Invitation to this weeks public Colloquium Talk on Indoor Mapping by Michael Peter (Stgt.)

We cordially invite anybody this week to the first Colloquium Talk in this semester of our “5 years anniversary series”.
The title is:
Indoor mapping by reverse-engineering of existing plans and analysis of pedestrian traces
and the presentation is given by Michael Peter from Stuttgart University (IFP) on Thursday, Apr 23, 2015, 4.15 pm.
The location is Department of Geography, Heidelberg University, Lecture Hall, Berliner Straße 48 (EG).

In the recent years, an increased interest in indoor positioning and mapping could be noticed, in the scientific community as well as with advertising companies and commercial map providers. Due to the vast amount of unmapped indoor areas, the ease of frequent rearrangements, and the high manual effort of current indoor mapping methods, a transfer of the approaches in use by Volunteered Geographic Information communities to indoor mapping would be preferable. The mapping concept used in OpenStreetMap bases on GPS traces as well as aerial images released to the community by their owners. These two data sources can be replaced for the mapping of interior structures by photographed evacuation plans and pedestrian traces resulting e.g. from a foot-mounted, inertial measurement based positioning system. The talk will give a review of related work in indoor positioning and mapping, before presenting an image processing pipeline developed for the extraction of polygonal models from photographed evacuation plans, including the geo-referencing by automatic mapping to a model of the building’s external shell. Furthermore, methods for the accuracy improvement of the pedestrian traces using the coarse model derived from the photographed evacuation plan will be presented. It will be shown how the traces can be employed for the semi-automatic and automatic semantic annotation and geometric refinement of the coarse model, e.g. by combining them with a hand-held low-cost range camera. Lastly, an indoor mapping method solely based on pedestrian traces and supported by a grammar for building interiors will be presented.

GIScience HD