A Tightly-Coupled GNSS/INS/MM Integrated System Based on Binary Search Algorithm for Train Localization Applications

Abstract

This paper proposed a novel integration system fusing Global Navigation Satellite System (GNSS), Inertial Navigation System (INS) and Map Matching (MM) method to provide continuous and accurate position, velocity and attitude (PVA) for the train. The tightlycoupled (TC) integration strategy is firstly applied to fuse the pseudorange and pseudorange rate from the GNSS and INS. In the TC integration strategy, the Kalman filter (KF) is used to estimate the accelerometer and gyroscope biases and then the estimated sensors biases are taken as feedback to correct the INS measurements, whereas the estimated PVT errors are used to obtain the integration solutions. To further improve the position accuracy, a map matching (MM) method based in binary search algorithm is proposed. Firstly, the nearest point-of-interest (POI) are calculated in binary search algorithm. Compare the distance between the POI to the two adjacent points to determine the nearest straight track lines. Based on the two nearest straight track lines, the correct track number is chosen according to the vertical distance from the TC GNSS/INS integration position to the two track lines. The perpendicular projection could be done to compute the perpendicular distance between the integrated point and the nearest straight track line, and the final position could be then updated. In order to evaluate the performance of the proposed system, a practical experiment was conducted on Beijing-Shenyang high-speed railway in China. The test route includes GNSS open-sky environment (over four visible satellites), GNSS partly blocked scenarios (less than or equal to four and greater than zero) and GNSS outages (without visible satellites). The DRMS value of the TC GNSS/INS integration is 2.76m, which has improved 38% compared to that of the LC GNSS/INS integration 4.47m. It is proved that the TC GNSS/INS integration is able to generate more accurate position information than the LC GNSS/INS integration. With the aided MM, the DRMS of the TC triple integration is 2.04m, which has a further improvement and meets the position accuracy requirement. In conclusion, the experimental results indicate that the proposed TC GNSS/INS/MM could provide continuous and accurate position solutions in both GNSS open-sky and GNSS-difficult scenarios for the train localization applications.