Robot gluing quality online detection and real-time correction method and system

Abstract

The invention belongs to the technical field of gluing quality detection, and provides a robot gluing quality online detection and real-time correction method and system, and the technical scheme is that the method comprises the steps: extracting the edge point cloud of a glue line based on a constructed three-dimensional glue shape data set, calculating the multi-dimensional geometric features of the glue line based on the edge point cloud, and preliminarily determining the abnormal point cloud data; constructing a defect point cloud data set based on the preliminarily determined abnormal point cloud data, training a defect identification model based on the constructed defect point cloud data set, and processing the real-time point cloud based on the trained defect identification model to obtain the category and spatial position of the defect; according to the category and the spatial position of the defect, automatically adjusting the gluing process parameter or path, and carrying out secondary gluing to correct the defect based on the adjusted gluing process parameter or path; and acquiring the variation trend of the glue shape parameters in the gluing process, predicting the probability of defect occurrence based on the established mapping relationship between the defect characteristics and the process parameters, and generating corresponding prevention measures.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of glue coating quality detection, and particularly relates to a robot glue coating quality online detection and real-time correction method and system. BACKGROUND

[0002] The statements in this section merely provide background information related to the present application and do not necessarily constitute the prior art.

[0003] In the robot glue coating process, quality defects such as glue breakage, glue leakage, uneven glue shape, and trajectory deviation frequently occur. If not discovered and corrected in time, it will lead to product scrap or rework, seriously affecting production efficiency and yield. At present, the quality detection means in this field mainly relies on offline inspection after glue coating. This method has strong lag and lacks online feedback and immediate correction capability. Although some existing technologies use visual systems for online monitoring, there are still problems such as limited detection accuracy, insufficient coverage of various defects, weak adaptability of defect identification, and single correction strategy, which makes it difficult to meet the needs of high-precision and high-robustness glue coating quality control. SUMMARY

[0004] In order to solve at least one of the technical problems in the background art, the present application provides a robot glue coating quality online detection and real-time correction method and system, which can detect glue coating quality in real time, intelligently identify defect types, and quickly adjust glue coating parameters to realize closed-loop correction, so as to improve production stability and glue coating consistency.

[0005] In order to achieve the above purpose, the present application adopts the following technical solutions: The first aspect of the present application provides a robot glue coating quality online detection and real-time correction method, comprising the following steps: Constructing a three-dimensional glue shape data set based on the obtained point cloud data; Extracting edge point cloud of glue line based on the constructed three-dimensional glue shape data set, calculating multi-dimensional geometric features of glue line based on the edge point cloud, and preliminarily determining abnormal point cloud data according to the multi-dimensional geometric features of glue line; Constructing a defect point cloud data set based on the preliminarily determined abnormal point cloud data, training a defect recognition model based on the constructed defect point cloud data set, and processing real-time point cloud based on the trained defect recognition model to obtain the category and spatial position of the defect; According to the category and spatial position of the defect, automatically adjusting the glue coating process parameters or path, and correcting the defect by secondary glue coating based on the adjusted glue coating process parameters or path; Obtaining the trend of the change of glue shape parameters in the glue coating process, predicting the probability of defect occurrence based on the mapping relationship between the established defect features and process parameters, and generating corresponding preventive measures when the probability exceeds the set threshold.

[0006] Further, the edge point cloud of the glue line is extracted from the constructed three-dimensional glue shape dataset by using point cloud gradient change detection or curvature threshold method.

[0007] Further, the multi-dimensional geometric features of the glue line include the width of the glue line at different sections, the height of the glue line, the uniformity of the glue line, and the total volume of the glue line. The width of the glue line at different sections is calculated according to the Euclidean distance between the left and right edge point clouds. The height of the glue line is calculated according to the distance between the highest point of the section point cloud and the reference surface. When calculating the uniformity of the glue line, the continuous section width and height are statistically analyzed, and the standard deviation or coefficient of variation is used to evaluate the uniformity of the glue line. When calculating the total volume of the glue line, the total volume of the glue line is obtained by integrating the cross-sectional area, which is used to evaluate whether the colloid is excessive or insufficient.

[0008] Further, when constructing the defect recognition model, the geometric features and local texture features of the glue line are extracted based on the constructed defect point cloud dataset, and a mapping relationship with the defect category label is established. The geometric features and local texture features of the glue line are used as the input of the defect detection network, and the defect category label is obtained by combining the mapping relationship.

[0009] Further, when automatically adjusting the glue application process parameters or path according to the category and spatial position of the defect, the adjustment principles include: If the width of the glue line at different sections is greater than the set width, the glue pressure is automatically reduced or the robot running speed is appropriately increased to reduce the glue output per unit length; if the width or height of the glue line is less than the set height, the glue pressure is increased or the running speed is slowed down. When there are glue breakage and glue leakage abnormalities, a glue supplement command is triggered at the defect position to make the needle nozzle stay temporarily and supplement the colloid to restore the continuity of the glue line. When there is a trajectory deviation anomaly, the deviation of the point cloud glue line center from the planned trajectory is compared to generate a local correction path.

[0010] Further, the defect and parameter data recorded in the glue application production process include the type, position coordinates, and severity of the defect; the glue application process parameters at the corresponding time; the correction measures taken and the results.

[0011] Further, the change trend of the glue shape parameters in the glue application process includes the continuous decrease of the width / height indicating that the glue breakage will occur soon, the gradual deviation of the glue line centerline indicating the trajectory deviation, and the unstable or excessive fluctuation of the colloid surface indicating the collapse.

[0012] The second aspect of the present application provides a robot glue coating quality online detection and real-time correction system, comprising: a data acquisition module configured to construct a three-dimensional glue shape data set based on the acquired point cloud data; an abnormal data determination module configured to extract edge point cloud of the glue line based on the constructed three-dimensional glue shape data set, calculate multi-dimensional geometric features of the glue line based on the edge point cloud, and preliminarily determine abnormal point cloud data according to the multi-dimensional geometric features of the glue line; a defect identification module configured to construct a defect point cloud data set based on the preliminarily determined abnormal point cloud data, train a defect identification model based on the constructed defect point cloud data set, and process real-time point cloud based on the trained defect identification model to obtain a category and a spatial position of the defect; a defect correction module configured to automatically adjust a glue coating process parameter or a path according to the category and the spatial position of the defect, and correct the defect by secondary glue coating based on the adjusted glue coating process parameter or the path; a prediction measure generation module configured to acquire a variation trend of a glue shape parameter in the glue coating process, predict a probability of the defect occurring based on a mapping relationship between the established defect feature and the process parameter, and generate a corresponding prevention measure when the probability exceeds a set threshold.

[0013] The third aspect of the present application provides a computer readable storage medium.

[0014] A computer readable storage medium having a computer program stored thereon, the program being executed by a processor to implement the steps of the robot glue coating quality online detection and real-time correction method as described above.

[0015] The fourth aspect of the present application provides a computer device.

[0016] A computer device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor implements the steps of the robot glue coating quality online detection and real-time correction method as described above when executing the program.

[0017] Compared with the prior art, the present application has the following advantages: The present application can identify complex and boundary fuzzy defects by extracting the specific multi-dimensional geometric features of the abnormal point cloud, and can correct the glue coating process by combining with the adaptive adjustment of the process parameters, so that the parameters can be adjusted in advance when the unstable trend of the glue line is detected, and the preventive control of the defects is realized, thereby improving the production stability and the glue coating consistency.

[0018] The advantages of the additional aspects of the present application will be partially given in the following description, partially will become obvious from the following description, or will be understood by the practice of the present application. Attached Figure Description

[0019] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0020] Figure 1 This is a flowchart of a method for online detection and real-time correction of robot adhesive application quality provided by an embodiment of the present invention. Detailed Implementation

[0021] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0022] It should be noted that the following detailed description is illustrative and intended to provide further explanation of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0023] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0024] Example 1 like Figure 1 As shown, this embodiment provides a method for online detection and real-time correction of robot adhesive application quality, including the following steps: Step 1: Construct a 3D gelatin dataset based on the acquired point cloud data; In this embodiment, a structured light camera continuously acquires and updates point clouds at different locations along the adhesive application trajectory, generating a dynamic 3D adhesive shape data stream. This 3D adhesive shape data will serve as input for subsequent defect detection and correction, enabling online monitoring. Specifically, the steps include: Step 101: Obtain point cloud data; In this embodiment, a 3D structured light camera is installed at the end of the glue-applying robot, maintaining a fixed relative position with the glue-applying nozzle. During the glue-applying process, the camera scans the glue line position in real time, acquiring continuous frames of three-dimensional point cloud data to ensure comprehensive coverage of the glue's shape, cross-section, and spatial position.

[0025] Step 102: Preprocess the acquired point cloud data; In this embodiment, the collected point cloud data is denoised, down-sampled and ROI cropped to filter out background points and interference points, and only the glue line and its neighborhood point cloud are retained. The Moving Least Squares (MLS) smoothing algorithm is used to process the glue line surface point cloud to eliminate random noise.

[0026] Step 103, constructing three-dimensional glue shape data based on the preprocessed point cloud data; The geometric parameters of the glue shape are the focus of the glue application quality detection. The actual measured parameter data is compared with the standard data, and if the difference exceeds a certain range, it is judged that there is a quality problem. Therefore, in this embodiment, the three-dimensional modeling is established based on the geometric parameters calculated by the segmented point cloud, which reflects the actual measurement results.

[0027] Specifically, it includes: By point cloud cross-section slicing method: divide the point cloud every fixed step along the glue line direction, extract the glue cross-section data; fit the cross-section curve for the cross-section point cloud, calculate the geometric parameters such as glue line width, height, cross-sectional area; reconstruct the continuous cross-section into a three-dimensional model of the glue line to realize complete modeling of the glue shape.

[0028] Step 2: Extract the edge point cloud of the glue line based on the constructed three-dimensional glue shape data set, calculate the multi-dimensional geometric features of the glue line based on the edge point cloud, and preliminarily determine the abnormal point cloud data according to the multi-dimensional geometric features of the glue line; Specifically, it includes the following steps: Step 201, extracting the edge point cloud of the glue line from the constructed three-dimensional glue shape data set; In this embodiment, the point cloud gradient change detection or curvature threshold method is used to extract the edge point cloud of the glue line; The edge point cloud reflects the boundaries of the two sides of the glue line, and is the core basis for measuring the width, uniformity and offset of the glue line.

[0029] Step 202, calculating the multi-dimensional geometric features of the glue line based on the edge point cloud; In this embodiment, the multi-dimensional geometric features of the glue line include the width of the glue line at different cross-sections, the height of the glue line, the uniformity of the glue line and the total volume of the glue line; Specifically, the width of the glue line at different cross-sections is calculated according to the Euclidean distance between the left and right edge point clouds; The height of the glue line is calculated according to the distance between the highest point of the cross-section point cloud and the reference surface; When calculating the uniformity of the glue line, the width and height of the continuous cross-section are statistically analyzed, and the standard deviation or coefficient of variation is used to evaluate the uniformity of the glue line; When calculating the total volume of the glue line, the total volume of the glue line is obtained by integrating the cross-sectional area, which is used to evaluate whether the glue is excessive or insufficient.

[0030] Step 203: Based on the multi-dimensional geometric features of the adhesive line, preliminarily determine the abnormal point cloud data; In this embodiment, the principle for anomaly detection and judgment is as follows: If the width or height of the adhesive line exceeds the corresponding process threshold at different cross-sections, or if the uniformity of the adhesive line is lower than the set standard, it is marked as a potential defect area.

[0031] Step 3: Construct a defect point cloud dataset based on the initially determined abnormal point cloud data, train the defect identification model based on the constructed defect point cloud dataset, and process the real-time point cloud based on the trained defect identification model to obtain the defect category and spatial location. Specifically, the steps include the following: Step 301: Construct a defect point cloud dataset based on the initially identified abnormal point cloud data; In this embodiment, the abnormal point cloud data includes a large number of different types of defect point cloud data, including glue breakage, air bubbles, glue line misalignment, glue collapse, and excessive / inadequate width. Each type of defect point cloud data is labeled to construct a defect point cloud training set.

[0032] Step 302: Train the defect identification model based on the constructed defect point cloud dataset; When constructing the defect recognition model, the structural feature extraction module first extracts the geometric features (width, height, volume) and local texture features (surface smoothness, edge continuity) of the glue line simultaneously based on the constructed defect point cloud dataset to improve recognition capabilities. Specifically, the Vision Transformer architecture can be adopted, which can more comprehensively extract the necessary information from the point cloud data and establish a mapping relationship between structural features and defect category labels. Then, defect identification is performed. Specifically, a deep learning network is used for identification, such as the YOLOv10 defect detection network. The input of the defect detection network is structural features, and the output is the defect category label.

[0033] Step 303: Process the real-time point cloud based on the trained defect recognition model to obtain the defect category; During the online detection phase, real-time point clouds are input into the trained deep learning model to automatically determine the defect category and severity. Compared with traditional threshold-based methods, this model can identify complex and vaguely defined defects, significantly improving detection accuracy.

[0034] Finally, the corresponding defect category (such as glue breakage, uneven glue line, air bubbles, etc.) and specific location coordinates are input; the defect data and process parameters (glue application pressure, moving speed, needle temperature, etc.) are stored together for subsequent correction and defect prediction model training.

[0035] Step 4: Based on the type and spatial location of the defect, automatically adjust the glue application process parameters or path, and perform a second glue application based on the adjusted glue application process parameters or path; After completing defect detection and classification, this invention ensures the stability and consistency of the adhesive application process through a real-time correction strategy, specifically including the following steps: Step 401: Automatically adjust the glue application process parameters or path according to the type and spatial location of the defect; In this embodiment, the adjustment principles include: When the width or height of the adhesive line is abnormal at different cross sections, if the width of the adhesive line at different cross sections is greater than the set width, the dispensing pressure will be automatically reduced or the robot running speed will be appropriately increased to reduce the amount of adhesive dispensed per unit length; if the width or height of the adhesive line is less than the set height, the dispensing pressure will be increased or the running speed will be slowed down to ensure complete adhesive filling.

[0036] When glue breakage or leakage occurs, a glue replenishment command is triggered at the defect location, causing the nozzle to pause briefly and replenish the glue, restoring the continuity of the glue line. When an abnormal trajectory deviation occurs, the deviation between the center of the adhesive line in the point cloud and the planned trajectory is compared to generate a local correction path; the robot adjusts its motion trajectory in real time according to the offset vector to bring the adhesive line back to the theoretical center line.

[0037] Step 402: Perform secondary adhesive application to correct defects based on the adjusted adhesive application process parameters or path; During the initial glue application, the system records the location, type, and severity of defects. During the second glue application, the robot only applies glue to the defective area locally, rather than reworking the entire trajectory, thus significantly improving production efficiency. The correction path sets transition zones at both ends of the defective section and uses smooth trajectory interpolation to avoid new defects appearing at the glue application point.

[0038] Then, the defect type, point cloud features, correction parameters, and repair results are stored in the database; a mapping table between defects and adhesive application parameters is established, such as "when the width is too wide, the flow rate is reduced by 10% and the speed is increased by 5%".

[0039] After correction, the camera will re-inspect the system. If the standard is still not met, the system will further adjust the parameters based on the feedback until the process requirements are met. After successful repair, the final parameters will be recorded as an experience reference for similar defects in the future.

[0040] With the accumulation of historical data, defect precursors can be predicted using defect-parameter mapping and deep learning models. When the system detects an unstable trend in the adhesive line, parameters can be adjusted in advance to achieve preventive control of defects.

[0041] Step 5: Obtain the changing trend of glue shape parameters during the glue application process. Based on the established mapping relationship between defect features and process parameters, predict the probability of defect occurrence. When the probability exceeds the set threshold, generate corresponding preventive measures. Specifically, the steps include the following: Step 501: Obtain the changing trend of adhesive parameters during the adhesive application process; In this embodiment, obtaining the changing trend of adhesive parameters during the coating process specifically includes: A continuous decrease in width / height may indicate impending glue breakage; a gradual shift in the glue line centerline may indicate trajectory deviation; and instability or excessive fluctuation in the glue surface may indicate collapse. These precursor features are extracted in real time by visual sensors and point cloud data and compared with historical defect features in the database.

[0042] Step 502: Record the defects and parameter data in the glue coating process, analyze the defects and parameter data in the glue coating process, and establish the mapping relationship between defect characteristics and process parameters; The recording of defects and parameter data during the adhesive coating process includes: In this embodiment, during the glue coating process, the following information is automatically recorded: the type of defect (e.g., glue breakage, excessive width, offset, etc.), location coordinates, and severity; the corresponding glue coating process parameters (glue dispensing pressure, nozzle movement speed, nozzle height, ambient temperature / humidity, etc.); and the corrective measures taken and their results (success / failure). This data is uniformly stored in a defect parameter database, providing data support for subsequent modeling.

[0043] Among them, the defects and parameter data in the glue coating production process are analyzed to establish the mapping relationship between defect characteristics and process parameters; for example: glue line too wide - glue pressure too high / speed too low; glue break - insufficient glue flow or air pressure fluctuation.

[0044] By introducing machine learning methods based on regression and classification (such as random forest, support vector machine, and lightweight neural network), patterns are extracted from multi-dimensional data to achieve nonlinear correlation modeling between defects and parameters. As production data continues to accumulate, the model parameters will be continuously optimized to adapt to the coating process under different batches, materials, or environmental conditions.

[0045] Step 503: Based on the established mapping relationship between defect features and process parameters, predict the probability of defect occurrence. When the probability exceeds the set threshold, generate corresponding preventive measures. Specifically, based on the established mapping relationship between defect features and process parameters, the probability of defect occurrence is predicted, including: using deep learning or time series prediction models to predict the probability of upcoming defect types. When the predicted probability exceeds the set threshold, preventive measures are triggered immediately, such as: adjusting the dispensing pressure by ±5%; fine-tuning the nozzle height by 0.1 to 0.3 mm; changing the moving speed; or adding glue in advance.

[0046] This invention corrects abnormal trends in advance, thereby preventing defects from actually occurring; if the prediction accuracy continues to improve, the system can achieve the goal of "zero-defect glue application"; the results of preventive adjustments are also recorded in the database for continuous training and optimization of the prediction model.

[0047] Example 2 This embodiment provides an online detection and real-time correction system for the adhesive coating quality of a robot, including: The data acquisition module is used to construct a 3D gel model dataset based on the acquired point cloud data; The abnormal data identification module is used to extract edge point clouds of glue lines based on the constructed 3D glue pattern dataset, calculate the multi-dimensional geometric features of glue lines based on the edge point clouds, and preliminarily identify abnormal point cloud data based on the multi-dimensional geometric features of glue lines. The defect identification module is used to construct a defect point cloud dataset based on the initially determined abnormal point cloud data, train the defect identification model based on the constructed defect point cloud dataset, and process the real-time point cloud based on the trained defect identification model to obtain the category and spatial location of the defect. The defect correction module is used to automatically adjust the glue application process parameters or path according to the type and spatial location of the defect, and perform secondary glue application to correct the defect based on the adjusted glue application process parameters or path. The predictive measures generation module is used to obtain the changing trend of glue parameters during the glue application process. Based on the established mapping relationship between defect features and process parameters, it predicts the probability of defect occurrence. When the probability exceeds the set threshold, it generates corresponding preventive measures.

[0048] It should be noted that the specific implementation of the robot glue coating quality online detection and real-time correction system in this embodiment of the invention is similar to the specific implementation of the robot glue coating quality online detection and real-time correction method in this embodiment of the invention. Please refer to the description in the method section for details. To reduce redundancy, it will not be repeated here.

[0049] Example 3 This embodiment provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the above-described method for online detection and real-time correction of robot adhesive application quality.

[0050] Example 4 This embodiment provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps in the above-described method for online detection and real-time correction of robot adhesive application quality.

[0051] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of hardware embodiments, software embodiments, or embodiments combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage and optical storage) containing computer-usable program code.

[0052] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0053] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0054] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0055] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0056] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for online detection and real-time correction of adhesive application quality in robots, characterized in that, Includes the following steps: A 3D gelatin dataset was constructed based on the acquired point cloud data; Edge point clouds of the glue lines are extracted based on the constructed 3D glue pattern dataset. Multi-dimensional geometric features of the glue lines are calculated based on the edge point clouds. Abnormal point cloud data are preliminarily identified based on the multi-dimensional geometric features of the glue lines. A defect point cloud dataset is constructed based on the initially identified abnormal point cloud data. A defect identification model is trained based on the constructed defect point cloud dataset. The real-time point cloud is then processed based on the trained defect identification model to obtain the defect category and spatial location. Based on the type and spatial location of the defect, the glue application process parameters or path are automatically adjusted, and a second glue application is performed to correct the defect based on the adjusted glue application process parameters or path. The system acquires the changing trends of adhesive parameters during the coating process, predicts the probability of defects based on the established mapping relationship between defect features and process parameters, and generates corresponding preventive measures when the probability exceeds a set threshold.

2. The method for online detection and real-time correction of robot adhesive application quality as described in claim 1, characterized in that, When extracting the edge point cloud of the glue line from the constructed 3D glue dataset, point cloud gradient change detection or curvature thresholding method is used to extract the edge point cloud of the glue line.

3. The method for online detection and real-time correction of robot adhesive application quality as described in claim 1, characterized in that, The multi-dimensional geometric characteristics of the adhesive line include its width, height, uniformity, and total volume at different cross-sections. The width of the adhesive line at different cross-sections is calculated based on the Euclidean distance between the point clouds on the left and right edges; The height of the adhesive line is calculated based on the distance between the highest point of the cross-sectional point cloud and the reference plane; When calculating the uniformity of the adhesive line, statistical analysis is performed on the width and height of the continuous cross section, and the standard deviation or coefficient of variation is used to evaluate the uniformity of the adhesive line. When calculating the total volume of the adhesive line, the total volume of the adhesive line is obtained by integrating the cross-sectional area, which is used to assess whether the adhesive is excessive or insufficient.

4. The method for online detection and real-time correction of robot adhesive application quality as described in claim 1, characterized in that, The defect identification model is constructed by simultaneously extracting the geometric features and local texture features of the glue line based on the constructed defect point cloud dataset, and establishing a mapping relationship with the defect category label. The geometric features and local texture features of the adhesive line are used as input to the defect detection network, and the defect category label is obtained by combining the mapping relationship.

5. The method for online detection and real-time correction of robot adhesive application quality as described in claim 1, characterized in that, When automatically adjusting the adhesive application process parameters or path based on the type and spatial location of the defect, the adjustment principles include: If the width of the adhesive line in different cross sections is greater than the set width, the dispensing pressure will be automatically reduced or the robot's running speed will be appropriately increased to reduce the amount of adhesive dispensed per unit length; if the width or height of the adhesive line is less than the set height, the dispensing pressure will be increased or the running speed will be reduced. When glue breakage or leakage occurs, a glue replenishment command is triggered at the defect location, causing the nozzle to pause briefly and replenish the glue, restoring the continuity of the glue line. When an abnormal trajectory deviation occurs, the deviation between the center of the point cloud adhesive line and the planned trajectory is compared to generate a local correction path.

6. The method for online detection and real-time correction of robot adhesive application quality as described in claim 1, characterized in that, The recorded defect and parameter data during the adhesive coating production process include the type, location coordinates, and severity of the defects; The adhesive application process parameters at the corresponding time point; the correction measures adopted and the results.

7. The method for online detection and real-time correction of robot adhesive application quality as described in claim 1, characterized in that, The changing trends of the adhesive parameters during the coating process include: a continuous decrease in width / height indicating impending adhesive breakage; a gradual shift in the center line of the adhesive line indicating trajectory deviation; and instability or excessive fluctuation of the adhesive surface indicating collapse.

8. A robot adhesive application quality online detection and real-time correction system, characterized in that, include: The data acquisition module is used to construct a 3D gel model dataset based on the acquired point cloud data; The abnormal data identification module is used to extract edge point clouds of glue lines based on the constructed 3D glue pattern dataset, calculate the multi-dimensional geometric features of glue lines based on the edge point clouds, and preliminarily identify abnormal point cloud data based on the multi-dimensional geometric features of glue lines. The defect identification module is used to construct a defect point cloud dataset based on the initially determined abnormal point cloud data, train the defect identification model based on the constructed defect point cloud dataset, and process the real-time point cloud based on the trained defect identification model to obtain the category and spatial location of the defect. The defect correction module is used to automatically adjust the glue application process parameters or path according to the type and spatial location of the defect, and perform secondary glue application to correct the defect based on the adjusted glue application process parameters or path. The predictive measures generation module is used to obtain the changing trend of glue parameters during the glue application process. Based on the established mapping relationship between defect features and process parameters, it predicts the probability of defect occurrence. When the probability exceeds the set threshold, it generates corresponding preventive measures.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by the processor, the program implements the steps in the online detection and real-time correction method for robot adhesive application quality as described in any one of claims 1-7.

10. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps in the robot adhesive application quality online detection and real-time correction method as described in any one of claims 1-7.

Images