transmission. These tests were performed on the reference station owned by the Department of Satellite Geodesy and Navigation. The receivers set up at well-known coordinates allowed continuous observation first in DGPS mode and then in RTK mode. We sent the corrections from the second of Olsztyn s stations located 2.7 kilometers from the rover. The observation intervals varied from several to 24 hours. In each case the recording interval was set at one second; therefore, a single hour of observation yielded 3600 epochs to analyze. The static tests produced favorable and promising results, although several s we observed gaps in the corrections received. The failure to receive a few corrections can quickly affect the accuracy of RTK positioning. In general the wireless GPRS connection showed good stability. FIGURE 6 shows a graph of accuracy obtained after nearly nine hours of observation. The maximum deviation from the true position is approximately six centers for the entire observation span. Even five or six very short gaps in delivery of corrections can be seen clearly on the graph. The remainder of the positions had access to corrections and gave an accuracy of about two centers. Overall, the results of the nine-hour RTK test showed a satisfactory continuity of data transmis- 4 GPS World September

4 sion and very good positioning accuracy. Dynamic Field Test For the dynamic test we selected a large unused field in Stawiguda, a village located 10 kilometers from the reference station. To get an interesting digital terrain model, we chose a very hilly and irregular section of about five hectares as a test field. To prepare for the measurement project we made the first measurements with EGNOS corrections using a Thales Mobile Mapper receiver. The guidelines of the project called for measurement profiles every 15 meters. The profiles were designed perpendicular to the road, which delimited our test field on the west side (FIGURE 7). The mapping receiver converted the designed profiles into base map readable format. For the purpose of the dynamic experiment, one of the mapping receivers was used for navigation over designed profiles, which proved very useful for this kind of navigation. The team prepared a set of GPS receivers to be used for the experimental measurements. The main receiver connected to the GPRS modem and working in RTK mode was an Ashtech Z-Extreme. We also had a precise geodetic backup receiver (Ashtech Z-Surveyor) connected to the same GPS antenna by a splitter and working in postprocessing mode. For comparing different GPS equipment and measurement techniques, we used a mapping receiver with a post-processing option connected to the external antenna. We used a second mapping receiver for navigation purposes. Recording intervals were set to one second for every GPS receiver taking part in the experiment. The university s reference station transmitted DGPS/RTK corrections in RTCM format while also recording raw observation data. The second reference station was set up at the test field as a backup for further calculations. On September 21, 2005, the team assembled the GPS equipment on an all-terrain vehcicle (ATV), which allowed us to drive over difficult terrain quickly and easily (FIGURE 8). Although the mapping receiver proved to be a good navigational aid, staying on track over the profiles required some practice. The practical part of the test, the field measurement of five hectares of land, lasted approximately 90 minutes, after which all of the collected measurements were processed and analyzed. Data Analysis Comparison of the accuracy of different GPS techniques required a true or reference positioning of the vehicle. We used four independent calculations of on-the-fly (OTF) dynamic positioning to best determine its position. Two reference stations and two moving rovers gave four independent results. These reference positions were computed as an average of four autonomous OTF post-processed positions, determined 2.0 Reference positioning accuracy 0.06 Reference positioning accuracy :10:01 11:28:21 11:45:18 12:02:22 12:21:08 12:38:34 FIGURE 9 Reference positioning accuracy for the full measurement :10:01 11:21:18 11:31:01 11:40:58 11:50:38 FIGURE 10 Reference positioning accuracy for the first 40 minutes of test 4 RTK positioning accuracy 0.20 RTK positioning accuracy :10:01 11:20:58 11:30:21 11:40:10 11:49:30 11:10:01 11:20:58 11:30:21 11:40:10 11:49:30 FIGURE 11 Accuracy of real- phase (RTK) positioning FIGURE 12 Accuracy of real- phase (RTK) positioning (zoomed) September 2006 GPS World 5

5 10 EGNOS positioning accuracy 4 DGPS positioning accuracy :11:00 13:22:57 13:35:06 13:49:51 FIGURE 13 Mapper accuracy with EGNOS corrections (local ) -4 13:11:00 13:22:57 13:35:06 13:49:51 FIGURE 14 Mapper accuracy with DGPS post-processed corrections (local ) FIGURE 15 Digital terrain model generated with the use of RTK data in independent calculations. The three redundant observations made possible the determination of OTF accuracy. A standard deviation error of reference positioning was computed for each second of the move for each coordinate B, L, and H. FIGURE 9 shows the results. The graph clearly indicates a change in results. The first part shows good positioning; the second part (in which, due to bad position dilution of precision and the low number of satellites) shows that accuracy drops to meters horizontally and meters vertically. FIGURE 10 focuses on the first 40 minutes of the field measurement. There the achieved positioning accuracy of the range of one center horizontally and two centers vertically gives a good reference for further analyses, and it was used for the next comparisons. We examined the accuracy of RTK positioning with the use of corrections sent via GSM/GPRS. FIGURE 11 is a graph showing RTK solution, in which we observed a few short gaps in the fixed solution. The gaps in receiving corrections gave a float solution with errors reaching as high as 3.5 meters for the vertical coordinate. Fortunately the gaps were very short and the fixed solution was regained after a few epochs. Considering the values with RTK fixed solution, we achieved very promising results with an accuracy of 3D positioning of about five centers (FIGURE 12). This clearly shows that GPRS can be used successfully for DGPS/RTK correction transmission. However, some work is still needed to make the transmission more reliable and resistant to intermittent gaps. We also examined the accuracy of the receiver for GIS mapping. Its compact size and resistance to atmospheric conditions made it very convenient to use. Achieved accuracies show that it can be successfully used in many applications that do not require centerlevel accuracy. The accuracy of real- positioning with the use of EGNOS corrections was generally within five meters, reaching 10 meters in two cases (FIG- URE 13). The positioning accuracy of the GIS receiver can be easy increased with the DGPS method either in real or postprocessed. Obviously the DGPS correction can be sent using GPRS transmission, and DGPS is far less sensitive to short gaps in transmission. In our experiment the emphasis was on RTK solution, and DGPS was examined 6 GPS World September

6 in post-processed mode. DGPS solution gave very good results with the 3D accuracy of better than 2 meters (FIGURE 14), which is satisfactory for many practical users of GPS/GIS systems. DTM Generation We used Surfer software to generate the digital terrain model (FIGURES 15 AND 16). RTK solution was fundamental in the model; however, it was combined with the phase postprocessed solution for the epochs that contained teletransmission gaps. The generation of a model using RTK measurements is very effective. In our example, the field measurement of a hilly area of about five hectares lasted only about 90 minutes thanks to use of the ATV for the experiment. The fact that the ATV and GPS equipment can be operated by one person provides even more ease of use. Conclusions The experimental network of GPS reference stations in northeast Poland transmitting DGPS/RTK corrections using GPRS technology is in the testing phase now. The fundamental assumption of the system is the possibility of receiving of DGPS/RTK data by a variety of users from a freely chosen reference station. Corrections can be received at any place of Polish territory with GSM coverage. Distribution of corrections is possible using different GSM operators in Poland. The system is open to any further updates, improvements, and modifications. The cost of using such a system is quite reasonable. One hour of DGPS corrections receiving (with two-seconds intervals) is approximately 0.2 PLN ($0.06 USD), and one hour of RTK corrections receiving (with two-second intervals) is around _1 PLN ($0.30 USD). The system can be used in any real- application of marine, air, or land navigation; geodetic precise positioning; and many more applications where reliable real- positioning is needed. The generation of a DTM model with the use of the proposed system is very efficient and cost-effective. The use of an ATV improves speed and efficiency. FIGURE 16 Digital terrain model generated with the use of RTK data {{WHAT ARE THE DIFFERENCES BETWEEN FIGURES 15 AND 16?}} Manufacturers (format later)biatel Thales Ashtech ADAM CIECKO is a member of scientific staff in the Department of Satellite Geodesy and Navigation, University of Warmia and Mazury in Olsztyn, Poland, where he earned his PhD degree in geodesy in BARTLOMIEJ OSZCZAK is a member of technical staff in the Department of Satellite Geodesy and Navigation at the same institution; he earned his master of science in geodesy and cartography in STANISLAW OSZCZAK chairs the Department of Satellite Geodesy and Navigation and is a full professor of the University of Warmia and Mazury. MORE ONLINE For further reading on this and related topics, see this column online article/ September 2006 GPS World 7