Research On Visual Recognition And Flexible Grasping Manipulator Of Fruit Picking Robot

Robot picking is one of the key technologies in the modernization of agriculture and forestry,and it is an essential prerequisite for achieving successful picking using machine vision technology to accurately identify the fruits growing in the natural environment.In the process of fruit picking,the target of fruit picking robot is easy to be affected by uneven illumination,overlapped and sheltered,which leads to the problem of false detection.In addition,the traditional rigid grasping manipulator is easy to scratch the peel of fruit and damage the fruit in the process of picking task.In this paper,the grasping of spherical fruit in the natural environment is taken as the research object,and the target recognition,location and grasping technology of fruit picking robot are studied.In addition,an effective edge extraction method based on geometry is proposed to reconstruct the actual contour of fruit target.Meanwhile,a three-finger flexible grasping manipulator is designed for the fruit picking.The main research work is as follows.1.Image recognition and location of fruit target: reconstructing the actual contour of overlapped and sheltered spherical fruit in the natural environment.Firstly,the recognition of fruit target is achieved by normalized color difference algorithm,and the segmentation boundaries of overlapped fruits are obtained by distance transform and watershed algorithm.Then the pixel coordinates of outer contour of fruit are acquired by boundary tracking algorithm,and an effective edge extraction method based on geometry is proposed.Finally,according to the pixel coordinates of effective edge,the actual contour of fruit is reconstructed based on the least square method.It can provide effective image information for spatial location.2.Spatial location and 3D(three-dimensional)reconstruction of fruit target:according to the results of image location,the corresponding 3D coordinates are obtained based on the binocular camera.Firstly,the binocular stereo vision system is built,and the binocular camera is calibrated according to the calibration parameters.Then the depth information of the center point of fruit is obtained by stereo matching algorithm.Finally,according to the spatial coordinates corresponding to image location,the 3D model is reconstructed based on the least square method.It can provide effective size and spatial parameters for the flexible grasping manipulator.3.Design of flexible grasping manipulator: a three-finger flexible grasping manipulator for fruit picking is designed,and the research is carried out from the aspects of structure design,control mode and application performance.Firstly,according to the design requirements,a three-finger flexible grasping manipulator is designed,and the mechanical properties and grasping process of the designed structure are simulated by using finite element analysis software ANSYS.Then some parts of the flexible grasping manipulator are produced and assembled.Finally,the experiments of object grasping for nine kinds of fruits with large,medium and small sizes are carried out to verify the feasibility of this design.4.Software platform of fruit target picking system: combining with Open CV,Py Qt and other software libraries,and the software platform for fruit target recognition location and reconstruction is developed.The results of image recognition,reconstruction and spatial positioning of overlapped and sheltered fruit are intuitively displayed on the platform.Then the fruit picking experiment is carried out by using AUBO-i10 manipulator and flexible grasping manipulator.The recognition and location algorithm of overlapped and sheltered fruit in natural environment proposed in this paper can reconstruct the actual contour of fruit target and realize image location.Moreover,the binocular stereo vision system can reconstruct the3 D model of fruit target and realize the spatial location.Furthermore,the designed three-finger flexible grasping manipulator can provide enough grasping force while ensuring safe interaction to achieve grasping of the fruit target effectively.Therefore,the research results of this paper can be applied to the system design of grasping manipulator of fruit picking robot,which have certain guiding significance and engineering application prospect.