Indoor-Outdoor Detection and Routing Using Bayesian Networks

Abstract

The purpose of this Master's thesis is to do research on methods for indoor-outdoor detection and to implement an Android application for the navigation through indoor and outdoor environments. Due to the length and complexity of the project, it has been divided in two different parts: the outdoor navigation with the indoor-outdoor detection and the indoor guidance. Here we are going to explain the first part, beginning with the study of recent papers about outdoor positioning and detection techniques. On the one hand the popular method to locate yourself outdoors is the GPS that has proved to be accurate and robust for path planning. On the other hand current studies of indoor-outdoor detection are based on the light intensity, the magnetic field, the cell tower strength, the number of satellites and the signal to noise ratio of the satellites. We have used those parameters to build a Bayesian network that uses their conditional probabilities to get the probability of being outdoors. Then we have implemented the outdoor navigation and the detection algorithm with the indoor guidance in the Android application. The indoor positioning system was provided by IndoorAtlas that uses the geomagnetic disturbances of buildings to locate the user indoors. To do that, the system requires a geomagnetic map of the building that has to be made beforehand. For this reason the application has been built only for a certain shopping center that was already mapped. With all this, the final Android application is able to guide the user from a store inside the mall to a place outdoors and vice versa using a combined indoor and outdoor guidance.