Computationally Efficient Mapping for a Mobile Robot with a Downsampling Method for the Iterative Closest Point

Abstract

This paper proposes a computationally efficient method for generating a three-dimensional environment map and estimating robot position. The proposed method assumes that a laser range finder mounted on a mobile robot can be used to provide a set of point cloud data of an environment around the mobile robot. The proposed method then extracts typical feature points from the point cloud data using an intensity image taken by the laser range finder. Subsequently, feature points extracted from two or more different sets of point cloud data are correlated by the iterative closest point algorithm that matches the points between the sets, creating a large map of the environment as well as estimating robot location in the map. The proposed method maintains an accuracy of the mapping while reducing the computational cost by downsampling the points used for the iterative closest point. An experimental demonstration using a mobile robot test bed confirms the usefulness of the proposed method.