Research On Target Recognition Autonomous Landing System Of Small Rotor UAV Based On Visual Navigation

In recent years,with the improvement of hardware equipment performance and the development of artificial intelligence,computer vision related algorithms have been developed rapidly,especially target recognition and tracking landing direction.Autonomous landing of small rotor UAV Based on vision navigation is still a hot topic.At the same time,compared with GPS navigation,vision navigation UAV autonomous landing can improve the accuracy of UAV landing.In this paper,the open source,economic and easy to operate visual navigation mode is used to study how to improve the accuracy of autonomous landing of UAV target recognition,which has certain engineering application value and practical significance.The related work and conclusions are as follows:1.Based on the principle of open source and economy,the visual navigation UAV target recognition autonomous landing control system is designed,and the airborne mode assisted by the UAV Ground Station is determined.The system is divided into four parts and the computer vision module,autonomous landing control module and four rotor UAV platform are selected.2.According to the requirements of the subject,the landing mark is designed independently and the design details are elaborated in detail.According to the actual outdoor landing scene and the characteristics of the marker,a three-level height identification and positioning scheme is proposed.The positioning coordinate range is divided into three levels of height to reduce and the specific implementation process of the scheme is introduced.3.According to the content of the scheme,the three-level recognition and location combination algorithm of identifier is determined,which are threshold detection,houffcircle detection and April tags tag recognition.Aiming at the problem of low real-time of identifier detection in the combination algorithm,the program improvement method of dynamically reducing the scanning area in threshold detection and the improvement method of image preprocessing in houffcircle detection are proposed.In the actual test,the improved method significantly improves the real-time identification and positioning of the markers,so as to reduce the time delay of the UAV autonomous landingflight control.Through the complex background experiment,the feasibility of the combined algorithm of identification and location is verified.4.According to the scheme,the autonomous landing control strategy of UAV is designed and the autonomous landing flight test is carried out.The autonomous landing control module adjusts the direction by calculating the relative position between the image center and the mark center,and outputs the control command to the flight controller,so as to realize the autonomous landing of the UAV.The flight data are obtained through the related flight test,which proves that the autonomous landing control system designed in this paper can effectively improve the landing accuracy.In this paper,the autonomous landing of small rotor UAV with visual navigation is realized.