An indoor pedestrian navigation algorithm based on smartphone mode recognition

In recent years, the rapid development of smartphone-based navigation has been proven to have great application prospects. However most smartphone-based navigation technologies only applicable to a fixed mode, which has a significant decline in user experience. Therefore, we design a multi-mode smartphone recognition method based on neural network. On the basis of it, a heading correction method based on mode-change detection is proposed to reduce the heading angle error of smartphone navigation in multi-mode. In addition, we study an intelligent step length estimation method to improve the accuracy of pedestrian navigation. And the experimental results demonstrate the effectiveness of the proposed methods, i.e. the average heading error and the maximum positioning error obtained by the proposed algorithm is 94.5% and 91.6% less than these of traditional PDR algorithm respectively.