A Factor Graph Optimization Approach for Orbit Determination

This paper examines the application of Factor Graph Optimization (FGO) to the problem of orbit determination (OD) in spacecraft navigation systems, areas traditionally addressed by methods such as the Extended Kalman Filter (EKF) and Nonlinear Least Squares (NLS). We present a factor graph formulation for OD based on GNSS pseudorange measurements, in cohort with the orbit dynamics model that would define a spacecrafts motion. Through simulations and real-world data from the GRACE-FO mission, the capabilities of FGO are assessed and compared with the EKF. The results indicate marginally improved orbital tracking and estimation accuracy over the EKF, and support the potential of FGO as a viable alternative for the problem of spacecraft OD. The presented factor graph formulation demonstrates its flexibility for OD, and its ability to easily incorporate diverse measurement sources for future work in this domain.