Abstract: The GPS/INS loose integrated navigation system applied in UAVs is affected by external noise interference, causing the measurement noise to constantly change in the filtering process, so that the filtering accuracy is reduced. Aimed at this problem, an algorithm based on variational Bayesian extended Kalman filtering (VB-EKF) is proposed. The algorithm used EKF to expand the state function and measurement function of the nonlinear system into a linear equation, and the data of two different navigation systems were fused, so as to avoid the divergence of navigation and positioning of the single system. Considering the time-varying characteristics of measurement noise in the combined system, variational Bayesian algorithm was introduced to improve and effectively solve the problem of the system filtering accuracy decline. The simulation results show that compared with EKF algorithm, VB-EKF algorithm can effectively improve the filtering stability, thereby improving the accuracy of system navigation and positioning.