Indoor Network Generation Using Building Evacuation Maps

Abstract

With the development of indoor positioning technology, there is a growing demand for location-based services such as indoor navigation systems. For such services, it is essential to construct an indoor network of walkable spaces. In this study, we propose a method to construct an indoor network using evacuation maps that can be easily obtained in most buildings. Since the proposed method extracts information for movement represented in the evacuation maps through the deep learning, indoor network can be generated efficiently. In addition, we designed the attribute tables of the network for utilization to actual services directly. We verified the appropriateness of the methodology by constructing an indoor network of parts of the Seoul National University. Keywords: Indoor Network, Indoor Spatial Information, Evacuation Map, Indoor Network Attribute, Conditional Generative Adversarial Network (cGAN)