| This paper studies online inspection technology for robot glue spreading quality based on machine vision, which aims to provide an efficient real-time inspection method for robot glue spreading quality of car engine hood to help manufacturers better control glue spreading quality while reducing their production cost.During the study, the overall scheme of online inspection system for glue spreading quality is determined and key equipment of hardware system is designed according to the performance and technical requirements of the system at first. Then image processing and analysis algorithms is studied aimed at the characteristics of robot glue spreading of car engine hood. For real-time online inspection, information prediction algorithm is proposed based on generalized regression neural network. And then an online quality inspection software is developed according to functional requirements of the system. Finally, the performance verification test of on-line vision inspection system is conducted in factory.The research results of this paper include image processing and analysis algorithms and a hardware platform with an application software of on-line vision inspection system. To facilitate the adjustment of the installation gesture of visual device, a fixing mechanism is designed based on the support structure of dial gauge, and the specific models of key equipment are determined to meet the system performance, and an hardware platform of on-line vision inspection system is build up. The image processing algorithm of inspection technology for glue spreading quality based on machine vision is studied and the most appropriate method in this paper is determined by comparing and analyzing common treatment methods. Aimed at the problem that it is impossible to get the camera calibration information of any point on the glue spreading track, the calibration method which uses dedicated calibration object to conduct discrete calibration and predict calibration information based on generalized regression neural network is designed. In order to get the camera calibration information at the specific location of glue spreading, calibration object diameter and position information detection algorithm is presented. The collection of robot status information is realized based on MOTOCOM32 software with Ethernet and the reading and storage of system data is completed using SQL Server and ADO technology. The online inspection software for glue spreading quality is designed based on MFC and has many functions including acquisition, processing, analysis and display of images, display and storage of the analysis results, and user operation interface.The results of field experiment show that the fixing mechanism designed for visual device can realize the flexible adjustment and is easy to operate for user; the calibration object diameter and position information detection algorithm proposed is designed appropriately and has high detection efficiency; the calibration accuracy and reliability of the calibration method designed based on neural network prediction both satisfy the system inspection requirements; the online visual inspection system for glue spreading quality has high detection accuracy and short detection time achieving the expected goal. |
PDF Full Text Request