Real-time gluing detection system and use method thereof

Abstract

The invention relates to a real-time gluing detection system and a use method thereof, a contour acquisition module comprises a plurality of line structure light contours, a first control interface and a first network controller, the line structure light contours are installed on glue guns, and the first network controller is used for controlling the acquisition time sequence of the contours in a multi-contours network; the robot module comprises a six-axis robot and a second control interface, and the first processing module controls the six-axis robot to advance according to a specified track through the second control interface and obtains a real-time pose of the six-axis robot; the glue gun module comprises a glue gun and a material feeding mechanism, and the material feeding mechanism carries out on-off and flow control on the glue gun according to the control signal sent by the first processing module; the first processing module comprises a CPU, an ROM and an RAM; and the display module comprises a display card and a display. According to the invention, the requirement of high-precision detection can be met, and a rapid and automatic detection process can be realized, so that the accuracy and efficiency of gluing quality detection are improved.

Description

Technical Field

[0001] This invention relates to adhesive coating technology, and in particular to a real-time adhesive coating detection system and its usage method. Background Technology

[0002] In the automotive and lithium battery industries, adhesive application technology is widely used in the manufacturing process of various components. The size, shape, location, and quality of the adhesive application are extremely important for preventing leaks and ensuring the structural strength of the product. With the development of 3D vision inspection technology, the demand for 3D measurement of adhesives has led to the application of 3D vision inspection in adhesive application inspection. Among these methods, line structured light, due to its high precision, is often used after adhesive application, employing a motion mechanism (such as a robot or servo drive) to scan the adhesive profile using a line structured light profilometer. However, this method requires additional vision cycles, which cannot meet the increasingly demanding requirements of customers for inspection cycle speed and accuracy.

[0003] Currently, most products on the market that utilize real-time adhesive application inspection are based on 2D imaging technology. Products using 3D technology suffer from high costs and excessive integration, making functional decomposition difficult. Traditional adhesive application inspection methods primarily rely on manual sampling, resulting in significant variations in individual inspection results and being affected by the individual work conditions of the inspectors. Furthermore, existing machine vision-based non-contact adhesive application inspection solutions, such as 2D multi-view cameras, while capable of detecting adhesive width and breakage to some extent, still have limitations in detecting crucial parameters such as adhesive height. Summary of the Invention

[0004] The purpose of this invention is to provide a real-time adhesive coating inspection system and its usage method, which can meet the requirements of high-precision inspection and realize a fast and automated inspection process, thereby improving the accuracy and efficiency of adhesive coating quality inspection.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a real-time glue application detection system, comprising a contour acquisition module, a robot module, a glue gun module, a first processing module, and a display module;

[0006] The contour acquisition module includes several line structured light contour meters, a first control interface and a first network controller mounted on the line structured light contour meters. The line structured light contour meters are mounted on glue guns. The first network controller is used to control the acquisition timing of the contour meters in a multi-contour meter network. The first processing module triggers the contour meters to start capturing contours through the first control interface.

[0007] The robot module includes a six-axis robot and a second control interface mounted on the six-axis robot. The first processing module controls the six-axis robot to move along a specified trajectory through the second control interface and obtains the real-time pose of the six-axis robot.

[0008] The glue gun module includes a glue gun and a feeding mechanism. The feeding mechanism controls the switching on and off of the glue gun and the flow rate according to the control signal sent by the first processing module.

[0009] The first processing module includes a CPU, a ROM, and a RAM. The CPU is used to perform operations, the RAM is used to cache contour data, point cloud data, and pose data, and the ROM is used to store programs and data.

[0010] The display module includes a graphics card and a monitor, used to display result data and point clouds in real time.

[0011] Furthermore, the structured light profiler is located on the rear side or both sides in the forward direction of the glue gun, or on the four sides in the front, rear, left, and right of the glue gun.

[0012] A method of using the real-time adhesive application detection system according to claim 1.

[0013] The method of use includes the following steps:

[0014] Step S100: Turn on the line structured light profiler to acquire profile data in real time, store it in the profile buffer area, acquire the robot's real-time pose from the robot module, send out control signals, and turn on the glue gun.

[0015] Step S200: Acquire contour data and robot real-time pose multiple times within the interval time Δt. The pose processing module differentiates the pose based on the number of acquired contours and maps the contours to the poses one by one.

[0016] Step S300: In the glue gun positioning module, obtain the glue gun coordinates corresponding to multiple real-time poses, and obtain the glue gun forward direction;

[0017] Step S400: In the first point cloud generation module, based on the hand-eye calibration results and the differentiated pose, the contour is generated into a point cloud and then unified into the robot coordinate system to generate the first point cloud and store it in the first point cloud cache area.

[0018] Step S500: When the number of point clouds cached in the first point cloud cache area reaches a specified number, the first point cloud is rotated and cut, and the processed first point cloud is stitched together in the second point cloud generation module to generate the second point cloud.

[0019] Step S600: In the second point cloud processing module, a glue shape ROI is generated based on the glue gun pose, the point cloud within the ROI is cut out, the glue shape is judged to be qualified, and the result is returned to the display module.

[0020] Step S700: After the glue application is completed and all data is obtained, determine whether there is glue breakage or glue accumulation, and return the results to the display module.

[0021] Define the coordinate system of the line structured light profiler as O cam The robot's base coordinate system is O. base The coordinate system of the robot flange end is O. end The coordinate system of the calibration plate is O. calobj The tool coordinate system at the probe tip is O tool The coordinate system of the glue gun nozzle is O. gun The line structured light profiler and probe are installed at the end of the robot flange, and their rigid body transformation relationship with the end of the robot flange is fixed; step S100 further comprises:

[0022] Step S101: Start executing from time t0, initial pose of the robot. During the time interval from t0 to t1 = t0 + Δt, the contour acquisition module continuously acquires contour data in real time and stores it in the contour cache area;

[0023] This represents the coordinate system O at the end of the robot flange at time t0. end To the robot base coordinate system O base The pose; the pose includes a translation component and a rotation component, wherein the translation component is represented as... The three coordinates are represented by three Euler angles [rx, ry, rz].

[0024] Step S102: The robot starts running and obtains the robot's pose at time t1. Will Set as Set as

[0025] This represents the coordinate system O at the end of the robot flange at time t1. end To the robot base coordinate system O base The position;

[0026] Step S103: Send a control signal to turn on the glue gun.

[0027] Furthermore, step S200 further comprises: the profilometer acquires n contour data points within a time interval Δt, and the pose processing module... and Generate n poses on average between the two poses. The formula for generating the formula is as follows:

[0028]

[0029] Where rx i ,ry i ,rzi remain unchanged. Indicates the coordinate system O at the end of the robot flange end To the robot base coordinate system O base The nth pose.

[0030] Furthermore, step S300 further comprises: obtaining the glue gun positioning module and The corresponding glue gun is in O base The coordinates below are given by the following formula:

[0031]

[0032] in Indicates the glue gun calibration result O gun Relative to O end Translation relationship;

[0033] Calculate vectors The angle θ between the y-axis and the x-axis is given by the following formula:

[0034]

[0035] in The y-axis unit vector. Then the transformed pose P is obtained by rotating θ around its own Z-axis. θ = [0,0,0,0,0,θ];

[0036] The calibration results of the glue gun include the following steps:

[0037] Step a: Use the TCP tip-touch four-point method to calibrate the probe tip coordinate system.

[0038] Step b: Calibrate O using the TCP cusp-touch three-point method calobj and O base Rigid body transformation relations

[0039] Step c: Move the glue gun to O calobj The origin is set so that the glue outlet touches any point on the target with known coordinates. Let this point be A, and the translation relationship of point A relative to the target coordinate system is as follows: At this time, robot O end and O base The rigid body transformation relationship between them is The calibration result is O. gun Relative to O end Translation relationship for:

[0040]

[0041] Furthermore, in step S400, "simultaneously storing in the first point cloud cache area" further means:

[0042] Compare the angle between the current profilometer direction and the glue gun's forward direction, and calculate... and The included angle ρ is calculated using the following formula:

[0043] glue gun in O base The coordinates below are

[0044]

[0045] Contouring at O base The coordinates below are

[0046]

[0047] Assuming there are i profilometers, ρ can be obtained. i i = 1, 2, ...;

[0048] When ρ i When the angle is less than 90°, the current profilometer is considered to be ahead of the direction of travel, and the current first point cloud is placed in the front point cloud region of the first point cloud buffer region; when ρ i When the angle is greater than 90°, it is assumed that the current profilometer is behind the direction of travel, and the current first point cloud is placed in the back point cloud region of the first point cloud buffer region.

[0049] Furthermore, step S400 further comprises: retrieving the contours from the contour buffer region to obtain a single contour point cloud, where n contours correspond to n poses, and for a single contour point cloud p i Transform to O using the following formula base Down:

[0050]

[0051] in and Representing O cam and O base First, obtain the point set of a single contour, then convert the point cloud of n single contours. After merging, the first point cloud is obtained, and the first point cloud is stored in the first point cloud cache area; This represents the result of the profilometer calibration.

[0052] Furthermore, step S500 further comprises:

[0053] Step S501: Repeat steps S100 to S400. After the first loop is completed, the loop operation of S100 is to... Set as And wait for the second pose. Then Set as Then, steps S200, S300, and S400 are executed sequentially to obtain a new first point cloud. Steps S100 to S400 are repeated until the number of first point clouds reaches the number m specified by the user.

[0054] The formula for the size of m is as follows:

[0055]

[0056] Where d represents the distance between the nozzle and the linear structured light plane, v represents the speed of the motion mechanism, Δt represents the shortest time interval for the robot module to acquire the real-time flange pose of the robot, and [·] represents rounding up;

[0057] Step S502: For each first point cloud point set p base Perform the following pose transformation.

[0058] p base′ =P θ p base

[0059] Based on ROI, select p base′ The points in the diagram have the following ROI range:

[0060]

[0061]

[0062] Where W and H are the width and height of the ROI specified by the user. and Represent the y-component of translation between two coordinates; p base′ The filtered point clouds are merged to obtain the second point cloud.

[0063] Furthermore, S600 further comprises: after obtaining the second point cloud, dividing the second point cloud along the y-direction according to the minimum detection length specified by the user to obtain several third point clouds; when the normal glue width specified by the user is w, cutting the third point cloud along the x-direction into three parts, the second part being wider than w; after performing planar fitting on the first and third part of the point cloud to obtain the glue application plane, subtracting the glue application plane from the second part of the point cloud to obtain the point cloud of the glue portion, then calculating the glue height and glue width, and then according to the threshold set by the user, when the calculated glue width is less than the minimum glue width threshold or the glue height is less than the minimum glue width threshold, the third point cloud is divided into three parts. If the minimum glue height threshold is met, the current third point cloud is determined to be a glue break. If the glue width is greater than the maximum glue width threshold or the glue height is greater than the maximum glue height threshold, the current third point cloud is determined to be a glue pile. The same processing is performed on each third point cloud. If any third point cloud results in a glue pile or glue break, the first processing result is considered abnormal. If all third point cloud results are normal, the first processing result is considered normal, and finally the first processing result is output to the display module. If the current first processing result is considered to be a glue break or glue pile, the start and end coordinates of the glue gun corresponding to the current second point cloud are recorded. and

[0064] S700 further comprises: until all glue point clouds have obtained the first processing result, obtaining the first processing quantity of glue breakage or glue accumulation in the first processing result; when the number of first processing results with continuous glue breakage reaches the glue breakage threshold given by the user, it is determined that glue breakage exists; when the number of first processing results with continuous glue accumulation reaches the glue accumulation threshold given by the user, it is determined that glue accumulation exists; if glue breakage or glue accumulation exists, the starting coordinates and ending coordinates of the glue gun corresponding to the abnormality are obtained, the Euclidean distance is calculated and accumulated to obtain the length of the abnormal part, and returned to the display module as the second processing result.

[0065] The beneficial effects of this invention are as follows: This invention can meet the requirements of high-precision detection and realize a fast and automated detection process, thereby improving the accuracy and efficiency of adhesive coating quality detection. At the same time, the system should also be able to adapt to different types of adhesive coating conditions, including non-linear adhesive coating trajectories such as corners and curves, to comprehensively improve the quality and reliability of adhesive coating detection.

[0066] By automating the real-time monitoring process, the problems of large individual test results and the influence of individual work status in traditional manual sampling are overcome, thus improving the accuracy and stability of test results.

[0067] The use of a line structured light profilometer for scanning during the adhesive application process avoids the visual time-consuming problem associated with traditional adhesive application methods that require scanning after the adhesive application is completed. This meets the increasingly demanding requirements of customers for both inspection time and accuracy.

[0068] By optimizing the layout design of the profilometer and adopting a grid pattern or other reasonable layout, it is ensured that the adhesive application can be effectively detected in any trajectory direction, which effectively solves the pain point of detecting non-linear adhesive application trajectories such as corners and curves, and improves the comprehensiveness of the detection.

[0069] By using the profilometer data before and after the glue gun, accurate glue height values ​​can be obtained even for uneven glue application surfaces. Attached Figure Description

[0070] Figure 1 This is a schematic diagram of the overall system architecture;

[0071] Figure 2 A schematic diagram of the installation of a single profilometer;

[0072] Figure 3 A schematic diagram showing the installation of two profilometers;

[0073] Figure 4 A schematic diagram showing the installation of four profilometers;

[0074] Figure 5 This is a flowchart of the system operation in Example 1;

[0075] Figure 6 This is a schematic diagram of the third point cloud;

[0076] Figure 7 Example 2: System operation flowchart.

[0077] The components include: 1. profilometer, 2. glue gun, 3. connecting mechanism, 4. glue gun fixing plate, 5. profilometer height adjustment mechanism, 6. profilometer fixing mechanism, 7. glue tube, and 8. smooth surface. Detailed Implementation

[0078] The invention will now be further described with reference to the accompanying drawings.

[0079] Please see Figures 1 to 7 The present invention provides an embodiment: a real-time glue application detection system, including a contour acquisition module, a robot module, a glue gun module, a first processing module and a display module;

[0080] The contour acquisition module includes several line structured light contour units 1, a first control interface and a first network controller mounted on each line structured light contour unit 1. Each line structured light contour unit 1 is mounted on a glue gun 2. The first network controller controls the acquisition sequence of the contour units 1 in a network configuration, ensuring that the light emitted by each contour unit 1 does not interfere with each other. A first processing module triggers the contour unit 1 to begin contour acquisition via the first control interface. The first processing module obtains the contour data from the contour unit 1 via the first control interface. The line structured light contour unit 1 is mounted on the glue gun 2 via a contour unit fixing mechanism 6 and a contour unit height adjustment mechanism 5. The position of the line structured light contour unit 1 can be adjusted via the contour unit height adjustment mechanism 5. The contour unit fixing mechanism 6 and the contour unit height adjustment mechanism 5 are well-known structures to those skilled in the art and will not be explained in detail here. The glue gun 2 is fixed to a connecting mechanism 3 via a glue gun fixing plate 4. The connecting mechanism 3 is used to connect with a robot. A glue tube 7 is located at the glue outlet of the glue gun 2.

[0081] The robot module includes a six-axis robot and a second control interface mounted on the six-axis robot. The first processing module controls the six-axis robot to move along a specified trajectory through the second control interface and obtains the real-time pose of the six-axis robot.

[0082] The glue gun module includes a glue gun 2 and a feeding mechanism. The feeding mechanism controls the switching on and off of the glue gun 2 and the flow rate according to the control signal issued by the first processing module. The control signal can be transmitted through an I / O board or other similar functional media.

[0083] The first processing module includes a CPU, a ROM, and a RAM. The CPU is used to perform operations, the RAM is used to cache contour data, point cloud data, and pose data, and the ROM is used to store programs and data.

[0084] The display module includes a graphics card and a monitor, used to display result data and point clouds in real time.

[0085] Please continue reading. Figures 1 to 4 As shown, in one embodiment of the present invention, the structured light profilometer 1 is disposed on the rear side or the front and rear sides of the glue gun 2 in the forward direction, or on the four sides of the glue gun 2. Several line structured light profilometers 1 need to be installed around the glue gun 2. Different installation methods can be used for different needs, using one or two profilometers 1 depending on the specific glue type. For glue application requirements where the motion mechanism travels in one direction and the glue application surface is flat, a single profilometer 1 can be installed after the travel direction. The glue application trajectory passes through the light plane 8 emitted by the profilometer 1, and the profilometer 1 can obtain the cross-sectional profile of the glue type, such as... Figure 2As shown, when using one line structured light profiler 1, it is positioned on the rear side of the forward direction. For applications where the motion mechanism travels in two opposite directions or in one direction, and the adhesive application surface is uneven, two profilers 1 can be installed on opposite sides of the forward direction, with the adhesive application trajectory passing through the light plane 8, as shown. Figure 3 The two units shown are positioned on the front and rear sides of the glue gun 2 in the forward direction.

[0086] For applications requiring adhesive application in any direction of travel of the motion mechanism, four contour applicators 1 can be mounted in a cross shape around the glue gun 2. Figure 4 As shown, when four line structured light profilers are used on the front, back, left, and right sides, the line structured light profilers are distributed in a "well" shape around the glue nozzle at 360°, ensuring that glue application can be detected in any trajectory direction, thus solving the pain point of detecting non-linear glue application trajectories such as corners and curves.

[0087] The profilometer 1 can generate point clouds in real time from the acquired contour data and detect problems such as glue breakage and glue accumulation, and finally return the processing results in real time. It can perform real-time glue pattern detection for various linear coating processes, adapt to various coating conditions, including non-linear coating trajectories such as corners and curves, and quickly output the detection results after coating is completed, effectively solving the problem of lack of real-time performance in current 3D coating detection technology.

[0088] Please see Figures 1 to 7 The present invention provides another embodiment: a method of using the real-time adhesive coating detection system according to claim 1.

[0089] The method of use includes the following steps:

[0090] Step S100: Turn on the line structured light profiler 1, acquire profile data in real time, store it in the profile buffer area, acquire the robot's real-time pose from the robot module, send out control signals, and turn on the glue gun 2.

[0091] Step S200: Acquire contour data and robot real-time pose multiple times within the interval time Δt. The pose processing module differentiates the multiple poses based on the number of acquired contours and maps the contours to the poses one by one.

[0092] Step S300: In the glue gun positioning module, obtain the coordinates of glue gun 2 corresponding to multiple real-time poses, and obtain the forward direction of glue gun 2;

[0093] Step S400: In the first point cloud generation module, based on the hand-eye calibration results and the differentiated pose, the contour is generated into a point cloud and then unified into the robot coordinate system to generate the first point cloud and store it in the first point cloud cache area.

[0094] Step S500: When the number of point clouds cached in the first point cloud cache area reaches a specified number, the first point cloud is rotated and cut, and the processed first point cloud is stitched together in the second point cloud generation module to generate the second point cloud.

[0095] Step S600: In the second point cloud processing module, a glue shape ROI is generated based on the 2-position pose of the glue gun, the point cloud within the ROI is cut out, the glue shape is judged to be qualified, and the result is returned to the display module.

[0096] Step S700: After the glue application is completed and all data is obtained, determine whether there is glue breakage or glue accumulation, and return the results to the display module.

[0097] Define the coordinate system of the line structured light profiler as O. cam The robot's base coordinate system is O. base The coordinate system of the robot flange end is O. end The coordinate system of the calibration plate is O. calobj The tool coordinate system at the probe tip is O tool The coordinate system of the glue gun nozzle 2 is O. gun The line structured light profiler 1 and the probe are installed at the end of the robot flange, and their rigid body transformation relationship with the end of the robot flange is fixed; step S100 further comprises:

[0098] Step S101: Start executing from time t0, initial pose of the robot. During the time interval from t0 to t1 = t0 + Δt, the contour acquisition module continuously acquires contour data in real time and stores it in the contour cache area;

[0099] This represents the coordinate system O at the end of the robot flange at time t0. end To the robot base coordinate system O base The pose; the pose includes a translation component and a rotation component, wherein the translation component is represented as... The three coordinates are represented by three Euler angles [rx, ry, rz].

[0100] Step S102: The robot starts running and obtains the robot's pose at time t1. Will Set as Set as During time intervals t0 to t1, the old pose is 0 and the new pose is 1. During time intervals t1 to t2, the old pose is 1 and the new pose is 2, and so on. The values ​​of the old and new poses are refreshed every time interval. This represents the coordinate system O at the end of the robot flange at time t1. end To the robot base coordinate system O baseThe position;

[0101] Step S103: Send a control signal to turn on glue gun 2.

[0102] Please continue reading. Figures 1 to 7 As shown, in one embodiment of the present invention, step S500 further comprises:

[0103] Step S200 further comprises: during the time interval Δt, the contour instrument 1 collects n contour data, and the pose processing module, according to... and Generate n poses on average between the two poses. The formula for generating the formula is as follows:

[0104]

[0105] Where rx i ,ry i ,rz i remain unchanged. Indicates the coordinate system O at the end of the robot flange end To the robot base coordinate system O base The nth pose.

[0106] Please continue reading. Figures 1 to 7 As shown, in one embodiment of the present invention, step S500 further comprises:

[0107] Step S300 further comprises: obtaining the glue gun positioning module and Corresponding glue gun 2 in O base The coordinates below are given by the following formula:

[0108]

[0109] in This indicates the calibration result of glue gun 2 is O. gun Relative to O end Translation relationship;

[0110] Calculate vectors The angle θ between the y-axis and the x-axis is given by the following formula:

[0111]

[0112] in The y-axis unit vector. Then the transformed pose P is obtained by rotating θ around its own Z-axis. θ =[0,0,0,0,0,θ].

[0113] Please continue reading. Figures 1 to 7As shown, in one embodiment of the present invention, the calibration result of the glue gun includes the following steps:

[0114] Step a: Use the TCP tip-touch four-point method to calibrate the probe tip coordinate system.

[0115] Step b: Calibrate O using the TCP cusp-touch three-point method calobj and O base Rigid body transformation relations

[0116] Step c: Move glue gun 2 to O calobj The origin is set so that the glue outlet touches any point on the target with known coordinates. Let this point be A, and the translation relationship of point A relative to the target coordinate system is as follows: At this time, robot O end and O base The rigid body transformation relationship between them is The calibration result is O. gun Relative to O end Translation relationship for:

[0117]

[0118] In step S400, "simultaneously storing in the first point cloud cache area" further means:

[0119] Compare the angle between the current direction of the profilometer 1 and the forward direction of the glue gun 2, and calculate. and The included angle ρ is calculated using the following formula:

[0120] glue gun 2 in O base The coordinates below are

[0121]

[0122] Contouring instrument 1 in O base The coordinates below are

[0123]

[0124] Assuming there are i profilometers 1, ρ can be obtained i i = 1, 2, ...;

[0125] When ρ i When the angle is less than 90°, it is assumed that the current profiler 1 is ahead of the direction of travel, and the current first point cloud is placed in the front point cloud region of the first point cloud buffer region; when ρ i When the angle is greater than 90°, it is assumed that the current profiler 1 is behind the direction of travel, and the current first point cloud is placed into the back point cloud region of the first point cloud buffer region.

[0126] Please continue reading. Figures 1 to 7 As shown, in one embodiment of the present invention, step S500 further comprises:

[0127] Step S400 further comprises: retrieving contours from the contour buffer region to obtain a single contour point cloud, where n contours correspond to n poses, and for a single contour point cloud p i Transform to O using the following formula base Down:

[0128]

[0129] in and Representing O cam and O base First, obtain the point set of a single contour point cloud, then combine the n single contour point clouds. After merging, the first point cloud is obtained, and the first point cloud is stored in the first point cloud cache area; This represents the result of the profilometer 1 calibration.

[0130] Please continue reading. Figures 1 to 7 As shown, in one embodiment of the present invention, step S500 further comprises:

[0131] Step S501: Repeat steps S100 to S400. After the first loop is completed, the loop operation of S100 is to... Set as And wait for the second pose. Then Set as Then, steps S200, S300, and S400 are executed sequentially to obtain a new first point cloud. Steps S100 to S400 are repeated until the number of first point clouds reaches the number m specified by the user.

[0132] The formula for the size of m is as follows:

[0133]

[0134] Where d represents the distance between the nozzle and the linear structure light plane 8, v represents the speed of the motion mechanism, Δt represents the shortest time interval for the robot module to acquire the real-time flange pose of the robot, and [·] represents rounding up;

[0135] Step S502: For each first point cloud point set P base Perform the following pose transformation.

[0136] p base′ =P θ p base

[0137] Based on ROI, select p base′ The points in the diagram have the following ROI range:

[0138]

[0139] Where W and H are the width and height of the ROI specified by the user. and Represent the y-component of translation between two coordinates; p base′ The filtered point clouds are merged to obtain the second point cloud.

[0140] Please continue reading. Figures 1 to 7 As shown, in one embodiment of the present invention, step S600 further comprises: after obtaining the second point cloud, dividing the second point cloud along the y-direction according to the minimum detection length specified by the user to obtain several third point clouds; when the normal glue width specified by the user is w, cutting the third point cloud along the x-direction into three parts, the second part being wider than w; after performing planar fitting on the first and third part of the point cloud to obtain the glue coating plane, subtracting the glue coating plane from the second part of the point cloud to obtain the point cloud of the glue portion, then calculating the glue height and glue width, and then according to the threshold set by the user, when the calculated glue width is less than the minimum glue width threshold or If the glue height is less than the minimum glue height threshold, the current third point cloud is determined to be a glue break. If the glue width is greater than the maximum glue width threshold or the glue height is greater than the maximum glue height threshold, the current third point cloud is determined to be a glue pile. The same processing is performed on each third point cloud. If any third point cloud results in a glue pile or glue break, the first processing result is considered abnormal. If all third point cloud results are normal, the first processing result is considered normal. Finally, the first processing result is output to the display module. If the current first processing result is considered to be a glue break or glue pile, the start and end coordinates of glue gun 2 corresponding to the current second point cloud are recorded. and

[0141] Please continue reading. Figures 1 to 7 As shown, in one embodiment of the present invention, S700 further comprises: obtaining the number of first processing results obtained for all glue point clouds; calculating the first processing results in the state of glue breakage or glue accumulation; determining that glue breakage exists when the number of first processing results with continuous glue breakage reaches the glue breakage threshold given by the user; determining that glue accumulation exists when the number of first processing results with continuous glue accumulation reaches the glue accumulation threshold given by the user; if glue breakage or glue accumulation exists, obtaining the starting coordinates and ending coordinates of glue gun 2 corresponding to the abnormal part, calculating the Euclidean distance and accumulating it to obtain the length of the abnormal part, and returning it to the display module as the second processing result. Specific Implementation Example 1:

[0143] Preparatory steps before starting glue application and testing:

[0144] Let the coordinate system of the line structured light profiler 1 be O. cam The robot's base coordinate system is O.base The coordinate system of the robot flange end is O. end The coordinate system of the calibration plate is O. calobj The tool coordinate system at the probe tip is O tool The coordinate system of the glue gun nozzle 2 is O. gun The line structured light profiler 1 and the probe are installed at the end of the robot flange, and their rigid body transformation relationship with the end of the robot flange is fixed.

[0145] The homogeneous transformation matrix H and the pose P can be transformed into each other. The pose includes a translation component and a rotation component, where the translation component is represented as follows: The three coordinates are represented by three Euler angles [rx, ry, rz].

[0146] (1) Line structured light profilometer 1-hand eye calibration

[0147] First, hand-eye calibration of the line structured light profilometer 1 is performed. This step uses the method described in the applicant's previously filed invention patent, "A Hand-Eye Calibration Method Based on a Line Structured Light Profilometer" (Publication No.: CN120194646A). The calibration result is O. cam and O end Rigid body transformation relations Homogeneous matrix Convert to pose

[0148] (2) Calibration of glue gun 2

[0149] Then, perform calibration of the glue gun 2. Glue gun 2 calibration can be performed based on steps one and two of the hand-eye calibration method, using the same special target as in the patent. Step one of the method describes: using the TCP tip-touch four-point method to calibrate the probe tip coordinate system. Step 2: Calibrate O using the TCP cusp-touch three-point method calobj and O base Rigid body transformation relations Based on this, step three: move glue gun 2 to O. calobj The origin is set so that the glue outlet touches any point on the target with known coordinates. Let this point be A, and the translation relationship of point A relative to the target coordinate system is as follows: At this time, robot O end and O base The rigid body transformation relationship between them is The calibration result is O. gun Relative to O end Translation relationship for

[0150]

[0151] (3) Calculate the angle between glue gun 2 and line structure light profiler 1.

[0152] For Example 2, this step needs to be performed in addition to (1) and (2), while it is not required in Example 1.

[0153] According to the calibration results of glue gun 2, glue gun 2 is currently at O base The coordinates below are

[0154]

[0155] Similarly, based on the calibration results of the line structured light profiler 1, the current line structured light profiler 1 is at O base The coordinates below are

[0156]

[0157] Since the robot flange generally does not rotate during the adhesive application process, it is assumed that the angular portion of all poses remains unchanged. Furthermore, for planar adhesive application, the Z-coordinate remains constant or its change is negligible; therefore, the O-coordinate is used in the calculation. base It is simplified to a two-dimensional plane with XY coordinates.

[0158] Then from O gun Pointing to O cam The vector is

[0159]

[0160] Assuming the shortest time interval for obtaining the real-time flange pose of the robot from the robot module is Δt, the robot's initial pose... Known.

[0161] S100: This includes steps S101, S102, and S103, which are executed simultaneously. Assuming the system starts execution at time t0, during the time interval from t0 to t1 = t0 + Δt, S101 is executed: the contour acquisition module continuously acquires contour data in real time and stores it in the contour buffer area; S102 is executed: the robot starts running and obtains the robot's pose at time t1. Will Set as Set as Execute S103: Send a control signal to open glue gun 2.

[0162] S200: Assuming that the linear structured light profilometer 1 collects n contour data points within time Δt, the pose processing module... and Generate n poses on average between the two poses. The formula for generating the formula is as follows:

[0163]

[0164] Where rx i ,ry i ,rz i It remains unchanged.

[0165] S300: Obtained from the glue gun positioning module and Corresponding glue gun 2 in O base The coordinates below are given by the following formula:

[0166]

[0167] Calculate vectors The angle θ between the y-axis and the x-axis is given by the following formula:

[0168]

[0169] in The y-axis unit vector. Then the transformed pose P is obtained by rotating θ around its own Z-axis. θ =[0,0,0,0,0,θ].

[0170] S400: Extract the contour from the contour buffer area and generate a single contour point cloud. For n contours and n poses, for a single contour point cloud i, transform it to O according to the following formula. base Down:

[0171]

[0172] in and Representing O cam and O base The point set of the lower point cloud. Then, the point clouds of n single contours are... After merging, the first point cloud is obtained and stored in the first point cloud cache area.

[0173] S500: Execute steps S501 and S502 simultaneously.

[0174] Execute S501: Loop through S100 to S400 until the number of the first point cloud reaches the user-specified number m. The loop for S100 is as follows: after completing the first loop, ... Set as And wait for the second pose. Then Set as Then, S200, S300, and S400 are executed sequentially to obtain a new first point cloud, and the subsequent loops follow the same pattern.

[0175] The formula for m is as follows:

[0176]

[0177] Where d represents the distance between the nozzle and the linear structured light plane 8, v represents the speed of the motion mechanism, Δt represents the shortest time interval for the robot module to acquire the real-time flange pose of the robot, and [·] represents rounding up.

[0178] Execute S502: For each first point cloud point set p base Perform the following pose transformation.

[0179] p base′ =P θ p base

[0180] Based on ROI, select p base′ The points in the diagram have the following ROI range:

[0181]

[0182] Where W and H are the width and height of the ROI specified by the user. and This represents the y-component of translation between two coordinates. Let p... base′ The filtered point clouds are merged to obtain the second point cloud.

[0183] S600: After obtaining the second point cloud, the second point cloud is segmented along the y-direction according to the minimum detection length specified by the user, resulting in several third point clouds. Taking one of the third point clouds as an example, assuming the normal glue width specified by the user is w, the third point cloud is cut into three parts along the x-direction, as follows: Figure 6 As shown. The second portion is slightly wider than w, and can be taken as... After performing planar fitting on the first and third point clouds, the glue-coating plane is obtained. The glue-coating plane is then subtracted from the second point cloud to obtain the point cloud of the glue portion. Subsequently, the glue height and glue width are calculated to obtain the processing result.

[0184] Alternatively, the second point cloud can be used to extract a cross-section along the y-direction according to the minimum detection length specified by the user. After obtaining the cross-sectional profile, similar to the point cloud processing, the profile is divided into three parts: the two ends are the profile of the glue-coated plane cross-section, and the middle part is the profile of the glue cross-section. The glue cross-section profile is compared with the straight line of the glue-coated plane cross-section profile to obtain the glue height and glue width, thus obtaining the processing result.

[0185] This process is performed on each third point cloud. If any third point cloud shows either glue buildup or glue breakage, the first processing result is considered abnormal. If all third point cloud results are normal, the first processing result is considered normal. Finally, the first processing result is output to the display module.

[0186] S700: Repeat the above steps until all adhesive point clouds have obtained the first processing result. Calculate the first processing results in the state of adhesive breakage or adhesive buildup. If the number of consecutive adhesive breakage first processing results reaches the user-given adhesive breakage threshold, adhesive breakage is determined to exist; if the number of consecutive adhesive buildup first processing results reaches the user-given adhesive buildup threshold, adhesive buildup is determined to exist. If adhesive breakage or adhesive buildup exists, obtain the starting coordinates of the abnormal part, calculate the length of the abnormal part, and return it to the display module as the second processing result. Specific Implementation Example 2:

[0188] S110: Same as S100.

[0189] S210: Same as S200.

[0190] S310: Same as S300.

[0191] S410: Based on S400, after obtaining the first point cloud, calculate... and The included angle ρ is calculated using the following formula:

[0192]

[0193] Assuming there are i profilometers 1, ρ can be obtained i ,i=1,2,...

[0194] When ρ i When the angle is less than 90°, it is assumed that the current profiler 1 is ahead of the direction of travel, and the current first point cloud is placed in the front point cloud region of the first point cloud buffer region; when ρ i When the angle is greater than 90°, it is assumed that the current profiler 1 is behind the direction of travel, and the current first point cloud is placed into the back point cloud region of the first point cloud buffer region.

[0195] S510: Execute steps S511 and S512 simultaneously.

[0196] Execute S511: Same as S501

[0197] Execute S512: Same as S512. Note that this operation is performed on the front point cloud and the back point cloud simultaneously to obtain the second point cloud (front point cloud and back point cloud).

[0198] S610: After obtaining the second point cloud, the point clouds before and after the second point cloud are simultaneously segmented along the y-direction according to the minimum detection length specified by the user, resulting in several point clouds before and after the third point cloud. Taking the point clouds before and after the third point cloud within the same ROI as an example, the point cloud before the third point cloud is fitted with a plane to obtain the adhesive application plane. The point cloud of the adhesive portion is obtained by subtracting the adhesive application plane from the point cloud after the third point cloud. Then, the adhesive height and adhesive width are calculated to obtain the processing result.

[0199] Alternatively, the cross-section can be extracted along the y-direction of the second point cloud according to the minimum detection length specified by the user. The outline of the adhesive plane cross-section obtained from the point cloud before the third point cloud is compared with the outline of the adhesive interface obtained from the point cloud after the third point cloud to obtain the adhesive height and adhesive width and thus the processing result.

[0200] This process is performed on each third point cloud. If any third point cloud shows either glue buildup or glue breakage, the first processing result is considered abnormal. If all third point cloud results are normal, the first processing result is considered normal. Finally, the first processing result is output to the display module.

[0201] S710: Same as S700.

[0202] In summary, this invention not only meets the requirements for high-precision detection but also enables a rapid and automated detection process, thereby improving the accuracy and efficiency of adhesive coating quality inspection. Furthermore, the system should be adaptable to different types of adhesive coating conditions, including non-linear coating trajectories such as corners and curves, to comprehensively enhance the quality and reliability of adhesive coating inspection.

[0203] By automating the real-time monitoring process, the problems of large individual test results and the influence of individual work status in traditional manual sampling are overcome, thus improving the accuracy and stability of test results.

[0204] The line structured light profilometer 1 is used to scan directly during the glue application process, which avoids the problem of visual cycle time occupation in the traditional glue application process where the profilometer 1 is used to scan after the glue application is completed. This meets the customer's increasingly higher requirements for detection cycle time and accuracy.

[0205] By optimizing the layout design of the profilometer 1 and adopting a grid pattern or other reasonable layout, it is ensured that the adhesive application can be effectively detected in any trajectory direction, which effectively solves the pain point of detecting non-linear adhesive application trajectories such as corners and curves, and improves the comprehensiveness of the detection.

[0206] Using the data from the profilometer 1 before and after the glue gun 2, accurate glue height values ​​can be obtained even for uneven glue application surfaces.

[0207] The above description is only a preferred embodiment of the present invention and should not be construed as a limitation of this application. All equivalent changes and modifications made in accordance with the scope of the patent application of the present invention should be covered by the present invention.

Claims

1. A real-time adhesive application detection system, characterized in that: It includes a contour acquisition module, a robot module, a glue gun module, a first processing module, and a display module; The contour acquisition module includes several line structured light contour meters, a first control interface and a first network controller mounted on the line structured light contour meters. The line structured light contour meters are mounted on glue guns. The first network controller is used to control the acquisition timing of the contour meters in a multi-contour meter network. The first processing module triggers the contour meters to start capturing contours through the first control interface. The robot module includes a six-axis robot and a second control interface mounted on the six-axis robot. The first processing module controls the six-axis robot to move along a specified trajectory through the second control interface and obtains the real-time pose of the six-axis robot. The glue gun module includes a glue gun and a feeding mechanism. The feeding mechanism controls the switching on and off of the glue gun and the flow rate according to the control signal sent by the first processing module. The first processing module includes a CPU, a ROM, and a RAM. The CPU is used to perform operations, the RAM is used to cache contour data, point cloud data, and pose data, and the ROM is used to store programs and data. The display module includes a graphics card and a monitor, used to display result data and point clouds in real time.

2. The real-time adhesive application detection system according to claim 1, characterized in that: The structured light profiler is located on the rear side or both sides in the forward direction of the glue gun, or on the four sides in front, behind, left, and right of the glue gun.

3. A method of using the real-time adhesive coating detection system according to claim 1, characterized in that: The method of use includes the following steps: Step S100: Turn on the line structured light profiler to acquire profile data in real time, store it in the profile buffer area, acquire the robot's real-time pose from the robot module, send out control signals, and turn on the glue gun. Step S200: Acquire multiple contour data and robot real-time pose within the interval time Δt. The pose processing module differentiates the pose based on the number of acquired contours and maps the contours to the poses one by one. Step S300: In the glue gun positioning module, obtain the glue gun coordinates corresponding to multiple real-time poses, and obtain the glue gun forward direction; Step S400: In the first point cloud generation module, based on the hand-eye calibration results and the differentiated pose, the contour is generated into a point cloud and then unified into the robot coordinate system to generate the first point cloud and store it in the first point cloud cache area. Step S500: When the number of point clouds cached in the first point cloud cache area reaches a specified number, the first point cloud is rotated and cut, and the processed first point cloud is stitched together in the second point cloud generation module to generate the second point cloud. Step S600: In the second point cloud processing module, a glue shape ROI is generated based on the glue gun pose, the point cloud within the ROI is cut out, the glue shape is judged to be qualified, and the result is returned to the display module. Step S700: After the glue application is completed and all data is obtained, determine whether there is glue breakage or glue accumulation, and return the results to the display module.

4. The method of using the real-time adhesive coating detection system according to claim 3, characterized in that: Define the coordinate system of the line structured light profiler as O cam The robot's base coordinate system is O. base The coordinate system of the robot flange end is O. end The coordinate system of the calibration plate is O. calobj The tool coordinate system at the probe tip is O tool The coordinate system of the glue gun nozzle is O. gum The line structured light profiler and probe are installed at the end of the robot flange, and their rigid body transformation relationship with the end of the robot flange is fixed; step S100 further comprises: Step S101: Start executing from time t0, initial pose of the robot. During the time interval from t0 to t1 = t0 + Δt, the contour acquisition module continuously acquires contour data in real time and stores it in the contour cache area; This represents the coordinate system O at the end of the robot flange at time t0. end To the robot base coordinate system O base The pose; the pose includes a translation component and a rotation component, wherein the translation component is represented as... The three coordinates are represented by three Euler angles [rx, ry, rz]. Step S102: The robot starts running and obtains the robot's pose at time t1. Will Set as Set as This represents the coordinate system O at the end of the robot flange at time t1. end To the robot base coordinate system O base The position; Step S103: Send a control signal to turn on the glue gun.

5. The real-time adhesive coating detection system and its method of use according to claim 3, characterized in that: Step S200 further comprises: During a time interval Δt, the contour analyzer acquires n contour data points, and the pose processing module... and Generate an average of n poses between two poses. For i = 0, ..., n-1, the generation formula is as follows: Where rx i ,ry i ,rz i It remains unchanged.

6. The real-time adhesive coating detection system and its method of use according to claim 3, characterized in that: Step S300 further comprises: obtaining the glue gun positioning module and The corresponding glue gun is in O base The coordinates below are given by the following formula: in Indicates the glue gun calibration result O gun Relative to O end Translation relationship; Calculate vectors The angle θ between the y-axis and the x-axis is given by the following formula: in The y-axis unit vector. Then the transformed pose P is obtained by rotating θ around its own Z-axis. θ = [0,0,0,0,0,θ]; The calibration results of the glue gun include the following steps: Step a: Use the TCP tip-touch four-point method to calibrate the probe tip coordinate system. Step b: Calibrate O using the TCP cusp-touch three-point method calobj and O base Rigid body transformation relations Step c: Move the glue gun to O calobj The origin is set so that the glue outlet touches any point on the target with known coordinates. Let this point be A, and the translation relationship of point A relative to the target coordinate system is: At this time, robot O end and O base The rigid body transformation relationship between them is The calibration result is O. gun Relative to O end Translation relationship for 7. The real-time adhesive coating detection system and its method of use according to claim 1, characterized in that: In step S400, "and store in the first point cloud cache area" is further defined as follows: Compare the angle between the current profilometer direction and the glue gun's forward direction, and calculate... and The included angle ρ is calculated using the following formula: glue gun in O base The coordinates below are Contouring at O base The coordinates below are Assuming there are i profilometers, ρ can be obtained i i = 1, 2, ...; When ρ i When the angle is less than 90°, the current profilometer is considered to be ahead of the direction of travel, and the current first point cloud is placed in the front point cloud region of the first point cloud buffer region; when ρ i When the angle is greater than 90°, it is assumed that the current profilometer is behind the direction of travel, and the current first point cloud is placed in the back point cloud region of the first point cloud buffer region.

8. The real-time adhesive coating detection system and its method of use according to claim 3, characterized in that: Step S400 further comprises: retrieving contours from the contour buffer region to obtain a single contour point cloud, where n contours correspond to n poses, and for a single contour point cloud p i Transform to O using the following formula base Down: in and Representing O cam and O base First, obtain the point set of a single contour point cloud, then combine the n single contour point clouds. After merging, the first point cloud is obtained, and the first point cloud is stored in the first point cloud cache area; This represents the result of the profilometer calibration.

9. The real-time adhesive coating detection system and its method of use according to claim 3, characterized in that: Step S500 further comprises: Step S501: Repeat steps S100 to S400. After the first loop is completed, the loop operation of S100 is to... Set as And wait for the second pose. Then Set as Then, steps S200, S300, and S400 are executed sequentially to obtain a new first point cloud. Steps S100 to S400 are then executed. S400 keeps looping until the number of first point clouds reaches the user-specified number m; The formula for the size of m is as follows: Where d represents the distance between the nozzle and the linear structured light plane, v represents the speed of the motion mechanism, Δt represents the shortest time interval for the robot module to acquire the real-time flange pose of the robot, and [·] represents rounding up; Step S502: For each first point cloud point set p base Perform the following pose transformation. p base' =P θ p base Based on ROI, select p base' The points in the diagram have the following ROI range: Where W and H are the width and height of the ROI specified by the user. and Represent the y-component of translation between two coordinates; p base' The filtered point clouds are merged to obtain the second point cloud.

10. The real-time adhesive coating detection system and its method of use according to claim 3, characterized in that: Step S600 further comprises: after obtaining the second point cloud, dividing the second point cloud along the y-direction according to the minimum detection length specified by the user to obtain several third point clouds; when the normal glue width specified by the user is w, cutting the third point cloud along the x-direction into three parts, with the second part being wider than w; after performing planar fitting on the first and third point clouds to obtain the glue application plane, subtracting the glue application plane from the second point cloud to obtain the point cloud of the glue portion; then calculating the glue height and glue width; and then, according to the threshold set by the user, when the calculated glue width is less than the minimum glue width threshold or the glue height is less than the minimum glue width threshold, the third point cloud is cut into three parts along the x-direction, with the second part being wider than w; If the minimum glue height threshold is met, the current third point cloud is determined to be a glue break. If the glue width is greater than the maximum glue width threshold or the glue height is greater than the maximum glue height threshold, the current third point cloud is determined to be a glue pile. The same processing is performed on each third point cloud. If any third point cloud results in a glue pile or glue break, the first processing result is considered abnormal. If all third point cloud results are normal, the first processing result is considered normal, and finally the first processing result is output to the display module. If the current first processing result is considered to be a glue break or glue pile, the start and end coordinates of the glue gun corresponding to the current second point cloud are recorded. and Step S700 further comprises: until all glue-type point clouds have obtained the first processing result, obtaining the number of first processing results in the first processing result that are in the glue breakage or glue accumulation state; when the number of first processing results with continuous glue breakage reaches the glue breakage threshold given by the user, it is determined that there is glue breakage. When the number of first processing results of continuous glue stacking reaches the glue stacking threshold given by the user, it is determined that glue stacking exists; if there is glue breakage or glue stacking, the starting coordinates and ending coordinates of the glue gun corresponding to the abnormal part are obtained, the Euclidean distance is calculated and accumulated to obtain the length of the abnormal part, and returned to the display module as the second processing result.

Images