Intelligent generation of indoor topology (i-GIT) for human indoor pathfinding based on IFC models and 3D GIS technology

Abstract

Abstract Indoor topology is the basis for most indoor location-based services, such as wayfinding in shopping malls, escape route planning, and passenger transfer in transport stations. Currently, however, indoor networks for wayfinding by people have not yet been developed to a standardized representation as street maps have been, but many studies have attempted to develop indoor network models for indoor LBS. A complete indoor network includes both floor-level paths and non-level paths, such as stairs, elevators, escalators, ramps, and so on. Most existing studies, however, have explored floor-level paths; few studies have considered non-level paths. Many studies have used 3D building models based on Industry Foundation Classes (IFC) to develop algorithms for generating paths, but most of them require the IFC model to contain semantic elements for the spatial relationships among building components, which means extra efforts are needed when preparing the IFC Model from the original building model. Moreover, most algorithms dealing with floor-level paths use simplified floor plans as the input, which are different from the real data retrieved directly from IFC models. All of these requirements and assumptions add to the cost of producing an indoor network. In this study, we propose an approach called i-GIT to produce a graph-based indoor network including floor-level and non-level paths from IFC-based building models. i-GIT requires only the geometric information of IFC data models to automatically identify indoor space boundaries as well as to produce six categories of indoor paths. The innovation of this study is that it includes novel algorithms to produce non-level paths, and introduces the polygon regularization on indoor space boundaries to reduce the number of excessive nodes before applying Constrained Delaunay Triangulation to produce level paths. Three campus buildings and a Taiwan High Speed Rail (THSR) station were chosen to validate the proposed approach. Multiple OD pairs in one of test cases were randomly selected to find the shortest routes using the paths generated by i-GIT. Compared with paths manually drawn using the principles of Medial Axis Transform (MAT) and the route lengths from actual measurement, the average path availability is higher than 96%, while the average error of route lengths based on generated paths is approximately 5%. The results showed that i-GIT was able to accurately and effectively produce a complete indoor network.