Research On Visual Vertical Autonomous Landing System Of Rotor Unmanned Aerial Vehicle

Autonomous landing is the most important step in the rotor UAV's four phases of take-off,cruising,hovering and landing.And in recent years,the autonomous navigation based on visual has also become a research hotspot in the field of unmanned aerial vehicles.Visual navigation has the advantages of high accuracy and low cost,which can make up the shortcoming of traditional navigation like GPS / INS in fixed-point landing.In this paper,a fusion algorithm of pose estimation for the hierarchical cooperation objective is studied,which is applying for the visual landing of the rotor UAVs.The main work and innovations are summarized as below:In order to solve the problem that the view of airborne camera is limited during the UAV landing processing,a cooperative target based on the hierarchical strategy is designed.And the location recognition algorithm is realized based on cooperative target.The level of cooperative target is determined through the comparison of the area ratio,and the feature for the pose estimation is detected through the straight line detection.To estimate UAV's position and attitude,the Tsai calibration method based on projection and the fast four-point method based on geometric are studied and analyzed.Then,considering the characteristic of the Tsai method and fast the four-point estimate method,the fusion method to estimate UAV's position and attitude is proposed.Using the fusion method,the run time and the estimation error are reduced,which compared to using the Tsai method and the fast four-point method.Lastly,in order to test the validity of the object recognition algorithm and the precision of the fusion pose estimation method,the autonomous landing simulation experiment on the four-rotor UAV is carried out.Experimental results show that the application range of position estimation is increased based on the hierarchical objective,and the precision of pose estimation is improved by using the fusion method.The proposed visual navigation system can meet the technical specification requirements of the rotor UAVs visual landing.