Research on Drogue Detection Algorithm for Aerial Refueling (IEEE/CSAA GNCC)

Unmanned aerial vehicle(UAV) has become more and more widely used in military and non-military applications, but endurance has become a major limiting factor in the development of UAV. In order to increase the UAV's endurance, this paper uses a vision guidance method to achieve automatic aerial refueling technology. This paper proposes an automatic refueling technology based on a deep-learning algorithm for detection. The Faster RCNN algorithm can simultaneously recognize the fuel receiver and the Drogue at long-distance and returns the center coordinate value of the receiver. If the pixel area of the identified Drogue is greater than a certain threshold, the central coordinate value of the Drogue is returned. The experiment results show that the average accuracy of the algorithm reaches 67.75%, and the real-time performance is about 5HZ.