Exploration and mapping of unknown polygonal environments based on uncertain range data

Abstract

We consider problem of exploration and mapping of unknown indoor environments using laser range finder. We assume a setup with a resolved localization problem and known uncertainty sensor models. Most exploration algorithms are based on detection of a boundary between explored and unexplored regions. They are, however, not efficient in practice due to uncertainties in measurement, localization and map building. The exploration and mapping algorithm is proposed that extends Ekman's exploration algorithm by removing rigid constraints on the range sensor and robot localization. The proposed algorithm includes line extraction algorithm developed by Pfister, which incorporates noise models of the range sensor and robot's pose uncertainty. A line representation of the range data is used for creating polygon that represents explored region from each measurement pose. The polygon edges that do not correspond to real environmental features are candidates for a new measurement pose. A general polygon clipping algorithm is used to obtain the total explored region as the union of polygons from different measurement poses. The proposed algorithm is tested and compared to the Ekman's algorithm by simulations and experimentally on a Pioneer 3DX mobile robot equipped with SICK LMS-200 laser range finder.