Extended Kalman filter-based mobile robot localization with intermittent measurements

In this paper, a theoretical study on extended Kalman filter (EKF)-based mobile robot localization with intermittent measurements is examined by analysing the measurement innovation characteristics. Even if measurement data are unavailable and existence of uncertainties during mobile robot observations, it is suggested that the mobile robot can effectively estimate its location in an environment. This paper presents the uncertainties bounds of estimation by analysing the measurement innovation to preserve good estimations although some measurements data are sometimes missing. Theoretical analysis of the EKF is proposed to demonstrate the conditions when the problem occurred. From the analysis of measurement innovation, Jacobian transformation has been found as one of the main factors that affects the estimation performance. Besides that, the initial state covariance, process and measurement noises must be kept smaller to achieve better estimation results. The simulation and experimental results obtained are showing consistent behaviour as proposed in this paper.