INFDok - Dokumentenvolltextserver des Fachbereichs Informatik

Kurzfassung auf Englisch:

Heavy transports are very important undertakings that due to their cost and risk factors require careful planning. The project 3D-Heavy Transport Route Finder (3D-HTRF) was developed by Gustav Seeland GmbH and SICK AG, which resulted in a car equipped with laser range finders and highly accurate, but expensive inertial navigation systems to build maps of possible transportation routes for quicker and cheaper scouting of routes. This work takes a look at the accuracy of the generated maps, as well as suggest and evaluates a method to further improve the maps. FastSLAM is a method from the field of robotics and originally developed for localization and mapping. We adopted the FastSLAM algorithm for 3D application and for using it to improve the match of laser scan data in a post processing step. The evaluation shows that, while the basic concept works, FastSLAM cannot help improving the maps of the 3D-HTRF project using the existing hardware setup. The reasons why are discussed and suggestions on how to overcome the problems faced are made.