Space Models
Space models for indoor navigation must include both a geometric and a topological representation of the physical space.
Geometric Models
To aid the perception and visualization of indoor spaces it was necessary to introduce the visualization of floorplans into the system. To make this possible, a process was created to convert the existing SVG floorplan files into GeoJSON files. These files were then seamlessly integrated and georeferenced in their respective real-world locations and, with the aid of map visualization tools, made available to the end-user so that they could perceive their location within the available indoor spaces.
This technique of visualization also proved useful to aid with the tracking of the progress of creating the radio mapping. This was achieved by adding the location and quantities of the fingerprinting samples in clusters on top of the floorplan layers. Through this method of visualization, it is possible to perceive where the fingerprints are concentrated and which areas need to be further mapped.
The outcomes of the geometric modelling task include:
- a process to generate geometric models of the space in a flexible and easy-to-use format (GeoJSON) from SVG, or other CAD formats, files;
- a webservice to provide access to the geometric model of each floor of each building of the University of Minho, covering the campi of Azurém and Gualtar;
- an Application Programming Interface (API) to facilitate the access to the geometric models, and that can be used in the development of web or mobile applications;
- a web application that provides a Graphical User Interface (GUI) for exploring all the geometric models.
Topological Models
Topological models: A topological model of a space represents how the elements of the space are connected (e.g. a particular classroom is accessible from a particular corridor), and is the fundamental part of a navigation system. While geometric models are often available in one form or another (e.g. the building blueprints in CAD format), topological models are not. Building topological models manually is an expensive and time-consuming task.
The approach adopted in this project was to resort to crowdsourcing, that is, engage the space users in building the topological model using a participatory collaborative method. Using a specially crafted smartphone application <http://where.dsi.uminho.pt/>, users are invited to collect data (read QR Codes, detected steps, orientation) while walking between two locations. The collected data is then used to estimate the corresponding trajectories, which are uploaded to a server for further processing. Merging all the individual trajectories from all the participating users ends up in a “organic” topological model of the space. The benefits of a collaborative approach is that the resulting model is a representation of the routes that frequent visitors actually take between two locations, and also the possibility of having the model being updated as structural changes are introduced into the space.