Automatic extraction of indoor navigation graph from building information model

Abstract

There is a growing demand for accurate and reliable indoor location-based service (indoor-LBS) to help pedestrians to navigate in an unfamiliar indoor environment. Accurate indoor navigation graph is crucial for indoor pedestrian navigation. Manual methods used to derive the indoor navigation graph is time-consuming, labor-intensive, and easily affected by human error. Automatic method of generating the indoor navigation graph is urgently needed for the market on indoor navigation application in high demand. A Building Information Modeling (BIM) model that contains rich geometric and semantic information is the ideal source for automatic generation of the indoor navigation graph. However, most of the information in the BIM model is for depicting architectural component, which is not directly related to indoor navigation. The data structure of the BIM model is also too complex for facilitating indoor navigation as BIM model is not design for indoor navigation applications. Data transformation is required to derive navigation information from BIM models. Automatic decomposition of space into fine-grained subspaces is a long-standing problem for the automatic generation of navigation graph. The open indoor space and closed indoor space have significant differences regarding their shapes and dimensions. The closed indoor space is often regarded as the network space, and the open indoor space is viewed as the scene space. Space decomposition is the prerequisite for the implementation of different modeling strategies to various types of indoor space. However, at present, there is no clear subdivision criterion to guide the automatic decomposition the indoor space. This study proposed an automatic polygon decomposition method for generating a proposed hybrid skeleton-grid navigation graph (HSGNG) that combines grid graph with skeleton graph from the BIM model. HSGNG is built upon the automatic identification of open indoor space and closed indoor space. A clear space decomposition criteria is provided which is based on the initial map matching rate and indoor positioning accuracy. These two kinds of indoor spaces are transformed into two different types of graph models and link together to form the hybrid graph model for indoor pedestrian navigation. The information required from BIM models for indoor navigation is identified. An automatic method is proposed to extract the navigation information in the BIM model. The obtained navigation information from the BIM model is then transformed into the data format supported by the GIS platform. The manipulation of the data in the BIM model with the Application Programming Interfaces (APIs) provided by the BIM platform is demonstrated. The proposed algorithm is tested with a BIM model of a large shopping mall. Experimental results show that the proposed method can automatically generate the HSGNG model from BIM model with different distance threshold values. This study explicitly quantifies the initial map matching problem with the graph-based navigation model. Mathematic definitions of open indoor space and closed indoor space are provided to facilitate the automatic decomposition of indoor space with different distance threshold value. HSGNG model is designed for indoor pedestrian navigation based on the decomposition result of indoor space. An automatic approach is proposed to automatically generate the HSGNG model from the BIM model.