Position And Attitude Estimation Of Small Aircraft Based On Binocular Vision

As an emerging navigation method,visual navigation has different applications in various industries.In agricultural production,visual navigation is used to navigate agricultural machinery;in the field of robotics,many tasks in dangerous environments are completed by visual navigation robots;in mobile vehicle applications,visual navigation is used to perceive the surrounding environment of the vehicle.With the development of computer vision and hardware performance,visual navigation as a low-cost navigation method has gradually developed in the aviation field,especially in the landing of UAVs,and has been widely used.This paper mainly studies the application of visual navigation in the landing of fixed-wing UAVs,and this paper focuses on the lack of actual scale information of objects in monocular vision and the difficulty of system initialization,and studies the problem of pose estimation based on binocular vision during landing.The research work of this article is mainly carried out from the following aspects:1.Research on binocular vision runway ranging algorithm.The content of binocular vision ranging in this article mainly includes the following aspects: 1)Calibrate the binocular camera to ensure the accuracy of subsequent experimental results;2)Correct the calibrated image to make it correspond to the left and right eye images.The points are in the same image row;3)Stereo matching.The stereo matching method based on the SGBM algorithm is used to acquire the depth image.4)Obtain the three-dimensional information of the target from the acquired depth image.2.Research on the detection algorithm of horizon and runway edge line.This paper relies on the simulated landing scene images taken by the camera during the experiment as the research material,and uses the runway boundary and the horizon as the feature extraction targets in the image processing.The extraction of the runway boundary mainly uses a contour recognition algorithm,and the extraction of the horizon uses a color-based sky segmentation algorithm,and the accuracy of the algorithm is verified through actual scene images.3.Estimation of attitude angle.This paper estimates the roll angle,pitch angle and yaw angle of the aircraft through the extracted horizon and runway boundary parameter equations.For the above research work,this article uses a simulated runway to conduct experiments,and through the above algorithm to process 4 sets of runway images,the calculation results of distance and attitude angle are obtained.The calculation results and actual measurement data are compared and analyzed to verify the accuracy of the pose estimation system in this paper.In addition,this article also tested the distance measurement of monocular vision,and verified the advantages of binocular vision by comparing monocular and binocular.Finally,this paper designs and develops the pose estimation system software through QT programming,and realizes the integration of functions,which is convenient for daily use and subsequent system hardware development.