Research On GNSS/INS Integrated Navigation Algorithms Based On Factor Graph Theory

With the rapid development of intelligent driving vehicles,the requirements for high-precision,low-latency,and high-robust navigation of integrated navigation systems are getting higher and higher.However,single navigation systems all have limitations:the positioning accuracy of Global Navigation Satellite System(GNSS)can be severely degraded in urban canyons with high-rise buildings.Inertial Navigation Systems(INS)can have error accumulation over time.Therefore,the combination of GNSS and INS can improve the stability of the integrated navigation system.However,the localization accuracy of traditional Bayesian filtering algorithm is limited.At the same time,GNSS signal interruption may occur when passing through tunnels or high-rise buildings.Therefore,this paper takes the integrated navigation system of intelligent driving vehicles as the research object,and conducts research on the integrated navigation algorithm based on factor graph theory for the accuracy of traditional algorithms and the navigation problem in the absence of GNSS signals.The main work of this paper is as follows:(1)Aiming at the positioning accuracy and robustness of the integrated navigation system,this paper constructs the system model based on the extended Kalman filter algorithm,the unscented Kalman filter algorithm and the factor graph algorithm respectively.The differences between the three algorithms in the joint optimization process are analyzed through theoretical derivation.Subsequently,this paper proposes a joint GNSS/INS optimization algorithm based on factor graphs,with the objectives of minimizing the positioning error and maximizing the robustness of the system,and solving for the navigation state by multiple iterations,relinearization and effective use of historical information.Finally,simulation experiments are conducted with real data in two urban canyons to verify that the proposed algorithm can achieve on-demand use of different sensing information with high localization accuracy and robustness at low complexity.(2)To address the problem of GNSS interruption during the travel of intelligent vehicles,this paper investigates the application of Autoregressive Integrated Moving Average Model(ARIMA),moving average model and long short-term memory model in predicting GNSS position information,respectively.Then,a GNSS sequence prediction algorithm based on the ARIMA model is proposed,which minimizes the prediction error by optimizing the model parameters and selecting the optimal sequence.Finally,it is experimentally verified that the proposed algorithm can effectively predict GNSS location information and has superior prediction capability than other models.(3)For the problem of navigation accuracy degradation and discontinuity of the combined navigation system after GNSS interruption,this paper further proposes the combined GNSS/INS navigation algorithm with ARIMA model assistant factor graph.After a GNSS interruption,the GNSS information is predicted using the saved ARIMA prediction model,and the prediction results are jointly optimized with INS information via factor graphs to minimize the impact of the interruption problem on the navigation system.Finally,simulation experiments are conducted with real measurement data to verify that the proposed algorithm can effectively cope with GNSS interruptions in harsh environments at low cost and ensure the continuous and high-precision output of the navigation system.