An Indoor Mobile Robot Localization in perspective of Analysis and Performance using Unscented Kalman Filter

This paper describes a method in an indoor environment for the estimation and position, using an Unscented Kalman Filter (UKF). The UKF algorithm applied for the position estimation proposing a new measurement uncertainty model that fixes the error covariance according to the distance measurement. In addition, this approach sets the non-diagonal component of the error covariance matrix for the uncertainty of the speed information and the measurement uncertainty to a value other than zero. This method is evaluated through an experiment using a wheel-type mobile robot with an LRF sensor in an indoor environment. In this experiment, we differentiate the estimation execution of the proposed approach with a conventional method that does not employ an adaptive uncertainty model. Moreover, the results improved the estimation performance by setting the non-diagonal component of the error covariance to a value other than zero. The main emphasis of this paper is to implement a practical UKF method for location estimation of a mobile robot and analyze it with better performance.