Research On Information Collection And Obstacle Detection System Of Seedling Box Of Unmanned Rice Transplanter

As the process of urbanization accelerates and the agricultural population is greatly reduced,and agricultural machinery accidents occur frequently,food production and safety issues have increasingly become the focus of social attention.As a crop with short planting period and large labor demand,rice urgently needs to develop efficient unmanned planting equipment.In recent years,there has been a research upsurge of unmanned rice transplanter at home and abroad,but it mainly focuses on automatic driving and navigation control.However,no one is involved in fault diagnosis and residual quantity monitoring of seedling box and stone monitoring in seedling,and little is involved in field obstacle detection.These problems seriously hinder the continuous safe operation of unmanned rice transplanter.In this paper,based on the camera,reflective photoelectric switch,laser radar and other sensors of the control system of the unmanned rice transplanter,fault diagnosis,residual quantity and stone monitoring of the seedling box seedlings,identification and positioning of dynamic and static obstacles are carried out in the unstructured farmland working environment.The main research contents are as follows:This paper analyzes the working requirements of the information collection and obstacle detection system of the seedling box of the unmanned rice transplanter.The system is mainly composed of cameras,reflective photoelectric switches,laser radar and other equipment.Seedling box fault diagnosis and remaining quantity monitoring research design.Establish a camera model and obtain camera parameters to correct image distortion.According to the real-time and accuracy requirements of the image,the image is preprocessed by using the combination strategy gray-scale,median filtering and maximum between-class variance method(otsu).Obtain the seedling information of each row of the seedling box based on the background subtraction method,diagnose whether the seedlings of each row of the seedling box are faulty and monitor the remaining amount of each row.Use GUI to develop real-time image monitoring interface,define data transmission format and write data transmission program from Matlab to driving controller,and design image data processing flow.Research and design of intelligent online monitoring of stones in seedlings.The rotation speed of the add-on mechanism is monitored by a reflective photoelectric switch.In order to avoid the interference of dust and mud on the photoelectric switch,an automatic dust removal device with a dust cover and nozzle as the core is designed to perform level conversion and pulse shaping of the photoelectric signal.Analyze the speed change of the separation mechanism to determine whether it is "stuck" by foreign objects such as stones,and use sound and light alarm and remote transmission technology to remind the operator to clear the fault.Research and design of obstacle detection for unmanned rice transplanter in farmland.Establish the absolute position relationship between obstacles and fields through coordinate model conversion,analyze obstacle point cloud data,perform median filtering and body posture correction on the data,analyze different target detection algorithms,and use nearest neighbor clustering algorithm to process data to obtain obstacles The location information and size of the object are detected and matched according to the data correlation evaluation coefficient,the motion state of the obstacle is determined,the data processing flowchart of the obstacle detection is designed,and the obstacle detection program is written and debugged.The reliability test of the unmanned rice transplanter monitoring system includes the stability test of the image acquisition platform,the diagnostic accuracy and real-time test of the seedling image,the monitoring accuracy test of the stones in the seedling,and the static and dynamic performance test of obstacle detection.The test results show that the image acquisition platform has good stability,real-time diagnosis and accuracy;the test shows that the automatic dust removal device can effectively improve the monitoring accuracy of the monitoring system in a dusty environment;in the static performance test of the obstacle,it is found that the horizontal distance increases with the test distance The deviation increases,but the maximum within the test range does not exceed 8 cm,and the longitudinal average deviation is 1.54 cm.In the dynamic performance test,the accuracy of dynamic obstacle detection reaches 85%.The results show that the information collection and obstacle detection system of the seedling box of the unmanned transplanter have good reliability.