The 3D localization algorithm of INS/UWB based on PF assisted by LS-SVM

Abstract

In indoor positioning, Inertial Navigation System (INS) and Ultra Wide Band (UWB) combined navigation and positioning are generally used instead of Global Position System (GPS) technology.For fusing the INS and UWB data seamlessly, Least Squares Support Vector Machines (LS-SVM) assisted Particle Filter (PF) filtering algorithm is presented in this paper. LS-SVM is used to map each state in the filtering process, and the particle filtering algorithm is used to optimize the navigation model. By using the proposed algorithm, the distance between INS and UWB reference node can be fused, and the measured data of INS can be corrected, and LS-SVM assisted PF filter is more stable. Moreover, an experiment has been done to confirm the proposed scheme. Experimental results indicate that the model designed about LS-SVM assisted PF is efficient to fuse INS and UWB data tightly. By comparison with PF, the proposed algorithm is able to decrease the error of position and add the localization accuracy.