Using Crowdsourced Geodata for Agent-Based Indoor Evacuation Simulations

Abstract: Crowdsourced geodata has been proven to be a rich and major data source for environmental simulations and analysis, as well as the visualization of spatial phenomena. With the increasing size and complexity of public buildings, such as universities or hotels, there is also an increasing demand for information about indoor spaces. Trying to stimulate this growing demand, both researchers and Volunteered Geographic Information (VGI) communities envision to extend established communities towards indoors. It has already been showcased that VGI from OpenStreetMap (OSM) can be utilized for different applications in Spatial Data Infrastructures (SDIs) as well as for simple shortest path computations inside buildings. The here presented research now tries to utilize crowdsourced indoor geodata for more complex indoor routing scenarios of multiple users. Essentially, it will be investigated if, and to what extent, the available data can be utilized for performing indoor evacuation simulations with the simulation framework MATSim. That is, this paper investigates the suitability of crowdsourced indoor information from OSM (IndoorOSM) for evacuation simulations. Additionally, the applicability of MATSim for agent-based indoor evacuation simulations is conducted. The paper discusses the automatic generation simulation-related data, and provides experimental results for two different evacuation scenarios. Furthermore, limitations of the IndoorOSM data and the MATSim framework for indoor evacuation simulations are elaborated and discussed.