Research On Vision-Based Autonomous Tracking System And Attitude Control Of UAV

With the development of the field of automatic control,the application prospects of drone systems are becoming more and more broad.In recent years,the domestic and international civil drone market has developed rapidly.The intelligentization of drones is one of the hot research directions in the current research.Other aspects have extremely high research value in intelligent control,visual tracking,three-dimensional track planning,multi-machine air traffic management,etc.In this paper,the four-rotor UAV is used as the control object,and its detection and tracking of known paths are realized based on the research contents of visual processing algorithm,embedded software development and aircraft control.Firstly,this paper analyzes the structure of the four-rotor UAV and its flight principle,and establishes the coordinate system and the dynamic model of the UAV.Based on the autonomous tracking task,introduce the software development environment,build a ground station platform,and propose an overall design.Secondly,this paper carries out Matlab/Simulink simulation based on the mathematical model of the four-rotor UAV.The three control schemes of design cascade PID,nonlinear auto disturbance rejection control and linear auto disturbance rejection are applied to the attitude control simulation of quadrotor UAV.Then,this paper introduces the SURF feature detection and matching algorithm to realize the detection and matching of the UAV tracking line and locate the path.Based on the features of the trajectory edge,three edge feature detection operators such as Sobel,Canny and Laplace are introduced.Based on the corner feature of the landing point,three corner feature detection operators such as Harris,Shi-Tomasi and FAST are introduced.The definition and the accuracy ratio of six detection operators are compared.Finally,this paper builds a UAV test platform.Based on this experimental platform,the linear auto-disturbance suppression control algorithm is transplanted into the PX4 native firmware attitude control library,and the hover experiment is combined with the path detection algorithm to successfully complete the UAV autonomy.The tracking test has achieved good results.