Abstract:

This paper proposes a graph optimization-based real-time kinematic global navigation satellite system (GNSS) positioning approach, which consists of two stages of factor graph optimization (FGO). The first stage computes float solutions of navigation states including the carrier phase integer ambiguities, where we characterize the time evolution of integer ambiguities with an adaptive ambiguity model to accommodate cycle slips. By exploring the time-correlated constraint inherent in the integer ambiguity evolution, we achieve integer fixation with higher accuracy. The second-stage FGO takes the solutions from the first stage as prior and performs another graph optimization to obtain the fixed solutions of positions and velocities. Monte Carlo simulation results demonstrate that our proposed approach can achieve statistically smaller root mean square error in position estimates compared to Kalman filter- based method and is more robust to cycle slips.