This paper proposes an intelligent positioning enhancement methodology on the basis of tightly coupled GNSS-INS integration. The basic idea of this paper is to make full use of the visible satellite observations for measurement update in occluded environments, and then execute the positioning enhancement based on intelligent models. According to the position error characteristics for a varying number of visible satellites in different occluded environments, different positioning enhancement strategies are adopted. For the case involving two or three visible satellites, a joint estimation model for satellite position and pseudo-range based on multi-task learning is designed in order to supplement the satellite observations, and thereby satisfying 4 satellites to achieve accurate localization solving. For the case involving zero visible satellites or one visible satellite, a hybrid temporal neural network, which is compatible with INS and/or satellite inputs, is designed for localization error prediction and compensation. With the purpose of verifying the feasibility and effectiveness of the proposed methodology, road-test experiments with various driving scenarios were performed. The experimental results demonstrate the superiority of the proposed methodology over all the other employed methods.
Vehicle Positioning, Intelligent Enhancement, Tightly Coupled, Multi-task Learning, Input Compatibility.
BaiDan LI, GuoXing RUAN, MinCong TANG, WenXi YANG, "Intelligent Positioning Enhancement Methodology for Tightly Coupled GNSS/INS in Different Occluded Environments", Studies in Informatics and Control, ISSN 1220-1766, vol. 35(2), pp. 111-123, 2026. https://doi.org/10.24846/v35i2y202610
All articles published by Studies in Informatics and Control (SIC) are published in accordance with the terms and conditions of the Creative Commons Attribution License (CC BY-NC-SA 4.0).