Outlier Detection for 3D-Mapping-Aided GNSS Positioning

Qiming Zhong, Paul Groves

Peer Reviewed

Abstract: This paper takes 3D-mapping-aided (3DMA) GNSS as an example and investigates the outlier detection for pattern matching based positioning. Three different test statistics, two in the measurement domain and one in the position domain, are presented. Two 3D city maps with different levels of detail were used, one of which contained two obvious errors, to demonstrate the performance of 3DMA GNSS positioning in the presence of errors in the mapping data. The experiments tested were conducted alongside busy roads in the London Borough of Camden, where a total of 8 sets of 2-minute static pedestrian navigation data were collected with a u-blox EVK M8T GNSS receiver. The results confirm that both 3D mapping errors and temporary environmental changes (such as passing vehicles) can have a significant negative impact on the performance of 3DMA GNSS positioning. After applying outlier detection, single-epoch 3DMA GNSS algorithm reduces the horizontal RMS position error by approximately 15% compared to that without outlier detection. The filtering algorithm attenuates the effects of temporary environmental changes, providing an improvement of about 15% over single-epoch positioning, while the outlier algorithm further reduces the RMS error to a comparable level to that of using high-accuracy maps, about 4.7m.
Published in: Proceedings of the 35th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2022)
September 19 - 23, 2022
Hyatt Regency Denver
Denver, Colorado
Pages: 2104 - 2126
Cite this article: Zhong, Qiming, Groves, Paul, "Outlier Detection for 3D-Mapping-Aided GNSS Positioning," Proceedings of the 35th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2022), Denver, Colorado, September 2022, pp. 2104-2126. https://doi.org/10.33012/2022.18332
Full Paper: ION Members/Non-Members: 1 Download Credit
Sign In