Simultaneous localization and mapping in domestic environments

Abstract

This paper describes an accurate and robust algorithm for simultaneous localization and map building (SLAM). The objective of SLAM is to enable a mobile robot to build an internal representation (map) of an unexplored environment while simultaneously using that map to navigate. An extended Kalman filter (EKF) approach is used to process the information acquired by the sonar sensors mounted on the robot. A method for recovering from failures of the SLAM algorithm is presented for increasing the robustness of the general EKF method. Real experiments are presented considering a Nomadic SuperScout mobile robot navigating in a domestic environment.