At present, 87 % of people's activities are in indoor environment; indoor navigation has become a research issue. As the building structures for people's daily life are more and more complex, many obstacles influence humans' moving. Therefore it is essential to provide an accurate and efficient indoor path planning. Nowadays there are many challenges and problems in indoor navigation. Most existing path planning approaches are based on 2D plans, pay more attention to the geometric configuration of indoor space, often ignore rich semantic information of building components, and mostly consider simple indoor layout without taking into account the furniture. Addressing the above shortcomings, this paper uses BIM (IFC) as the input data and concentrates on indoor navigation considering obstacles in the multi-floor buildings. After geometric and semantic information are extracted, 2D and 3D space subdivision methods are adopted to build the indoor navigation network and to realize a path planning that avoids obstacles. The 3D space subdivision is based on triangular prism. The two approaches are verified by the experiments.

The demand for indoor navigation is increasingly urgent in many applications such as safe management of underground spaces or location services in complex indoor environment, e.g. shopping centres, airports, museums, underground parking lot and hospitals. Indoor navigation is still a challenging research field, as currently applied indoor navigation algorithms commonly ignore important environmental and human factors and therefore do not provide precise navigation. Flexible and detailed networks representing the connectivity of spaces and considering indoor objects such as furniture are very important to a precise navigation. In this paper we concentrate on indoor navigation considering obstacles represented as polygons. We introduce a specific space subdivision based on a simplified floor plan to build the indoor navigation network. The experiments demonstrate that we are able to navigate around the obstacles using the proposed network. Considering to well-known path-finding approaches based on Medial Axis Transform (MAT) or Visibility Graph (VG), the approach in this paper provides a quick subdivision of space and routes, which are compatible with the results of VG.