A New Method of Seamless Navigation for the INS/WSN-Integrated System

Abstract

This work presents a new method using Kalman filter (KF) and Least Squares Support Vector Machine (LS-SVM) for the inertial navigation systems (INS)/wireless sensors network (WSN) integrated navigation. In this mode, when the ultrasonic-based WSN is working well, LS-SVM is trained for the mapping between the position measured by INS and the corresponding error. Once the ultrasonic-based WSN is outage, the LS-SVM is used to predict the error of position, which is the unavailable measurement vector of the integrated filter when the ultrasonic-based WSN is outage. Thus, the filter in this mode is able to work where there is no data from the ultrasonic devices. The results show that the proposed method is able to provide continuous navigation information when the data of indoor positioning system is outage, and it is effective to reduce the probability of the estimating outliers.