A New Algorithm of Tightly-Coupled GNSS/INS Integrated Navigation Based on Factor Graph

摘要

GNSS/INS tightly coupled system has become the highlight of integrated navigation system because of its proper computational complexity and superior navigation performance. However, frequent short-time loss of lock of GNSS signal has impacted on the positioning accuracy and robustness of the tightly coupled GNSS/INS integration greatly in the city canyons and other complex environment. The typical tightly coupled integration algorithm, such as Extended Kalman Filter, will be often divergent in the case of measurement outlier or fault. In this paper, the joint weight matrix of the internal parameters of GNSS receiver is proposed. The pseudo-range and pseudo-range rate measurement covariance matrixes are adjusted by the joint weight matrix in real time. The improved factor graph algorithm proposed in this paper has high navigation accuracy and system stability when the GNSS signal is outlier and frequent short-time loss of lock. Simulation and experiment results show that the improved factor graph algorithm has low computational complexity and higher stability than classical Extended Kalmam Filter algorithm and factor graph algorithm, and is more suitable in the actual engineering fields.