Visual servoing quadrotor control in autonomous target search

Abstract

The use of visual input for quadrotor unmanned aerial vehicle control calls for fast algorithms to identify and localize target objects in the video image. Localization of target is done using color histogram. Advantage of this method is not only capable to detect the target but it also give information on size that we use to calculate its position relative to the quadrotor. Once the target found, the quadrotor is guided using visual feedback control for the approach and then the precise pose position is maintained. In the desired position in front of the target, an angle of orientation is calculated using the detection and positioning technique. This angle controls yaw movement of the quadrotor, providing an ability to orientate its position when it arrives with a certain angle relative to target plane. Experiments were done in the indoor environment in sufficient light condition. A search order is defined based on patterns inside target images. The quadrotor takes off, then locates, approaches and orientates itself successively on four patterns found on four targets, and finally lands.