A novel strategy of localization based on EKF for mobile robot

This paper focuses on the accurate localization based on WSN (wireless sensor network) for indoor mobile robot using NLOS (non-line of sight) identification. For the traditional localization three measurement circles from observation are usually obtained to estimate the robot position, which brings up the larger errors in positioning. In order to resolve this problem, the position distributions calculated from multiple measurements are used to estimate the mobile robot location. Intersections of each two measurement circles are computed and distributions are fitted by using Gaussian. Expectations are taken as the observation values for extended Kalman Filter (EKF), which is applied to optimizing of localization. Also NLOS identifications are proposed to find the potential NLOS measurements which will severely deteriorate observation results. Efficiency of the presented localization algorithm with NLOS identification is illustrated via simulation experiments.