Research On The Basis Of The Visual-oriented Robot To Crawl Positioning Technology

Due to the characteristics of its high efficiency, stable and reliable, robots are widelyused in the field of industrial automation.With the development of machine vision technologyand it was introduced into the field of robots, robots started to have their own eyes. Thisprogress not only expanded the scope of its application, but also improved the intelligent levelof industrial robots.There are two main methods of the application of machine vision to guide industrialrobots’ hand movements: one is Eye-in-hand and the other is Eye-to-hand. The former isflexible to use, but because of constantly changes of image field, it is more complicated toimage calibration. The method Eye-in-hand is more diffcult. Compared to the former,Eye-to-hand method is easier, the image sensor don’t move with the robot and imagesprocessing is easy, so it is more widely used. In the current industry field, the common isrobots grasping plane objects by the guide of two-dimensional image recognition results.Yet there is little uses of robots grasping3D objects based on the recognition ofthree-dimensional images, and such technology is badly needed. Currently, Besides usingmachine vision technology of structured light scanning to acquire the three-dimensionalimformation, the main method is by the use of binocular vision techonolgy. As this methodneeds the same cameras and complicated calibration, it is only suitable for laboratory research.In this paper, the techonolgy of vision guide robots to grasp3D objects was studied.Proposed a method based on monocular vision technology, which used3D geometryimformation and coordinate posture information in robot working coordinate system ofobjects, guided robots to grasp these objects. It improved the traditional way of cameracalibration by using LabVIEW platform of Vision Development and Vision Builder forAutomated Inspection components, optimizes of the calibration method of camera system,simplifies the calibration process. In LabVIEW VBAI environment, designing the processesof image acquisition, pre-processing, post-processing and computational geometry algorithms.For regular3D objects, according to its single-frame stereoscopic image can identify thetarget three-dimensional shape, size, position and orientation information, build aexperimental verification platform with six-axis multi-joint industrial robot, a industrial camera, workstations and target objects, established coordinates system for the cameracalibration, image aquisition, the size and shape of the target objects recognition experiments.The teaching capture program of industrial robots was writed, and through teachingprogramming method to move the robot hand tools to capture movement relative objects,Verified the identification of the target pose with an acceptable accuracy. The development ofpractical system in this research that the vision guide the robot to grasp the3D object providesgood practice reference for the field of industrial automation using robots grasp3D objects bythe guide of machine vision.