Research Of Simultaneous Localization And Mapping For Quadrotor UAV Based On Monocular Camera

In disaster rescue,unmanned aerial vehicles(UAVs)gradually show more and more advantages.In recent years,GPS-based navigation and positioning of UAV have achieved significant development.However,these technologies can only be used in outdoor environment.So navigation and positioning autonomously in GPS-denied indoor environment becomes a new challenge.This thesis uses quadrotor UAV control technology and monocular camera to realize autonomous positioning,which has high scientific and application value.Firstly,according to the knowledge of computer vision,three kinds of landmark extraction method for visual SLAM are used and analyzed in this thesis.The NCC algorithm is used for the signpost matching.And mismatch outliers are removed by using RANSAC algorithm.In addition,the extended Kalman algorithm is studied in details and some simulations are committed.Secondly,visual SLAM coordinate system modeling is presented in this thesis,EKF and PTAM monocular SLAM theory are analyzed.Considering the real-time requirements,the PTAM monocular SLAM algorithm for the quadrotor UAV is adopted.Thirdly,hardware and software are developed based on AR.Drone quadrotor.The navigation and position control system on ground station are realized based on Robot Operating System(ROS).The control system uses PID controller with limiter.The delays of data processing and communication between UAV and ground station are eliminated when states are predicted.Based on the observation model of feature points,the positioning system of quadrotor UAV is built.Finally,in order to verify the monocular SLAM system based on quadrotor UAV,this thesis carried out actual experiments the indoor unknown environments.To increase quadrotor UAV's view field,calibrated fisheye lens are added.The monocular SLAM the maps showed the hovering and trajectory performance.In addition,combined with IARC task,the target detection and discrimination function are achieved and the target tracking and collision experiments are conducted.The results of experiment show that the system designed in this thesis can meet the needs of the indoor mapping and positioning for quadrotor UAV.