Factor Graph Optimization-Based GNSS PPP-RTK: An Alternative Platform to Study Urban GNSS Precise Positioning
Shaoming Xin, Jianghui Geng, Li‐Ta Hsu
Abstract
The GNSS PPP-RTK technique which integrates the superiorities of PPP and RTK is increasingly attracting people's attention in recent years. This paper proposes a factor graph optimization (FGO)-based GNSS PPP-RTK method to provide an alternative platform to study urban GNSS precise positioning. Instead of a batch FGO estimator, the proposed method uses the sliding window method to meet the real-time requirement. Pseudorange, carrier-phase measurements, atmospheric corrections at each epoch, and Doppler velocity and atmospheric variations between the adjacent epochs are integrated to estimate receiver positions. We evaluated the performance of the proposed method by carrying out static and kinematic experiments in different scenarios. Two static experiments based on the data from a base station in an open sky environment and a low-cost GNSS receiver in a typical urban environment can both obtain cm-level positioning results. Two kinematic experiments using a low-cost GNSS receiver experiment in two complex urban canyons in Hong Kong achieve the mean horizontal positioning error of the real-time results of 1.606 meters and 4.055 meters, respectively. Compared with existing classic PPP-RTK methods based on Kalman filters, the proposed method improves the availability of the positioning result. It mitigates in part large positioning errors, ensuring that the maximum horizontal positioning error is less than 10 m in complex urban environments.