This paper applies two Artificial Intelligence (AI) techniques, fuzzy logic and expert system, to enhance the Kalman filter-based MEMS INS/GPS integration. For better INS error control, the expert knowledge on vehicle dynamics is utilized to simplify dynamics models and to extend measurement update schemes in the velocity filter. To optimize position fusion, a fuzzy inference system is developed to provide GPS signal degradation information for modification of the innovation-based adaptive measurement covariance in the position filter. The effectiveness of the proposed AI-based enhancement methods is demonstrated through several field tests.