Global Position system (GPS) is a positioning system with superior long-term error performance, while Dead Reckoning (DR) system has good positioning precision in short-term, through advantage complementation, a GPS/DR integration provides position data with high reliability for vehicle navigation system. This paper focuses on the design of the embedded GPS/DR vehicle integrated navigation system using the nonlinear Kalman filtering approach. The signal's observation gross errors are detected and removed at different resolution levels based on statistic 3sigma- theory, and navigation data are solved with Extended Kalman filter in real-time, the fault tolerance and precision of the vehicle integrated navigation system are improved greatly.