| Autonomous navigation of agricultural machinery is an important direction of agricultural automation and agricultural modernization,an important part of achieving precision agriculture and intelligent agriculture,and an important way to solve the shortage of rural labor and agricultural technicians.The development,application and popularization of computer technology,big data,cloud computing and 5G technology provide favorable conditions for its development.The autonomous navigation technology of agricultural machinery includes agricultural machinery positioning,operation path planning and path tracking control.In this paper,the field navigation system of cluster crops is designed and researched.The specific research contents are as follows:(1)According to the field operation conditions and system functional requirements in hilly areas,the overall scheme design of the field navigation system for cluster crops was carried out.The system adopted the combined navigation mode combining machine vision and GNSS(global navigation satellite system)technology.The system was mainly composed of visual navigation system,GNSS,chassis and sensors,chassis control module and system software.According to the requirements of the navigation system,a systematic chassis was designed,including the design of the amount and layout of the chassis gear train,the power drive and transmission structure,the steering mode and steering mechanism.The extended-range engine plus lithium battery were selected as the power source of the system,and the chassis structure of the front wheel deflection steering driven by the brushless DC motor was used as the carrier of the navigation system.Based on the simplified two-wheeled vehicle model,the chassis was kinematically analyzed,the chassis kinematics model was established,and the chassis steering mechanism was kinematically analyzed to obtain the correspondence between the actual front wheel angle of the chassis and the front wheel angle in the kinematic model.(2)Parameter calibration was performed on the camera,and the pixel coordinate system,image coordinate system,camera coordinate system and world coordinate system were established.The calibration module provided by ZED was used to calibrate the camera's internal parameters,and the mapping relationship between the point on the image of the visual system and the point on the navigation coordinate system was obtained.For cluster crops,an operation path fitting algorithm was designed.The 2GRB color space was used to grayscale the crop image collected by the navigation system,and the binarization of gray-scale images was performed by OSTU automatic threshold.And then the morphological filtering of binary image was performed by first closing operation and then opening operation.According to the characteristics of crop planting,the parameters of camera installation and the parameters of camera,the automatic threshold value of contour area was designed,and the connected domain processing was performed on the image.Followed by this,the moment of the pre-processed image was calculated and then the contour of the image was segmented in the v direction.And then the number and position of the first row of crops were found out and the image was divided into regions by columns.The contour area was served as the weight to perform weighted least squares fitting for the contour centroid of each region to obtain the position and orientation of the crop rows finally.(3)Selecting the north-east coordinate system as the navigation coordinate system,the increment of the abscissa,ordinate and heading angle of the agricultural machine chassis as the state quantity,the chassis running speed and the front wheel Angle as the control quantity,the systematic filter was designed by using the extended Kalman filter.By inputing or collecting the boundary path of the plot,the forward-looking distance of the pure tracking algorithm could be determined through the curvature of the operation path and operating speed,then followed by tracking the boundary path,adjusting the camera angle to collect the position and orientation information of all crop rows in the plot.And then the oxen-led method was used to plan the operation path of crop plots.Finally,according to the relationship between crop row spacing P and chassis steering radius R,three turning methods of semicircular turn,bow turn and pear turn were selected as the generation method of transition path between crop rows to generate the operation path of crop plot.(4)Based on Microsoft visual studio 2017 platform,combined with openCV visual library and MFC,using C language,C++,JavaScript for mixed programming,the development of system software was completed,the human-machine interaction interface was designed.Using the JavaScript API of Amap,the system webpage script was written,and the CWebBrowser2 class was designed,the real-time location of the chassis was displayed on the human-machine interaction interface through the map.The Vision class was designed for preprocessing,morphological filtering and operation path fitting of images acquired by Vision system.The Navigation class was designed for job path planning and tracking,and the Serial-port class was used for serial communication with the chassis control module.Using the Keil5 SDK provided by STM32 company,the software design of the chassis control module was completed finally.(5)The navigation system was systembuilt and debugged,and the systematic field test was carried out on a 6m×10m rectangular plot.The block boundary path was tracked at a speed of 5km/h with an average tracking error of 0.07 m and a maximum tracking error of 0.2m.Crop row information was collected and operation path planning was performed.The operation path was tracked at a speed of 5km/h with an average tracking error of 0.08 m and a maximum tracking error of 0.14 m.The navigation system designed in this paper has realized the autonomous navigation of cluster crops and the navigation errors has met the requirements of the operation. |
PDF Full Text Request