Simultaneous Mobile Robot Localization and Mapping using an Adaptive Curvature-based Environment Description

摘要

This paper presents an algorithm for simultaneous localization and mapping (SLAM) of office-like environments to use with cnventional 2D laser range finders which is based on the extended Kalman filter (EKF) approach. This system employs the set of landmarks extracted from a novel curvature-based environment description. Landmarks include straight-line segments and corners, defined as the intersection of previously detected line segments. Therefore, these corners can be associated to real features of the environment or to virtual ones. In order to provide precise feature estimation, uncertainties will be represented and propagated from single range reading measurements to all stages involved in the feature estimation process. Experimental results provided by the EKF-SLAM algorithm show the ability of the proposed set of landmarks to correctly characterize structured environments.