The multi-path effect caused by occlusion and small number of visible stars in the vehicle global satellite navigation system during driving result in poor positioning accuracy, an interactive multi model vehicle integrated navigation algorithm based on EM is proposed. In this article, the mixed Gaussian distribution model is used to describe the error distribution of GNSS multipath effect, and the SINS / GNSS integrated navigation subsystem information fusion method based on EM is proposed to estimate the offset error of multipath effect. The SINS / OD integrated navigation model based on zero speed constraint is formulated. Meanwhile, the interactive multi model algorithm is utilized to realize the interactive fusion of navigation models in the case of GNSS signal loss, which improves the accuracy of the vehicle integrated navigation system. Vehicle experiment results show that the proposed algorithm can effectively improve the navigation accuracy under the conditions of GNSS multipath effect and information loss. When the offset of Gaussian mixture model of multipath effect is 10 m, the offset estimation error is less than 0. 5 m, and the maximum horizontal positioning error is 2 m, which is 84. 62% lower than that of traditional interactive multi model algorithm.