A Novel FGO-RTK Method for Reliable Urban GNSS Positioning: Enhanced by Ambiguity Outlier DIA

Global Navigation Satellite System (GNSS) positioning methods based on factor graph optimization (FGO) have received significant attention. Compared to traditional filter-based approaches, these methods demonstrate superior positioning performance in complex urban environments. In existing FGO-based real-time kinematic (RTK) methods, although multi epoch joint solution strategies are employed, ambiguity resolution and propagation processes are only performed at the adjacent epoch level, failing to utilize the information from the entire window segment. The constant and integer properties of carrier phase ambiguities during multi-epoch optimization have not been fully explored and utilized. We propose a novel FGO-RTK framework that incorporates an ambiguity outlier detection, identification, and adaptation (ADIA) mechanism. First, unreliable solutions are detected through ambiguity-domain and state-domain checks. Second, potential cycle slips are identified via forward-backward window information repair. Then, the factor graph is adapted using cycle slip correction factors. Static and kinematic test results demonstrate that the proposed method reduces positioning root mean square (RMS) error from over 1 m to approximately 20 cm, improves the ambiguity fixing rate by over 10%, and delivers more robust and reliable solution in real urban environments.