Research On Autonomous Landing UAV Technology Based On Cooperation Objectives And Vision-guided

Visually guided autonomous landing UAV has very important theoretical and practical value,how to accurate segmentation the corresponding cooperation and the method of obtaining the relativeposition are important part of autonomous landing. In this paper, the above issues related to researchwork carried out.First, this paper compared the existing UAV landing pose estimation methods, including thosemethods based on the projection matrix and based on the geometric characteristics, and then throughsimulation and physical experiments were carried out a comparative study, the experimental resultsshowed that the accuracy of based on the projection matrix compared to the method based on thegeometric characteristics of higher than80%. Through the analysis of the actual experiment, westudied the effect of the number of the target point and the shape of cooperation accuracy, When thenumber of feature points as nine highest precision, and little effect on the shape. Through the study ofalgorithms pose settlement, the same set of feature points having the same ratio between the differentfocal lengths, the relationship will bring this ratio Pose solving process. Comparative experimentsshow that the proposed method improves the accuracy of more than60%.Second, the actual effective detection of existing cooperative target distance is limited and in thecase of long-distance estimation algorithm based on projection matrix error is larger, in response tothese problems, the cooperation of segmented was designed, namely cooperation objectives through acombination of different sizes of the same shape of the goals. Not only in the case of a close situationsbut also distance situations, and design the corresponding segmented cooperative target recognitionmethod, compared with the existing cooperation objectives, the new extraction method of infraredcooperative easier.Finally, previous experiments did not get landing UAV relative distance throughout and difficultto synchronize the time, for the above-described problems, we build a six-rotor experimental platform,including DGPS, IMU and serial data memory etc, through three switch on the camera, DGPS andIMU data synchronization, and the existing cooperation objectives and segmented infraredcooperation flight tests were conducted experimental results, experimental results showed thatsegmented infrared cooperation’s accuracy improved a lot, especially in the case of long-distance,increased by more than50%. The modified algorithm is verified not only improve the accuracy of the height of more than50%, while the longitudinal distance Ty accuracy is improved by more than50%.