Research On Visual Servo Control System Of Target Spraying Robot

Target spraying technology is of great significance to reduce pesticide residues,protect environment and workers.In this paper,a target spraying robot scheme based on visual servo,aimed at greenhouse crop target spray,was proposed and a test system was estimated.And visual servo control algorithms based on the scheme were studied.The main content is as follow:(1)Design of a target spraying robot scheme and construction of a testing system.The scheme is based on a vision structure blend of monocular scene and monocular hand-eye.And based on the scheme,the mechanical arm structure of spraying robot was designated,the forward and inverse kinematics of position and velocity and workspace were analyzed.The main structural parameters were optimized based on the workspace which was needed,and the mechanical arm type was obtained by making the optimized consequence as a foundation.The trajectory interpolation method of spray manipulator in Cartesian space and joint space was studied.The target spraying robot testing system was established,and the software and hardware design was finished.(2)Research on background segmentation and feature extraction of crop image.Taking gourd seedling in early growing season as research objects,the segmentation effect of common operators in various color spaces on sample image was compared and it was found that CIVE and 2G-R-B operators of RGB space had the best segmentation effect.To find better operator,the optimization in RGB space using genetic algorithm,which took overlapping area rate,distance of means,standard deviation and algorithm computing time in the sample image gradation histogram as indexes,was performed.And a better segmentation operator was obtained.In view of the fact that segmentation effect of the operator is not satisfied when crop represent a small part in the image,a strategy combining the operator + Ostu segmentation with a operator in Lab space + fixed threshold segmentation was proposed.And to obtain target features,such as the centroid and enclosing circle of crop,some algorithms for extracting connected regions and enclosing circle were studied.(3)Research on visual servo control of the spraying robot.For speed controller used by the spraying robot,two feedback features,include a set of position deviations of nozzle based on enclosing circle of crop canopy and a set of moments whose expected values did not change with the change of objectives,were designed.The Laboratory experiments showed that the controller with two kinds of feedback characteristics had good performance on target and tracking control.Furthermore,for the continuous motion pattern of spray robot,a novel algorithmic for the motion planning and control of target was proposed based on the improved potential field method.The velocity potential field,field constraint and joint limit position constraints were introduced based on the traditional potential field method,which could plan a better spraying trajectory for the spraying robot and improved the robustness of target control.For the intermittent motion pattern of spraying robot,on the basis of obtaining feasible path between each two crops' spray position,enumeration method and improved nearest neighbor method were combined to plan spraying path.This method was verified by experiments of controlling spraying arm to target crops.