For this reason, the first step has been to compare the results of the former algorithm of dead reckoning with the first implementation of Kalman filtering with Q constant. The results are reported in Fig. 6, where the typical trend path is reported together with the trend of the mean error (estimated on the basis of 8 indoor navigations) on the same path of Fig. 6a. In Fig. 7, navigation screens taken during the navigation are depicted. The results depicted in Fig. 6b for Kalman (2) show a relevant improvement with respect to the basis solution of dead reckoning[5].