Automatic warehouse barcode generation system based on warehouse management

Through the automated warehousing barcode generation system, point cloud data and artemisinin optimization algorithm are used to solve the problem of low space utilization in traditional warehousing management, and more efficient shelf space utilization is achieved.

CN118691209BActive Publication Date: 2024-11-19NANJING KUOYOU INFORMATION TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411171019.2
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-26
Publication Date
2024-11-19
Estimated Expiration
2044-08-26

AI Technical Summary

Technical Problem

Traditional warehousing management relies on manual operations, is inefficient and has low space utilization. The existing technology fails to effectively utilize the vertical space of the shelves, resulting in low space utilization.

Method used

Using an automated warehousing barcode generation system based on warehousing management, through point cloud data collection and processing, combined with artemisinin optimization algorithm, it automatically identifies and calculates the three-dimensional remaining storage space in the shelf area and the three-dimensional storage requirements of the goods to be put into storage to generate optimized Warehousing plan.

Benefits of technology

Improve the utilization rate of warehouse space, and achieve more efficient use of shelf space through automated identification and allocation of goods.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118691209B_ABST
    Figure CN118691209B_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of warehouse management, and discloses an automatic warehouse barcode generation system based on warehouse management, comprising: a point cloud data processing module, which processes first point cloud data of multiple viewing angles of all shelf areas to obtain second point cloud data; a remaining storage space calculation module, which calculates the remaining storage space of the shelf area according to the second point cloud data; a goods required space calculation module, which obtains the storage space required for goods to be stored; a warehouse plan generation module, which generates a warehouse plan according to the remaining storage space of the shelf area and the storage space required for the goods to be stored by an artemisinin optimization algorithm; the invention automatically identifies and calculates the three-dimensional remaining storage space of the shelf area by combining three-dimensional point cloud scanning with a neural network model, and automatically identifies the three-dimensional required storage space of the goods to be stored by a target detection technology, and realizes automatic allocation of the goods to be stored by an artemisinin optimization algorithm, thereby improving space utilization.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of warehouse management, and more specifically, to an automatic warehouse barcode generation system based on warehouse management. Background Art

[0002] Traditional warehouse management is highly dependent on manual operations. From space planning, goods in and out management to inventory counting, most of them rely on manual experience and manual records. This method is not only inefficient but also has low utilization rate of warehouse space.

[0003] The existing technology uses a Gaussian filter to detect the edges of the shelf area in the image, and binarizes the image through the maximum inter-class difference method. The binarized background area is calculated through a rectangular box as the remaining storage space of the shelf area, and the storage plan with the lowest storage cost and transportation cost is obtained based on the storage space required for the goods to be stored in the warehouse in combination with the ant colony algorithm.

[0004] However, the above solution only considers the two-dimensional plane space of the shelf area, but does not consider the vertical space of the shelf area, resulting in low space utilization. Summary of the invention

[0005] The present invention provides an automated warehouse barcode generation system based on warehouse management, which solves the technical problems in the above-mentioned background technology.

[0006] The present invention provides an automatic warehouse barcode generation system based on warehouse management, comprising:

[0007] A point cloud data acquisition module, which is used to collect first point cloud data from multiple perspectives of all shelf areas in the warehouse;

[0008] Each shelf area is assigned a unique positive integer number;

[0009] The entrance plane composed of different rows and columns of each shelf area is assigned a unique positive integer number;

[0010] The position of a data point of the first point cloud data is represented by a combination of a horizontal coordinate, a vertical coordinate and a vertical coordinate;

[0011] A point cloud data processing module, which is used to process the first point cloud data of multiple perspectives of all shelf areas to obtain second point cloud data;

[0012] The second point cloud data has the same representation as the first point cloud data;

[0013] A remaining storage space calculation module, which is used to calculate the remaining storage space of all shelf areas according to the second point cloud data;

[0014] A cargo space calculation module is used to calculate the storage space required for the cargo to be stored based on the image of the cargo to be stored;

[0015] A storage plan generation module, which is used to generate a storage plan based on the remaining storage space of all shelf areas and the storage space required for all goods to be stored through an artemisinin optimization algorithm;

[0016] Allocate incoming goods to rows and columns of different shelf areas according to the storage plan;

[0017] The storage barcode generation module is used to generate a storage barcode for each item to be stored according to the storage plan.

[0018] Further, the first point cloud data of multiple perspectives of all shelf areas are processed to obtain second point cloud data, including the following steps:

[0019] Step S201, calculating the deviation distance of each data point in the first point cloud data of each viewing angle by using a Gaussian filter, and directly removing the data point when the deviation distance is greater than or equal to a first preset distance, wherein the first preset distance is a custom parameter;

[0020] Step S202, performing point cloud registration on first point cloud data of multiple perspectives by using a generalized iterative closest point algorithm to obtain first intermediate point cloud data;

[0021] Step S203, performing point cloud fusion on the first intermediate point cloud data of multiple perspectives by using a surface reconstruction algorithm to obtain second intermediate point cloud data;

[0022] Step S204: performing point cloud thinning on the second intermediate point cloud data by using a density-based clustering algorithm to obtain second point cloud data.

[0023] Furthermore, the calculation formula of the Gaussian filter includes:

[0024] ;

[0025] ;

[0026] ;

[0027] in Represents the deviation distance of the i-th data point of the first point cloud data, and Respectively represent the positions of the i-th data point and the j-th data point of the first point cloud data, represents the updated position of the i-th data point of the first point cloud data, It means that a set of data points within a spherical range is generated with the i-th data point of the first point cloud data as the center and the third preset distance as the radius. and Respectively represent the weight coefficient and Euclidean distance between the i-th data point and the j-th data point of the first point cloud data, represents the standard deviation of the Gaussian kernel, e represents a natural constant, and the standard deviation of the Gaussian kernel and the third preset distance are both custom parameters.

[0028] Further, performing point cloud registration on the first point cloud data of multiple perspectives by using a generalized iterative closest point algorithm to obtain first intermediate point cloud data includes the following steps:

[0029] Step S301, randomly selecting first point cloud data from two perspectives as source point cloud data and target point cloud data respectively;

[0030] Step S302, searching, in the target point cloud data, for each data point in the source point cloud data, a data point with the smallest Euclidean distance thereto as a data pair;

[0031] Step S303, establishing a local coordinate system for each data pair, calculating the covariance matrix of the data pair in the local coordinate system, and using the covariance matrix to calculate the Mahalanobis distance between two data points of the data pair;

[0032] Select the data point of the data pair in the target point cloud data as the origin, and the horizontal axis component, vertical axis component and vertical axis component of the normal vector of the origin as the horizontal axis, vertical axis and vertical axis of the local coordinates;

[0033] The Mahalanobis distance between two data points of a data pair The calculation formula is as follows:

[0034] ;

[0035] in The vector representation of the position difference of the two data points representing the data pair, Represents the covariance matrix of the data pair in the local coordinate system, and T represents the transpose operation;

[0036] Step S304, specifying the sum of the Mahalanobis distances between two data points of all data pairs as the loss function, and updating the transformation matrix by a gradient descent algorithm until the iteration termination condition is met;

[0037] The transformation matrix is ​​expressed as: ;

[0038] Where rotate represents a rotation matrix of size 3×3, and translate represents a translation vector of size 3×1;

[0039] The iteration termination conditions include: reaching the maximum number of iterations or the loss value obtained by the loss function within three consecutive iterations is lower than the preset loss threshold, where the preset loss threshold is a custom parameter;

[0040] Step S305, randomly select the first point cloud data of the remaining perspectives as the source point cloud data, select the target point cloud data of step S301 as the target point cloud data, and repeat steps S302 to S304 until the first intermediate point cloud data of all perspectives is obtained for point cloud registration.

[0041] Further, calculating the remaining storage space of all shelf areas according to the second point cloud data includes the following steps:

[0042] Step S401, identifying the shelf area and the stored goods inside the shelf area in the second point cloud data by using a point cloud semantic segmentation model;

[0043] Step S402, extracting four key points of the entrance plane composed of each row and column of all shelf areas through the point cloud semantic segmentation model, and calculating the area of ​​the entrance plane;

[0044] Step S403, calculating the minimum Euclidean distance between the entrance plane composed of each row and column of all shelf areas and the goods already in storage, and multiplying it by the area of ​​the entrance plane to obtain the remaining storage space of all shelf areas.

[0045] Furthermore, the sample labels of the training samples of the point cloud semantic segmentation model are obtained by manual annotation. The sample labels of the training samples include: the category and position of the data points. The difference between the value output by the point cloud semantic segmentation model in each iteration and the sample label of the training sample is specified as the loss function, and the weight parameters and bias parameters of the point cloud semantic segmentation model are updated by the gradient descent algorithm.

[0046] Further, the image of the goods to be stored is input into the space prediction model, and the storage space required for the goods to be stored is output;

[0047] The spatial prediction model includes: a first hidden layer and a second hidden layer;

[0048] The first hidden layer is built based on the object detection model to identify the eight corner points of the goods to be stored;

[0049] The sample labels of the training samples of the first hidden layer are obtained by manual labeling;

[0050] The second hidden layer is used to calculate the storage space required for the goods to be stored according to the eight corner points of the goods to be stored;

[0051] The calculation formula for obtaining the storage space required for the goods to be stored based on the eight corner points of the goods to be stored is as follows:

[0052] ;

[0053] in Represents the Euclidean distance between two adjacent corner points, max represents the maximum value, width represents the width, and represents The maximum value of the Euclidean distance between two adjacent corner points of the middle corner point. Height represents the height and represents the distance between the middle corner point and the corner point connected to the middle corner point. The middle corner point is represented by The two corner points of The two corner points of are adjacent to each other. Zoom represents the zoom factor and is a custom parameter with a value range of 0 to 1. It represents the Euclidean distance between the camera that takes the image of the goods to be stored and the goods to be stored.

[0054] Furthermore, a storage plan is generated by an artemisinin optimization algorithm according to the remaining storage space of all shelf areas and the storage space required for all goods to be stored, including the following steps:

[0055] Step S501, randomly generate an initialization population with U individuals that meets the constraint conditions, each individual in the initialization population is represented by a vector code, and the current iteration number t is initialized to 0, and the maximum iteration number Max is initialized to Q, where U and Q are custom parameters;

[0056] The constraint condition is: the remaining storage space of all shelf areas is greater than or equal to the storage space required for all goods to be stored;

[0057] The length of the individual vector code is M, where M represents the number of goods to be stored. The mth dimension value of the vector code is represented by a real number code. The integer part of the real number code represents the unique positive integer number of the shelf area to which the goods to be stored with a unique positive integer number m are assigned. The decimal part of the real number code represents the unique positive integer number of the entrance plane composed of rows and columns of the shelf area.

[0058] Step S502, calculating the fitness value of each individual in the initialized population through a fitness function;

[0059] The calculation formula of the fitness function is as follows:

[0060] ;

[0061] Where Fit represents the fitness value of an individual. and They represent the storage cost and transportation distance of the goods to be stored in an individual with a unique positive integer number m. and denote the first weight coefficient and the second weight coefficient respectively, and These are all custom parameters. The storage cost of the goods to be stored is manually marked, and the value range is 1 to 10. The larger the value, the higher the storage cost.

[0062] Step S503, generating a first random number in the range of 0 to 1, and determining that the first random number is less than the probability coefficient K, then proceeding to step S504, otherwise proceeding to step S505;

[0063] The calculation formula of probability coefficient K is as follows:

[0064] ;

[0065] Where t represents the current number of iterations, and Max represents the maximum number of iterations;

[0066] Step S504, updating the vector codes of U individuals in the initialization population through the calculation formula of the comprehensive elimination stage;

[0067] Step S505, updating the vector codes of U individuals in the initialization population through the calculation formula of the local clearing stage;

[0068] Step S506, updating the vector codes of U individuals of the initialized population by the calculation formula of the post-consolidation stage;

[0069] Step S507, the current number of iterations is increased by 1. When it is determined that the current number of iterations is greater than or equal to the maximum number of iterations, the individual with the largest fitness value in the initialized population is used as the storage solution, otherwise, the process returns to step S502 to continue execution.

[0070] Furthermore, the calculation formula of the artemisinin optimization algorithm includes:

[0071] The calculation formula for the comprehensive elimination stage includes:

[0072] ;

[0073] ;

[0074] Where 1≤i≤U, and Respectively represent the j-th dimension value of the vector encoding of the i-th individual in the initialization population with the current iteration number t+1 and t, represents the j-th dimension value of the vector encoding of the individual with the largest fitness value in the initialization population with the current iteration number t, c represents the first attenuation coefficient, represents a second random number ranging from 0 to 1, and e represents a natural constant;

[0075] The calculation formula for the local clearance stage includes:

[0076] ;

[0077] ;

[0078] ;

[0079] Where 1≤i≤U, represents the vector code of the i-th individual in the initialization population with the current iteration number t+1, , and They represent the vector codes of the b1th, b2th, and b3th individuals in the initialization population with the current iteration number t, respectively. and They represent the fitness value and normalized fitness value of the i-th individual in the initialization population with the current iteration number t, and They represent the maximum and minimum fitness values ​​of individuals in the initialization population with the current iteration number t, d represents the second attenuation coefficient, Indicates a third random number ranging from 0 to 1;

[0080] The calculation formula for the post-consolidation phase is as follows:

[0081] ;

[0082] Where 1≤i≤U, and Respectively represent the j-th dimension value of the vector encoding of the i-th individual in the initialization population with the current iteration number t+1 and t, Represents the j-th dimension value of the vector encoding of the individual with the largest fitness value in the initialization population with the current iteration number t, Indicates a fourth random number ranging from 0 to 1.

[0083] Furthermore, the unique positive integer number of the shelf area to which each incoming cargo is allocated and the unique positive integer number of the corresponding entrance plane composed of rows and columns are converted into an eight-bit binary code, and the two eight-bit binary codes are concatenated to generate a storage barcode.

[0084] The beneficial effects of the present invention are as follows: the present invention automatically identifies and calculates the three-dimensional remaining storage space of the shelf area through three-dimensional point cloud scanning combined with a neural network model, and automatically identifies the three-dimensional storage space required for goods to be stored through target detection technology, and realizes automatic allocation of goods to be stored through an artemisinin optimization algorithm, thereby improving space utilization. BRIEF DESCRIPTION OF THE DRAWINGS

[0085] Figure 1 It is a schematic diagram of an automatic warehouse barcode generation system based on warehouse management of the present invention;

[0086] Figure 2 is a flow chart of the present invention for processing first point cloud data of multiple perspectives of all shelf areas to obtain second point cloud data;

[0087] Figure 3 It is a flow chart of obtaining first intermediate point cloud data by performing point cloud registration on first point cloud data of multiple perspectives through a generalized iterative closest point algorithm of the present invention;

[0088] Figure 4 is a flow chart of calculating and obtaining the remaining storage space of all shelf areas according to the second point cloud data of the present invention;

[0089] Figure 5 It is a flow chart of the present invention for generating a storage plan through an artemisinin optimization algorithm according to the remaining storage space of all shelf areas and the storage space required for all goods to be stored.

[0090] In the figure: point cloud data acquisition module 101, point cloud data processing module 102, remaining storage space calculation module 103, cargo required space calculation module 104, storage plan generation module 105, storage barcode generation module 106. DETAILED DESCRIPTION

[0091] The subject matter described herein will now be discussed with reference to example embodiments. It should be understood that the discussion of these embodiments is only to enable those skilled in the art to better understand and implement the subject matter described herein, and the functions and arrangements of the elements discussed may be changed without departing from the scope of protection of the contents of this specification. Each example may omit, replace or add various processes or components as needed. In addition, the features described relative to some examples may also be combined in other examples.

[0092] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in one or more embodiments of the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in one or more embodiments of the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0093] like Figure 1-Figure 5 As shown, the automatic warehouse barcode generation system based on warehouse management includes:

[0094] Point cloud data collection module 101, which is used to collect first point cloud data of multiple perspectives of all shelf areas in the warehouse;

[0095] Each shelf area is assigned a unique positive integer number;

[0096] The entrance plane composed of different rows and columns of each shelf area is assigned a unique positive integer number;

[0097] The position of a data point of the first point cloud data is represented by a combination of a horizontal coordinate, a vertical coordinate and a vertical coordinate;

[0098] A point cloud data processing module 102, which is used to process the first point cloud data of multiple perspectives of all shelf areas to obtain second point cloud data;

[0099] The second point cloud data has the same representation as the first point cloud data;

[0100] A remaining storage space calculation module 103, which is used to calculate the remaining storage space of all shelf areas according to the second point cloud data;

[0101] A cargo space calculation module 104 is used to calculate the storage space required for the cargo to be stored based on the image of the cargo to be stored;

[0102] A storage plan generation module 105, which is used to generate a storage plan according to the remaining storage space of all shelf areas and the storage space required for all goods to be stored by using an artemisinin optimization algorithm;

[0103] Allocate incoming goods to rows and columns of different shelf areas according to the storage plan;

[0104] The storage barcode generation module 106 is used to generate a storage barcode for each of the goods to be stored according to the storage plan.

[0105] In one embodiment of the present invention, first point cloud data of multiple viewing angles of the shelf area are collected by a three-dimensional laser scanner.

[0106] In one embodiment of the present invention, Figure 2 As shown, the first point cloud data of multiple perspectives of all shelf areas are processed to obtain the second point cloud data, including the following steps:

[0107] Step S201, calculating the deviation distance of each data point in the first point cloud data of each viewing angle by using a Gaussian filter, and directly removing the data point when the deviation distance is greater than or equal to a first preset distance, wherein the first preset distance is a custom parameter;

[0108] Step S202, performing point cloud registration on the first point cloud data of multiple perspectives by using GICP (generalized iterative closest point algorithm) to obtain first intermediate point cloud data;

[0109] Step S203, performing point cloud fusion on the first intermediate point cloud data of multiple perspectives by using a surface reconstruction algorithm to obtain second intermediate point cloud data;

[0110] The surface reconstruction algorithm may be a Poisson surface reconstruction algorithm or a Laplace surface reconstruction algorithm, etc., which will not be described in detail here;

[0111] Step S204, performing point cloud thinning on the second intermediate point cloud data by using a density-based clustering algorithm to obtain second point cloud data;

[0112] Traverse each data point of the second intermediate point cloud data, and generate a spherical range with the data point as the center and the second preset distance as the radius, calculate the Euclidean distance between the remaining data points in the spherical range and the center point, and take the median as the distance threshold to remove data points less than the distance threshold.

[0113] It should be noted that the Gaussian filter helps to suppress noise, making the overall point cloud data more continuous and regular; point cloud registration unifies multi-perspective point cloud data into the same coordinate system to facilitate subsequent point cloud fusion; point cloud thinning through density-based clustering algorithms can reduce subsequent calculations while retaining point cloud feature data points.

[0114] In one embodiment of the present invention, the calculation formula of the Gaussian filter includes:

[0115] ;

[0116] ;

[0117] ;

[0118] in Represents the deviation distance of the i-th data point of the first point cloud data, and Respectively represent the positions of the i-th data point and the j-th data point of the first point cloud data, represents the updated position of the i-th data point of the first point cloud data, It means that a set of data points within a spherical range is generated with the i-th data point of the first point cloud data as the center and the third preset distance as the radius. and Respectively represent the weight coefficient and Euclidean distance between the i-th data point and the j-th data point of the first point cloud data, represents the standard deviation of the Gaussian kernel, e represents a natural constant, and the standard deviation of the Gaussian kernel and the third preset distance are both custom parameters.

[0119] In one embodiment of the present invention, the first intermediate point cloud data may also be obtained by performing point cloud registration on the first point cloud data of multiple perspectives using an ICP (Iterative Closest Point) algorithm.

[0120] In one embodiment of the present invention, the second point cloud data may also be obtained by performing point cloud thinning on the second intermediate point cloud data using an FPS (Furthest Point Sampling) algorithm.

[0121] In one embodiment of the present invention, Figure 3 As shown, performing point cloud registration on first point cloud data of multiple perspectives by using a generalized iterative closest point algorithm to obtain first intermediate point cloud data includes the following steps:

[0122] Step S301, randomly selecting first point cloud data from two perspectives as source point cloud data and target point cloud data respectively;

[0123] Step S302, searching, in the target point cloud data, for each data point in the source point cloud data, a data point with the smallest Euclidean distance thereto as a data pair;

[0124] Step S303, establishing a local coordinate system for each data pair, calculating the covariance matrix of the data pair in the local coordinate system, and using the covariance matrix to calculate the Mahalanobis distance between two data points of the data pair;

[0125] Select the data point of the data pair in the target point cloud data as the origin, and the horizontal axis component, vertical axis component and vertical axis component of the normal vector of the origin as the horizontal axis, vertical axis and vertical axis of the local coordinates;

[0126] The Mahalanobis distance between two data points of a data pair The calculation formula is as follows:

[0127] ;

[0128] in The vector representation of the position difference of the two data points representing the data pair, Represents the covariance matrix of the data pair in the local coordinate system, and T represents the transpose operation;

[0129] Step S304, specifying the sum of the Mahalanobis distances between two data points of all data pairs as the loss function, and updating the transformation matrix by a gradient descent algorithm until the iteration termination condition is met;

[0130] The transformation matrix is ​​expressed as: ;

[0131] Where rotate represents a rotation matrix of size 3×3, and translate represents a translation vector of size 3×1;

[0132] The iteration termination conditions include: reaching the maximum number of iterations or the loss value obtained by the loss function within three consecutive iterations is lower than the preset loss threshold, where the preset loss threshold is a custom parameter;

[0133] Step S305, randomly select the first point cloud data of the remaining perspectives as the source point cloud data, select the target point cloud data of step S301 as the target point cloud data, and repeat steps S302 to S304 until the first intermediate point cloud data of all perspectives is obtained for point cloud registration.

[0134] In one embodiment of the present invention, Figure 4 As shown, calculating the remaining storage space of all shelf areas according to the second point cloud data includes the following steps:

[0135] Step S401, identifying the shelf area and the stored goods inside the shelf area in the second point cloud data by using a point cloud semantic segmentation model;

[0136] Step S402, extracting four key points of the entrance plane composed of each row and column of all shelf areas through the point cloud semantic segmentation model, and calculating the area of ​​the entrance plane;

[0137] Step S403, calculating the minimum Euclidean distance between the entrance plane composed of each row and column of all shelf areas and the goods already in storage, and multiplying it by the area of ​​the entrance plane to obtain the remaining storage space of all shelf areas.

[0138] In one embodiment of the present invention, the sample labels of the training samples of the point cloud semantic segmentation model are obtained by manual annotation, and the sample labels of the training samples include: the category and position of the data point, and the difference between the value output by the point cloud semantic segmentation model in each iteration and the sample label of the training sample is specified as the loss function, and the weight parameters and bias parameters of the point cloud semantic segmentation model are updated by the gradient descent algorithm.

[0139] In one embodiment of the present invention, the point cloud semantic segmentation model may be PointNet++ or RandLA-Net, etc.

[0140] In one embodiment of the present invention, an image of goods to be stored is input into a space prediction model, and the storage space required for the goods to be stored is output;

[0141] The spatial prediction model includes: a first hidden layer and a second hidden layer;

[0142] The first hidden layer is built based on the object detection model to identify the eight corner points of the goods to be stored;

[0143] The sample labels of the training samples of the first hidden layer are obtained by manual labeling;

[0144] The target detection model can be YOLO or Faster R-CNN, etc.

[0145] The second hidden layer is used to calculate the storage space required for the goods to be stored according to the eight corner points of the goods to be stored;

[0146] The calculation formula for obtaining the storage space required for the goods to be stored based on the eight corner points of the goods to be stored is as follows:

[0147] ;

[0148] in Represents the Euclidean distance between two adjacent corner points, max represents the maximum value, width represents the width, and represents The maximum value of the Euclidean distance between two adjacent corner points of the middle corner point. Height represents the height and represents the distance between the middle corner point and the corner point connected to the middle corner point. The middle corner point is represented by The two corner points of The two corner points of are adjacent to each other. Zoom represents the zoom factor and is a custom parameter with a value range of 0 to 1. It represents the Euclidean distance between the camera that takes the image of the goods to be stored and the goods to be stored.

[0149] In one embodiment of the present invention, Figure 5As shown, a storage plan is generated by the artemisinin optimization algorithm according to the remaining storage space of all shelf areas and the storage space required for all goods to be stored, including the following steps:

[0150] Step S501, randomly generate an initialization population with U individuals that meets the constraint conditions, each individual in the initialization population is represented by a vector code, and initialize the current iteration number t to 0, and initialize the maximum iteration number Max to Q, where U and Q are custom parameters, preferably, U is set to 20, and Q is set to 200;

[0151] The constraint condition is: the remaining storage space of all shelf areas is greater than or equal to the storage space required for all goods to be stored;

[0152] The length of the individual vector code is M, where M represents the number of goods to be stored. The mth dimension value of the vector code is represented by a real number code. The integer part of the real number code represents the unique positive integer number of the shelf area to which the goods to be stored with a unique positive integer number m are assigned. The decimal part of the real number code represents the unique positive integer number of the entrance plane composed of rows and columns of the shelf area.

[0153] For example, if the first dimension value of the vector code is 21.31, it means that the code of the shelf area assigned to the goods to be stored with the number 1 is 21, and the entrance plane composed of the rows and columns of the shelf area is 31;

[0154] Step S502, calculating the fitness value of each individual in the initialized population through a fitness function;

[0155] The calculation formula of the fitness function is as follows:

[0156] ;

[0157] Where Fit represents the fitness value of an individual. and They represent the storage cost and transportation distance of the goods to be stored in an individual with a unique positive integer number m. and denote the first weight coefficient and the second weight coefficient respectively, and These are all custom parameters. The storage cost of the goods to be stored is manually marked, and the value range is 1 to 10. The larger the value, the higher the storage cost.

[0158] Step S503, generating a first random number in the range of 0 to 1, and determining that the first random number is less than the probability coefficient K, then proceeding to step S504, otherwise proceeding to step S505;

[0159] The calculation formula of probability coefficient K is as follows:

[0160] ;

[0161] Where t represents the current number of iterations, and Max represents the maximum number of iterations;

[0162] Step S504, updating the vector codes of U individuals in the initialization population through the calculation formula of the comprehensive elimination stage;

[0163] Step S505, updating the vector codes of U individuals in the initialization population through the calculation formula of the local clearing stage;

[0164] Step S506, updating the vector codes of U individuals of the initialized population by the calculation formula of the post-consolidation stage;

[0165] Step S507, the current number of iterations is increased by 1. When it is determined that the current number of iterations is greater than or equal to the maximum number of iterations, the individual with the largest fitness value in the initialized population is used as the storage solution, otherwise, the process returns to step S502 to continue execution.

[0166] In one embodiment of the present invention, the calculation formula of the artemisinin optimization algorithm includes:

[0167] The calculation formula for the comprehensive elimination stage includes:

[0168] ;

[0169] ;

[0170] Where 1≤i≤U, and Respectively represent the j-th dimension value of the vector encoding of the i-th individual in the initialization population with the current iteration number t+1 and t, represents the j-th dimension value of the vector encoding of the individual with the largest fitness value in the initialization population with the current iteration number t, c represents the first attenuation coefficient, represents a second random number ranging from 0 to 1, and e represents a natural constant;

[0171] The calculation formula for the local clearance stage includes:

[0172] ;

[0173] ;

[0174] ;

[0175] Where 1≤i≤U, represents the vector code of the i-th individual in the initialization population with the current iteration number t+1, , and They represent the vector codes of the b1th, b2th, and b3th individuals in the initialization population with the current iteration number t, respectively. and They represent the fitness value and normalized fitness value of the i-th individual in the initialization population with the current iteration number t, and They represent the maximum and minimum fitness values ​​of individuals in the initialization population with the current iteration number t, d represents the second attenuation coefficient, Indicates a third random number ranging from 0 to 1;

[0176] The calculation formula for the post-consolidation phase is as follows:

[0177] ;

[0178] Where 1≤i≤U, and Respectively represent the j-th dimension value of the vector encoding of the i-th individual in the initialization population with the current iteration number t+1 and t, Represents the j-th dimension value of the vector encoding of the individual with the largest fitness value in the initialization population with the current iteration number t, Indicates a fourth random number ranging from 0 to 1.

[0179] It should be noted that the comprehensive elimination stage simulates the process of artemisinin's first comprehensive attack on malarial parasites, aiming to conduct a global search, explore a wide area of ​​the solution space, avoid falling into the local optimum too early, maintain the diversity of individuals in the initialized population, and increase the possibility of finding the global optimal solution; the local elimination stage is similar to artemisinin's targeted attack on the remaining parasites, and an in-depth search in the neighborhood of known high-quality solutions can speed up the algorithm's convergence to the optimal solution; the post-consolidation stage is equivalent to the consolidation therapy in the later stage of treatment, ensuring that all parasites are completely eliminated to prevent recurrence and avoid falling into the same local optimal solution again.

[0180] In one embodiment of the present invention, the unique positive integer number of the shelf area assigned to each to-be-warehoused cargo and the unique positive integer number of the corresponding entrance plane composed of rows and columns are converted into an eight-bit binary code, and the two eight-bit binary codes are concatenated to generate a storage barcode.

[0181] The above describes an embodiment of the present embodiment, but the present embodiment is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present embodiment, ordinary technicians in this field can also make many forms, all of which are within the protection of the present embodiment.

Claims

1. The automatic warehouse barcode generation system based on warehouse management is characterized by: include: A point cloud data acquisition module, which is used to collect first point cloud data from multiple perspectives of all shelf areas in the warehouse; Each shelf area is assigned a unique positive integer number; The entrance plane composed of different rows and columns of each shelf area is assigned a unique positive integer number; The position of a data point of the first point cloud data is represented by a combination of a horizontal coordinate, a vertical coordinate and a vertical coordinate; A point cloud data processing module, which is used to process the first point cloud data of multiple perspectives of all shelf areas to obtain second point cloud data; The second point cloud data has the same representation as the first point cloud data; A remaining storage space calculation module, which is used to calculate the remaining storage space of all shelf areas according to the second point cloud data; A cargo space calculation module is used to calculate the storage space required for the cargo to be stored based on the image of the cargo to be stored; A storage plan generation module, which is used to generate a storage plan based on the remaining storage space of all shelf areas and the storage space required for all goods to be stored through an artemisinin optimization algorithm; Allocate incoming goods to rows and columns of different shelf areas according to the storage plan; A storage barcode generation module is used to generate a storage barcode for each incoming cargo according to the storage plan; Processing the first point cloud data of multiple perspectives of all shelf areas to obtain second point cloud data includes the following steps: Step S201, calculating the deviation distance of each data point in the first point cloud data of each viewing angle by using a Gaussian filter, and directly removing the data point when the deviation distance is greater than or equal to a first preset distance, wherein the first preset distance is a custom parameter; Step S202, performing point cloud registration on first point cloud data of multiple perspectives by using a generalized iterative closest point algorithm to obtain first intermediate point cloud data; Step S203, performing point cloud fusion on the first intermediate point cloud data of multiple perspectives by using a surface reconstruction algorithm to obtain second intermediate point cloud data; Step S204, performing point cloud thinning on the second intermediate point cloud data by using a density-based clustering algorithm to obtain second point cloud data; The storage plan is generated by the artemisinin optimization algorithm according to the remaining storage space of all shelf areas and the storage space required for all goods to be stored, including the following steps: Step S501, randomly generate an initialization population with U individuals that meets the constraint conditions, each individual in the initialization population is represented by a vector code, and the current iteration number t is initialized to 0, and the maximum iteration number Max is initialized to Q, where U and Q are custom parameters; The constraint condition is: the remaining storage space of all shelf areas is greater than or equal to the storage space required for all goods to be stored; The length of the individual vector code is M, where M represents the number of goods to be stored. The mth dimension value of the vector code is represented by a real number code. The integer part of the real number code represents the unique positive integer number of the shelf area to which the goods to be stored with a unique positive integer number m are assigned. The decimal part of the real number code represents the unique positive integer number of the entrance plane composed of rows and columns of the shelf area. Step S502, calculating the fitness value of each individual in the initialized population through a fitness function; The calculation formula of the fitness function is as follows: ; Where Fit represents the fitness value of an individual. and They represent the storage cost and transportation distance of the goods to be stored in an individual with a unique positive integer number m. and denote the first weight coefficient and the second weight coefficient respectively, and These are all custom parameters. The storage cost of the goods to be stored is manually marked, and the value range is 1 to 10. The larger the value, the higher the storage cost. Step S503, generating a first random number in the range of 0 to 1, and determining that the first random number is less than the probability coefficient K, then proceeding to step S504, otherwise proceeding to step S505; The calculation formula of probability coefficient K is as follows: ; Where t represents the current number of iterations, and Max represents the maximum number of iterations; Step S504, updating the vector codes of U individuals in the initialization population through the calculation formula of the comprehensive elimination stage; Step S505, updating the vector codes of U individuals in the initialization population through the calculation formula of the local clearing stage; Step S506, updating the vector codes of U individuals of the initialized population by the calculation formula of the post-consolidation stage; Step S507, the current number of iterations is increased by 1. When it is determined that the current number of iterations is greater than or equal to the maximum number of iterations, the individual with the largest fitness value in the initialized population is used as the storage solution, otherwise, the process returns to step S502 to continue execution.

2. The automatic warehouse barcode generation system based on warehouse management according to claim 1 is characterized in that: The calculation formula of Gaussian filter includes: ; ; ; in Represents the deviation distance of the i-th data point of the first point cloud data, and Respectively represent the positions of the i-th data point and the j-th data point of the first point cloud data, represents the updated position of the i-th data point of the first point cloud data, It means that a set of data points within a spherical range is generated with the i-th data point of the first point cloud data as the center and the third preset distance as the radius. and Respectively represent the weight coefficient and Euclidean distance between the i-th data point and the j-th data point of the first point cloud data, represents the standard deviation of the Gaussian kernel, e represents a natural constant, and the standard deviation of the Gaussian kernel and the third preset distance are both custom parameters.

3. The automatic warehouse barcode generation system based on warehouse management according to claim 1 is characterized in that: Performing point cloud registration on the first point cloud data of multiple perspectives by using a generalized iterative closest point algorithm to obtain first intermediate point cloud data includes the following steps: Step S301, randomly selecting first point cloud data from two perspectives as source point cloud data and target point cloud data respectively; Step S302, searching, in the target point cloud data, for each data point in the source point cloud data, a data point with the smallest Euclidean distance thereto as a data pair; Step S303, establishing a local coordinate system for each data pair, calculating the covariance matrix of the data pair in the local coordinate system, and using the covariance matrix to calculate the Mahalanobis distance between two data points of the data pair; Select the data point of the data pair in the target point cloud data as the origin, and the horizontal axis component, vertical axis component and vertical axis component of the normal vector of the origin as the horizontal axis, vertical axis and vertical axis of the local coordinates; The Mahalanobis distance between two data points of a data pair The calculation formula is as follows: ; in The vector representation of the position difference of the two data points representing the data pair, Represents the covariance matrix of the data pair in the local coordinate system, and T represents the transpose operation; Step S304, specifying the sum of the Mahalanobis distances between two data points of all data pairs as the loss function, and updating the transformation matrix by a gradient descent algorithm until the iteration termination condition is met; The transformation matrix is ​​expressed as: ; Where rotate represents a rotation matrix of size 3×3, and translate represents a translation vector of size 3×1; The iteration termination conditions include: reaching the maximum number of iterations or the loss value obtained by the loss function within three consecutive iterations is lower than the preset loss threshold, where the preset loss threshold is a custom parameter; Step S305, randomly select the first point cloud data of the remaining perspectives as the source point cloud data, select the target point cloud data of step S301 as the target point cloud data, and repeat steps S302 to S304 until the first intermediate point cloud data of all perspectives is obtained for point cloud registration.

4. The automatic warehouse barcode generation system based on warehouse management according to claim 1 is characterized in that: Calculating the remaining storage space of all shelf areas according to the second point cloud data includes the following steps: Step S401, identifying the shelf area and the stored goods inside the shelf area in the second point cloud data by using a point cloud semantic segmentation model; Step S402, extracting four key points of the entrance plane composed of each row and column of all shelf areas through the point cloud semantic segmentation model, and calculating the area of ​​the entrance plane; Step S403, calculating the minimum Euclidean distance between the entrance plane composed of each row and column of all shelf areas and the goods already in storage, and multiplying it by the area of ​​the entrance plane to obtain the remaining storage space of all shelf areas.

5. The automatic warehouse barcode generation system based on warehouse management according to claim 4 is characterized in that: The sample labels of the training samples of the point cloud semantic segmentation model are obtained by manual annotation. The sample labels of the training samples include: the category and position of the data point. The difference between the value output by the point cloud semantic segmentation model in each iteration and the sample label of the training sample is specified as the loss function, and the weight parameters and bias parameters of the point cloud semantic segmentation model are updated by the gradient descent algorithm.

6. The automatic warehouse barcode generation system based on warehouse management according to claim 1 is characterized in that: Input the image of the goods to be stored into the space prediction model, and output the storage space required for the goods to be stored; The spatial prediction model includes: a first hidden layer and a second hidden layer; The first hidden layer is built based on the object detection model to identify the eight corner points of the goods to be stored; The sample labels of the training samples of the first hidden layer are obtained by manual labeling; The second hidden layer is used to calculate the storage space required for the goods to be stored according to the eight corner points of the goods to be stored; The calculation formula for obtaining the storage space required for the goods to be stored based on the eight corner points of the goods to be stored is as follows: ; in Represents the Euclidean distance between two adjacent corner points, max represents the maximum value, width represents the width, and represents The maximum value of the Euclidean distance between two adjacent corner points of the middle corner point. Height represents the height and represents the distance between the middle corner point and the corner point connected to the middle corner point. The middle corner point is represented by The two corner points of The two corner points of are adjacent to each other. Zoom represents the zoom factor and is a custom parameter with a value range of 0 to 1. It represents the Euclidean distance between the camera that takes the image of the goods to be stored and the goods to be stored.

7. The automatic warehouse barcode generation system based on warehouse management according to claim 1 is characterized in that: The calculation formula of the artemisinin optimization algorithm includes: The calculation formula for the comprehensive elimination stage includes: ; ; Where 1≤i≤U, and Respectively represent the j-th dimension value of the vector encoding of the i-th individual in the initialization population with the current iteration number t+1 and t, represents the j-th dimension value of the vector encoding of the individual with the largest fitness value in the initialization population with the current iteration number t, c represents the first attenuation coefficient, represents a second random number ranging from 0 to 1, and e represents a natural constant; The calculation formula for the local clearance stage includes: ; ; ; Where 1≤i≤U, represents the vector code of the i-th individual in the initialization population with the current iteration number t+1, , and They represent the vector codes of the b1th, b2th, and b3th individuals in the initialization population with the current iteration number t, respectively. and They represent the fitness value and normalized fitness value of the i-th individual in the initialization population with the current iteration number t, and They represent the maximum and minimum fitness values ​​of individuals in the initialization population with the current iteration number t, d represents the second attenuation coefficient, Indicates a third random number ranging from 0 to 1; The calculation formula for the post-consolidation phase is as follows: ; Where 1≤i≤U, and Respectively represent the j-th dimension value of the vector encoding of the i-th individual in the initialization population with the current iteration number t+1 and t, Represents the j-th dimension value of the vector encoding of the individual with the largest fitness value in the initialization population with the current iteration number t, Indicates a fourth random number ranging from 0 to 1.

8. The automatic warehouse barcode generation system based on warehouse management according to claim 1 is characterized in that: The unique positive integer number of the shelf area assigned to each incoming cargo and the unique positive integer number of the corresponding entrance plane composed of rows and columns are converted into an eight-bit binary code, and the two eight-bit binary codes are concatenated to generate a storage barcode.