Research On Autonomous Landing Method Of Multi-rotor UAV Based On Binocular Vision

In recent years,the UAV has been widely concerned in surveying and mapping,aerial photography,line inspection of power,air reconnaissance and other fields.With the deepening of people's understanding about computer vision system and the rapid development of various complex algorithms,it is possible that computer vision is applied to the UAV.And computer vision has the advantages of large amount of information and high independence,it can provide rich external information for the UAV's intelligent flight.During the entire intelligent flight,the safe landing is a very important stage.With the autonomous landing system,the UAV can autonomously land on high-altitude buildings and be charged by itself for the purpose of autonomous operation.Compared with the traditional navigation methods,it has high precision and freedom from electronic interference.This dissertation focuses on the study of autonomous landing detection about multi-rotor UAV based on binocular vision.Firstly,a calibration method of Three-Parameter Model for parallel binocular camera is proposed.Then,the landmarks are analyzed and the QR code is selected as the landing landmark.The two methods of positioning QR code from close range and long distance are put forward from verifying the following experiment.During the autonomous landing of UAV,the conversion relation of each coordinate system is a very important link to complete the estimation about the attitude of UAV.Finally,the implementation of autonomous landing system based on binocular vision are completed,and the experimental verification is completed.The dissertation is summarized,and the future work is prospected.