The Design And Control Of Automatic Pick-up System Of UAV With Manipulator Based On Binocular Vision

Because the four-rotor uav can be miniaturized,has good mobility and is easy to control,and can be used in aerial photography of scenery,environmental monitoring,military reconnaissance,entertainment toys and other aspects,it is greatly favored by all kinds of people,greatly promoting the research and development of the rotor UAV.With the continuous development of automation,the machine becomes more and more intelligent.As a very popular rotor UAV system,it is urgent to carry out intelligent upgrade and expand its application range.Research has four rotor unmanned aerial vehicle(UAV)system with mechanical arm system together,make the unmanned aerial vehicle(UAV)have the ability to capture and carry objects,this paper will combine the four rotor unmanned aerial vehicle(UAV)system,mechanical arm system and a binocular vision system three system makes unmanned aerial vehicle(UAV)at the same time have the ability to grab objects and identify the tracking ability of the object,finally realizes the binocular indoor guidance,automatic identification track the whole function of autonomous flight of grab objects.Because this is a multi-system fusion,their mutual influence and some unpredictable problems are indispensable,so it is a great challenge to ensure the stable and coordinated operation of the system.The main research contents of this paper are as follows:In this paper,the UAV system,robotic arm system and binocular vision system are firstly built.This paper design the unmanned aerial vehicle(UAV)mechanical structure and its dynamical system,and then check out the design of unmanned aerial vehicle(UAV)load capacity meets the demand of paper practice,after the analysis of the mechanical arm system design,final design for the single degree of freedom mechanical arm,only responsible for fetching task,simplify the complex system modeling difficulty,reduce the interference between each other.And analyzes four rotor dynamics of accurate kinematics mathematical model is established,and by measuring instrument and Solidworks 3D modeling software modeling by the established mathematical model of the important parameters,and then to establish the mathematical model is divided into three attitude Angle and a high position control four channel integral saturation resistance cascade PID algorithm is designed respectively,and then set up in MATLAB simulation model of control platform,the simulation debugging to get four channel simulation response time were 0.402,0.451,1.191,1.231.Then,the binocular vision system was further studied.The object was designed as a light cylinder with yellow as the main color and green line as the auxiliary winding to increase the recognition efficiency.The algorithm combining color filter recognition and tracking with CAMshift algorithm tracking was proposed to identify and track the target object.HSV color space was used to identify the yellow and green colors to determine the target area.After the target was detected,BM stereo matching algorithm was used to calculate the three-dimensional spatial coordinates of the target object relative to the binocular camera coordinate system.The coordinate transformation matrix between the camera coordinate system and the four-rotor UAV coordinate system is derived,and the center point coordinate of the target object is converted to the three-dimensional coordinate of the body coordinate system,and then the MAVLink aircraft communication protocol transplanted to the image processing controller raspberry 4B is transmitted to the UAV controller.Realization of binocular destination automatic calibration.Here,pixel guidance is proposed: the motion is decomposed into yaw Angle motion and linear motion,that is,the yaw Angle is first rotated to make the pixel point of the target object in the center of the pixel coordinate system,and then the vertical and horizontal linear motion is carried out simultaneously.When tracking the path,PID algorithm is used in the horizontal and vertical directions to modify the trajectory by monitoring the target pixel coordinates,and then the modified control information is transmitted to the UAV through MAVLink protocol,and the target object is finally captured.Finally,the fuzzy PID control algorithm was established for the control of the whole system.Finally,simulation experiments and anti-interference simulation experiments were carried out on the four channels through MATLAB.After the interference was generated,the roll Angle,pitch Angle,yaw Angle and height were stabilized at 0.9s,0.81 s,1.24 s and0.91 s,respectively.Finally,the experimental platform is constructed to complete the tasks of grasping and delivering target objects in a small range.