Bo Li, Beihang University, China; Zhiqiang Dan, Beijing Hualong Tong Science and Technology Co,.Ltd, China; Kun Fang, Kai Guo, Zhipeng Wang, Beihang University, China; Yanbo Zhu, Aviation Data Communication Corporation, China

View Abstract Sign in for premium content

Abstract:

Although Global Navigation Satellite System (GNSS) can provide global positioning service, its performance is unstable in the urban environments due to the reception of non-line-of-sight (NLOS) and multipath signals caused by building blockages and reflections. Therefore, we propose a real-time LiDAR aided GNSS positioning fault monitoring algorithm suitable for the regions affected by the NLOS and multipath. In this algorithm, the position of the autonomous car is determined by the GNSS/Inertial Navigation System (INS) integrated navigation system, while the target in the surrounding environment is detected by LiDAR. According to the differences of the target positions detected by LiDAR at different epochs, a monitor is constructed to detect GNSS positioning faults. The test statistic is constructed based on the difference between the detected target position and its true position. An adaptive sliding window method and an adaptive weight sequence are designed to construct the threshold, which is used to determine whether there is GNSS fault. The performance of the proposed algorithm is verified with simulation experiments. When the step error is 5, 10 and 20 m, the percentage of false alarm epochs are 0.05%, 0.05% and 0.057%, respectively. All GNSS positioning fault can be detected in real time without missed detection epochs. The proposed algorithm is compared with the residual chi-square test. The percentage of false alarm epochs in the proposed algorithm is used to calculate the fault detection threshold of the residual chi-square method. Results show that when the step error is 5 or 10 m, the residual chi-square test cannot detect the GNSS positioning faults. By contrast, when the error is increased to 20 m, this method can detect the fault. However, the percentage of false alarm epochs is 1.4%, which is much higher than that of the proposed algorithm. On the other hand, when a slope error of 5 and 10 m is added, GNSS positioning faults can be detected with the proposed algorithm with the percentage of missed detection epochs is 6% and 2%, respectively. When the slope error is 20 m, there is no missed detection epoch. For the residual chi-square test method, the GNSS positioning fault cannot be detected when the error is 5 or 10 m. When the error is 20 m, the percentage of missed detection epochs is 52%. The simulation results show that in the presence of step or slope error, the proposed algorithm performs better than the residual chi-square test method.