Vision tracking system for UAVS formation

In this thesis, a vision tracking system for a UAV formation without inter-vehicle communication is proposed. A monocular camera, as a sensor, is installed on the follower UAV in this system, which is to track the leader UAV to maintain a relative distance from the leader UAV with the help of markers on the leader. At first, a target detection algorithm is used to extract the coordinates of the specified markers and then, a pose estimation algorithm is used to calculate the relative distance and the pose of the leader. After that, a prediction part is proposed to predict the location of the leader in the next image frame of the camera based on Kalman filter. After that, the prediction result is fed back to the target detection algorithm, which decreases the possibility of detecting targets wrongly by reducing the size of detection area in the whole image. At the end of this thesis, some experiments are designed and conducted to verify the effectiveness of this system.MASTER OF ENGINEERING (EEE