Swarm based construction of large scale maze-like environments

This paper presents two approaches to pattern based exploration and mapping of large scale, maze-like environments. These approaches fit the needs of single robot as well as multirobot exploration. A further aspect is the subsequent localization of participating robots within previously generated maps. The construction of these maps relies mostly on the extraction and merging of environmental describing patterns. Two different algorithmic solutions are presented within this work. Both solutions were evaluated by either real hardware implementations or exhaustive simulations.