The Performance of RTK GPS Mapping In Urban Environments

Transcription

1 Presented at GNSS 2004 The 2004 International Symposium on GNSS/GPS Sydney, Australia 6 8 December 2004 The Performance of RTK GPS Mapping In Urban Environments InSu Lee Linlin Ge Satellite Navigation And Positioning Group (SNAP) School of Surveying and Spatial Information Systems University of New South Wales, Sydney, Australia Tel: Fax: i.lee@unsw.edu.au l.ge@unsw.edu.au YoungJin Lee Department of Urban Information & Cadastral Engineering Kyungil University, Gyungbuk, Korea Tel/Fax: yjlee@kiu.ac.kr WoonYong Park Department of Civil Engineering, Donga-A University, Busan, Korea Tel: Fax: upark@daunet.donga.ac.kr ABSTRACT In the past, classical surveying techniques by EDM (Electromagnetic Distance Meter), and Total Station (TS), was used for making a plan map. They contained lots of problems, such as atmospheric condition, costs and especially a line of sight. Nowadays, Real-Time Kinematic (RTK) GPS techniques, was introduced in the fields of engineering survey. This study aims at the availability assessment of RTK positioning for making a map under the urban environments. TS surveying was used to check the performance on the result of RTK positioning. Besides, a field test was performed, in the spring and winter season to investigate which factors affect the accuracy of RTK positioning. That result was that the number of visible satellites being tracked rather than a mass of green foliage has an effect on the results of RTK positioning directly. It is required that the research into multipath effect by trees and/or building in RTK positioning should be more carried out. KEYWORDS: GPS, Urban environments, Multipath, RTK positioning 1. INTRODUCTION 1

2 Global Positioning System (GPS) has become more and more important for GPS geodesy, land surveying, cadastral surveying and maritime surveying, etc. But unfortunately, on their own, they do not guarantee the centimetre level of accuracy (Sven Martin et al., 2000). To gain more accurate coordinates of positions, differential GPS technique is required. This technique allows even centimetre level accurate positioning using the so-called integer ambiguity resolution technique. The basic concept of differential technique is to mitigate the main error sources, such as ionospheric and tropospheric delay, orbit errors and satellite clock errors by receiving satellite signals at a well-known location. All common errors between this reference receiver and the user receiver are cancelled out (Herber Landau et al., 2001). Nowadays, this differential technique is used in real time as well as in post-processing of data. And real-time data transfer is routinely possible, which enables real time computation of baseline vectors (B. Hofmann-Wellenhof et al., 199) and has led to the Real-time Kinematic (RTK) technique. And RTK positioning could play important roles in wide applications such as the mobile mapping, the emergency communication, and the prevention of disasters, etc. In emergency, what is the most import thing is locating an occurrence spot of event quickly. RTK positioning could make a much contribution for locating. Also RTK positioning was implemented as tool of mapping method of Digital Map (Hong SE et al, 2002) and the analysis of road alignment and construction GIS (Jang SK et al, 2003). But there are lots of problems (e.g., multipath by trees and building, and a signal derangement, etc) to be solved in advance, especially under and/or near the urban environments. The main aim of this paper is to assess the availability of RTK positioning to make a map and to investigate the horizontal positions accuracy of its under a urban environment, such as multipath effects. 2. The Concept of RTK positioning Real Time Kinematic positioning is the dynamic GPS positioning technique Using short observation time, this system provides precise results in real time. To achieve higher positioning accuracies (decimetre or centimetre level) in real time, the double differencing technique should be implemented using carrier phase data. At any epoch of measurement, differencing carrier phase measurements (which can be made on either the L1 or L2 carrier) made simultaneously by receivers k and i, to satellites s and q, yields the following "doubledifference" observable: sq Φ ki = s k - q k - s q i - i = s k - q k - s i - q i +. N sq sq ki + ' ki (1) Note that the is the true geometric range from one receiver to one satellite, is the carrier sq phase measurements, and, the double-differenced ambiguity term. N k i The use of (double-differenced) carrier phase is problematic because any data processing scheme must ensure the estimation of two classes of parameters: the baseline parameters sq x ki, y ki, z ki (contained within the geometric range term ρ ki )), and the ambiguity sq parameter (for each pair of independent satellites s, q). N k i 2

3 Ambiguity resolution is a process, which can only be applied to static GPS carrier phase data, in recent years algorithms have been developed to carry out ambiguity resolution "on-the-fly", that is, while one of the receivers is in motion. Once the ambiguities have been resolved, all carrier phase data can be converted to a range-like observable that can be used for precise (centimetre-level) instantaneous positioning. Nowadays kinematic carrier phase-based positioning can be carried out in real-time if an appropriate communications link is provided over which the carrier phase data collected at a static base receiver can be made available to the rover receiver's onboard computer; to generate the double-differences, resolve the ambiguities and perform the position calculations. (Rizos C, 1999). This is defined as Real Time Kinematic (RTK) GPS positioning. Total Station is one of the representative and very popular equipments of engineering survey. Generally, distance is measured by EDM, and angle, by transit or theodolite, but TotalStation can process the combination functions. Nowadays, further, computation and drawing in batch processing could be with TotalStation. Figure 1 presents a graphical overview of RTK positioning. In Figure 2, the application tools of TS were shown. Figure 1. A graphic overview of RTK positioning Figure 2. Various work of Total Station 3. Field Observations To assess the availability of RTK positioning in various conditions, a field-testing was performed twice, in winter and in spring season, respectively. Herein, RTK1 means a performance implemented in winter and RTK2, that in spring. Figure 3 present a photograph of a test site taken in winter and in spring season. In winter season, trees are naked of their leaves and in spring, full of a mass of green foliage. GPS equipment consisted of a pair of Trimble 400SSi receivers, dual-frequency with firmware allowing RTK observations. The accuracy in horizontal was the orders of 20mm+2ppm in horizontal, when more than 5 satellite tracked, at least 2epoch. On the other hand, TotalStation, Leica TC 1000 model has the accuracy of order of 1-sec in angle measurement and orders of 2mm+2ppm in distance measurement. 3

4 To see the availability of RTK positioning on the varying number of satellites in view, topographic mapping using TS was progressed at the same time. Figure 3. Test site of RTK1 and RTK2 4. Processing and Assessment Figure 4. Point distribution of RTK1 and RTK2 This section deals with the accuracy of positions measured, the variation of Position Dilution of Precision (PDOP) along epoch-by-epoch with respect to the number of visible satellite. Figure 4 presents the graphic overview of points distribution measured on RTK1 performance in winter and RTK2 performance in spring, respectively. At first stage, the rate of fixed solution and float solution of carrier phase integer ambiguity was investigated to check the quality of RTK measurements. The flags in Figure 4 indicate that positions have a lot of fixed solution of carrier phase integer ambiguity, but in Figure 4, hardly see a flag. In RTK1, among 18 points measured, 169 points had float solutions of integer ambiguity and the rest of them had the fixed solutions (4.4%). On the other hand, in RTK2, 150 points had float solutions of integer ambiguity among the 314 points measured, and the rest of them had the fixed solutions (4.%). To gain the fixed solutions of integer ambiguity, at least 5 satellites must be tracked or the quality of signals from satellites also have to be favourable. Hence, because a good geometry were tracked at RTK2 for the surveyed time, these results were very likely. A table 1 shows the accuracy in horizontal and vertical of points measured. 4

5 ACCURACY Horizontal Vertical CASE MEAN RMS (m) Mean RMS (m) RTK1 RTK2 Table 1. Accuracy of points position At second stage, the variation of position dilution of precision (PDOP) with respect to the number of visible satellites being tracked was investigated. Dilution of precision (DOP) is a measure of the quality of the GPS data being received from the satellites. DOP is a mathematical representation for the quality of the GPS position solution. The main factors affecting DOP are the number of satellites being tracked and where these satellites are positioned in the sky. As shown in Figure 5, Mean number of visible satellite in RTK1 and in RTK2 was 4.85 and 5.22, respectively and PDOP, 5.2 and 5.41, respectively. Even if more satellites in RTK2 were tracked, variation of PDOP over epochs was rather high SVs Point ID - epochs SVs PDOP PDOP SVs Point ID - epochs Figure 5. PDOP VS. SVs of RTK1 and RTK SVs PDOP PDOP And Even if, in winter season, there was no trees leaf at a test-site, Mean RMS of points positions in horizontal and vertical in RTK2 was lower than that in RTK1. This may be due to the number of satellites being tracked. So a mass of green foliage, in spring, also had no so much effect on the accuracy of point position. Vertical Horizontal Vertical Horizontal RMS(m) RMS(m) Point ID - epochs Point ID - epochs Figure. 6 RMS of RTK1 and RTK2 In Figure 6, both of lines indicate the accuracy in horizontal and vertical of points measured, 5

6 respectively. When comparing with both each other, the graphic overview of Figure 6 presents big variation (where the accuracy of points position change unsteady on epoch by epoch) than that of Figure 6, despite there are few obstacles, such as tree leaf, etc. Especially, at the initial, the middle, and the end part along epochs, big jump were shown. These are the reason why more obstacles existed at that sites GPS rover antenna mounted on a stakeout rod was located near and/or under big trees with a mass of foliage. In the previous phrase, mainly focused on the analysis associated with the number of satellites in view, PDOP and the accuracy in horizontal and vertical of points measured. At this stage, the garden map of a testing-site made by RTK system (Figure 8) and Total Station (Figure ) were shown. The procedure for these works is as follows; downloads the Cartesian coordinate of points from each system s storage memory to PC, edits the data using commercially available Computer Aided design (CAD), depicts the following a garden map. These garden maps were used as reference tool to assess the availability of RTK system approximately under a worse environment like an urban garden. In Figure, G1~G12 indicate some portions of a garden map by Total Station. When comparing Figure 8 with Figure 8, the garden map of Figure 8 looks like that of Figure except some portions. But, still, many mismatch remain on some portions of a garden map (e.g., G9, G12, and G13) between Figure and Figure 8. Figure. A garden map by Total Station Figure 8. A garden map by RTK1 and RTK2 6

7 These results were mainly due to not tracking many satellites at surveyed time because GPS rover antenna was located close to big buildings, etc. In case of G8 and G10, even though many satellites are tracked at that time, PDOP was very high and float solution of a carrier phase integer ambiguity (where ambiguity resolution is not fixed) are occurred frequently. Additionally, even if the distance between reference station and rover is within 100 m, a radio link was interrupted now and then. So this could result in Single Point Positioning (SPP) using code only because rover did not obtain any kinds of observations from reference station. At last stage, three maps based on the same coordinate reference frame were overlapped and then, extracted the Local Cartesian Coordinates of the edges of each portion using CAD, which have similar shapes. Due to a limited space, every coordinates extracted by this way were excluded in this paper. Table 2 shows only RMS of the difference of Local Cartesian Coordinates between three systems; say RTK1, RTK2, and TotalStation according to portions. SYSTEM RTK1-Minus-TS RTK2-Minus-TS RTK1-Minus-RTK2 PORTION RMS (m) RMS (m) RMS (m) G G G G G G G G Table 2. RMS of the Coordinate Difference RTK1, RTK2, and Total Station And overall RMS of RTK1-Minus-TS, RTK2-Minus-TS, and RTK1-Minus-RTK2 were 2.811m, 2.245m, and 1.6m, respectively. Especially, these large RMS is be caused by inaccurate selection of points to compare the coordinates of the specific position on each map. Maybe, this was one reason, and another one was factors such as trees leaves, building, etc to interrupt RTK positioning directly. In Table 2, portion G, G9, G10, and G11 showed large. These results say that under the urban environments, RTK positioning is not suitable for precision positioning, but possible for a rough mapping. So, in this kind of environment, the combination of RTK positioning with topographic mapping using Total Station, Plane table, etc. can be a good alternative. 5. Conclusions This paper deals with the availability assessment of RTK system under the urban environments. This result has showed that fixed solution of carrier phase integer ambiguity has a high rate in case that the number of satellite tracked increased, nevertheless, PDOP is still in high value. And even though it has a mass of green foliage at test site in spring, many satellites are tracked except for under the worst environments, e.g. under trees and near buildings. Through the comparison of a garden map produced using TotalStation with that using RTK positioning, there are a few mismatch on some portion of map produced using data measured near building and under trees directly, but the approximate drawing of a garden map was possibly done. Researches are further required to overcome the drawbacks that are

8 revealed in this investigation. RTK positioning integrated with other sensors or TS terrestrial techniques could make a much contribution to various applications, such as disasters prevention, emergency mapping, etc. ACKNOWLEDGEMENTS The author wishes to thank to Hsing-Chung Chang in supplying a suggestion and polishing up a paper. References B. Hofmann-Wellenhof, H. Lichtenegger, and J. Collins (199) GPS: Theory and Practice, Herbert Landau, Ulrich Vollath, Alois Deking, and Christial Palels (2001) Virtual reference Station Networks Recent Innovations by Trimble, Technical Paper, Trimble Terrasat GmbH. Hong, SungEon, and Park, SooHong (2002) A Study on the Mapping of Digital Cadastral Map using RTK-GPS, The Korean Society of Cadastre, 12(2): (Korean). Jang, SangKyu, Hong SoonHeon, and Kim GaYa (2003) The Analysis of Road Alignment and Construction GSIS Using RTK and GPS, Korean Society of Surveying, Geodesy, Photogrammetry and Cartography, 21(4): (Korean). Lee InSu, Youn KyungCheul, Park WoonYong, and Lee KeeBoo (2003) The Comparison of Precision of RTK according to Satellites Visibility, Proceeding of Korean Society of Geodesy, Photogrammetry, and Cartography, Dong-A University, (Korean). Langley, R.B (1994) RTCM SC-104 DGPS standards, GPS World 5(5): Rizos C (1999) GPS enhancement,s In: Chris Rizos (Ed), Notes on Basics GPS Positioning and Geodetic Concepts, School of Surveying and Spatial Information, University of New South Wales, 15~16, electric version (in Sven Martin, and Cord-Hinrich Jahn (2000) Experiences in set-up and usage of a geodetic real-time differential correction network, Earth Planet Space, 52: Timothy R. Lemmon, and George P. Gerdan (1999) The Influence of the Number of Satellites on the Accuracy of RTK GPS Positions, Department of Land Information RMIT University, Melbourne, Victoria, THE AUSTRALIAN SURVEYOR, Vol. N