Deciphering city skies: AI reveals a GNSS error

Facing the persistent challenge of non-line of sight (NLOS) errors in global navigation satellite systems (GNSS) urban navigation, researchers have introduced an innovative solution powered by artificial intelligence (AI). By leveraging the Light Gradient Boosting Machine (LightGBM), this method analyzes multiple characteristics of the GNSS signal to accurately identify and differentiate NLOS errors. This advancement promises to significantly improve the accuracy and reliability of GNSS-based positioning systems, making it an essential advancement for urban navigation, where accuracy is essential.

In urban environments, global navigation satellite systems (GNSS) often face signal obstructions caused by tall buildings, vehicles, and other structures. These obstacles lead to non-line-of-sight (NLOS) errors that lead to positioning inaccuracies, particularly problematic for technologies such as autonomous vehicles and intelligent transportation systems. The need for effective, real-time solutions to detect and mitigate these NLOS errors has never been more urgent, as reliable GNSS positioning is vital for the development of smart cities and transportation networks.

Published (DOI: 10.1186/s43020-024-00152-7) In Satellite navigation on November 22, 2024, This study presents a state-of-the-art machine learning approach to combat NLOS errors in urban GNSS systems. Researchers from Wuhan University, Southeast University and Baidu have developed a solution using the Light Gradient Boosting Machine (LightGBM), a powerful AI-based model designed to detect and exclude inaccuracies related to NLOS. The model’s performance was validated through real-world dynamic experiments conducted in Wuhan, China, proving its effectiveness in harsh urban environments.

The research highlights an advanced method for identifying NLOS errors in GNSS systems using the LightGBM machine learning model. This method involves using a fisheye camera to label GNSS signals as line of sight (LOS) or NLOS, based on satellite visibility. The researchers then analyzed a range of signal characteristics, including signal-to-noise ratio, elevation angle, pseudo-range coherence, and phase coherence. By identifying correlations between these characteristics and signal types, the LightGBM model was able to accurately distinguish between LOS and NLOS signals, achieving an impressive 92% accuracy. Compared to traditional methods such as XGBoost, this approach provided superior performance in terms of accuracy and computational efficiency. The results show that excluding NLOS signals from GNSS solutions can lead to substantial improvements in positioning accuracy, especially in urban canyons where obstacles are common.
Dr Xiaohong Zhang, lead researcher, commented: “This method represents a big step forward in improving GNSS positioning in urban environments. Using machine learning to analyze multiple signal characteristics, we showed that excluding NLOS signals can significantly improve accuracy. and the reliability of satellite navigation systems. This has profound implications for applications such as autonomous driving and smart city infrastructure.

This research has immense potential for industries that rely on GNSS technology, including autonomous vehicles, drones and urban planning. By improving the detection and exclusion of NLOS errors, this method can improve the accuracy of GNSS systems, making navigation safer and more efficient in densely populated cities. As cities become smarter and more connected, these advancements will play a crucial role in supporting the next generation of transportation and navigation technologies.

###

References

DOI

10.1186/s43020-024-00152-7

Original source URL

Funding information

This research was supported by the National Science Fund for Distinguished Young Researchers of China (Grant No. 42425003), the National Natural Science Foundation of China (Grants No. 42274034, 42388102), the Major Program (JD) of the province of Hubei (Grant No. 2023BAA026), the Hubei Luojia Laboratory Special Fund (Grant No. 2201000038), and the special fund of High Precision Positioning Technology Cooperative Joint Laboratory of Wuhan University and Baidu Map Beidou.

About Satellite navigation

Satellite navigation (E-ISSN: 2662-1363; ISSN: 2662-9291) is the official journal of Aerospace Information Research Institute, Chinese Academy of Sciences. The journal aims to report innovative ideas, new results or advances on theoretical techniques and applications of satellite navigation. The journal welcomes original articles, reviews and commentaries.