In pedestrian inertial navigation

zero velocity detection is premise to reset velocity errors and estimate navigation error. Aiming at problem that time interval of zero velocity is short and it is difficult to detect in pedestrian moving process

a zero velocity detection algorithm based on human foot movement characteristics is proposed

where walking motion is abstracted into a hidden Markov model with 4 hidden states and 15 observations

and construction mechanism of model is described. The model parameters are trained and optimized by Baum—Welch algorithm and detection accuracy is improved. Experimental results show that the proposed method has better effect on zero velocity detection

and the positioning error of pedestrian inertial navigation system using this method is about 0. 73 % of the travel distance

and the positioning precision is high.