The invention discloses a navigation method of a human occupied vehicle. According to the method, firstly, the dynamical model equation of the human occupied vehicle is established. Through utilization of an iterative SRUKF method, the navigation data of the dead reckoning system and the long baseline positioning system of the human occupied vehicle are filtered and estimated, and therefore precise and reliable navigation data is obtained in real time. The method has good timeliness and reliability and can be used for guiding the human occupied vehicle to execute tasks in a deep see environment. According to the method, the navigation precision of the human occupied vehicle can be improved, and the system robustness can be improved.