Analysis of the Performance of Extended Kalman Filtering in SLAM Problem

Abstract

Simultaneous Localization and Mapping (SLAM) is an essential task for autonomous robot navigation. Due to the presence of sensor noise and odometry error, pose estimation requires a stochastic filter in SLAM problem. In this paper, a differential drive model based EKF-SLAM problem is studied. From simulation and experimental results, it is observed that EKF, for the system considered, fails to compensate the initial heading error. An observability analysis of the underlying non-linear system and EKF model has been provided. The effect of observability on the inconsistency of EKF for the SLAM problem is studied.