Research On Technology Of UAV's Automatic Landing On Ship Based On Computer Vision

For the purpose of implementing UAV's landing on the deck automatically and accurately under all-weather, a new program of UAV's automatic landing on the ship's deck was proposed, whose main navigation is GPS/SINS combined with the height instrument and assistant navigation is computer vision/SINS combined with the height instrument based on the cooperative object on the runway of the ship.namely ,we set a cooperative object on the runway of the ship , then the cooperative object can be recognized by using computer vision , ultimately UAV can precisely navigate and land automatically . This technology also resolves the potential dangerous problem that GPS of UAV's navigation system based on GPS/SINS is controlled by the other country and easily interfered.8 ~12μm was identified as the band of the visual system on the UAV after analyzing these elements : the landing requirements and conditions of UAV, the transmission property of visible light , infrared irradiation in atmosphere, radiation of the ship, rapidity and veracity of controlling of the cooperative object's temperature,et al. By means of theory and experiments we also confirmed that the maximal temperature range of the cooperatieve object corresponding spectrum radiation efficiency range is 100 ~186 C .Additionally, we have developed a cooperative object on the ground which can be controlled to change its temperature .In this paper, we propose a kind of design idea.we automatically activated the cooperative object through wireless communication with the cooperative object on the runway of the ship.During the cooperative object's recognition period, we use affine moment invariants to regnition the cooperative object. And we optimized the best assemble of character from the affine moment invariants. We woke out a method how to extract the character points from the image of the cooperative object, then introduce how to use these points and relation of the frame of axes interrelated with UAV to compute angle of pitching, angle of roll, angle of yaw relative to the frame of axes of the runway of the ship. And finally we have finished the basal research of UAV's navigating and landing on the deck automatically.The result shows that the average recognition time is 17.2ms using affine moment and can improve the reliability and real-time characteristic of UAV's autonomous landing dramatically.