TY - JOUR TI - Advances in instantaneous and dynamic localization in indoor environments DO - https://doi.org/doi:10.7282/T3GX4D5G PY - 2014 AB - The knowledge of the exact location of an object or person is an important enabler for a wide variety of applications. Localization, itself, has a long and distinguished history from sextants to the geodesic tools of modern civil engineers. The recent general availability of global positioning systems has enabled the development of new applications from car navigation and emergency assistance systems to location targeted advertisements. However these systems are largely limited to outdoor spaces preventing the accurate determination of location inside of a building. In this dissertation, we consider the issues of both the instantaneous and dynamic localization of people in an indoor environment where GPS signals are not available. We begin by studying instantaneous localization and the variety of new techniques that share the common approach of utilizing a number of fixed reference points with known locations in order to determine the location of the target. Since indoor environments usually prevent a direct line of sight from the target to a known point we need to infer the target’s location from secondary measurements with varying reliability. We then outline a new method for instantaneous localization whose output not only provides the location of the target but also the confidence of the localization based on the environmental variability. We then examine how these factors extend to the complex scenarios of dynamic localization. In these scenarios instantaneous localization steps would need to be performed at an unrealistically high frequency in order to extract the desired smooth trajectory of a moving object. We find that by passing the results of these measurements over time through a sampling-importance-resampling particle filter can reduce the influence of noise and allow interpolation to predict the current, and sometimes future, location of the target. This particle filter utilizes a probabilistic reasoning technique to extract continuous tracking information from the periodic instantaneous location input as well as to integrate knowledge about the environment and the target’s capabilities in order to provide it’s location. We then present the experimental results from scenarios taken from multiple indoor locations demonstrating the accuracy and precision of the localization method. KW - Computer Science KW - Indoor positioning systems (Wireless localization) LA - eng ER -