Robust Factor Graph Optimization for GNSS/INS Tightly Coupled Integration: A Flexible Strategy for Urban Navigation Resilience

Intelligent transportation systems, including autonomous vehicles, require precise and reliable navigation and positioning systems to ensure safety. However, global navigation satellite system (GNSS) observations are degraded by signal interference, such as obstructions and reflections from high-rise buildings. Navigation methods struggle to effectively process GNSS outliers caused by multipath effects and non-line-of-sight (NLOS) reception in urban environments under changing scenarios. To address this issue, this study proposes a robust factor graph optimization (FGO) approach based on a flexible strategy. Initially, a quality evaluation method utilizing the Mahalanobis distance of pseudorange residual test statistics is developed, enabling smooth and dynamic adjustment of the measurement fusion weights for each satellite. Furthermore, by integrating both satellite quantity and quality data, a flexible strategy is proposed that fully accounts for the differences in measurement quality and the reference values of each satellite across different scenarios. This integration with the navigation environment improves the robustness of the tightly coupled navigation system. Three vehicle-mounted experiments in urban canyon environments were conducted, and the results indicate that, compared to conventional methods, the root mean square errors of horizontal positioning were reduced by 13.24%, 26.06%, and 33.92%, respectively. The proposed method effectively mitigates the impact of GNSS outliers and enhances the reliability of GNSS-based navigation systems for applications such as autonomous vehicles, unmanned aerial vehicles (UAVs) and urban transportation systems.