Design Of The Quad-rotor Positioning System Based On Vision

The quad-rotor has a simple structure,low manufacturing cost requirements,good control stability,and has received very extensive attention in the fields of commerce,agriculture,and military,and has an excellent application prospect.However,at present,quad-rotors are basically controlled by a remote controller or are automatically controlled by GPS.They are affected by visual errors or precision limitations of the control personnel and cannot be applied in many high-precision applications.This dissertation focuses on the visual autonomous flight control of a quadrocopter,and uses a robot control system(ROS)and an open source computer vision library(OpenCV)to design a quad-rotor positioning system control system based on computer vision.The system adopts the "PC-side/quad-rotor" operating mode,uses the basic directional control of the quadrotor,adopts the robot control system ROS as the development environment,and uses C++ programming,open source computer vision library OpenCV to process images,and other related technologies.Computer vision-controlled quad-rotor tracking and design were studied.First,according to the characteristics of the quadrotors and ROS systems,the Linux,ROS,OpenCV software environment and the ROS-based quad-rotor positioning system framework were established.The corresponding workspaces,themes,nodes,and C++programs were established.The system was built.The information processing capability of the UAV platform can be greatly improved,and the scalability of the program of the quad-rotor visual positioning system is also improved.Secondly,the target graphics adopted by various large-scale competitions and well-known laboratories were analyzed.The graphics were divided and designed as necessary.At the same time,based on the consideration of the practicality of the system,spherical markers,ground H-shaped markers,and ground circular markers were established.Thirdly,the image acquisition,preprocessing and graph fitting processing of the related target graphics are studied.The Hough circle detection theory,corner fitting algorithm and least squares ellipse fitting algorithm are respectively introduced.Then the target graphics of the adaptive system are improved.According to the characteristics of the ROS system,the algorithm for calculating the relative position of the current aircraft and the image center and the conversion instruction format algorithm are proposed.These algorithms have less computational complexity and require lower processor performance,and solve the problem of monocular vision that cannot measure the distance between the quadrotor and the target.Finally,using the positioning control system platform built,the visual positioning experiment of the AR.Drone quad-rotor was performed.It has been verified that this design achieves a high-accuracy,high-flying quad-rotor aircraft visual positioning automatic flying mission,which satisfies the practical application requirements and achieves the design goals.