Development Of Visual System For Precision Automatic Weeding Robot

Mechanical weeding instead of chemical weeding is one of the important links to solve the problem of food safety.Mechanical weeding robot has the characteristics of reducing labor intensity,automatic control and convenient operation,so it has a broad application prospect in the field of agriculture.Fast and accurate separation of weeds and location of crops are keys to automatic weeding.This paper presents a vision processing method for weeding robot,which provides reliable real-time data supports for autonomous navigation and automatic weeding.The main works completed are as follows:(1)Based on the analysis of the precision automatic weeding robot functions,the embedded vision method for crop root positioning and navigation processing is put forward.By extracting the position coordinates of the crop and the radius of the protection area,the plant protection area and the weeding robot working area are defined.The autonomous navigation of the weeding robot is realized by extracting the relative deviation and position offset of the robot with the crop row ridge.The TCP/IP communication is developed to contact the visual system with the control system.(2)In view of the motion conditions of the weeding robot,the camera selection is carried out,and the parameters of the camera's internal parameters,the lens distortion and the image plane size correspondence are obtained to correct the image distortion.An embedded visual system test platform is built for the running environment of the visual system,and the system transplantation and environment configuration are completed.(3)In view of the problem of visual navigation in the case of sparse distribution of crops,an isometric sampling is proposed to find the feature points of the boundarys of crop rows,and then the boundarys are determined by line fitting,and the middle line of boundary lines is obtained to compare with the axis of the robot to get the relative deviation and position offset between the row of the robot and the crop row.(4)In order to realize fast weeds separation,achieve the crop coordinate and extract protection area radius,a method of extracting crop contour information by using area method to screen weed information and retain the information of crop contours is used after image segmentation,and then the crop feature information is extracted from the coordinate and the protection area radius.(5)In order to ensure the reliability of the communication between the control system and the visual system,the heartbeat detection mechanism is designed in the communication module,which enables the control system to find the visual system abnormality in time and provide the solution.In order to ensure the timeliness of communication,the asynchronous communication process between the server of the visual system and the control system client is realized by the idea of multithread programming.(6)Based on the CLion IDE and C++,the development of visual system software with concurrent design idea of multi process and multi thread program is completed and the visual system software is transplanted into the embedded platform.(7)The embedded visual system is tested.The average time of the navigation module processing an 800×600 image is 0.375 s,and the calculation error of the robot's deflection is within ±0.5° within 10°.The average time-consuming of the plant feature extraction module handling a 1280×720 image is 0.409 s,and the radius extraction relative error of the protected area is less than 5%.The longest abnormal response time of the visual system is about 1.5s.The test results show that the speed of the image processing and the response speed of the vision system accord with the weeding robot's speed of 800mm/s.