A Construction Method of Pedestrian Road Networks for Agent-Based Evacuation Simulation Koichi Asakura, Yuki Ikeya and Toyohide Watanabe Abstract We propose a method for constructing pedestrian road networks with pub- licly available geospatial data sets. Our method consists of two algorithms: estimation of road widths and deployment and connection of pedestrian road segments. In the two algorithms, we used two types of geospatial data: the set of road center line segments and the set of road edge segments. The experimental results show that the estimation algorithm of road widths can calculate precise road widths and the con- struction algorithm of pedestrian road networks works well even for road networks with a complicated connecting relationship. Keywords Pedestrian road network · Estimation of road width · Geospatial data · Geographic information system (GIS) 1 Introduction Pre-planning for evacuation is necessary to prevent people from being affected by disasters such as big earthquakes. Japan is known as an earthquake-prone country. Many local governments in Japan provide evacuation information as hazard maps [2, 6]. We can receive information from hazard maps about seismic intensity, ground liquefaction, etc., for a potential earthquake. Many roads in disaster areas cannot be used for evacuation because of earthquake damage. Thus, the flow of people in the area is carefully taken into account in order to make good evacuation plans. There has been a lot of research done on agent-based evacuation simulations [3, 7, 8]. Road networks are often used in urban-level simulations to calculate the move- ments of agents [1, 9]. Connection relationships among intersections and capacities K. Asakura (B ) · Y. Ikeya Daido University, Nagoya 457-8530, Japan e-mail: asakura@daido-it.ac.jp T. Watanabe Nagoya Industrial Science Research Institute, Nagoya 464-0819, Japan e-mail: watanabe@nagoya-u.jp © Springer Nature Singapore Pte Ltd. 2019 Y.-W. Chen et al. (eds.), Innovation in Medicine and Healthcare Systems, and Multimedia, Smart Innovation, Systems and Technologies 145, https://doi.org/10.1007/978-981-13-8566-7_52 573