|Abstract:||Currently, the availability of global navigation satellite systems (GNSSs) is anticipated to improve because of the presence of various positioning satellites. The Asian region, in particular, will benefit from multi-GNSSs, using the United State’s Global Positioning System (GPS), Russia’s Global Navigation Satellite System (GLONASS), Europe’s global navigation satellite system (GALILEO), China’s Navigation Satellite System (BeiDou), and Japan’s Quasi-Zenith Satellite System (QZSS). It has been said that the performance of real time kinematic GPS (RTK-GPS) using low-cost single-frequency receivers is insufficient for practical applications. In these days, even the GNSS chipset of a mobile phone can utilize the multi-GNSS constellation, which includes GPS, GLONASS, GALILEO, BeiDou, and QZSS, to continuously provide robust position information everywhere. It is indeed no longer a special capability for many people, and position information is quite a valuable source for making our life better. From such a point of view, an RTK-GNSS (not to be confused with RTK-GPS) using low-cost single-frequency receivers has been paid attention to for new applications, such as navigating unmanned aerial vehicles. Although the performance difference between a geodetic-level receiver and a low-cost single-frequency receiver is still significant, users don’t need a geodetic-level receiver under a relatively open sky condition and a short baseline. The performance of instantaneous RTK-GNSS with a low-cost single-frequency receiver under these conditions is almost perfect. In previous studies, we showed that the RTK performance using a dual-frequency GPS receiver was similar to that when using a low-cost single-frequency GPS/BeiDou/QZSS receiver. These results are probably derived from the fact that the number of usable observations in the ambiguity resolution is almost the same for both types of receiver. The dual-frequency GPS receiver can use 20 observations, in terms of L1 and L2 observation, if there are 10 satellites. On the other hand, a single-frequency GPS/BeiDou/QZSS receiver can also use 20 observations just because there are 20 satellites in view. In addition, the difficulty of treating GLONASS observations in the RTK algorithm seem to disappear, even when using the low-cost receiver, as long as the same type of receiver is used in both the reference and the rover. We will discuss this issue further in the details of the paper. Our goal is to provide continuous decimeter-level accuracy using low-cost receivers in urban environments for future intelligent transportation system (ITS) applications. We previously presented the code and velocity-based differential positioning method using a low-cost single-frequency receiver by a loosely coupled Kalman filter (Kubo et al., ION2015). The velocity information was deduced from the Doppler frequency. The maximum horizontal errors were about 2-3 m in all tests, and the standard deviations were below 50 cm in normal urban environments in Tokyo. In the current work, we integrated an RTK-GNSS with the previous code and velocity-based positioning method using a Kalman filter to further improve the accuracy. The target of this study was to reduce the maximum horizontal errors to within 1.5 m, for lane recognition. In order to achieve this performance, we modified the RTK-GNSS algorithm itself in terms of float solutions in the ambiguity resolution. It is well known that providing good float solutions enables the performance of an RTK to improve. If we use normal code-based positions as float solutions, over 20-30 m errors are frequently seen near buildings. To overcome this issue, we produced new float solutions, meaning the outputs of the code and velocity-based integrated results mentioned earlier (Kubo et al., ION2015), and these float solutions were used for ambiguity resolution. Furthermore, to make use of the multi-GNSS effect in terms of the number of usable satellites, we utilized two low-cost receivers simultaneously. The constellation of the first receiver was set to GPS/BeiDou/QZS/GALILEO, while the constellation of the second receiver was chosen to be GPS/GLONASS/QZS/GALILEO. The reason we did this is simply that the currently available commercial low-cost receiver we used is not able to provide BeiDou and GLONASS observations simultaneously. However, in Asian countries, both BeiDou and GLONASS satellites are very important for increasing the number of usable satellites in urban areas. Two tests were conducted to evaluate the proposed method, using two sets of low-cost receivers. For the receiver setting, “Set1” refers to the GPS/BeiDou/QZS/GALILEO constellation, while “Set2” is the GPS/GLONASS/QZS/GALILEO constellation. The first test was conducted under a relatively open sky environment. The second test was conducted under an urban environment. The duration of each test was about 30 minutes. The receivers used in this test were u-blox M8T in both the reference and the rover, and the data rate was set to 5 Hz. The length of the baseline was less than 5 km. For clarification of the test results, we prepared geodetic-level receivers–Trimble NetR9 (all constellations and frequencies are available)–in both the reference and the rover, a fiber-optic gyro (FOG), and a speed sensor to produce at least 10 cm-level reference positions without RTK solutions. All raw data were post-processed (in the same manner as the real-time implementation) to obtain position results. The fix rates of the RTK-GNSSs are shown in the following Table 1 for different receivers. The results were as we expected; many incorrect fixes can be seen in the case of the u-blox receiver (for both Set1 and Set2) in urban areas. To avoid these large errors due to inaccurate fixes, a suitable number of least satellites was selected (Tokura et al., ION2015). It was simple to produce the reference positions by the integration of the RTK-GNSS results using the FOG/speed sensor. Table 1. RTK-GNSS Results (Fix Rate) NetR9 U-blox Set1 U-blox Set2 Open Sky 100 % 99.2 % 94.1 % Urban Areas 91.7 % 39.1 % 28.3 % Table 2 lists the maximum horizontal errors, with the 99th percentile of all horizontal errors in parentheses, for a comparison of the two methods. The first method is our conventional method (Kubo et al., ION2015). It is the integration of code and velocity-based differential positions. The second method is our proposed method, which is the integration of the RTK-GNSS with the conventional differential method. Although the maximum horizontal error exceeded 1 m, the overall results were largest for the conventional method; the 99th percentile of all horizontal errors was actually only 0.95 m for the case of the combined results for the integrated method. The merit of combining two receivers was also confirmed. The reason we did not see a large improvement with the RTK-GNSS method is that most of the epochs were deduced from code and velocity information. Even when there are 20-30% correct fixed positions, the rest of the 70-80% were deduced, not from cm-level accuracy but from decimeter-level accuracy. Table 2. Maximum Horizontal Errors and 99th percentile U-blox Set1 U-blox Set2 U-blox Combined Conventional 1.89 m (1.43 m) 2.22 m (1.73 m) 1.79 m (1.32 m) +RTK-GNSS 1.81 m (1.12 m) 1.92 m (1.42 m) 1.41 m (0.95 m) We have investigated the current performance of low-cost single-frequency receivers in urban areas. The results clearly show that the use of multi-GNSSs enables us to achieve 1-2 m maximum horizontal error by integrating the RTK-GNSS information with the conventional code and velocity information. This will surely contribute to the use of GNSS for future ITS applications.|
Proceedings of the 29th International Technical Meeting of The Satellite Division of the Institute of Navigation (ION GNSS+ 2016)
September 12 - 16, 2016
Oregon Convention Center
|Pages:||1891 - 1913|
|Cite this article:||
Higuchi, Motoki, Kubo, Nobuaki, "Achievement of Continuous Decimeter-Level Accuracy Using Low-Cost Single-Frequency Receivers in Urban Environments," Proceedings of the 29th International Technical Meeting of The Satellite Division of the Institute of Navigation (ION GNSS+ 2016), Portland, Oregon, September 2016, pp. 1891-1913.
ION Members/Non-Members: 1 Download Credit