Buch, Englisch, 268 Seiten, PB, Format (B × H): 148 mm x 210 mm, Gewicht: 406 g
Reihe: Informatik
Buch, Englisch, 268 Seiten, PB, Format (B × H): 148 mm x 210 mm, Gewicht: 406 g
Reihe: Informatik
ISBN: 978-3-8439-3035-2
Verlag: Dr. Hut
This dissertation examines the problem of exploring arbitrary indoor environments with a mobile robot. The goal is to construct a complete and correct map in minimal time that supports service tasks like transportation or surveillance. Based on a common framework for an integrated exploration, four different strategies are developed and evaluated in a virtual office scenario and in its real world counterpart. They differ in the frequency of perception and decision taking as well as in the quality of the derived map. It is analyzed how online map updates and unrestricted navigation paths reduce detours while approaching a next best view position, whether continuously adapting the exploration trajectory increases efficiency, and whether extracting the topology of rooms and doors avoids oscillations between distant open frontiers. A hybrid combination of a frequently updated topological map and a fallback to a highly available grid map yields the best results in cluttered areas with dynamic obstacles.
