Indoor localization and point cloud generation for building interior modeling

Indoor modeling can be defined as the study of how photorealistic 3D model can be realized by localization and mapping for describing indoor environment which is beyond 2D model while providing useful information on the location. In this paper, a 3D localization method for an IGS(Information Gathering System) is presented for the point cloud generation of building interiors. First, a backpack IGS which is worn by a human operator and gathers information on environment along hallways is introduced. Next, a travel distance of the IGS is defined, which is utilized for estimating 3D poses of IGS. Then a 3D pose estimation method for localization of the IGS is proposed based on the SLAM method composed of two kinds of 2D SLAM methods, which extract the most reliable poses from the sensors on IGS. The proposed method is evaluated on challenging data of various indoor environments, and it is shown that it effectively localizes the IGS in complex indoor spaces with staircases.