3D reconstruction of a static indoor environment by fusion of sonar and video data

Abstract

This paper proposes a new system to construct a 3D model of a real in door environment by using merging sonar and video data. Data is collected by a moving robot exploring the environm entand the 3D model is constructed on-line. Hence. while the robot navigates in the real environm ent, any remote PC may navigate in a virtual copy of such environment. The main advantage of the proposed system is that the combination of two different sources of information allows fast and computationally cheap construction of the model. The system has been successfully tested in unstructured realenvironments.