The Research And Application Of The Key Technoiogies In Visual Casting Fins Extraction Of The Industrial Robot Intelligent Grinding System

With the development of social industrialization,more and more robots are used in factory.However,the proportion of manual processing with hammer and normal hand-held grinders in casting cleaning is still very height.It not only increases the burden on the enterprise but also causes serious damage of the workers’ health.Therefore,it is of great significance to study the technology of intelligent visual grinding system.In this paper,the key visual technologies of the industrial robot intelligent grinding system in the casting production line are focused on.The visual image processing is applied to improve the grinding efficiency by controlling the motion of robot with perceiving the position and size of the casting fins.The poor working environment of casting production and low cost camera in low profit casting industry cause some challenges for the research in this paper.The main contents of this paper include the following points.Firstly,the rich geometric edge and the terrible working environments are considered in the extraction of casting fins.An evaluation index of edge detection effect of casting is designed based on screening the tradition edge detection operators.It evaluates the operators from missed detection rate and minimum variance error rate.Then,the experiments were carried out to compare the design with subjective evaluation.The experimental results show that the evaluation index consistent with subjective evaluation.And it can provide reliability of the visual system in production line.Secondly,a geometrical edge optimization method for castings is proposed based on the previous step.Since it has a lot of interfering information and a large amount of valid geometric edge in castings,the bifurcation and the rectangular perimeter are applied to optimize the image match and detection.Specifically,the SIFT algorithm and SURF algorithm are used to extract the feature points of casting images,and the effect of the two algorithms are analyzed.The geometric edge optimization is used to enrich the feature points by removing the interference in the background based on the geometric edge characteristics of the casting image.The experiments of global and local processing of image match show that it can obtain more accurate matching results by the geometric optimization feature points.Finally,a processing flow of extracting casting fins is designed by mapping image space the physical space based on the above results.The calibration image of the standard castings is processed to calibrate the region of interest and the base line.Then the image matching,image edge detection,edge geometric spatial optimization and castings fins edge location searching are processed to map the fins form the image to space by the calibrated parameters.Also,the visual extractions are applied in the industrial robots intelligent grinder system.The experiments show that it can effectively reduce the impact force by the visual castings fins position detection.