Automatic control system and method of mechanical arm for automated assembly line

Abstract

The present application provides a robot arm automatic control system and method for an automated assembly line, which relates to the field of intelligent manufacturing. The robot arm automatic control system and method are used to monitor and collect the workpiece image of the current workpiece in real time through a camera, and introduce an image processing and analysis algorithm based on artificial intelligence and machine vision at the back end to perform multi-level and multi-scale feature extraction on the workpiece image, so as to detect the workpiece type by using the kernel feature joint interactive information between the shallow shape features and deep structural features of the workpiece, so as to select the appropriate type of fixture. In this way, the type of fixture that is suitable for the workpiece can be automatically selected during the robot arm control process of the automated assembly line, which can not only improve the reliability of the robot arm's work, but also enhance the flexibility and response speed of the entire automated assembly line, and provide a basis for the grasping control of workpieces of different sizes and types.

Description

Technical Field

[0001] The present application relates to the field of intelligent manufacturing, and more specifically, to a robotic arm automatic control system and method for an automated assembly line. Background Art

[0002] With the rapid development of industrial automation, automated assembly lines play an increasingly important role in the manufacturing industry. As the core component of automated assembly lines, the performance of robotic arms directly affects production efficiency and product quality. Especially in assembly line environments, robotic arms need to perform a series of precise operations to handle workpieces of different types and sizes. In order to improve production efficiency and reduce human errors, it is essential to realize the automation of robotic arms and efficient and precise control.

[0003] Traditional robot control methods often rely on fixed programs and manual settings, which requires separate programming for each different task, and when the types of products on the production line change, the robot control solution is not flexible and adaptable enough. In addition, traditional control solutions often have difficulty in ensuring high-precision positioning and grasping requirements when processing workpieces with complex geometric shapes or large size variations.

[0004] Therefore, an optimized automatic control solution for a robotic arm used in an automated assembly line is desired. Summary of the invention

[0005] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides a robot arm automatic control system and method for an automated assembly line, which collects the workpiece image of the current workpiece through real-time monitoring by a camera, and introduces an image processing and analysis algorithm based on artificial intelligence and machine vision at the back end to perform multi-level and multi-scale feature extraction on the workpiece image, so as to utilize the kernel feature joint interactive information between the shallow shape features and deep structural features of the workpiece to detect the workpiece type, so as to select the appropriate type of fixture. In this way, the type of fixture that is suitable for the workpiece can be automatically selected during the robot arm control process of the automated assembly line, which can not only improve the reliability of the robot arm's work, but also enhance the flexibility and response speed of the entire automated assembly line, and provide a basis for the grasping control of workpieces of different sizes and types.

[0006] According to one aspect of the present application, there is provided a method for automatic control of a robot arm for an automated assembly line, comprising:

[0007] Acquiring a target height parameter input by an operator, wherein the target height parameter is determined based on a size of a current workpiece and a height requirement of a working area;

[0008] Acquire a workpiece image of the current workpiece captured by a camera;

[0009] Based on the workpiece image, determining an adaptive type of fixture, comprising the steps of: performing multi-scale feature extraction and feature enhancement processing on the workpiece image to obtain a shallow shape enhancement feature of the workpiece and a deep structure enhancement feature of the workpiece; performing kernel feature joint interactive encoding on the shallow shape enhancement feature of the workpiece and the deep structure enhancement feature of the workpiece to obtain a workpiece shape-structure joint interactive feature; determining a workpiece type label based on the workpiece shape-structure joint interactive feature, and determining the adaptive type of fixture;

[0010] Performing image recognition on the workpiece image to obtain position data of the current workpiece in the working image;

[0011] Converting the position data of the current workpiece in the working image into three-dimensional coordinates of the robot arm coordinate system to obtain the three-dimensional coordinates of the current workpiece;

[0012] Based on the three-dimensional coordinates of the current workpiece, a motion trajectory of the robot arm is determined.

[0013] According to another aspect of the present application, there is provided a robotic arm automatic control system for an automated assembly line, comprising:

[0014] A target height parameter acquisition module, used to acquire a target height parameter input by an operator, wherein the target height parameter is determined based on the size of the current workpiece and the height requirement of the working area;

[0015] A current workpiece image acquisition module is used to acquire a workpiece image of the current workpiece captured by a camera;

[0016] An adaptive fixture type determination module is used to determine an adaptive fixture type based on the workpiece image; wherein the adaptive fixture type determination module is used to: perform multi-scale feature extraction and feature enhancement processing on the workpiece image to obtain a shallow shape enhancement feature and a deep structure enhancement feature of the workpiece; perform kernel feature joint interactive encoding on the shallow shape enhancement feature and the deep structure enhancement feature of the workpiece to obtain a workpiece shape-structure joint interactive feature; determine a workpiece type label based on the workpiece shape-structure joint interactive feature, and determine the adaptive fixture type;

[0017] A workpiece position data recognition module, used for performing image recognition on the workpiece image to obtain the position data of the current workpiece in the working image;

[0018] A workpiece position data conversion module, used for converting the position data of the current workpiece in the working image into three-dimensional coordinates of the robot arm coordinate system to obtain the three-dimensional coordinates of the current workpiece;

[0019] The robot arm motion trajectory determination module is used to determine the motion trajectory of the robot arm based on the three-dimensional coordinates of the current workpiece.

[0020] Compared with the prior art, the present application provides a robot arm automatic control system and method for an automated assembly line, which uses a camera to monitor and collect the workpiece image of the current workpiece in real time, and introduces an image processing and analysis algorithm based on artificial intelligence and machine vision at the back end to perform multi-level and multi-scale feature extraction on the workpiece image, so as to use the kernel feature joint interactive information between the shallow shape features and deep structural features of the workpiece to detect the workpiece type, so as to select the appropriate type of fixture. In this way, the type of fixture that is suitable for the workpiece can be automatically selected during the robot arm control process of the automated assembly line, which can not only improve the reliability of the robot arm's work, but also enhance the flexibility and response speed of the entire automated assembly line, and provide a basis for the grasping control of workpieces of different sizes and types. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0022] Figure 1 It is a flow chart of a method for automatic control of a robotic arm for an automated assembly line according to an embodiment of the present application.

[0023] Figure 2 The present invention is a flowchart of determining a fixture of an adapted type based on the workpiece image in a method for automatically controlling a robot arm for an automated assembly line according to an embodiment of the present application.

[0024] Figure 3 The present invention is a data flow diagram for determining a suitable type of fixture based on the workpiece image in a method for automatically controlling a robot arm for an automated assembly line according to an embodiment of the present application.

[0025] Figure 4 The present invention is a flowchart of performing multi-scale feature extraction and feature enhancement processing on the workpiece image in the robot arm automatic control method for an automated assembly line according to an embodiment of the present application to obtain the workpiece shallow shape enhancement feature and the workpiece deep structure enhancement feature.

[0026] Figure 5The present invention is a flowchart of performing kernel feature joint interactive encoding on the shallow shape enhancement feature and the deep structure enhancement feature of the workpiece in the robot arm automatic control method for the automated assembly line according to the embodiment of the present application to obtain the workpiece shape-structure joint interactive feature.

[0027] Figure 6 The present invention is a flowchart of inputting the workpiece shallow shape enhancement feature map and the workpiece deep structure enhancement feature map into the kernel feature distillation network to obtain the workpiece shallow shape kernel feature vector and the workpiece deep structure kernel feature vector in the robot arm automatic control method for an automated assembly line according to an embodiment of the present application.

[0028] Figure 7 It is a system block diagram of a robotic arm automatic control system for an automated assembly line according to an embodiment of the present application. DETAILED DESCRIPTION

[0029] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0030] Since traditional robot control methods usually rely on fixed programs and manual settings, this method leads to a lack of flexibility and adaptability in the robot control solution when the product types on the production line change.

[0031] In response to the above technical problems, the present application proposes a method for automatic control of a robotic arm for an automated assembly line. Figure 1 FIG. 1 is a flow chart of a method for automatically controlling a robotic arm for an automated assembly line according to an embodiment of the present application. Figure 1As shown, according to an embodiment of the present application, a robot arm automatic control method for an automated assembly line includes: S110, obtaining a target height parameter input by an operator, wherein the target height parameter is determined based on the size of the current workpiece and the height requirement of the working area; S120, obtaining a workpiece image of the current workpiece captured by a camera; S130, based on the workpiece image, determining an adaptive type of fixture, including the steps of: performing multi-scale feature extraction and feature enhancement processing on the workpiece image to obtain a shallow shape enhancement feature of the workpiece and a deep structure enhancement feature of the workpiece; performing kernel feature joint interactive encoding on the shallow shape enhancement feature of the workpiece and the deep structure enhancement feature of the workpiece to obtain a workpiece shape-structure joint interactive feature; determining a workpiece type label based on the workpiece shape-structure joint interactive feature, and determining the adaptive type of fixture; S140, performing image recognition on the workpiece image to obtain the position data of the current workpiece in the working image; S150, converting the position data of the current workpiece in the working image into three-dimensional coordinates of the robot arm coordinate system to obtain the three-dimensional coordinates of the current workpiece; S160, determining the motion trajectory of the robot arm based on the three-dimensional coordinates of the current workpiece.

[0032] In the above-mentioned method for automatic control of a robotic arm for an automated assembly line, the step S110 obtains a target height parameter input by an operator, wherein the target height parameter is determined based on the size of the current workpiece and the height requirement of the working area. It should be understood that the size of the workpiece may vary due to production batches, material differences, or changes in processing technology. The height of the working area also needs to be determined based on the working range and technical parameters, combined with the size of the current workpiece. Through the target height parameter input mechanism, the workpiece acquisition process can be made more efficient.

[0033] In the above-mentioned robot arm automatic control method for an automated assembly line, the step S120 is to obtain a workpiece image of the current workpiece captured by a camera. It should be understood that the image acquired by the camera can display the size, shape and surface state of the workpiece, and can analyze the captured workpiece image to extract key dimensional data of the workpiece, reduce errors caused by manual measurement, and improve production efficiency.

[0034] In particular, in the robot arm automatic control method for the automated assembly line, it is crucial to analyze the workpiece image to determine the appropriate type of fixture, which is one of the key links in the entire automated process and robot arm automatic control. This is because modern production lines often need to handle workpieces of various types, sizes, weights, and even materials. Different types of workpieces require fixtures of different designs to ensure that the robot arm can grasp and place them stably and safely. In addition, some precision machining processes have very high accuracy requirements for the positioning of the workpiece. The wrong type of fixture may cause the workpiece to shift or be damaged during the handling process of the robot arm, thereby affecting the quality of subsequent processes.

[0035] Based on this, in the process of determining the suitable type of fixture, the technical concept of this application is to collect the workpiece image of the current workpiece through real-time monitoring by a camera, and introduce an image processing and analysis algorithm based on artificial intelligence and machine vision at the back end to perform multi-level and multi-scale feature extraction on the workpiece image, so as to use the kernel feature joint interactive information between the shallow shape features and deep structural features of the workpiece to detect the workpiece type, so as to select the suitable type of fixture. In this way, the type of fixture that adapts to the workpiece can be automatically selected during the robot arm control process of the automated assembly line, which can not only improve the reliability of the robot arm's work, but also enhance the flexibility and response speed of the entire automated assembly line, and provide a basis for the grasping control of workpieces of different sizes and types.

[0036] Specifically, in the above-mentioned robot arm automatic control method for an automated assembly line, the step S130 determines the type of fixture that is suitable based on the workpiece image. It should be understood that the workpiece image provides detailed information about the workpiece, including its shape, size and surface features. By analyzing the workpiece image, the characteristics of the workpiece can be understood more accurately, so as to select the most suitable type of fixture. Among them, a fixture that can be firmly clamped is required for workpieces with complex shapes or smooth surfaces, and a fixture with high strength and good stability is required for larger or heavier workpieces. In this way, the safety and stability of the workpiece during processing can be significantly improved.

[0037] More specifically, Figure 2 The present invention is a flowchart of determining a fixture of an adapted type based on the workpiece image in a method for automatically controlling a robot arm for an automated assembly line according to an embodiment of the present application. Figure 3 The following is a data flow diagram of determining a fixture of an adapted type based on the workpiece image in a method for automatically controlling a robot arm for an automated assembly line according to an embodiment of the present application. Figure 2 and Figure 3As shown, the step S130, based on the workpiece image, determines the adaptive type of fixture, including: S131, performing multi-scale feature extraction and feature enhancement processing on the workpiece image to obtain the shallow shape enhancement feature of the workpiece and the deep structure enhancement feature of the workpiece; S132, performing kernel feature joint interactive encoding on the shallow shape enhancement feature of the workpiece and the deep structure enhancement feature of the workpiece to obtain the workpiece shape-structure joint interactive feature; S133, determining the workpiece type label based on the workpiece shape-structure joint interactive feature, and determining the adaptive type of fixture.

[0038] Specifically, Figure 4 The present invention is a flowchart of performing multi-scale feature extraction and feature enhancement processing on the workpiece image in the robot arm automatic control method for an automated assembly line according to an embodiment of the present application to obtain the workpiece shallow shape enhancement feature and the workpiece deep structure enhancement feature. Figure 4 As shown, the step S131, performing multi-scale feature extraction and feature enhancement processing on the workpiece image to obtain the workpiece shallow shape enhancement feature and the workpiece deep structure enhancement feature, includes: S1311, inputting the workpiece image into a workpiece image multi-scale feature scanner based on a hollow pyramid network to obtain a workpiece shallow shape feature map and a workpiece deep structure feature map; S1312, inputting the workpiece shallow shape feature map and the workpiece deep structure feature map into a feature enhancement module based on a converter structure respectively to obtain the workpiece shallow shape enhancement feature map as the workpiece shallow shape enhancement feature and the workpiece deep structure enhancement feature map as the workpiece deep structure enhancement feature.

[0039] It should be understood that in the technical solution of the present application, first of all, considering that in the automated assembly line, the workpieces that the robot needs to grasp may have different sizes and shapes, the features of a single scale are not enough to describe and understand all the details and characteristics of the workpiece. Therefore, in order to be able to more comprehensively and accurately understand and depict the feature information of the workpiece that needs to be grasped, in the technical solution of the present application, the workpiece image is input into the multi-scale feature scanner of the workpiece image based on the hollow pyramid network to obtain the shallow shape feature map of the workpiece and the deep structure feature map of the workpiece. It is worth mentioning that the shallow shape feature map of the workpiece contains information about the edge, texture and shape of the workpiece in the image, which can help identify the basic outline, overall shape and surface characteristics of the workpiece, while the deep structure feature map of the workpiece contains more deep structural features about the object, which helps to understand the structural semantics of the workpiece, which is particularly important for the identification of workpiece types.

[0040] Then, in order to further refine and enhance the expression of workpiece features in these feature maps so as to better serve subsequent tasks, such as workpiece identification and classification tasks, in the technical solution of the present application, the workpiece shallow shape feature map and the workpiece deep structure feature map are further input into the feature enhancement module based on the converter structure to obtain the workpiece shallow shape enhancement feature map and the workpiece deep structure enhancement feature map. It should be understood that the converter structure is particularly good at capturing long-distance dependencies in the data. For visual tasks, this means that it can better understand the correlation between different parts of the workpiece shallow shape feature map and the workpiece deep structure feature map, which is very important for objects that require global understanding (such as complex workpieces). In addition, by using a feature enhancement module based on the converter structure, the useful information in the original workpiece shallow shape feature map and the workpiece deep structure feature map can be further enhanced in the process of long-distance association understanding of different parts in the feature map, while suppressing irrelevant or noise information. This enhanced feature map will more clearly highlight the key features of the workpiece, thereby improving the accuracy of subsequent workpiece type detection tasks.

[0041] Specifically, Figure 5 The present invention is a flowchart of performing kernel feature joint interactive coding on the shallow shape enhancement feature of the workpiece and the deep structure enhancement feature of the workpiece in the automatic control method of the robot arm for the automated assembly line according to the embodiment of the present application to obtain the workpiece shape-structure joint interactive feature. Figure 5As shown, the step S132, performing kernel feature joint interactive coding on the shallow shape enhancement feature of the workpiece and the deep structure enhancement feature of the workpiece to obtain the workpiece shape-structure joint interactive feature, includes: S1321, inputting the shallow shape enhancement feature map of the workpiece and the deep structure enhancement feature map of the workpiece into the kernel feature distillation network to obtain the shallow shape kernel feature vector of the workpiece and the deep structure kernel feature vector of the workpiece; S1322, performing global correlation coding on the shallow shape kernel feature vector of the workpiece and the deep structure kernel feature vector of the workpiece to obtain the global coding matrix between the workpiece shape-structure kernel features; S1323, The global encoding matrix between the workpiece shape-structure kernel features is input into the sparse mask unit to obtain the sparse weight matrix associated between the workpiece shape-structure kernel features; S1324, the workpiece shallow shape enhancement feature map and the workpiece deep structure enhancement feature map are input into the position-by-position association interaction network to obtain the workpiece shape-structure position-by-position interaction fusion feature map; S1325, based on the sparse weight matrix associated between the workpiece shape-structure kernel features, the workpiece shape-structure position-by-position interaction fusion feature map is subjected to feature significant sparse modulation to obtain the workpiece shape-structure joint interaction feature map as the workpiece shape-structure joint interaction feature. It should be understood that since the workpiece shallow shape enhancement feature map and the workpiece deep structure enhancement feature map respectively contain information about the shallow shape feature and deep structure feature of the workpiece, both features play an important role in the subsequent workpiece type detection task. In order to comprehensively consider the interaction between the surface morphology (shallow shape) and the internal structure (deep structure) of the workpiece, and extract the most representative and distinguishing features therefrom, so as to provide a basis for subsequent workpiece type identification and detection, in the technical solution of the present application, the workpiece shallow shape enhancement features and the workpiece deep structure enhancement features are further subjected to kernel feature joint interactive encoding to obtain workpiece shape-structure joint interactive features. The kernel feature joint interactive encoding process can extract the most representative and distinguishing feature information from the original workpiece shallow shape enhancement feature map and the workpiece deep structure enhancement feature map through multi-level processing steps, and on this basis effectively fuse these two multi-scale features at different levels of the workpiece to obtain a new feature representation, which includes both appearance characteristics and internal structures, thereby providing a more comprehensive description of the workpiece.

[0042] Specifically, the step of joint interactive coding of kernel features includes: first, inputting the workpiece shallow shape enhancement feature map and the workpiece deep structure enhancement feature map into the kernel feature distillation network to obtain the workpiece shallow shape kernel feature vector and the workpiece deep structure kernel feature vector. The kernel feature distillation network utilizes the powerful representation ability of the neural network, which not only reduces redundant information, but also improves the expressiveness and robustness of the workpiece shallow shape features and the workpiece deep structure features. By identifying and enhancing the expressiveness of these features, it can best distinguish or describe the feature attributes of the essence of the data, that is, the core attribute information in the workpiece shallow shape and deep structure features. Then, the workpiece shallow shape kernel feature vector and the workpiece deep structure kernel feature vector are globally associated encoded to obtain the global encoding matrix between the workpiece shape-structure kernel features. It is worth mentioning that in the technical solution of the present application, the shallow shape kernel feature vector of the workpiece and the deep structure kernel feature vector of the workpiece are globally associated and encoded to obtain the global encoding matrix between the workpiece shape-structure kernel features, including: calculating the multiplication between the transposed vectors of the shallow shape kernel feature vector of the workpiece and the deep structure kernel feature vector of the workpiece to obtain the global encoding matrix between the workpiece shape-structure kernel features. That is, the shallow shape and deep structure kernel feature vectors of the workpiece are processed by the global association coding technology to construct a global encoding matrix between kernel features that can reflect the correlation and interaction between kernel features at different levels of the workpiece, which helps to better distinguish different workpieces, thereby more accurately performing the workpiece type recognition task. Then, the global encoding matrix between the workpiece shape-structure kernel features is input into the sparse mask unit to obtain the sparse weight matrix between the workpiece shape-structure kernel features. That is, the sparsity principle is introduced to filter out the less important workpiece shape-structure kernel association feature information, while highlighting those parts that are most critical to the overall information transmission, which can effectively condense the key information flow path in the original data. In this way, the system can significantly reduce computing costs while maintaining or even improving performance, which is particularly important for actual deployment, especially for applications in resource-constrained environments. At the same time, on another data processing link, the workpiece shallow shape enhancement feature map and the workpiece deep structure enhancement feature map are input into a position-by-position association interaction network to obtain a workpiece shape-structure position-by-position interactive fusion feature map. It is worth mentioning that in the technical solution of the present application, the workpiece shallow shape enhancement feature map and the workpiece deep structure enhancement feature map are input into a position-by-position association interaction network to obtain a workpiece shape-structure position-by-position interactive fusion feature map, including: multiplying the workpiece shallow shape enhancement feature map and the workpiece deep structure enhancement feature map by position point to obtain the workpiece shape-structure position-by-position interactive fusion feature map.In this way, it is possible to pay attention to the feature interaction between the shallow shape features and the deep structural features of the workpiece at the local detail level, and realize the information exchange and integration at each corresponding position across the feature maps through a carefully designed network structure, and finally form a workpiece shape-structure position-by-position interactive fusion feature map. Finally, based on the sparse weight matrix associated between the workpiece shape-structure kernel features, the workpiece shape-structure position-by-position interactive fusion feature map is subjected to feature-significant sparse modulation. It is worth mentioning that in the technical solution of the present application, based on the sparse weight matrix associated between the workpiece shape-structure kernel features, the workpiece shape-structure position-by-position interactive fusion feature map is subjected to feature-significant sparse modulation to obtain a workpiece shape-structure joint interactive feature map as the workpiece shape-structure joint interactive feature, including: multiplying each feature matrix along the channel dimension in the workpiece shape-structure position-by-position interactive fusion feature map by the sparse weight matrix associated between the workpiece shape-structure kernel features to obtain the workpiece shape-structure joint interactive feature map. This modulation process adjusts the contribution of each part according to the importance level indicated by the associated sparse weight matrix, so that the adjusted result (workpiece shape-structure position-by-position interactive fusion feature map) can better reflect the most representative feature combination in the original input and the most relevant to the subsequent workpiece type recognition task. By giving these features higher weights, the system can better adapt to the complex and changing needs of actual application scenarios, providing a basis for subsequent workpiece type detection and grasping control of workpieces of different sizes and types.

[0043] More specifically, Figure 6 The present invention is a flowchart of inputting the workpiece shallow shape enhancement feature map and the workpiece deep structure enhancement feature map into the kernel feature distillation network to obtain the workpiece shallow shape kernel feature vector and the workpiece deep structure kernel feature vector in the robot arm automatic control method for the automated assembly line according to the embodiment of the present application. Figure 6As shown, the step S1321, inputting the workpiece shallow shape enhancement feature map and the workpiece deep structure enhancement feature map into the kernel feature distillation network to obtain the workpiece shallow shape kernel feature vector and the workpiece deep structure kernel feature vector, includes: S1321-1, feature decomposition of the workpiece shallow shape enhancement feature map to obtain a sequence of workpiece shallow shape decomposition feature vectors; S1321-2, calculating the difference features of each workpiece shallow shape decomposition feature vector in the sequence of workpiece shallow shape decomposition feature vectors relative to other workpiece shallow shape decomposition feature vectors to obtain a workpiece shallow shape decomposition feature difference representation vector; S1321-3, normalizing each eigenvalue in the workpiece shallow shape decomposition feature difference representation vector to obtain a normalized workpiece shallow shape decomposition feature difference representation vector; S1321-4, using each eigenvalue in the normalized workpiece shallow shape decomposition feature difference representation vector as a weight, calculating the weighted sum of the sequence of the workpiece shallow shape decomposition feature vectors to obtain the workpiece shallow shape kernel feature vector.

[0044] More specifically, the step S1321-2 calculates the difference characteristics of each shallow shape decomposition feature vector of the workpiece in the sequence of shallow shape decomposition feature vectors of the workpiece relative to other shallow shape decomposition feature vectors of the workpiece to obtain a shallow shape decomposition feature difference representation vector of the workpiece, including: calculating the norm of the position difference between each shallow shape decomposition feature vector of the workpiece in the sequence of shallow shape decomposition feature vectors of the workpiece and the shallow shape decomposition feature vectors of other workpieces to obtain a sequence of workpiece shallow shape decomposition feature difference one-norm representation values; calculating the mean value of the sequence of workpiece shallow shape decomposition feature difference one-norm representation values ​​to obtain a workpiece shallow shape decomposition feature difference one-norm average representation value; vectorizing and arranging a plurality of the workpiece shallow shape decomposition feature difference one-norm average representation values ​​to obtain the workpiece shallow shape decomposition feature difference representation vector.

[0045] More specifically, the step S1321-3 normalizes each eigenvalue in the workpiece shallow shape decomposition feature difference representation vector to obtain a normalized workpiece shallow shape decomposition feature difference representation vector, including: calculating the positional exponential function value with a natural constant as the base for each eigenvalue in the workpiece shallow shape decomposition feature difference representation vector to obtain the workpiece shallow shape decomposition feature difference class support representation vector; calculating the weighted sum of each eigenvalue in the workpiece shallow shape decomposition feature difference class support representation vector to obtain the workpiece shallow shape decomposition feature difference class support sum; calculating the positional division between each eigenvalue in the workpiece shallow shape decomposition feature difference class support representation vector and the workpiece shallow shape decomposition feature difference class support sum to obtain the normalized workpiece shallow shape decomposition feature difference representation vector.

[0046] More specifically, step S1323, inputting the global coding matrix between the workpiece shape and the structure kernel features into a sparse mask unit to obtain a sparse weight matrix associated with the workpiece shape and the structure kernel features, includes: performing a sparse mask operation on the eigenvalues ​​of each position in the global coding matrix between the workpiece shape and the structure kernel features to obtain a sparse weight matrix associated with the workpiece shape and the structure kernel features; wherein, in response to the eigenvalue in the global coding matrix between the workpiece shape and the structure kernel features being greater than twice a preset threshold, multiplying the eigenvalue greater than twice the preset threshold by a constant 3; in response to the eigenvalue in the global coding matrix between the workpiece shape and the structure kernel features being less than or equal to the preset threshold, dividing the eigenvalue less than or equal to the preset threshold by a constant 3; in response to the eigenvalue in the global coding matrix between the workpiece shape and the structure kernel features being greater than the preset threshold and less than or equal to twice the preset threshold, taking the eigenvalue as the original value to obtain the sparse weight matrix associated with the workpiece shape and the structure kernel features.

[0047] More specifically, the workpiece shallow shape enhancement feature and the workpiece deep structure enhancement feature are subjected to kernel feature joint interaction encoding using the following feature joint interaction formula to obtain the workpiece shape-structure joint interaction feature;

[0048] Among them, the feature joint interaction formula is: ;in, and They are respectively the workpiece shallow shape enhancement feature map and the workpiece deep structure enhancement feature map, For kernel feature distillation, and are respectively the kernel feature vector of the shallow shape of the workpiece and the kernel feature vector of the deep structure of the workpiece, is a decomposition operation, are the first decomposition eigenvector, the second decomposition eigenvector, The decomposed eigenvectors, The decomposition eigenvectors and decomposition eigenvectors, is the one-norm of the vector, is the sequence of decomposed eigenvectors, is the number of vectors in the sequence of decomposed eigenvectors minus one, To decompose the eigenvalues ​​of each position in the feature difference representation vector, is the length of the decomposed feature difference representation vector, is the exponential operation, is the kernel eigenvector, is the global encoding matrix between the workpiece shape and structure kernel features, is the feature value of each position in the global encoding matrix between the workpiece shape-structure kernel features, is the preset threshold, is a sparse mask operation is the sparse weight matrix of the correlation between the workpiece shape and structure kernel features, It is the workpiece shape-structure position-by-position interactive fusion feature map. It is the point multiplication by position. is matrix multiplication, It is the workpiece shape-structure joint interaction feature map.

[0049] Specifically, the step S133, based on the workpiece shape-structure joint interaction feature, determines the workpiece type label, and determines the adapted type of fixture, including: inputting the workpiece shape-structure joint interaction feature map into a workpiece identifier based on a classifier to obtain a recognition result, and the recognition result is used to represent the workpiece type label; based on the workpiece type label, determining the adapted type of fixture. It should be understood that the workpiece shape-structure joint interaction feature map is input into the workpiece identifier based on a classifier to obtain a recognition result, and the recognition result is used to represent the workpiece type label. That is, the joint interaction feature between the shallow shape feature and the deep structure feature of the workpiece is used to perform classification processing, so as to identify and detect the workpiece type. Then, based on the workpiece type label, the adapted type of fixture is determined. In this way, the fixture type adapted to the workpiece can be automatically selected in the process of controlling the robot arm of the automated assembly line, which can not only improve the reliability of the robot arm, but also enhance the flexibility and response speed of the entire automated assembly line, and provide a basis for the grasping control of workpieces of different sizes and types. It is worth mentioning that in the technical solution of the present application, for workpieces with complex shapes or smooth surfaces, a firmly clamping fixture can be used, such as a fixture with tiny protrusions and rubber coating on the inner surface; for larger or heavier workpieces, a fixture with high strength and good stability can be used, such as a fixture with pneumatic or hydraulic clamping function.

[0050] Preferably, in an example of the present application, the workpiece shape-structure joint interactive feature map is input into a classifier-based workpiece identifier to obtain a recognition result, including: calculating the sum of the absolute values ​​of each feature value of the workpiece shape-structure joint interactive feature map to obtain a first workpiece shape-structure joint fine-grained perception spatial structure value, and calculating the square root of the sum of the squares of each feature value of the workpiece shape-structure joint interactive feature map to obtain a second workpiece shape-structure joint fine-grained perception spatial structure value; multiplying each feature value of the workpiece shape-structure joint interactive feature map with the first workpiece shape-structure joint fine-grained perception spatial structure value and the second workpiece shape-structure joint fine-grained perception spatial structure value to obtain a first workpiece shape-structure joint fine-grained perception structure reference value and a second workpiece shape-structure joint fine-grained perception structure reference value corresponding to each feature value; multiplying each feature value of the workpiece shape-structure joint interactive feature map with the scale of the workpiece shape-structure joint interactive feature map and the square root of the scale to obtain a first workpiece shape-structure joint fine-grained perception structure reference value corresponding to each feature value. a workpiece shape-structure joint fine-grained perception scale transformation value and a second workpiece shape-structure joint fine-grained perception scale transformation value; dividing the first workpiece shape-structure joint fine-grained perception structure reference value by the difference between the first workpiece shape-structure joint fine-grained perception space structure value and the first workpiece shape-structure joint fine-grained perception scale transformation value to obtain a first workpiece shape-structure joint fine-grained perception transformation adjustment value; dividing the second workpiece shape-structure joint fine-grained perception structure reference value by the difference between the second workpiece shape-structure joint fine-grained perception space structure value and the second workpiece shape-structure joint fine-grained perception scale transformation value to obtain a second workpiece shape-structure joint fine-grained perception transformation adjustment value; calculating a weighted sum of the first workpiece shape-structure joint fine-grained perception transformation adjustment value and the second workpiece shape-structure joint fine-grained perception transformation adjustment value to obtain each feature value of the optimized workpiece shape-structure joint interaction feature map; and inputting the optimized workpiece shape-structure joint interaction feature map into a classifier-based workpiece identifier to obtain a recognition result.

[0051] Here, the workpiece shape-structure joint interaction feature map is denoted as The optimization expression is: ;in, Represents the eigenvalues ​​of the joint interaction feature map of workpiece shape and structure, represents the first workpiece shape-structure joint fine-grained perception transformation adjustment value, represents the second workpiece shape-structure joint fine-grained perception transformation adjustment value, represents the first workpiece shape-structure joint fine-grained perception spatial structure value, represents the second workpiece shape-structure joint fine-grained perception spatial structure value, represents the number of eigenvalues ​​in the workpiece shape-structure joint interaction feature graph, represents the summation function, represents the absolute value function, represents the set of real numbers, represents the width of the feature map, represents the height of the feature map, Indicates the number of channels of the feature map, A first workpiece shape-structure joint fine-grained perceptual transformation adjustment feature map is represented by vectorizing a plurality of the first workpiece shape-structure joint fine-grained perceptual transformation adjustment values, A second workpiece shape-structure joint fine-grained perceptual transformation adjustment feature map is represented by vectorizing a plurality of the second workpiece shape-structure joint fine-grained perceptual transformation adjustment values, It means adding by position. represents the fine-grained perceptual transformation adjustment weighted value, represents positional multiplication, A feature map representing the optimized joint shape-structure interaction of the workpiece.

[0052] That is, the applicant of the present application took into account that the shallow shape enhancement feature map of the workpiece and the deep structure enhancement feature map of the workpiece respectively represent the global context enhancement features of the shallow and deep image semantic features of the workpiece image at different scales, and when they are passed through a feature significant joint interaction encoding module based on kernel-correlated sparse masks, the global context enhancement distribution differences of image semantic features at different scales and depths will also have the global attention perception differences of fine-grained spatial distribution deconstruction. Therefore, it is expected to improve the complex spatial structure regression convergence and generalization effect of the workpiece shape-structure joint interaction feature map based on the joint perceptual interaction of different spatial distribution structures.

[0053] Therefore, in the preferred example, for the spatial structure information of the feature set of the workpiece shape-structure joint interaction feature map in the high-dimensional space, the scale-based frame transformation of each feature value of the workpiece shape-structure joint interaction feature map is performed by using the class norm space structured representation of the workpiece shape-structure joint interaction feature map as a reference window, and the spatial structure-based frame attention weight adjustment of each feature value of the workpiece shape-structure joint interaction feature map is realized to ensure the invariance of the spatial transformation (translation, scaling and rotation) of the workpiece shape-structure joint interaction feature map under the feature space interaction, thereby improving the convergence and generalization effect of the classification regression of the feature set of the workpiece shape-structure joint interaction feature map under the complex spatial structure representation, and improving the accuracy of the recognition result obtained by inputting it into the workpiece identifier based on the classifier. In this way, the workpiece type can be detected more accurately, so as to select the fixture type suitable for the workpiece, which can not only improve the reliability of the robot arm, but also enhance the flexibility and response speed of the entire automated assembly line, and provide a basis for the grasping control of workpieces of different sizes and types.

[0054] In the above-mentioned robot arm automatic control method for automated assembly line, the step S140 performs image recognition on the workpiece image to obtain the position data of the current workpiece in the working image. It should be understood that image recognition is a computer vision technology that aims to automatically identify objects, scenes or features by analyzing and understanding the content in the image. Image recognition technology can accurately locate the workpiece, identify the edges and key features of the workpiece, and accurately determine the placement position of the workpiece, which is helpful for subsequent robot arm grasping. In a specific embodiment of the present application, first, the current workpiece in the working image is image preprocessed using denoising, grayscale conversion and contrast enhancement. Then, edge detection is used to extract the key features of the current workpiece, and the shape, contour and key points of the workpiece are analyzed. Then, feature matching is used to calculate the similarity between features to determine the position and posture of the workpiece. Finally, the position information is output in the form of coordinates, such as X, Y coordinates and the rotation angle of the workpiece. These data can be used to control the robot arm to ensure the accurate positioning of the workpiece during the processing process.

[0055] In the above-mentioned robot arm automatic control method for automated assembly line, the step S150 converts the position data of the current workpiece in the working image into the three-dimensional coordinates of the robot arm coordinate system to obtain the three-dimensional coordinates of the current workpiece. It should be understood that in the automation system, there are usually multiple coordinate systems, such as the camera coordinate system, the world coordinate system and the robot arm coordinate system. Among them, the robot arm usually uses its own coordinate system for motion control, and the position data in the workpiece image is based on the coordinate system of the camera. In order to enable the robot arm to accurately grasp the workpiece, the position data of the workpiece is converted into the coordinate system of the robot arm. In this way, the standards can be unified so that the various systems can work in coordination to ensure the accuracy of the motion instructions. In a specific embodiment of the present application, first, the definition of each coordinate system is clarified. Among them, the origin of the camera coordinate system is at the viewpoint of the camera, and the origin of the robot arm coordinate system is at the base of the robot arm. Then, the relationship between the coordinate systems is obtained by calibration technology, including the rotation matrix and the translation vector. Next, the two-dimensional coordinates (X, Y) and depth information (Z) of the current workpiece are converted into three-dimensional coordinates in the robot arm coordinate system. Finally, the converted three-dimensional coordinates are input into the control system of the robotic arm.

[0056] In the above-mentioned robot arm automatic control method for automated assembly line, the step S160 determines the motion trajectory of the robot arm based on the three-dimensional coordinates of the current workpiece. It should be understood that the motion trajectory based on the three-dimensional coordinates can eliminate position errors and avoid production problems caused by inaccurate trajectories, thereby reducing unnecessary movements of the robot arm during task execution, shortening operation time, and improving overall production efficiency.

[0057] Specifically, the step S160, based on the three-dimensional coordinates of the current workpiece, determines the motion trajectory of the robot arm, including: based on the three-dimensional coordinates of the current workpiece, calculating the motion trajectory of the robot arm to generate a reverse dynamics solution to ensure that the fixture of the robot arm can accurately reach the position of the current workpiece. It should be understood that the inverse kinematics solution is a key technology in the control of the robot arm, which is used to derive the angle or position required for each joint from the target position and posture of the end effector. By solving a set of equations, inverse kinematics converts the position and posture in three-dimensional space into joint angles, and this process takes into account the geometric structure and motion restrictions of the robot arm. In a specific embodiment of the present application, first, a kinematic model of the robot arm is established, including various joint types (such as rotational joints, translational joints) of the robot arm and their corresponding motion parameters (such as length, angle restrictions, etc.). In this way, the motion capability and constraints of the robot arm can be understood. Then, the path planning is performed using the three-dimensional coordinate information. Among them, linear interpolation, spline curve interpolation or other smooth trajectory generation methods can be selected to ensure that the fixture does not produce severe acceleration or deceleration during movement, thereby reducing damage and improving efficiency. Finally, the required joint angles are calculated through the inverse kinematics algorithm, and the motion trajectory of the robot arm is generated to ensure that the fixture can accurately move to the target position. In this way, the robot arm fixture can accurately grasp the workpiece and improve the safety and efficiency of automated operations.

[0058] In summary, the automatic control method of the robot arm for the automated assembly line based on the embodiment of the present application is explained, which collects the workpiece image of the current workpiece through real-time monitoring by a camera, and introduces an image processing and analysis algorithm based on artificial intelligence and machine vision at the back end to perform multi-level and multi-scale feature extraction on the workpiece image, so as to detect the workpiece type by using the kernel feature joint interactive information between the shallow shape features and deep structural features of the workpiece, so as to select the appropriate type of fixture. In this way, the type of fixture that is suitable for the workpiece can be automatically selected during the robot arm control process of the automated assembly line, which can not only improve the reliability of the robot arm's work, but also enhance the flexibility and response speed of the entire automated assembly line, and provide a basis for the grasping control of workpieces of different sizes and types.

[0059] Figure 7 : is a system block diagram of a robotic arm automatic control system for an automated assembly line according to an embodiment of the present application. Figure 7As shown, according to an embodiment of the present application, a robot arm automatic control system 100 for an automated assembly line includes: a target height parameter acquisition module 110, for acquiring a target height parameter input by an operator, wherein the target height parameter is determined based on the size of the current workpiece and the height requirement of the working area; a current workpiece image acquisition module 120, for acquiring a workpiece image of the current workpiece captured by a camera; an adaptation fixture type determination module 130, for determining an adaptation type of fixture based on the workpiece image; wherein the adaptation fixture type determination module is used to: perform multi-scale feature extraction and feature enhancement processing on the workpiece image to obtain a shallow shape enhancement feature and a deep structure enhancement feature of the workpiece; The shallow shape enhancement feature of the workpiece and the deep structure enhancement feature of the workpiece are jointly interactively encoded with kernel features to obtain the workpiece shape-structure joint interactive feature; based on the workpiece shape-structure joint interactive feature, the workpiece type label is determined, and the adapted type of fixture is determined; a workpiece position data recognition module 140 is used to perform image recognition on the workpiece image to obtain the position data of the current workpiece in the working image; a workpiece position data conversion module 150 is used to convert the position data of the current workpiece in the working image into the three-dimensional coordinates of the robot arm coordinate system to obtain the three-dimensional coordinates of the current workpiece; a robot arm motion trajectory determination module 160 is used to determine the motion trajectory of the robot arm based on the three-dimensional coordinates of the current workpiece.

[0060] Here, those skilled in the art will appreciate that the specific operations of each step in the above-mentioned robotic arm automatic control system for the automated assembly line have been described in detail above. Figures 1 to 6 The invention has been described in detail in the description of the automatic control method of a robotic arm for an automated assembly line, and therefore, its repeated description will be omitted.

[0061] As described above, the robotic arm automatic control system 100 for an automated assembly line according to an embodiment of the present application can be implemented in various terminal devices. In one example, the robotic arm automatic control system 100 for an automated assembly line can be integrated into the terminal device as a software module and / or a hardware module. For example, the robotic arm automatic control system 100 for an automated assembly line can be a software module in the operating system of the terminal device, or can be an application developed for the terminal device; of course, the robotic arm automatic control system 100 for an automated assembly line can also be one of the many hardware modules of the terminal device.

[0062] Alternatively, in another example, the robotic arm automatic control system 100 for the automated assembly line and the terminal device may also be separate devices, and the robotic arm automatic control system 100 for the automated assembly line may be connected to the terminal device via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0063] In summary, the robot arm automatic control system for the automated assembly line based on the embodiment of the present application is explained, which collects the workpiece image of the current workpiece through real-time monitoring by a camera, and introduces an image processing and analysis algorithm based on artificial intelligence and machine vision at the back end to perform multi-level and multi-scale feature extraction on the workpiece image, so as to detect the workpiece type by using the kernel feature joint interactive information between the shallow shape features and deep structural features of the workpiece, so as to select the appropriate type of fixture. In this way, the type of fixture that is suitable for the workpiece can be automatically selected during the robot arm control process of the automated assembly line, which can not only improve the reliability of the robot arm's work, but also enhance the flexibility and response speed of the entire automated assembly line, and provide a basis for the grasping control of workpieces of different sizes and types.

Claims

1. A method for automatically controlling a robot arm for an automated assembly line, characterized in that: include: Acquiring a target height parameter input by an operator, wherein the target height parameter is determined based on a size of a current workpiece and a height requirement of a working area; Acquire a workpiece image of the current workpiece captured by a camera; Based on the workpiece image, determining an adaptive type of fixture, comprising the steps of: performing multi-scale feature extraction and feature enhancement processing on the workpiece image to obtain a shallow shape enhancement feature of the workpiece and a deep structure enhancement feature of the workpiece; performing kernel feature joint interactive encoding on the shallow shape enhancement feature of the workpiece and the deep structure enhancement feature of the workpiece to obtain a workpiece shape-structure joint interactive feature; determining a workpiece type label based on the workpiece shape-structure joint interactive feature, and determining the adaptive type of fixture; Performing image recognition on the workpiece image to obtain position data of the current workpiece in the working image; Converting the position data of the current workpiece in the working image into three-dimensional coordinates of the robot arm coordinate system to obtain the three-dimensional coordinates of the current workpiece; Based on the three-dimensional coordinates of the current workpiece, a motion trajectory of the robot arm is determined.

2. The automatic control method of a robot arm for an automated assembly line according to claim 1, characterized in that: The workpiece image is subjected to multi-scale feature extraction and feature enhancement processing to obtain a shallow shape enhancement feature of the workpiece and a deep structure enhancement feature of the workpiece, including: Inputting the workpiece image into a workpiece image multi-scale feature scanner based on a hollow pyramid network to obtain a workpiece shallow shape feature map and a workpiece deep structure feature map; The workpiece shallow shape feature map and the workpiece deep structure feature map are respectively input into a feature enhancement module based on a converter structure to obtain a workpiece shallow shape enhancement feature map as the workpiece shallow shape enhancement feature and a workpiece deep structure enhancement feature map as the workpiece deep structure enhancement feature.

3. The automatic control method of a robot arm for an automated assembly line according to claim 2, characterized in that: Performing kernel feature joint interactive coding on the shallow shape enhancement feature of the workpiece and the deep structure enhancement feature of the workpiece to obtain the workpiece shape-structure joint interactive feature, including: Inputting the workpiece shallow shape enhancement feature map and the workpiece deep structure enhancement feature map into a kernel feature distillation network to obtain a workpiece shallow shape kernel feature vector and a workpiece deep structure kernel feature vector; Performing global correlation coding on the shallow shape kernel feature vector of the workpiece and the deep structure kernel feature vector of the workpiece to obtain a global coding matrix between the workpiece shape and structure kernel features; Inputting the global encoding matrix between the workpiece shape and the structure kernel features into the sparse mask unit to obtain the associated sparse weight matrix between the workpiece shape and the structure kernel features; Inputting the workpiece shallow layer shape enhancement feature map and the workpiece deep layer structure enhancement feature map into a position-by-position association interaction network to obtain a workpiece shape-structure position-by-position interaction fusion feature map; Based on the associated sparse weight matrix between the workpiece shape-structure kernel features, the workpiece shape-structure position-by-position interactive fusion feature map is subjected to feature significant sparse modulation to obtain a workpiece shape-structure joint interactive feature map as the workpiece shape-structure joint interactive feature.

4. The automatic control method of a robot arm for an automated assembly line according to claim 3, characterized in that: Inputting the workpiece shallow shape enhancement feature map and the workpiece deep structure enhancement feature map into a kernel feature distillation network to obtain a workpiece shallow shape kernel feature vector and a workpiece deep structure kernel feature vector, including: Performing feature decomposition on the workpiece shallow shape enhancement feature map to obtain a sequence of workpiece shallow shape decomposition feature vectors; Calculating the difference characteristics of each workpiece shallow shape decomposition feature vector in the sequence of the workpiece shallow shape decomposition feature vectors relative to other workpiece shallow shape decomposition feature vectors to obtain a workpiece shallow shape decomposition feature difference representation vector; Normalizing each eigenvalue in the workpiece shallow shape decomposition feature difference representation vector to obtain a normalized workpiece shallow shape decomposition feature difference representation vector; The workpiece shallow shape kernel feature vector is obtained by calculating the weighted sum of the sequence of the workpiece shallow shape decomposition feature vectors using each eigenvalue in the normalized workpiece shallow shape decomposition feature difference representation vector as a weight.

5. The automatic control method of a robot arm for an automated assembly line according to claim 4, characterized in that: Calculating the difference feature of each workpiece shallow shape decomposition feature vector in the sequence of the workpiece shallow shape decomposition feature vectors relative to other workpiece shallow shape decomposition feature vectors to obtain a workpiece shallow shape decomposition feature difference representation vector, including: Calculating a norm of positional differences between each shallow shape decomposition feature vector of the workpiece in the sequence of shallow shape decomposition feature vectors of the workpiece and the shallow shape decomposition feature vectors of other workpieces to obtain a sequence of values ​​representing the norm of differences in shallow shape decomposition features of the workpiece; Calculating the average value of the sequence of the workpiece shallow shape decomposition feature difference-norm representation values ​​to obtain the workpiece shallow shape decomposition feature difference-norm average representation value; A plurality of the workpiece shallow shape decomposition feature difference one-norm average representation values ​​are vectorized and arranged to obtain the workpiece shallow shape decomposition feature difference representation vector.

6. The automatic control method of a robot arm for an automated assembly line according to claim 5, characterized in that: Normalizing each eigenvalue in the workpiece shallow shape decomposition feature difference representation vector to obtain a normalized workpiece shallow shape decomposition feature difference representation vector, including: Calculating the exponential function value of each eigenvalue in the shallow shape decomposition feature difference representation vector of the workpiece based on the position and taking the natural constant as the base to obtain the shallow shape decomposition feature difference class support representation vector of the workpiece; Calculating the weighted sum of each eigenvalue in the workpiece shallow shape decomposition feature difference class support representation vector to obtain the workpiece shallow shape decomposition feature difference class support sum; The normalized workpiece shallow shape decomposition feature difference representation vector is obtained by calculating the positional division between each eigenvalue in the workpiece shallow shape decomposition feature difference class support representation vector and the sum of the workpiece shallow shape decomposition feature difference class support.

7. The automatic control method of a robot arm for an automated assembly line according to claim 6, characterized in that: The global encoding matrix between the workpiece shape and the structure kernel features is input into the sparse mask unit to obtain the associated sparse weight matrix between the workpiece shape and the structure kernel features, including: Performing a sparse mask operation on each position feature value in the global encoding matrix between the workpiece shape and structure kernel features to obtain a sparse weight matrix associated with the workpiece shape and structure kernel features; Among them, in response to the eigenvalue in the global coding matrix between the workpiece shape and the structural kernel features being greater than twice a preset threshold, the eigenvalue greater than twice the preset threshold is multiplied by a constant 3; in response to the eigenvalue in the global coding matrix between the workpiece shape and the structural kernel features being less than or equal to the preset threshold, the eigenvalue less than or equal to the preset threshold is divided by a constant 3; in response to the eigenvalue in the global coding matrix between the workpiece shape and the structural kernel features being greater than the preset threshold and less than or equal to twice the preset threshold, the eigenvalue is taken as the original value to obtain the associated sparse weight matrix between the workpiece shape and the structural kernel features.

8. The automatic control method of a robot arm for an automated assembly line according to claim 7, characterized in that: Based on the workpiece shape-structure joint interaction feature, determining the workpiece type label and determining the suitable type of fixture, including: Inputting the workpiece shape-structure joint interaction feature map into a classifier-based workpiece identifier to obtain a recognition result, wherein the recognition result is used to represent a workpiece type label; Based on the workpiece type label, the suitable type of fixture is determined.

9. The automatic control method of a robot arm for an automated assembly line according to claim 8, characterized in that: Based on the three-dimensional coordinates of the current workpiece, the motion trajectory of the robot arm is determined, including: based on the three-dimensional coordinates of the current workpiece, the motion trajectory of the robot arm is calculated to generate a retrograde dynamics solution to ensure that the clamp of the robot arm can accurately reach the position of the current workpiece.

10. A robotic arm automatic control system for an automated assembly line, characterized in that: include: A target height parameter acquisition module, used to acquire a target height parameter input by an operator, wherein the target height parameter is determined based on the size of the current workpiece and the height requirement of the working area; A current workpiece image acquisition module is used to acquire a workpiece image of the current workpiece captured by a camera; An adaptive fixture type determination module is used to determine an adaptive fixture type based on the workpiece image; wherein the adaptive fixture type determination module is used to: perform multi-scale feature extraction and feature enhancement processing on the workpiece image to obtain a shallow shape enhancement feature and a deep structure enhancement feature of the workpiece; perform kernel feature joint interactive encoding on the shallow shape enhancement feature and the deep structure enhancement feature of the workpiece to obtain a workpiece shape-structure joint interactive feature; determine a workpiece type label based on the workpiece shape-structure joint interactive feature, and determine the adaptive fixture type; A workpiece position data recognition module, used for performing image recognition on the workpiece image to obtain the position data of the current workpiece in the working image; A workpiece position data conversion module, used for converting the position data of the current workpiece in the working image into three-dimensional coordinates of the robot arm coordinate system to obtain the three-dimensional coordinates of the current workpiece; The robot arm motion trajectory determination module is used to determine the motion trajectory of the robot arm based on the three-dimensional coordinates of the current workpiece.

Images