Traditional dynamic ZUPT for onboard inertial navigation system (INS) highly requires vehicle maneuvering motion, but motions like turning just has weak constraint to the error on INS. A novel dynamic ZUPT algorithm based on detecting the motion state of the vehicle is proposed in this paper. First, precisely detecting the vehicle motion in real time is designed. Then, the corresponding filter for different motion is used to estimate the state error, in order to improve the adaptability of the proposed algorithm. The experiment results show that, compared with the traditional dynamic ZUPT method, the position accuracy is improved by 160m in 2.6 hours, so the validity of the proposed method is proven.