Research On Visual Integrated Navigation Method Of Agricultural Facility Robot

At present,the application of visual navigation in industrial scenes has become more and more mature,but the research in the agricultural greenhouse scene is still in infancy.On the one hand,the application of agricultural scenarios is far less critical than industrial situations.On the other hand,agricultural scenarios are more complex than industrial environments.However,autonomous navigation technology is an indispensable part of facility agrarian robots.The reliability and accuracy of the navigation performance directly determine the performance of the facility agrarian robots.Therefore,this paper proposes navigation combined visual method combining navigation line and QR code to solve the problems existing in the navigation work of agricultural greenhouse facilities,to realize the navigation accuracy of the robot in straight path or intersection.Also,the facility robot can also complete decision actions according to the instructions stored in the QR code.Besides,the facility robot can also complete decision-making actions according to the instructions stored in QR code,such as sprinkler opening and closing,sprinkler lifting and lowering,etc.To solve the influence of uneven illumination on the navigation picture in the natural environment,an improved MSR algorithm is designed to restore the shadow area of the navigation image to ensure the stability and reliability of the navigation system.Firstly,the improved MSR algorithn is used to restore the shadow area of the image before the navigation work.The navigation image is separated into the brightness component and color component.The brightness image is self-adaptive thresholding and MSR processing separately,and the processed luminance region detail information is merged with the result of the MSR algorithm processing.Based on the color component extracted from the original image,the color restoration of navigation image is completed.Finally,the processed navigation image takes proper account of both color balance and image enhancement.Additionally,the processing time will be reduced to about 0.1s by adding down-sampling and up-sampling algorithm,which meets the real-time requirement of the navigation system.A navigation method for a low-cost,portable and straightforward red navigation line is designed for the facility robot to travel on a straight path.To eliminate the influence of environmental debris such as soil,plant stems and leaves on navigation image,the image is preprocessed by normalization,binarization,and median filtering.An improved Hough transform is used to extract the red navigation line.In the course of navigation,the lateral deviation distance and heading angle are converted into wheel deviation in real time,so that aecurate and efficient navigation can be achieved.For the navigation line navigation,there will be a large deviation when turning at the intersection,the navigation line and QR code integrated navigation method is designed.When the facility robot makes a large turn or turns at an intersection,the QR code is used to complete the navigation work.At the same time,QR code also stores some decision-making information,such as the opening and closing of sprinklers,the lifting and retrieving of sprinklers,and so on.While realizing navigation,it further completes intellectualization.Finally,the software interface design of the navigation system is completed,and the greenhouse spray robot is used as the platform to conduct the experimental research and analysis of the navigation system.In order to verify the reliability of the improved MSR algorithm for image shadow restoration,300 images were collected at 8:00-10:00,10:00-12:00,12:00-14:00,14:00-16:00 and 16:00-18:00 in one day and the final success rate was about 95%.The facility spray robot was tested at the speed of 0.1 m/s,0.2m/s,0.4m/s and 0.6m/s.The lateral position deviation,heading angle deviation and steering angle deviation were recorded during driving.The experimental results show that the average lateral position deviation is 2.2cm,the maximum lateral position deviation is 3.9cm,the average heading angle deviation is 2.1°,and the average steering angle deviation is 5.6°,Therefore,the integrated visual navigation system has superior stability and can meet the working needs of greenhouse robots in facility agriculture.