Sole Dispensing Path Planning Based On SOM Neural Network

The footwear industry is an important industry in the national economy,and sole dispensing is an important process in the footwear industry.With the development of artificial intelligence technology,the field of sole dispensing can intelligently guide dispensing by introducing artificial intelligence technology.The machine's six-axis manipulator automatically dispenses glue,but the traditional algorithm has unclear recognition results in recognizing the contour of the edge of the shoe sole.It cannot form an effective dispensing path for planning,and it cannot complete the planning of the dispensing path within the specified time.It seriously affected the efficiency and accuracy of shoe sole dispensing production.Aiming at the problems of complex algorithm implementation,low path planning efficiency and unreasonable planning in the existing dispensing processing methods,this paper proposes a method for intelligently planning shoe sole dispensing paths based on the combination of Canny algorithm and SOM neural network.This method first obtains the sole model of the production line through visual technology and a line laser scanner,and extracts the edge trajectory points of the sole through recognition,and gradually standardizes the sole dispensing plan points to form a two-dimensional dispensing map mapping,and learns through competition for all sole dispensing path points.The algorithm is connected to form the path of sole dispensing,which is used for the motion trajectory control of the six-axis manipulator.The main research contents of this paper are as follows:By using the sole model data scanned by the CCD camera and the line laser sensor,the sole model image matrix is weighted and averaged to gray,and the second-order Gaussian convolution kernel is used to smooth the noise,and the smoothed by the first-order sobel operator gradient template The image is constrained by the extreme value of the gradient amplitude.The grayscale image is divided into two parts and the variance is calculated based on the principle of the maximum difference between the grayscale classes and the minimum difference within the class.The maximum variance value is iteratively selected as the high threshold,in order to take into account the time efficiency The factor sets the low threshold to one-half the high threshold,and finally uses the high and low thresholds to traverse the 8-connected pixel domain of each edge point to select effective edge points and retain the generated effective edge point matrix.Build a two-layer structure of the SOM neural network model,add vertex offset planning to the input layer to reduce the size of the point set,self-intersection to remove some unreasonable planning points,establish a two-dimensional coordinate system and indicate the coordinate position of each dispensing planning point,and use the TSP file The format specification generates the data of the two-dimensional dispensing point of the shoe sole and maps it into a two-dimensional dispensing plane dispensing map.Through the iteration of the competition layer,adjust the weights of the sole dispensing trajectory points,update the position of the winning neuron,keep approaching the dispensing planning point,and connect them in the order in which the winning neurons appear to form the dispensing path of the sole dispensing robot.This algorithm has been applied to the sports shoes of internationally renowned brands,and has achieved good results in the edge recognition of the sole model and the dispensing path planning.Through experiments and comparisons of the same type of measurement algorithms and dispensing methods,the path planning time is used to identify Time,dispensing uniformity,and planning rationality are compared for the evaluation system.Experiments show that the planning results based on Canny algorithm and SOM neural network model are generally superior to other sole dispensing planning algorithms in terms of planning time and dispensing uniformity.