This thesis contains a system design of a so called ``flying camera'' system. The camera position can be given without using special skills. The system is designed for indoor use, for example to record an event from a flying perspective.
It includes a quadrocopter as a UAV (Unmanned Aerial Vehicle) controlled by a computer. The device carries a high resolution camera transmitting video over Wi-Fi connection. The position and orientation of the camera can be controlled from a PC.
A PCB (Printed Circuit Board) is designed for the UAV to accomplish the required tasks of servo drive, video recording, and stability control. A Gumstix Overo COM (Computer-On-Module) is chosen for high-level computation, camera interfacing, and video compression. An STM32 F4 microcontroller completes the low-level task of stability control, sensor and actuator interfacing.
The position of the UAV is continuously measured using a stereo camera system tracking active markers on the quadrocopter. The images of the stereo vision are processed on a PC, where the UAV position is controlled using state feedback based control. Stereo vision measurement results are investigated in order to determine the overall accuracy. Simulation results are introduced of the designed position control.
An alternate position measurement system design is introduced using a LIDAR (Light Detection and Ranging) device. This device requires extremely accurate time delay measurement. The high-speed signal processing method is presented with 10mm accuracy.
The resulting system is easy to set up, and solves the problem of video recording from the air. The control of the flying camera does not require RC flying experience, because the position of the camera is controlled through a PC.