Research On Guiding Unmanned Helicopter Autonomous Landing A Ship Based On Vision

Because positioning accuracy is not high 、 2D landing target cannot fully recognition in clutter background and partial occlusion based research of domestic and foreign on autonomous landing a ship for unmanned helicopter, however, the research on the navigation precision, control technology and landing reliability have some strict requirements, Therefore, the study is a challenging and difficult task of national defense.The thesis takes the rotor unmanned helicopter as the object,describes some influence of the safety to land on a ship for the unmanned helicopter, which it is derived from the landing environment and the characteristics of the ship motion. And establish the nonlinear mathematical model for unmanned helicopter. Study on using image matching features of target recognition technology to expand the research on visual landing, the study and innovations include the following three parts:(1) For the method of the traditional classical control design the flight control law existing some shortcomings as long time 、 heavy workload and the control accuracy is not higher. Based on the research that provided a new method named signed domination to the design of the flight control system for unmanned helicopter. Firstly, according to the non-linear model fitting to make it linear and then handled by linear model, and realized it the spatial discretization. Secondly, we get signed domination used in it and produce some inputs of the signed domination, and then we get the structure of the signed domination system by using the symbolic controller. Finally,the signed domination algorithm is adopted in achieving a transform of the space state and a standard type of the Brunovsky to get the flight control law of the unmanned helicopter.(2) For the 2D landing target in complex background scene and partial occlusion caused a target recognition and other difficulties, this paper presents the method of target recognition technology based on SURF and FLANN fusion, Firstly, according to the acquisition of the target image using the median filtering and binary threshold segmentation technique for preprocessing to segment the target image region to be processed.Secondly, dealing with the target area using based on SURF method respectively to extract feature points of landing target and landing target in the scene, to generate the feature vector.Finally, using FLANN algorithm of feature points matching technology for two recent feature points on the image, to match feature points of two images based robust method. According to the feature points and matching distance to make target image matching and confirmation.Simulation results show that therecognition technology to deal with landing target under various scenarios can better identify the complete target.(3) In order to obtain the position and pose information of unmanned helicopter landing to the ship, the algorithm of the position and pose information is studied and detected target horizon and ship at the sea by using the Canny operator.Based on the information of the landing target feature points and the result of the horizon, the ship target are identified, estimation of relative position and pose information simulation results are obtained. Finally,based on the estimation algorithm of 3D reconstruction,we make a 3D simulation for the unmanned helicopter autonomous landing a ship,analyzed the result of the 3D simulation and mapped a 3D trajectory base-vision landing on a ship for the unmanned helicopter.The simulation results show that : the research of guiding unmanned helicopter landing a ship based on vision on the precision have meet the practical requirements.