ZibJana: A Sensor Fusion Framework for Improved Localization

Global Positioning System (GPS) is probably the most widely used technique for localization these days. While the location information obtained from GPS is often reliable enough for applications like Tomtom and Garmin in giving driving directions on the road, there are many other applications out there (for example, augmented reality applications), which demand more accurate position information than what can be obtained from GPS. In this paper we describe ZibJana, a localization application we developed on htc magic smart phone running Android 1.6. ZibJana collects sensor information from smart phone’s built-in GPS receiver, camera, and accelerometer. Then applying Kalman Filter it combines those data from different sources to obtain smart phone’s location estimation more accurately than what is obtainable from GPS only. Major contributions of this work include a sensor fusion framework for accuracy gain in GPS position estimation. This will be useful for developing augmented reality applications, ubiquitous games, and many other applications which require very fine level of accuracy in position estimation.