Research On Autonomous Relative Navigation Technology

Aerial refueling technology plays an important role in increasing the endurance of the receiver aircraft,increasing the payload of the receiver aircraft and increasing the aerial duration of the receiver aircraft,in which case it has become a very important technology for the air force of all countries.In the aerial refueling technology,the most urgent problem is how to realize the accurate docking of the receiving probe and the refueling drogue in the process of aerial refueling,i.e.,the autonomous relative navigation technology.At present,the probe-and-drogue refueling technology is widely used in our country.Therefore,in this paper,we conduct a series of researches on the autonomous relative navigation technology for probe-and-drogue refueling and they are mainly about the following aspects.Firstly,the existing aerial refueling technology at home and abroad are introduced,the boom-and-receptacle refueling and probe-and-drogue refueling are analyzed and compared in detail.In view of the problem of accurate docking in the process of aerial refueling,we detail the application of machine vision in the process of probe-and-drogue refueling.Moreover,we introduce the working process and model of the existing probe-and-drogue refueling visual navigation system and have a related simulation.Secondly,a tracking algorithm based on detection,namely Tracking-Learning-Detection(TLD)tracking algorithm,is studied.We firstly detail the principle,composition and workflow of the TLD tracking algorithm.In the light of the limitations of the existing probe-and-drogue refueling visual navigation system,we apply the TLD tracking to the autonomous relative navigation technology for probe-and-drogue refueling based on characteristics of TLD tracking algorithm that can track the target for a long time.And the related simulation experiment is carried out.Thirdly,aiming at the problem of tracking failure of TLD tracking algorithm when refueling drogue is deformed,an improved anti deformation TLD tracking algorithm is proposed.The algorithm combines the TLD tracking algorithm with the meanshift algorithm,and uses the processing results of the TLD tracking algorithm to modify the initial point of the meanshift algorithm.At the same time,the results of the meanshift algorithm are used to modify the data of the TLD tracking algorithm.At last,aiming at the problem of tracking failure of TLD tracking algorithm when refueling drogue is occluded,an improved anti occlusion TLD tracking algorithm is proposed.The algorithm combines the TLD tracking algorithm with the Kalman filtering algorithm.On the one hand,when the refueling drogue is not occluded,the results of the TLD tracking algorithm are adopted and the Kalman filtering algorithm is updated.On the other hand,when the refueling drogue is occluded,the prediction results of Kalman filtering algorithm are adopted and the TLD tracking algorithm is updated.The results of the simulation experiments show that the two improved algorithms proposed in this paper can improve the performance of the TLD tracking algorithm when the refueling drogue is deformed or occluded.