| dc.description.abstract | This thesis presents the creation of a payload system for a small unmanned aircraft, which is used for data collection and testing of detection and tracking theory. Theory for position estimation of on-ground or on-sea objects from a camera from height is presented, along with methods for reducing error associated with camera lens distortion errors, and synchronization errors between GPS/INS and camera measurements. An overview of methods for detecting and classifying objects is presented, along with a method for determining a ship's orientation in the water, allowing for direction (yaw) feedback to a nonlinear observer. It is shown that such a measurement greatly increases the accuracy of the observer. Theory for observing and tracking objects with known and unknown models is presented, using both linear and nonlinear Kalman filters. The observers and other theory were tested in simulations, and will be tested with real-world data after flight tests. The results are concluded in a paper submitted to ICUAS'15 by Leira and myself. | |
