Indoor Map Acquisition System Using Global Scan Matching Method and Laser Range Scan Data

Abstract

Simultaneous localization and mapping (SLAM) is the latest technique for constructing indoor maps. In indoor environment, a localization method using the features of the walls as landmarks has been studied in the past. The past study has a drawback. It cannot localize in spaces surrounded by featureless walls or walls on which similar features are repeated. To overcome this drawback, we developed an accuracy localization method that ignores the features of the walls. We noted the fact that the walls in a building are aligned along only two orthogonal directions. By considering a specific wall to be a reference wall, the location of a robot was expressed by using the distance between the robot and the reference wall. We developed the robot in order to evaluate the mapping accuracy of our method and carried out an experiment to map a corridor (40m long) that contained featureless parts. The map obtained had a margin of error of less than 2%.