Mikro hava araçlarının görsel etkin poz takibiyle kontrolü

Autonomous control of unmanned aerial vehicles requires fast and accurate measurements of vehicle's pose and velocity in real-time. Moreover, for indoor applications, collusion free control of robots demands knowledge of environment's 3D model. In this study, a visual odometry process based on pixel intensities is proposed for unmanned vehicle applications. In contrast to conventional methods, the proposed method does not uses visual features such as corners, edges or surfaces. In this approach, the motion estimation is obtained without using a feature descriptor or matching algorithm. Thus, the odometry system has better performance in real-time. For the real world application, camera's color and depth images are used. In order to minimize the error, an energy function has utilized. The process has been tested with an indoor dataset and pose of camera is tracked successfully.