The Technical Feature and Accuracy Analysis of Land-based MMS Application for Lancang River Basin Area in Yunnan Province of China

Transcription

1 The Technical Feature and Accuracy Analysis of Land-based MMS Application for Lancang River Basin Area in Yunnan Province of China Quanfa Day, Kun Shi, Renjun Ding Faculty of Land Resource Engineer, Kunming University of Science and Technology Kunming, Yunnan, , P.R.China Abstract Lancang River, which main basin area is in Yunnan province, is the beginning of the Mekong River. And Yunnan is the member of the GMS, including other five Southeast Asian countries. All those countries and Yunnan hope to develop economy and utilize resources in harmony with each other. Lancang River basin area is the emphasis in the developing plan of GMS. So it is urgent to establish necessary Geographic Information System in Lancang River basin area. Land-based MMS (Mobile Mapping System) is a 3S(GPS, GIS, RS) integrated surveying system to fulfill the real-time and periodic data collection for modern surveying engineering and GIS. Land-based MMS uses GPS and other navigation sensors, such as IMU, to provide the position and orientation of the vehicle. And the dead-reckoning CCD digital s are used for relative positioning from the vehicle. In addition, a computer with software is used to extract useful spatial and attribute data. This paper discusses the accuracy of MMS. Analysis testified that the Land-based MMS could apply to measure the large-scale map. It has prominent meanings in establishing Geographic Information System in Lancang River. Keywords: Land-based MMS, 3S Integration, Accuracy Analysis 1. Introduction Greater Mekong Subregion (GMS) consists of Cambodia, Vietnam, Laos, Thailand, Myanmar and China (Yunnan province). In the first leaders' summit of the GMS, the six countries endorsed major development plans along the Mekong River that would bring highways, power grids and telecommunications to those regions. All those plans need accurate and dense geo-spatial date in short time. That is to say, Geographic Information System (GIS) should build and update quickly. But some region, such as Lancang River basin area, is very lack of the original topographic maps. And traditional methods in data collection, including terrestrial surveys and aerial photogrammetric surveys, can t fulfill the demand completely. However, in the other hand, followed by the development of science and technology, the means of surveying and mapping are changed quickly. Many new solutions in geo-spatial data collection are emerging. One of the best solutions is Mobile Mapping System (MMS), a multi-sensor system that integrates various navigation and remote sensing technologies together on a common aerial or land-based carrier. It is also a type of 3S integration as well because of using the technologies of GPS, GIS and RS. 2. Constitution of Land-based MMS MMS can be air-based or land-based. Land-based MMS often use van as carrier though some of it use train, person or other carriers. Except the carrier and a computer including necessary software, the core components of land-based MMS are divided into two kinds of sensors, the navigation sensors and the mapping sensors. The navigation sensors in the carrier are employed to determine the position and orientation of the carrier namely, position and orientation of the mapping sensors, also in the carrier, is provided. Then the mapping sensors are used to determine the positions of object points relative to the sensors. So the spatial data of the objects is acquired in this procedure by computer. At the same time, the attribute data is extracted from the mapping image by intelligent software. Then all the data are recorded in the compute and entered in database of GIS. 3. Navigation Sensors As a navigation sensor, GPS receiver is a key component of land-based MMS. In fact, the development of land-based MMS is followed by the development of GPS. The realization of land-based Day, Q.; Shi, K. and Ding, R. 27

2 MMS depends primarily upon the precise positioning of GPS. The relation between them likes the relation between traditional control survey and traditional topographic mapping. For land-based MMS, GPS must provide the carrier position in order to provide the mapping sensors position. Moreover, GPS has secondary function. It can provide system synchronization and give coordinates in WGS84. Code-only GPS was employed in early low-precision land-based MMS. Obviously this system cannot be used for large-scale topographic mapping. But dual frequency GPS in recent system has resolved the problem. Other navigation sensor is IMU (Inertial Measurement Unit). Providing the mapping sensors orientation is the primary function of IMU. Because of the limitations of finance or technology, the accurate of IMU may be different. Some of land-based MMS use precise navigation-grade IMU. Some only use a lot of low-precision components of IMU----gyros, odometer. In general, the high-precision IMU is often fit for land-based MMS with high-precision GPS. Another function of IMU is positioning. When GPS receiver loses satellite signs, IMU can bridge GPS outages. GPS/IMU integrated system is a cooperate system. In the one hand, GPS controls the long-term error growth of the IMU. In the other hand, IMU corrects GPS cycle slips, and precisely interpolates between GPS fixes. 4. Mapping Sensors Digital (CCD) hadn t been used in aerial photogrammetry or RS before 1990s because of poor solution. But for land-based MMS, the small -to-object distances mean acceptable accuracy. With the advantage of direct data transmission to computer, it is impossible to integrate RS and GIS in land-based MMS. So CCD is the chief mapping sensor. According to the bundle aerial triangulation, two images about object are constructed stereo pair to solve the distance between and object. In order to get better imaging geometry, many MMS employ more than two CCD s. There are other mapping sensors, such as laser range finder (LRF) and laser scanner. Using these sensors, the distance between sensors and object can be acquired directly. In common circumstances, LRF is an assistant mapping sensor to CCD s. With the help of LRF, stereo pair is not necessary and only one CCD can be employed to survey. The first operational land-based MMS is GPSVan TM that was developed by the Center for Mapping at the Ohio State University. From then on, there have been over 10 land-based MMS developed by universities and companies. The principles of them have no different. The successful example is VISAT system. In china, WUMMS, developed by Wuhan University, is the first one. Table 1 shows differences in the constitution of four systems. It is obvious that the accuracy of the MMS is mainly related with the accuracy of the sensors. But by reason of technology and money, the most precise sensors are not always fit to MMS. It is worth emphasizing that different MMS can be fulfilled with different demands. Low-precision MMS can use in Investigation for appearance situation of the road and Investigation for utilities, like electricity pole, guard rail, traffic sign, tunnel, bridge, building along the road etc. In the following part of this paper, it is mainly discussed the feasible of land-based MMS used in large-scale topographic map. Table 1. Comparison of Chinese/Foreign Classic Land-based MMS System GPS Van VISAT WUMMS LD2000R Developers Ohio state University of Wuhan university Leador Corp. university Calgary Nation U.S.A Canada China China GPS Dual-frequency Dual-frequency Pseudo-range Pseudo-range carrier phase carrier phase Navigation Sensors Mapping Sensors IMU 2 gyros, 2 odometers Strapdown Inertial Navigation System None Inertial Navigation System DC 2 monochrome 8 monochrome 3 monochrome 4 monochrome CCD CCD CCD CCD Other 2 color VHS 1 color VHS Laser range 1 color VHS finder Accuracy 1 3 m m 1 3 m 0.5 m 28 Day, Q.; Shi, K. and Ding, R.

3 Figure 1. Land-based MMS mapping 5. Mapping algorithm There are three coordinate frames concerning in mobile mapping: mapping coordinate frame (M-frame), IMU body coordinate frame (b-frame) and sensor coordinate frame (c-frame). The order of mapping is to measure the position of the object P in the M-frame, r op (x p, y p, h p ), showing in Fig 1. r op =r oc + sr oc r cp (1) In the above equation, r oc (x c, y c, h c ) is the position of the CCD in the M-frame; s and R oc are respectively the scale factor and rotation matrix between the c-frame and the M-frame. As mention above, s can be solved by stereopair or LRF. r cp is the position of the object p in the c-frame. r cp = (x c p x c pp, (y c p y c pp)/k, -f) (2) Where x c p and y c p are the image coordinates of point p corresponding to the object P; and x c pp, y c pp are the principal point offsets from the CCD format center; k is a scale factor accounting for the non-squareness of the CCD pixels, f is the calibrated focal length of the lens in use. r oc = r og - R oi (r ig + r ic ) (3) R oc =R oi R ic (4) r og (x g, y g, h g ) is the position of the GPS antenna in the M-frame that is measured by GPS receiver. r ig (x b g, y b g, z b g)and r ic (x b c, y b c, z b c)are respectively the position of the GPS antenna and the CCD in the b-frame. R oi is respectively the rotation matrix with three parameters between the b-frame and the M-frame. Three parameters are determined by the orientation of IMU. R ic is respectively the rotation matrix with three parameters between the c-frame and the b-frame. R ic, r ig and r ic are determined by measuring the relative position of different sensors in the MMS. It can be deduced by equation (1), (3), (4) r op = r og +R oi (r ic -r ig + s R ic r cp ) (5) Because the MMS is moving during surveying, r og and R oi can be expressed in function with time, r og (t) and R oi (t). So the general expression of r op is r op = r og (t)+ R oi (t) (r ic -r ig + sr ic r cp ) (6) Accuracy Analysis There are many factor influenced MMS precision. Linearising equation (6) is δrop = δr og (t) + Positioning error δr oi (t)(r ic -r ig + s R ic r cp ) + Attitude error δr ic R oi (t) Calibration error in IMU/CCD offset δr ig R oi (t)+ Calibration error in IMU/GPS offset δs R oi (t) R ic r cp + Scale error δr ic R oi (t) s r cp + Calibration error in IMU/CCD rotations δr cp R oi (t) s R ic + Imaging error δt[ν(t)+ω(t) (r ic r ig + s R ic r cp )] Synchronization error (7) Day, Q.; Shi, K. and Ding, R. 29

4 There are eight errors consisted of the MMS error from equation (7). They can be divided in 3 kinds: the measure error, the calibration error of known quantity and the synchronization error. 6. Calibration error Among them, r ic, r ig and R ic are calibration errors. All sensors are installed according to the designing project when making a MMS and their mutual position will still calibrate after installing. So the influence of these errors is very limited to the total error. This kind of error is systematic error, and it can be solved if redundant observation permitted. 7. Measure error As the positioning error, r og (t) is decided by GPS's precision in normal circumstances. The dual-frequency GPS with RTK technology is employed in the newest MMS. The real-time dynamic position errors have promoted to ±10mm+1ppm (Horizontal error) and ±20mm+2ppm (Vertical error) and The Post-processed dynamic position error are even lower. But the positioning error of other navigation sensors is fairly high when GPS receiver loses satellite signs, for example SINS, horizontal error and Vertical error all surpasses ±200mm. In spite of the accuracy can improved by Kalman filtering, the effect is limited and unstable. Therefore the carrier positioning error especially in GPS signal interruption is very high to the whole precision of MMS. When the precise IMU served as the orientation sensor, the draft of IMU does not surpass 1 degree per hour. That is to say, the error R oi (t) resulted in does not achieve 1 cm in most circumstances. Contrasting with positioning error and imaging error, attitude error is not prominent. δr cp is imaging error of the CCD. CCD (Charge Coupled Device) is one kind of solid- form imagery sensor. Compared with the vidicons tradition, CCD changes light signal to pulse signal from many independent pixels (small photosensitive cell). Thus the image information from CCD can be directly handled. But owing to the electronic image distortion, namely the deviation on the pixels of body imagery, the accuracy of CCD is poor in long -to-object distance. As an example, an error of one pixel in the x image coordinate will introduce a 20cm error for objects 30m away. It is obviously that the influence to the total error of δs has the direct ratio with the -to-object distance, though the size of δs is not too large. Hence, it is considered δs and δr cp will be significant in the total error of MMS as the -to-object distance become longer. 8. Synchronization error When the various data do not produce at the same moment, the synchronization error is produced. At present, owing to the advance of timing technology, the influence of the synchronization error is not very notable. But it is still worth paying attention to it especially in a precise system. Takes GPS positioning as the example, if the precision of timing was 0.001second and the velocity of carrier is 18km/h, the synchronization error is 5mm. Table 2 VISAT System Absolute Accuracy(10 30m) Survey mode Horizontal Vertical Max RMS Max RMS GPS/IMU Ideal conditions 0.42m 0.13m 0.26m 0.08m IMU Only Worst conditions 0.87m 0.29m 0.50m 0.19m 9. Conclusion At present, the research and development of Land-based MMS is still going on, the precision analysis object indicated: First, Land-based MMS can fulfill the requirement of large-scale topographic mapping. The experiment accuracy of VISAT system is showed in Table 2. The root mean square (RSM) is less than ±20cm or ±30cm, which meets also the need of production accuracy of 1:500-scale map. The LD2000R system also fulfills the requirement of the 1:3000-scale mapping, which absolute positioning precision of the target point is 0.5m (Surveying distance: 5-40m). Second, the reliability of land-based MMS depends upon an effective filtering model in GPS/IMU positioning system. Positioning error is the basic error in MMS. But the error varies in a large size. So the consistency of mapping accuracy is not well. The research should improve Kalman filtering model. Third, how to solve the image error is the key of a further development of MMS. According to the errors of MMS, image error is one of the largest errors (the other is positioning error). Making out the principal of 30 Day, Q.; Shi, K. and Ding, R.

5 image distortion, perfecting solution model is the emphasis. In Lancang River basin, many works of development plan are being carried out. Because of the lack of necessary geo-spatial data, some of the works period becomes longer. It is believed that the using of land-base MMS can shorten the period and save the fund. By reason that the bank of a Lancang River is very important especially in building a GIS of Lancang River basin. The vessel-based MMS should be considered. For the vessel is swayed, the attitude sensors must be different from van-based MMS. This research is proceeding now. In view of the success of other MMS, vessel-based MMS is feasible. Acknowledgement The Foundation Committee of Nature Science from Yunnan Bureau of Science and Technology supports this research work from 2002 to The research grants number is 2002D0018M. The authors of this paper would like to be grateful to Division of International Cooperation of Kunming University of Science and Technology for their supports. References Ellum, C. M., and El-Sheimy, N., 2002, Land-based Integrated Systems For Mapping And GIS Applications, Survey Review, 36, 283: Feng, W., 2002, Close-range Photogrammetry, Wuhan University Publishing House Kaplan, E. D., 1996, Understanding GPS Principles and Application, Artech House, Inc. Li, D., et al, 2000, Vehicle Navigation System Design and Implementation Based on Integration of GPS and GIS, Journal of Wuhan Technical University of Surveying and mapping, 25(2): 95~98 Chen, X., Kusanagi, M., 2001, Automatic Recognition and Location of Road Signs from Terrestrial Color Imagery, in Proceeding of The 3 rd ISPRS Workshop on Dynamic and Multi-Dimensional GIS & The 10 th Annual Conference of CPGIS on Geoinformatics, May 23-25, Bangkok, Thailand. He, X., Chen, Y., 1998, Design of Minimax Robust Kalman Filter for an Integrated GPS/INS System, ACAT Geodaetica et Cartographica Sinica, 27(2): 177~184 Day, Q.; Shi, K. and Ding, R. 31