Indoor localization is used to locate people or objects within indoor environments such as buildings and rooms. The major applications of indoor localization involve the expansion of location-aware computing such as monitoring and surveillance in scientific labs, accessibility aids, targeted advertisements, inventory tracking etc. This thesis explores localization within the context of the scientific labs called cleanrooms. The philosophy behind indoor localization is first broken down into three parts: presence detection, movement tracking and proximity sensing. The unique challenges for each of these parts facing the cleanroom environments are evaluated and then solved using specific technologies: 1) RFID (Radio Frequency Identification) is used for presence detection to automatically detect the entry / exit of the person into the cleanroom; 2) BLE (Bluetooth Low Energy) is used to continuously track the movement of the person across multiple cleanrooms; 3) Computer vision through a single monocular camera is used to measure the relative proximity of a person to different areas of interest within a room. Our results with RFID provide 82.5% accuracy for presence detection, BLE yields 2.1m accuracy for movement tracking and 62% accuracy for presence detection while computer vision gives an accuracy of 71% for proximity sensing. A comparative study between the three types of sensors highlights the specific benefits and limitations of each technology. It shows that different sensors will have to be combined for full-scale indoor localization which will act as a building block towards the next generation of computing environments.
PDF
download
878987797
028-85220240
