UAV Automatic Docking Technology Based on Deep Learning

Deep learning was widely used in various fields in recent times. The effect it has shown in some areas even surpassed experts in the field. In this paper, we used deep learning technology and applied it to the drone automatic docking technology. The realization of this technology is conducive to people’s intelligent and efficient daily life in the future. The parking lots of the data set we used was taken in the Northern Arizona University, Flagstaff, USA. For some images in the data set, we also used data enhancement methods to improve our training efficiency. The final result of our experiment fully proved the correctness and effectiveness of our method. The recognition rate of the parking lot is as high as 96.7% including different weather, lighting conditions, and more, on this basis, the accuracy of the automatic docking technology of the drone that it introduces is 63.2%. And behind this, we sealed the technology of the automatic recognition of parking spaces in the drone into an application.