GLINS: GNSS-LiDAR-INS Integrated Navigation System
Jiahui Liu, Cheng Chi, Xin Zhang, Binlin Zhang, Dongen Li, Xingqun Zhan, Marcelo H. Ang
Abstract
Reliable navigation in complex urban environments is challenging due to frequent GNSS signal outages and multipath interference. A practical mitigation is to incorporate LiDAR-inertial odometry (LIO) to bound drift during GNSS gaps and maintain local accuracy. However, commonly adopted scan-to-map constraints may yield inconsistencies when global corrections conflict with a static map in a tightly coupled framework. To address these issues, this letter presents GLINS, a measurement-level tightly coupled GNSS-LiDAR-INS system that features a consistent estimator design with explicit landmark states. GLINS extracts stable landmarks from the LiDAR voxel map and treats them as explicit variables in the state estimator, allowing the map representation to co-evolve with global corrections. The proposed estimator is implemented as a keyframe-based sliding-window factor graph that tightly fuses raw GNSS pseudorange, carrier-phase, and Doppler measurements with IMU preintegration and LiDAR landmark factors. A robust GNSS module performs fault detection and exclusion and carrier-phase ambiguity resolution, further improving reliability. Experiments on a public benchmark and proprietary urban driving sequences demonstrate that GLINS achieves improved estimator consistency compared to representative baselines and delivers high-precision, drift-free trajectories across diverse urban scenarios.
BibTeX
@inproceedings{ral2026_glinsgnsslidarin,
title = {GLINS: GNSS-LiDAR-INS Integrated Navigation System},
author = {Jiahui Liu and Cheng Chi and Xin Zhang and Binlin Zhang and Dongen Li and Xingqun Zhan and Marcelo H. Ang},
booktitle = {RA-L 2026},
year = {2026}
}