EKF based SLAM with FIM Inflation

This paper deals with an analysis based on Fisher Information Matrix(FIM) for Extended Kalman Filter based Simultaneous Localization and Mapping(SLAM) problem. We show theoretically that the Cramer Rao Lower Bound is proportional to the number of landmarks, the magnitude of process and the measurement noises. In addition, we propose a method of adding a pseudo Positive semidefinite(PsD) matrix to the Fisher Information Matrix to decrease the computational cost in EKF based SLAM. The simulation results are convincing and realizes the improvement for EKF-based SLAM. Therefore, this method further improves the estimation in comparison with the normal EKF performance.