Source strategy analytics shows that people spend 80-90% of their time indoors, while 70% of cellular calls and 80% of data connections originate indoor. Many indoor locations are very complicated in structure and from the navigation point of view, e.g., malls, exhibition centers, museums, etc. We propose an indoor localization method to simplify navigation for indoor locations using mobile augmented reality. In this project, we use Wi-Fi radio field fingerprinting to approximate the current location of the user, and we use computer vision to significantly improve the accuracy of our localizations results. Using mobile augmented reality, we create a 3D visualization of the navigational information for the user. The outcome of this project will be an SDK for mobile indoor localization and an Android app. This project involves machine learning, data mining, computer vision using OpenCV, and Android programming. The students are expected to learn many theoretical and practical techniques in addition to programming mobile devices.
Required Skill: Java and NDK for Android programming and basic C++ programming
The students will learn the techniques for indoor localisation using both WiFi fingerprints and visual methods. They will get intensive training in Android programming and they will learn how to conduct system research.