For visual navigation purposes, a Simultaneous Localization and Mapping (SLAM) system is required in many robotic applications. A SLAM system estimates a map of an unknown environment and the position of the robot in the map. Current openly available SLAM systems assume, that the environment is mostly static. Yet, in real world environments there are many moving objects, like humans and cars. Thus, to use an openly available SLAM system effectively as real-world application, it should be able to estimate a map and the position of the robot despite such moving objects. Therefore, adapting an openly available SLAM system by an algorithm dealing with moving objects would expand its range of application widely. Different approaches, adapting an openly available SLAM system are implemented and evaluated regarding their performance.
Ausführliche Beschreibung der Arbeit. Hard- und Software Dokumentation. Bedienungsanleitung. Links. Fotos. usw.