Abstract. Algorithms for controlling an unmanned aerial vehicle (quadrocopter) are considered. The algorithms allow solving the problem of tracking moving objects with the use of video camera mounted on the quadrocopter. An algorithm of quadrocopter piloting along a predetermined path was developed. The algorithm applies an approximation of the original curve with small segments and it allows for simplification of the controlling by utilization of only two commands: forward movement and rotation about the vertical axis (yaw). An approach to object identification from its digital image, based on edge detection is presented. Methods of obtaining of the tracking object displacement while accounting for background inconstancy caused by the camera movements are analyzed. An algorithm for measuring distance to the tracking object using a single camera is developed. The algorithm is based on comparison of two frames taken from different altitudes, and it is a modification of the method of stereoscopic vision when the camera is hard mounted on the quadrocopter. For multiple targets in a frame, a method to measure distances to the targets is proposed. The method applicability does not depend on the targets position in the frame and therefore may be used for safe passing over obstacles along the route. Combined application of the above algorithms is reported to allow obtaining of the full set of parameters describing target position and movement, the parameters may serve as a base for creation of control commands for flying robot necessary to implement a valuable system of tracking moving objects with the use of unmanned aerial vehicles   

Keywords: quadrocopter, trajectory control, UAV, tracking, moving object, Vision Systems