In order to meet the requirements of high performance of INS/GPS integrated navigation system and further improve the calculation accuracy of Kalman filtering algorithm in the integrated navigation position, a reverse data processing and smoothed smoothing fusion algorithm are proposed in this paper. A forward and reverse fusion Kalman filtering algorithm is applied to the loose combination of INS/GPS integrated navigation system. The calculation method of the forward and reverse fusion Kalman filtering algorithm is analyzed and is compared with the normal forward Kalman filter algorithm. In the experiment section, the data collected by the vehicle sensors are used. Through analyzing the errors of the two kinds of algorithms, it is showed that the forward and reverse fusion filtering algorithm has better performance than that of the normal forward filtering algorithm in terms of positioning accuracy.