LS-SVM Assisted Multi-rate INS UWB Integrated Indoor Quadrotor Localization Using Kalman Filter

This paper focuses on the problem of positioning accuracy degradation caused by inconsistent sampling frequencies of INS/UWB navigation system. In order to achieve the same sampling frequency of INS and UWB, this paper proposes a data fusion algorithm combining extended Kalman filter (EKF), locally weighted linear regression (LWLR), least squares support vector machine (LS-SVM). First, during the UWB data sampling interval, the UWB data are fitted by LWLR, and the fitted UWB data and INS data are fused by EKF. Then, estimation error of EKF is optimized by LS-SVM. At last, the simulation results indiciate data fusion algorithm restrains divergence problem in the UWB sampling interval. And the positioning accuracy of indoor quadrotor INS/UWB navigation system has been increased through the algorithm.