Navigation of mobile robots using 6DOF SLAM accelerated by leveled maps

Abstract

This article is presenting a method for simultaneous localization an mapping (SLAM) of mobile robots in unknown environments. The method is using 3D range data which are acquired from a continuously inclined laser time of flight ranging sensor. In order to accelerate existing computational time-demanding Iterative Closest Point SLAM methods, the SLAM procedure is using a leveled map of the environment created from the 3D data which serves as a pre-estimation of the optimal translation and rotation in three degrees of freedom. This can eliminate the number of iterations of the final robust six degrees of freedom (6DOF) ICP registration, thus saving a lot of computational time. The criteria of the success of this method are the characteristics of the environment, especially the way the environment is structured, and the appropriate setting of the leveled map extraction. Results from the testing of the method in indoor environment are also presented. © Springer Science+Business Media B.V. 2008.