Automatic palletizing robot operation control method and system

Abstract

The present invention relates to the field of robot control, and more specifically, to an automatic palletizing robot operation control method and system, the method comprising: obtaining point cloud data of a product to be grasped, clustering the coordinates and normal vectors of the point cloud data, obtaining multiple cluster clusters, determining whether the cluster cluster is a boundary line, and evaluating the effective coefficient of the boundary line; using the effective coefficient of the boundary line as a weight to obtain the matching degree, extracting the intersection of the boundary line, obtaining the characteristic information of the boundary line connected by the intersection, and matching the intersection of the boundary line of the template product using the KM algorithm to obtain the comprehensive matching degree; using the product to be grasped as the template product with the highest comprehensive matching degree, and controlling the mechanical arm to grasp and transport it to a set position according to the template product with the highest comprehensive matching degree. The present invention ensures the accuracy of the product matching template by evaluating the effectiveness of the boundary line fitted by clustering, adapts to different grasping requirements, and improves the adaptability and flexibility of the system.

Description

Technical Field

[0001] The present invention relates to the field of robot control, and more specifically, to an automatic palletizing robot operation control method and system. Background Art

[0002] With the rapid development of logistics and manufacturing industries, the application scope and development trend of automatic palletizing robots are also constantly expanding. It imitates the movements of human arms to achieve tasks such as handling, palletizing, and depalletizing of items. It has been widely used in food and beverage, chemical and pharmaceutical, logistics and warehousing and other fields, providing efficient, reliable and intelligent solutions for the packaging, transportation and storage of various products.

[0003] The existing Chinese patent document with announcement number CN104058260B discloses a robot automatic palletizing method based on visual processing, which adopts a camera calibration algorithm for a large area space for calibration, then collects images of the workpiece, and then analyzes and processes the images to identify the type of workpiece and the center of mass position, and the palletizing robot classifies and palletizes them. The present invention provides a highly efficient and intelligent solution for production automation. This solution introduces machine vision into industrial production, realizes the task of unattended sorting and palletizing, saves a lot of manpower, and can perform quality inspection on workpieces, and the control accuracy is much higher than manual accuracy.

[0004] In the above scheme, the product to be grasped is identified by Hough-chain code, but Hough transform can only determine the direction of the straight line, and loses the length information of the line segment, which reduces the accuracy of recognition. When the existing method is matched, the shape features of the items to be stacked are not accurately extracted, which affects the subsequent matching results, resulting in misjudgment or grasping failure of the stacking robot. Summary of the invention

[0005] In order to solve the problem that inaccurate shape feature extraction of objects to be stacked affects subsequent matching results, resulting in misjudgment or grasping failure of the stacking robot, the present invention provides solutions in the following aspects.

[0006] In a first aspect, an automatic palletizing robot operation control method includes: obtaining point cloud data of a product to be grasped, clustering the coordinates and normal vectors of the point cloud data to obtain multiple cluster clusters, determining whether the cluster cluster is a boundary line, and evaluating the effective coefficient of the boundary line; using the effective coefficient of the boundary line as a weight to obtain a matching degree, extracting the intersection of the boundary line, obtaining feature information of the boundary line connected by the intersection, and using the KM algorithm to match the intersection of the boundary line of the template product to obtain a comprehensive matching degree; using the product to be grasped as the template product with the highest comprehensive matching degree, controlling the robot arm to grasp and transport it to a set position according to the template product with the highest comprehensive matching degree; wherein the matching degree Satisfies the following relationship: ; In the formula, Indicates the first The first of the boundary lines connected by the matching points The effective coefficient of the boundary line, Indicates the first The first of the boundary lines connected by the matching points The length of the boundary line is Indicates the template product The first of the boundary lines connected by the matching points The length of the boundary line is Indicates the first The angle between two boundary lines connecting the matching points, Indicates the template product The angle between two boundary lines connecting the matching points, represents the cosine function, Indicates An exponential function with base .

[0007] The effect is: by using point cloud data and normal vectors for clustering, the shape and structure of the product can be identified more accurately, thereby improving the recognition accuracy of the product to be grasped, and evaluating the effective coefficient of the boundary line, it is possible to distinguish between the real boundary line and the non-boundary line, and use it as the weight for matching in the KM algorithm, and taking into account the consistency of the boundary line length and angle, which is beneficial to improving the efficiency and accuracy of the robot arm's grasping. Through precise identification and matching, the overall operating performance of the palletizing robot, as well as the operating speed and stability, can be improved.

[0008] Preferably, obtaining the coordinates and normal vector of the point cloud data includes:

[0009] Each point in the point cloud data contains three-dimensional coordinates For each data point, find the points in its neighborhood according to the set search radius, use the local fitting plane method to calculate the covariance matrix of the local neighborhood points, perform eigenvalue decomposition on the covariance matrix, and use the eigenvector corresponding to the minimum eigenvalue as the normal vector of the data point.

[0010] The effect is that in cluster analysis, normal vector information can help distinguish different surfaces and boundaries, thereby improving the clustering effect, making points belonging to the same plane or surface more likely to be classified into the same cluster cluster; normal vector information can support the palletizing robot to consider more factors when making grasping decisions, such as surface inclination, grasping angle, etc., so as to select the best grasping strategy.

[0011] Preferably, the determining whether the cluster is a boundary line includes:

[0012] The plane and boundary line are distinguished according to the number of data points in the cluster: a threshold is set, and the data points in the cluster that are less than or equal to the set threshold are considered to be boundary lines. Conversely, the clusters with data points greater than the set threshold are considered to be planes on the product surface.

[0013] The effect is that boundary lines and plane areas can be more accurately identified based on the statistical properties of the number of data points, because boundary lines are usually composed of fewer points, while plane areas contain more points.

[0014] Preferably, obtaining the effective coefficient of the boundary line includes:

[0015] Based on each cluster, the average normal vector of all point cloud data is calculated, and the average of the square of the Euclidean distance between the normal vector of each data point and the average normal vector is calculated to obtain the normal vector variance of the cluster; the clusters that are boundary lines are screened out to obtain the normal vector variance of two adjacent point cloud data belonging to the surface plane of the product;

[0016] The least squares method is used to fit the boundary line of each cluster, and the sum of the squares of the distances between each data point in the cluster and the fitted straight line is calculated to obtain the residual sum of squares of the fitted boundary line in the cluster, and the result is normalized. The product of the ratio of the normal vector variance of the cluster and the normal vector variance of the point cloud data and the normalized residual sum of squares is used as the effective coefficient of the boundary line of the cluster.

[0017] The effect is: using the least squares method to fit the boundary line for each cluster and calculating the residual sum of squares, the accuracy of the boundary line can be evaluated. The smaller the residual sum of squares, the closer the fitted boundary line is to the actual data point, and the higher the effectiveness of the boundary line.

[0018] Preferably, obtaining the effective coefficient of the boundary line of the cluster further includes:

[0019] Take any cluster as the target cluster, calculate the average distance from all points in the target cluster to the cluster center, get the compactness of the target cluster, calculate the distance between the cluster centers of all clusters, and get the separation between all clusters;

[0020] Use the least squares method to fit the boundary line of each cluster, calculate the sum of the squares of the vertical distances between the data points in the target cluster and the data points on the corresponding fitting straight line, obtain the degree of boundary line fitting, and perform normalization processing;

[0021] The product of the ratio of compactness to separation and the normalized boundary line fitting degree is taken as the effective coefficient of the boundary line.

[0022] Preferably, the use of the KM algorithm to match the intersection of the template product boundary line to obtain a comprehensive matching degree includes:

[0023] Create a matching matrix based on the matching degree, where each row in the matching matrix represents the matching point of the product to be captured, and the list is the intersection of the boundary lines of the template product. Input the matching matrix into the KM algorithm, and output the comprehensive matching result, where the comprehensive matching result is the average of the matching degrees of the matching points of the product to be captured and the intersection points of the upper boundary of the template.

[0024] The effect is that by constructing a matching matrix and using the KM algorithm, it can be ensured that the matching point of each product to be captured can be matched with the most similar boundary intersection on the template product, thereby improving the matching accuracy.

[0025] Preferably, the controlling robot arm to grab the template product with the highest comprehensive matching degree and transport it to the set position includes:

[0026] After identifying the product type of the template product with the highest comprehensive matching degree, obtain product information, select the appropriate grabbing method, accurately perform the grabbing action, and transport it to the set location;

[0027] The product information includes: shape, weight and material, and the grasping methods include: suction cup and clamp.

[0028] In a second aspect, an automatic palletizing robot operation control system includes: a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the automatic palletizing robot operation control method described above is implemented.

[0029] The present invention has the following effects:

[0030] 1. The present invention ensures the accuracy of subsequent product matching templates by evaluating the effectiveness of the boundary lines fitted by clustering, and also reduces the adverse effects caused by inaccurate clustering; by constructing the boundary line intersection matching degree feature, the similarity between the product to be captured and the template product is obtained, and the matching algorithm is used to realize automatic recognition of different products, thereby improving the adaptability and flexibility of the system.

[0031] 2. The present invention improves the accuracy of matching by evaluating the effectiveness coefficient of clustering clusters and using the effectiveness as the weight for matching calculation. It can adapt to products of different shapes, weights and materials. The system can automatically identify the product type and select the most suitable grasping method, such as a suction cup or a clamp, to meet different grasping requirements, thereby improving the practicality and market competitiveness of the automatic palletizing robot. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0033] Figure 1 It is a method flow chart of steps S1 to S3 in an automatic palletizing robot operation control method according to an embodiment of the present invention.

[0034] Figure 2 The present invention is a structural block diagram of an automatic palletizing robot operation control system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0035] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0036] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0037] The specific implementation scenario is: place a 3D camera on the palletizing robot to collect point cloud data of the product to be grasped, where the 3D camera should be placed higher than half of the height of the product, with an inclination angle of 30° and a distance of 3 meters from the product to be grasped to ensure that product details are captured and the entire product is covered. Before the palletizing robot grasps the product to be grasped, the palletizing robot collects three-dimensional point cloud data of the current product to be grasped from three different angles in a stationary state, and obtains the overall point cloud data of the product to be grasped for analysis and identification.

[0038] Reference Figure 1 , an automatic palletizing robot operation control method includes steps S1 to S3, which are as follows:

[0039] S1: Obtain the point cloud data of the product to be grasped, cluster the coordinates and normal vectors of the point cloud data to obtain multiple clusters, determine whether the cluster is a boundary line, and evaluate the effective coefficient of the boundary line.

[0040] Each point in the point cloud data contains three-dimensional coordinates For each data point, find the points in its neighborhood according to the set search radius, use the local fitting plane method to calculate the covariance matrix of the local neighborhood points, perform eigenvalue decomposition on the covariance matrix, and use the eigenvector corresponding to the minimum eigenvalue as the normal vector of the data point.

[0041] The DBSCAN algorithm is used to cluster the point cloud data. The points are assigned to different clusters according to their coordinates and normal vectors. The cluster radius and the minimum number of points are preset to obtain multiple clusters.

[0042] Further explanation: the clustering radius in the DBSCAN algorithm is 1 and the minimum number of points is 5.

[0043] The plane and boundary line are distinguished according to the number of data points in the cluster: a threshold is set, and the data points in the cluster that are less than or equal to the set threshold are considered to be boundary lines. Conversely, the clusters with data points greater than the set threshold are considered to be planes on the product surface.

[0044] Exemplarily, the threshold is 100, which can be adjusted according to actual conditions.

[0045] It should be noted that in three-dimensional space, plane regions usually contain a large number of points because they cover a large area. In contrast, boundary lines, as the junction of two plane regions, usually contain only a small number of points because they are one-dimensional in space. The threshold is based on the local density characteristics of point cloud data, reflecting the difference in data point distribution between plane regions and boundary lines.

[0046] That is to say, when clustering point cloud data, the coordinates and normal vectors of the data points in the point cloud are used. The normal vectors of different faces on the surface of the product to be grasped and the normal vectors of the boundary line between two different faces are different. Therefore, in an ideal case, the clustering result should divide the data points belonging to different faces and the data points of the boundary line into different clusters. The change of the normal vector of the point cloud data of any plane on the surface of the product to be grasped should be relatively gentle, and the change of the normal vector of the point cloud data on the boundary line is more obvious. Therefore, the variance of the clustering cluster can be used to evaluate the effectiveness of the boundary line in the clustering result. The specific steps are as follows:

[0047] Based on each cluster, the average normal vector of the normal vectors of all point cloud data is calculated, and the average of the square of the Euclidean distance between the normal vector of each data point and the average normal vector is calculated to obtain the normal vector variance of the cluster;

[0048] Specifically, the normal vector variance of the cluster satisfies the following relationship:

[0049] ;

[0050] In the formula, Indicates The variance of the normal vectors of all point cloud data in the clusters, Indicates The number of data points in a cluster, Indicates The first The square of the Euclidean distance between the normal vectors of the data points and the mean normal vector.

[0051] Filter out the clusters that are boundary lines, and obtain the normal vector variance of two adjacent point cloud data belonging to the surface plane of the product;

[0052] Use the least squares method to fit the boundary line of each cluster, calculate the sum of the squares of the distances between each data point in the cluster and the fitted straight line, obtain the residual sum of squares of the fitted boundary line in the cluster, and normalize it;

[0053] In other words, the larger the residual sum of squares, the greater the deviation from the fitted straight line, and the less effective the boundary line obtained by clustering.

[0054] The product of the ratio of the normal vector variance of the cluster and the normal vector variance of the point cloud data and the normalized residual sum of squares is taken as the effective coefficient of the boundary line of the cluster.

[0055] Specifically, the effective coefficient of the cluster fitting boundary line satisfies the following relationship:

[0056] ;

[0057] In the formula, Indicates The effective coefficient of the boundary line of the clusters, Indicates The variance of the normal vectors of all point cloud data in the clusters, Indicates The clusters correspond to the adjacent The variance of the normal vector of the point cloud data belonging to the surface plane of the product to be grasped, Indicates The clusters correspond to the adjacent The variance of the normal vector of the point cloud data belonging to the surface plane of the product to be grasped, Indicates The residual sum of squares of the fitted boundary lines in the clusters, Indicates An exponential function with base .

[0058] That is to say, on the boundary line of the product to be grasped, the larger the variance of the normal vector of all the point cloud data in the cluster, the greater the effectiveness of the cluster cluster fitting boundary line; the smaller the residual sum of squares of the fitted boundary line in the cluster, the closer the data points in the cluster are to the boundary line of the product to be grasped, and the greater the effectiveness of the cluster cluster fitting boundary line.

[0059] In addition, in another embodiment, the effectiveness of the boundary line is obtained by clustering the compactness and separation of the clusters.

[0060] Take any cluster as the target cluster, calculate the average distance from all points in the target cluster to the cluster center, get the compactness of the target cluster, calculate the distance between the cluster centers of all clusters, and get the separation between all clusters;

[0061] Specifically, the tightness satisfies the following relationship:

[0062] ;

[0063] In the formula, Indicates The compactness of the clusters, Indicates The number of data points in a cluster, represents the data points in the cluster, Indicates The cluster centers of the clusters.

[0064] Specifically, the separability satisfies the following relationship:

[0065] ;

[0066] In the formula, represents the separation between all clusters, represents the total number of clusters, Indicates The cluster centers of the clusters, Indicates The cluster centers of the clusters.

[0067] It is further explained that the compactness reflects the average distance between the points inside the cluster and the cluster center. The smaller the value, the tighter the points inside the cluster and the higher the clustering quality. The separability reflects the distance between the centers of different clusters. The larger the value, the better the separation between the clusters and the clearer the clustering effect. The better the clustering effect, the more accurate the boundary data points can be separated from the plane data points to form a separate cluster.

[0068] Use the least squares method to fit the boundary line of each cluster, calculate the sum of the squares of the vertical distances between the data points in the target cluster and the data points on the corresponding fitting straight line, obtain the degree of boundary line fitting, and perform normalization processing;

[0069] Specifically, the boundary line fitting degree satisfies the following relationship:

[0070] ;

[0071] In the formula, Indicates The degree of fitting of the boundary lines of the clusters. represents the data points in the cluster, Represents the data points on the fitted line.

[0072] It is further explained that the degree of boundary line fitting reflects the accuracy of the boundary line and the quality of fitting. If the degree of fitting is low, it means that the boundary line fits the data points in the cluster well.

[0073] The product of the ratio of compactness to separation and the normalized boundary line fitting degree is taken as the effective coefficient of the boundary line.

[0074] Specifically, the boundary line effective coefficient satisfies the following relationship:

[0075] ;

[0076] In the formula, Indicates The effective coefficient of the boundary line of the clusters, Indicates The compactness of the clusters, represents the separation between all clusters, Indicates The degree of fitting of the boundary lines of the clusters. Indicates An exponential function with base .

[0077] Finally, the boundary line effectiveness coefficient combines the compactness, separation and boundary line fitting degree to provide a comprehensive indicator to evaluate the overall effectiveness of the boundary line, which is helpful to help identify the boundary line that best fits the data distribution characteristics, thereby improving the accuracy and efficiency of subsequent processing.

[0078] It should be noted that after obtaining the boundary line of the product to be grasped, in order to realize the accurate recognition of the product to be grasped by the palletizing robot, the extracted boundary line needs to be matched with the contour boundary of the template product. However, directly using the boundary line for matching, since the position of the boundary line is fixed, the matching result is not in line with reality and is not accurate enough. Therefore, the intersection of the boundary line is used as the matching point and matched with the intersection of the boundary line on the template, and the boundary line feature connected to the matching point is used as the matching basis.

[0079] At the same time, it needs to be explained that there are multiple types of template products, and the products to be matched need to be matched with the template products respectively to obtain the template product with the highest matching degree as the target that the palletizing robot needs to grasp accurately.

[0080] S2: Use the effective coefficient of the boundary line as a weight to obtain the matching degree, extract the intersection points of the boundary lines, obtain the feature information of the boundary lines connected by the intersection points, and use the KM algorithm to match the intersection points of the template product boundary lines to obtain the comprehensive matching degree.

[0081] Further explanation, in step S2, the intersection points of each boundary line are extracted as matching points, and multiple matching points can be obtained. Matching points, get the All boundary lines connected by matching points And the length sequence corresponding to the boundary line , and the angle sequence between every two boundary lines Similarly, it is also necessary to obtain the first The length sequence of the boundary lines corresponding to the intersection points Angle sequence with boundary line ;

[0082] In the matching process, in order to avoid inaccurate matching results caused by inaccurate fitted boundary lines, weighted matching is performed considering the effectiveness of the boundary lines. The larger the effectiveness coefficient of the boundary line, the greater the matching weight of the boundary line.

[0083] Specifically, the matching degree satisfies the following relationship:

[0084] ;

[0085] In the formula, Indicates the first The matching point and the template product The matching degree of the matching points, Indicates the first The first of the boundary lines connected by the matching points The effective coefficient of the boundary line, Indicates the first The first of the boundary lines connected by the matching points The length of the boundary line is Indicates the template product The first of the boundary lines connected by the matching points The length of the boundary line is Indicates the first The angle between two boundary lines connecting the matching points, Indicates the template product The angle between two boundary lines connecting the matching points, represents the cosine function, Indicates An exponential function with base .

[0086] In other words, the corresponding boundary line lengths and angles of the matching points and template intersections should be highly consistent, and the effective coefficient should be used as the weight of the boundary line matching. The larger the weight, the higher the boundary line matching priority, and the greater the impact on the calculated matching degree.

[0087] Create a matching matrix based on the matching degree, where each row in the matching matrix represents the matching point of the product to be captured, and the list is the intersection of the boundary lines of the template product. Input the matching matrix into the KM algorithm, and output the comprehensive matching result, where the comprehensive matching result is the average of the matching degrees of the matching points of the product to be captured and the intersection points of the upper boundary of the template.

[0088] It is further explained that the KM algorithm is a well-known technology in the art and will not be described in detail.

[0089] S3: The product to be grasped is regarded as the template product with the highest comprehensive matching degree, and the robot arm is controlled to grasp the template product with the highest comprehensive matching degree and transport it to the set position.

[0090] After identifying the product type of the template product with the highest comprehensive matching degree, obtain product information, select the appropriate grabbing method, accurately perform the grabbing action, and transport it to the set location;

[0091] Among them, product information includes but is not limited to: shape, weight and material, and grasping methods include but are not limited to: suction cup, clamp.

[0092] By accurately identifying product information and selecting the most appropriate grasping method, not only the grasping accuracy and efficiency are improved, but also the adaptability and reliability of the system are enhanced, providing strong support for automated production.

[0093] The present invention also provides an automatic palletizing robot operation control system. Figure 2 As shown, the system includes a processor and a memory, the memory stores computer program instructions, and when the computer program instructions are executed by the processor, an automatic palletizing robot operation control method according to the first aspect of the present invention is implemented.

[0094] The system also includes other components familiar to those skilled in the art, such as a communication bus and a communication interface. The configuration and functions of these components are known in the art and will not be described in detail here.

[0095] In the present invention, the aforementioned memory may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, apparatus or device. For example, a computer-readable storage medium may be any appropriate magnetic storage medium or magneto-optical storage medium, such as a resistive random access memory RRAM (Resistive Random Access Memory), a dynamic random access memory DRAM (Dynamic Random Access Memory), a static random access memory SRAM (Static Random-Access Memory), an enhanced dynamic random access memory EDRAM (Enhanced Dynamic Random Access Memory), a high-bandwidth memory HBM (High-Bandwidth Memory), a hybrid memory cube HMC (Hybrid Memory Cube), etc., or any other medium that can be used to store the required information and can be accessed by an application, a module, or both. Any such computer storage medium may be part of a device or accessible or connectable to a device. Any application or module described in the present invention may be implemented using computer-readable / executable instructions that may be stored or otherwise maintained by such a computer-readable medium.

[0096] In the description of this specification, "plurality" or "several" means at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.

[0097] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

Claims

1. An automatic palletizing robot operation control method, characterized in that: include: Obtain the point cloud data of the product to be captured, cluster the coordinates and normal vectors of the point cloud data to obtain multiple clusters, determine whether the cluster is a boundary line, and evaluate the effective coefficient of the boundary line; The effective coefficient of the boundary line is used as a weight to obtain the matching degree, the intersection points of the boundary lines are extracted, the characteristic information of the boundary lines connected by the intersection points is obtained, and the KM algorithm is used to match the intersection points of the boundary lines of the template product to obtain the comprehensive matching degree; The product to be grasped is regarded as the template product with the highest comprehensive matching degree, and the robot arm is controlled to grasp the template product with the highest comprehensive matching degree and transport it to the set position; Among them, the matching degree Satisfies the following relationship: ; In the formula, Indicates the first The first of the boundary lines connected by the matching points The effective coefficient of the boundary line, Indicates the first The first of the boundary lines connected by the matching points The length of the boundary line is Indicates the template product The first of the boundary lines connected by the matching points The length of the boundary line is Indicates the first The angle between two boundary lines connecting the matching points, Indicates the template product The angle between two boundary lines connecting the matching points, represents the cosine function, Indicates An exponential function with base .

2. The automatic palletizing robot operation control method according to claim 1 is characterized in that: Obtaining the coordinates and normal vector of the point cloud data includes: Each point in the point cloud data contains three-dimensional coordinates For each data point, find the points in its neighborhood according to the set search radius, use the local fitting plane method to calculate the covariance matrix of the local neighborhood points, perform eigenvalue decomposition on the covariance matrix, and use the eigenvector corresponding to the minimum eigenvalue as the normal vector of the data point.

3. The automatic palletizing robot operation control method according to claim 1, characterized in that: The step of determining whether the cluster is a boundary line includes: The plane and boundary line are distinguished according to the number of data points in the cluster: a threshold is set, and the data points in the cluster that are less than or equal to the set threshold are considered to be boundary lines. Conversely, the clusters with data points greater than the set threshold are considered to be planes on the product surface.

4. The automatic palletizing robot operation control method according to claim 1 is characterized in that: Obtaining the effective coefficient of the boundary line includes: Based on each cluster, the average normal vector of all point cloud data is calculated, and the average of the square of the Euclidean distance between the normal vector of each data point and the average normal vector is calculated to obtain the normal vector variance of the cluster; the clusters that are boundary lines are screened out to obtain the normal vector variance of two adjacent point cloud data belonging to the surface plane of the product; The least squares method is used to fit the boundary line of each cluster, and the sum of the squares of the distances between each data point in the cluster and the fitted straight line is calculated to obtain the residual sum of squares of the fitted boundary line in the cluster, and the result is normalized. The product of the ratio of the normal vector variance of the cluster and the normal vector variance of the point cloud data and the normalized residual sum of squares is used as the effective coefficient of the boundary line of the cluster.

5. The automatic palletizing robot operation control method according to claim 1, characterized in that: Obtaining the effective coefficient of the boundary line of the cluster also includes: Take any cluster as the target cluster, calculate the average distance from all points in the target cluster to the cluster center, get the compactness of the target cluster, calculate the distance between the cluster centers of all clusters, and get the separation between all clusters; Use the least squares method to fit the boundary line of each cluster, calculate the sum of the squares of the vertical distances between the data points in the target cluster and the data points on the corresponding fitting straight line, obtain the degree of boundary line fitting, and perform normalization processing; The product of the ratio of compactness to separation and the normalized boundary line fitting degree is taken as the effective coefficient of the boundary line.

6. The automatic palletizing robot operation control method according to claim 1, characterized in that: The KM algorithm is used to match the intersection of the template product boundary line to obtain a comprehensive matching degree, including: Create a matching matrix based on the matching degree, where each row in the matching matrix represents the matching point of the product to be captured, and the list is the intersection of the boundary lines of the template product. Input the matching matrix into the KM algorithm, and output the comprehensive matching result, where the comprehensive matching result is the average of the matching degrees of the matching points of the product to be captured and the intersection points of the upper boundary of the template.

7. The automatic palletizing robot operation control method according to claim 1, characterized in that: The control robot arm grabs the template product with the highest comprehensive matching degree and transports it to the set position, including: After identifying the product type of the template product with the highest comprehensive matching degree, obtain product information, select the appropriate grabbing method, accurately perform the grabbing action, and transport it to the set location; The product information includes: shape, weight and material, and the grasping methods include: suction cup and clamp.

8. An automatic palletizing robot operation control system, characterized in that: include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the automatic palletizing robot operation control method according to any one of claims 1 to 7 is implemented.

Images