Research On Binocular Visual Manipulator In Logistics Sorting

In the era of AI,new standards and new technologies are developing rapidly.With the aging of the population and the demand for industrial development,people gradually depend more on machines to accomplish tasks.Machine vision is born in this circumstance.In the field of logistics sorting,manpower can not meet the requirements of industrial development,therefore,it is imperative to apply machine vision technology in the logistics sorting task.It is the core problem in visual logistics sorting to complete three-dimensional position calculation of target in different environment.In this paper,binocular vision technology has been applied to the field of logistics sorting with the help of four axis high-precision manipulator in order to complete the sorting for the target color.Firstly,the binocular camera has been calibrated and the image has been preprocessed.The calibration includes the calculation of its internal and external parameters,the relation establishment between the left camera and the manipulator,and the extraction of the feature points.In the experiment,different color workpiece are regarded as the target to be detected.Through a series of processing such as color space conversion,threshold segmentation,boundary extraction and straight line fitting,the target workpiece is extracted.The specific principle is utilizing color recognition to roughly locate the area range of the cube in the image,and then this area is used as the interest region for mask processing.The contour of the workpiece is extracted from the interest region and finally its center position has been determined.Secondly,this paper presents an algorithm for feature point matching and for 3D coordinates of the targets locating.It includes matching the feature points in the left and right images,and verifying their accuracy through the experimental data.Then the 3D coordinates of the matching points are restored by the parallel optical axis ranging principle.On this basis,the three-dimensional calculating equation of non parallel optical axis is deduced,and the accuracy of three-dimensional coordinates is optimized.The target workpiece is translocated into the manipulator coordinate system Through the coordinate transformation,and then guiding the manipulator to carry out the sorting work.Finally,building a binocular vision mechanical arm logistics sorting system.The manipulator and binocular camera have been selected according to the sorting requirements.The sorting system based on Python has also been designed whose feasibility has been verified and whose inaccuracy has been analyzed in experiments.