Summary: | 碩士 === 國立成功大學 === 民航研究所 === 100 === In recent year, fixed-wing Unmanned Aerial Vehicle (UAV) system has been a popular platform for Intelligence, Reconnaissance and Surveillance (ISR) missions in both military and civil application. The UAV system equipped with a onboard CCD camera system to conduct such kind of mission. In order to provide better aerial images for missions such as road patrol or river patrol, the UAV have to track the ground path, terrain features or landscapes, It requires the remote sensing capability to provide sufficient navigation information from the onboard CCD camera images for the guidance and navigation of the UAV. Therefore, this thesis incorporates the image processing with the robust image estimation technique, called RANdom SAmple and Consensus (RANSAC) for the ground path tracking. The goal is to provide useful flight information for the UAV to perform the capability of vision-based ground-path tracking as part of the autonomous navigation.
The simulation results indicate that, the combination of techniques of image processing and RANSAC indeed provides an accurate reference heading for the UAV to conduct ground-path tracking flight. The real flight demonstrations were also made on two different UAV systems, Spoonbill and Swallow, UAV systems to ready verify the feasibility of the designed algorithm in flight path tracking using the onboard CCD camera system to conduct the vision based navigation.
|