Font Size: a A A

Autonomous Tracking And Landing For Vehicle Quadrocopters Based On Nested 2D Tag

Posted on:2017-05-11Degree:MasterType:Thesis
Country:ChinaCandidate:S G JinFull Text:PDF
GTID:2382330569498721Subject:Control engineering
Abstract/Summary:PDF Full Text Request
In recent years,the quadrocopter has been widely used in the field of military reconnaissance,power inspection,express delivery and so on,and more and more attention is paid to it both at home and abroad,because of its low cost,light weight,good maneuverability and strong concealment ability.At the same time,the future war is unmanned combat.It is an inevitable trend that quadrocopter and unmanned ground vehicle(UGV)are forming an integrated air-to-ground combat.Military applications of quadrocopter will become increasingly widespread.Aiming at the needs of quadrocopter land on UGV for charging or recycling,research on the pose estimation method of 2D tag,and make full use of the pose estimation information of the on-board sensor to control the quadrocopter tracking and landing on the UGV.In this paper,the recognition and detection of 2D tag,the data processing of pose estimation of 2D tag and the autonomous control of quadrocopter are researched on respectively,employing ROI,PnP,filtering,error compensation etc.,to estimate the pose and other state of quadrocopter during its flying,achieveing quadrocopter autonomous tracking and landing on the UGV.The major focuses and innovations are as follows:(1)Acceleration strategy of 2D tag algorithm based on ROIIn order to improve the running speed of 2D tag(apriltag)algorithm,the target tracking algorithm is used to select the region where the 2D tag exists,and to update the 2D tag in the image plane coordinate system.The number of pixels traversed by the algorithm in the image processing greatly improves the running speed of the 2D tag algorithm and leads to the real-time performance of the localization algorithm.(2)Anti-interference method of 2D tag algorithmIn the actual detection process,there may be some missing or partially obscured 2D tag which can not be used for position estimation.The principle of 2D tag pose estimation is well studied in detail,and the characteristics of PnP problem are fully utilized.The method of completeness missing vertices is used to construct the P4 P problem,which improves the robustness of 2D tag algorithm.(3)Precise position estimation methods for 2D tagIn depth research on the principle of detection and recognition of 2D tag,the nesting layout of 2D tag is designed.The method of linear regression is used to solve the position estimation error model of 2D tag with different sizes in nested 2D tag system.Position estimation is compensated by the error mode.Then the position estimation data are weighted fusion and smoothed.These methods improve the accuracy of the position estimation.(4)Autonomous tracking and landing strategy for vehicle quadrocopterThe position control loop,speed control loop,yaw control loop and height control loop are designed according to the characteristics of the vehicle quadrocopter.According to the actual situation of autonomous tracking and landing of vehicle quadrocopter,the flying state of the vehicle quadrocopter is divided into hover control,tracking control and landing control.Based on the software framework of autonomous tracking and landing of vehicle quadrocopter,the ROS node and the topic are designed,and the algorithm involved in the experiment is implemented.The vehicle quadrocopter is successfully and accurately landed on the ground mobile platform.(5)A human computer interaction method for controlling UAVA human computer interaction method was proposed to identify the manipulator's facial expressions and control the UAV through facial expressions.Eight kinds of facial expressions,such as anger,sadness,joy,surprise,contempt,fear,disgust and neutral,are corresponding to control the UAV up and down,yaw,roll and pitch.According to the principle of human face control UAV and software framework to achieve the ROS packet algorithm.According to the principle of human face control UAV and software framework to achieve the ROS packet algorithm.(6)UAV vision guidance deviceDepth research on the principle of detection and recognition of 2D tag,according to the characteristics of 2D tag black and white,designed a UAV visual guidance device,which can be used at night.The 2D tag is placed on the visual light source,and the brightness of the visual light source is adjusted so that the structure of the 2D tag is displayed on the light source completely.Ultimately,the realization of the quadrocopter nocturnal landing.
Keywords/Search Tags:2D tag, Nested, Vehicle quadrocopter, Tracking, Landing, ROS, Facial expressions
PDF Full Text Request
Related items