Research On Key Technology Of Visual Navigation On Rice Transplanter

As the world population continues to grow, the demand on food and other agriculturalproducts also grow. Traditional farming methods have become increasingly unsuited to meet therequirements of the current time. Along with the development of automation technology, sensortechnology and intelligent, mechanized farming method gradually replaced the traditionalmethod. Autonomous navigation technology is also conform to this trend, also it has beenapplied in agriculture in current time. This paper aiming at visual navigation of transplantingresearch done as follows:1) Extraction the center line of seedlings of cropThis article first use2G-R-B transform and threshold processing to get gray image, and thenuse the Meanshift algorithm to removal of seedling planting shadow, finaly use Hough toextraction the center line of seedlings of crop. Then comparison is made between this methodand the traditional method of extracting the experimental results. We analysis of two kinds ofextraction methods, and we discuss the causes of disadvantages of the traditional method ofextracting.2) calibration of cameraDoing a comparative study of MATLAB and OpenCV camera calibration. By analyzing thesize error of calibration results, the final choice is relatively stable in the MATLAB method.Through the calibration of camera intrinsic parameters were obtained, the key parameters toobtain combined with external parameters of navigation.3) control method of camera navigationBy the navigation parameters acquisition, we classify the path planning into three cases. Weintroduce a photoelectric encoder, and use it to measure the actual angle of transplanter wheels.The actual angle can be used as the feedback loop regulating.4) develop visual navigation experimentThe hardware platform and software platform is constructed to carry out the experiment onland on transplanting machine vision navigation system. The experiment is made to lay thegroundwork for the paddy field of vision navigation. After the experiment we make sure theparameter of PID control:kp=10、ki=0.025、kd=0.4. The experimental results show that, thecurrent average angle of the system error in1degrees, the average position error less than3.2mm, the angle of the standard deviation of about2.5degrees, offset the standard deviation isless than34.80mm. Through the analysis of the obtained data summary, we obtained thenavigation system, and project the next improvement scheme.