Research On Integrated Navigation System Of Unmanned Rice Transplanter Based On RTK And Machine Vision Technology

As more and more young and middle-aged people in rural areas flock to cities because they are unwilling to engage in agricultural work,some rural areas have already panicked in employing people.The increase in food production costs caused by rising rural labor costs has attracted more and more attention.As a seasonal crop with short planting cycle and high labor demand,rice is an urgent need to develop efficient and intelligent unmanned planting equipment to meet the high-quality operation requirements of modern agriculture.In recent years,there has been an upsurge in the research of unmanned rice transplanters based on satellite navigation at home and abroad,but it is mainly focused on automatic driving,and the path planning in the turning area of the field and the automatic seedling loading device have not been successfully developed.Still need to go to the field to replenish the seedlings to return to the point of operation.Breakpoint navigation path planning navigation,obstacle avoidance path planning guidance for obstacles such as trees and graveyards in the field,and auxiliary navigation after the accuracy of satellite navigation is reduced due to environmental factors,but there is little research.These problems seriously hinder the efficient and safe operation of the unmanned rice transplanter.In response to the above problems,this paper adopts Beidou-RTK technology,inertial navigation technology and visual navigation technology to carry out research on full path planning,breakpoint endurance path planning and obstacle avoidance path planning navigation,and developed a set of satellite navigation based on visual navigation.Auxiliary automatic integrated navigation system.The main research contents are as follows:Based on Beidou-RTK technology,inertial navigation technology and visual navigation technology,this paper builds a set of automatic integrated navigation system with satellite navigation as the main and visual navigation as the supplement.The entire navigation system is composed of PC industrial computer,attitude sensor,lidar,Beidou-RTK equipment,notebook computer and camera.The data transmission of each system uses CAN bus technology.Based on the QtCreator platform,this paper uses C++ to develop a set of navigation system software that integrates data reception(Beidou-RTK data and lidar data and visual navigation data),solution,path planning under different needs,and human-computer interaction interface.Based on the coordinates of the four vertices of the field,combined with the operation width of the rice transplanter,the whole path planning navigation including the turning area of the upper and lower fields is carried out;when the seedling quantity is insufficient,the breakpoint continuous navigation path planning and navigation;according to the obstacle information and the kinematic model of the rice transplanter And pure tracking algorithm for obstacle avoidance path planning and navigation.And use UI to develop navigation control interface.This paper develops a visual navigation system based on Visual Studio 2015 and Opencv using C++,and transmits the obtained navigation information to the satellite navigation system controller through the CAN bus.When the satellite navigation accuracy cannot meet the operational requirements,the visual navigation data is selected and sent to the computer.Man driving controller.According to the characteristics of the paddy field image,the super green algorithm is used to separate the seedling row from the background;the maximum between-class variance method(otsu)is used to binarize the image of the separated seedling row;the swelling corrosion operation is used to eliminate the gap between the seedlings;based on the maximum boundary point The algorithm extracts the boundary of the seedlings;the least square method is used to fit the line of the boundary seedlings;the visual navigation dead-reckoning deviation calculation first calibrates the selected camera internal parameters,distortion parameters and the transformation between the camera-transplanter coordinate system;-The calibration experiment of the rice transplanter coordinate system shows that the coordinate transformation error within 6meters of the preview distance is less than 1%.Test the stability and accuracy of the designed satellite and visual navigation system on cement ground.Among them,artificial turf is used to simulate the border of seedlings in visual navigation.The test results show that the entire navigation system is stable and reliable.The accuracy of satellite navigation is less than 2 cm,the accuracy of the seedling boundary line extraction algorithm of the visual navigation system is more than 90%,the average time is about 100 ms,and the navigation accuracy is less than 4 cm.The actual operating requirements of the machine.