On-Board Visual Relative Localization For Small Uavs

On-board relative localization is one of the crucial technologies for unmanned aerial vehicle (UAV) formation flight.In order to maintain the shape of a formation, it is necessary to maintain the distances between individual pairs of UAVs fixed, and all UAVs moves as a rigid entity. This paper presents a high-precision relative pose estimation method for UAV formation flight within the framework of ROS based on the camera ("standard load" of small UAV) and the deep learning algorithm. Indoor and outdoor experiments were conducted and the results show that the proposed algorithm can estimate the position of UAV member with accuracy at 0.18m.