Localization And Navigation In Complicated Environments Using UWB And IMU

It is hard to get accurate position due to serious attenuation of positioning signals and even interruption in complicated environments.A single kind of localization method can not solve this problem fundamentally,so combining advantages of different kinds of localization sensors is an effective method.Ultra-wide band(UWB)signals have high accuracy of distance measurement and ability to penetrate obstacles.And inertial measurement unit(IMU)presents new opportunities for position tracking independently through its accelerations and angles.Thus taking advantages of UWB and IMU to achieve high accurate localization and navigation is a good idea.Although UWB signals have ability to penetrate most of obstacles,positive nonline-of-sight(NLOS)error will be introduced.For the problem,this dissertation presents the K-Nearest Neighbor(K-NN)to mitigate NLOS error,which is based on machining learning and low complexity in computation.Comparing to Least Squares Support Vector Machines(LS-SVM),although the two algorithms can effectively mitigate NLOS error,the K-NN significantly saves the computational time for training samples and is more practical.This dissertation also designs a UWB localization system to demonstrate the UWB position accuracy and the K-NN algorithm in practical application.On the basis of the above research,we use Extended Kalman Filter(EKF)to fuse measurements of UWBs and IMU to achieve high accurate localization and tracking in the lack of distance measurements between agent and anchors.For the nonlinear problems of observation equations,the EKF utilizes the first-order Taylor series expansion to linearize nonlinear functions.The simulation results show that the 90% position error is under 0.6 meters in 40 m ? 40 m scenario through joint localization even only one lineof-sight(LOS)distance is obtained.The study demonstrates that information fusion between heterogeneous sensors is an effective method to improve the position accuracy in complicated environments.The K-NN for NLOS mitigation and the joint localization fusing information of UWBs and IMU proposed in this dissertation provide theoretical reference for improving positioning accuracy in complicated environments.