Industrial Robots Object Recognition And Location For Cigarette Packages

In the process of self-stacking cigarettes for industrial robots,the irregularity of cigarette packages brings difficulties to the process of industrial robots.Real-time feedback of target object information to industrial robots is one of the most effective solutions to solve such difficulties.The industrial robot combined with vision system gets a lot of information and has strong stability,it solves the problem that the robot cannot adjust itself when errors occur at the same time.So,the identification and positioning of industrial robots for cigarette packages studied in this thesis has important theoretical and practical significance.Based on the study of object contour recognition and binocular vision positioning,the industrial robot system extracts the number and location information of cigarette packages.The industrial robot pick-and-place system with visual control is designed.According to the accuracy and quick response requirements of industrial robot cigarette packages,this thesis has carried out in-depth research and implementation of cigarette package information extraction and system design.In the cigarette package identification,the traditional Sobel algorithm is fast but low accurate in the process of contour extraction,so this thesis uses fusion extended Sobel algorithm and K-L transform method to obtain edge extraction of image information,then the data analysis processing is performed on the extracted edge information.According to the structural characteristics of the fixed square of the cigarette package,the method that the external matrix minus contour information is introduced to identify the number of package packets.In the cigarette package positioning,traditional methods are slow and low accurate,in this thesis,a positioning strategy based on improved SURF(Speeded-Up Robust Feature)algorithm is proposed to quickly and accurately locate cigarettes.The SURF algorithm is used to match the feature points of the cigarette package information identified in the left and right pictures.Then the feature point pair anti-counterfeiting method proposed in this thesis is used to filter the matched feature point pairs.Finally,according to the principle of triangle measurement,the three-dimensional coordinates of the object are precisely located.After the completion of cigarette package identification and positioning,an industrial robot identification and positioning system for cigarette packages is designed.In addition,a cigarette package oriented identification and positioning experiment is implemented on the laboratory 6R industrial robot platform.Finally,the feasibility and superiority of this system scheme in the self-stacking of cigarette package is proved by comparing several kinds of industrial robots with visual control.