ROBUST, TIME-CRITICAL, EVIDENCE-BASED ADAPTIVE DATA FUSION
Degree granted by The University of Texas at Arlington
Abstract
Sensors have become inevitable part of many studies and working areas ranging from navigation, transportation and medical applications. A sensor can help a user in a variety of situations including dangerous, inaccessible, time and money-consuming circumstances. Applying multiple sensors simultaneously allows for improving the accuracy of measurement estimates for system states. As an example, a part of this study uses a GPS sensor to increase the accuracy of the position estimation obtained by an IMU in an indoor environment. The same GPS device with position outputs can also be studied to provide a new measuring dimension such as velocity. This way of sensors helping each other is covered in this study to achieve more accurate and robust estimates compared to when they contribute separately. This study is carried out to develop a new way of achieving better estimates of system states. An attempt to indicate the applicability and robustness is also provided for navigation purposes. This work is founded on evidence-based theory wherein pieces of evidence or facts are used to cross-check the outputs given by sources. It also applies the concept of meta-sensing in which proportional weights are assigned to the data from each sensor based on how close the data from each sensor is to the others. This causes the estimates to be more robust and reliable in a real-time environment. The obtained outcomes show more reliable decision-making under the defined scenarios.