Visual labeling mechanism device and automatic labeling method thereof

Abstract

The invention discloses a visual labeling mechanism device and an automatic labeling method thereof, and the method comprises the steps: obtaining a workpiece surface image in motion in real time through a visual assembly, extracting a pose parameter of a workpiece in combination with an image processing algorithm, generating a pose data set containing a coordinate parameter and a deflection angle, and inputting a motion prediction model. Calculating a future movement track of the workpiece in combination with the speed parameter of the conveying line, and generating an adjustment instruction set of the labeling head; the movement track of the multi-axis mechanical arm is controlled based on the adjusting instruction set, the attaching pressure of the label contact face is adjusted through the pressure feedback module, and labeling operation of the curved surface of the workpiece is completed; a line scanning camera module is used for scanning the geometric position of the labeled label, a real-time detection result is generated by comparing the geometric position with a preset quality standard, a compensation signal is triggered if position deviation exists, and compensation operation is conducted through a compensation pressure head so as to ensure that the labeling quality meets the preset standard; and through cooperative control of visual identification, track prediction and compensation, the stability and the fitting precision of dynamic labeling are effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of labeling, and in particular to a visual labeling mechanism and an automatic labeling method thereof. Background Art

[0002] In the lithium battery manufacturing sector, with the rapid development of downstream industries such as new energy vehicles and energy storage systems, the automation level of cylindrical lithium battery production lines continues to increase, and the production line cycle continues to accelerate, with single-line production rates generally reaching above 120-150 PPM. During the battery shelling or sorting process, in order to achieve product traceability and information identification, labels with QR codes or barcodes must be accurately affixed to the battery surface. Labeling accuracy is directly related to the stability of subsequent processes such as scanning, assembly, and quality inspection. Therefore, the development of a vision-guided automatic labeling system that can adapt to high-speed conveying, precise bonding, and multi-category compatibility has become a key link in the intelligent manufacturing equipment of cylindrical batteries.

[0003] Most existing labeling technologies rely on photoelectric sensors or simple positioning devices to achieve synchronized label application, controlling the labeling head by setting a fixed trigger position. However, these solutions are primarily suitable for static or low-speed conveying environments. Labeling positioning is typically based on two-dimensional plane coordinates and lacks the ability to identify and compensate for workpiece posture (such as rotational deflection angle). Under high-speed dynamic conveying, cylindrical batteries are prone to jitter, bouncing, or deflection, causing significant offsets in the labeling position, leading to skewed or duplicate labeling, and affecting production line yield.

[0004] Therefore, the existing technology has significant deficiencies in dynamic posture compensation control and is difficult to meet the reliability requirements in high-speed and high-precision labeling scenarios. Summary of the Invention

[0005] The purpose of the present invention is to provide a visual labeling mechanism device and an automatic labeling method thereof to solve the above technical problems.

[0006] To achieve this object, the present invention adopts the following technical solutions: An automatic labeling method for a visual labeling mechanism device, comprising: The surface image of the moving workpiece is acquired in real time through the visual component, and the pose parameters of the workpiece are extracted by combining the image processing algorithm to generate a pose data set containing coordinate parameters and deflection angles. The pose data set is input into the motion prediction model, and the future motion trajectory of the workpiece is calculated in combination with the conveyor line speed parameters to generate an adjustment instruction set for the labeling head; Based on the adjustment instruction set, the motion trajectory of the multi-axis robotic arm is controlled, and the fitting pressure of the label contact surface is adjusted through the pressure feedback module to complete the labeling operation on the workpiece surface; The line scan camera module scans the geometric position of the labeled part and compares it with the preset quality standards to generate real-time detection results. If there is a position deviation, a compensation signal is triggered and compensation is performed through the compensation head.

[0007] Optionally, the method of acquiring a surface image of a moving workpiece in real time through a visual component, extracting the pose parameters of the workpiece in combination with an image processing algorithm, and generating a pose data set including coordinate parameters and deflection angles comprises the following steps: When the workpiece enters the visual capture area of ​​the visual component, the shooting is triggered, and multi-view images of the front, side and top of the workpiece surface are collected according to the preset time sequence; Adopting the adaptive polarization light source module to adjust the polarization direction of the incident light in real time, eliminating the interference of metal reflection on the workpiece surface and generating original image data; The original image data is aligned and fused using an image stitching algorithm based on the optical flow method to generate a ring projection view with complete geometric features.

[0008] Optionally, the alignment processing is performed on the original image data, and an annular projection view with complete geometric features is generated by fusing the image data using an optical flow-based image stitching algorithm, and then the following steps are further included: Loading a pre-installed feature extraction module on the annular projection view, synchronously identifying the edge contour of the workpiece and positioning the reference mark, and outputting a geometric parameter matrix including the coordinates of the contour vertices and the reference angle; The geometric parameter matrix is ​​input into the pose calculation module, and the spatial coordinates (X, Y, Z), the deflection angle θ around the axis, and the radial offset Δr of the workpiece are calculated in combination with the binocular vision coordinate system conversion relationship to generate a standardized pose data set and transmit it to the motion controller.

[0009] Optionally, the pose data set is input into a motion prediction model, and the future motion trajectory of the workpiece is calculated in combination with the conveyor line speed parameters to generate an adjustment instruction set for the labeling head, including the following steps: The input pose data set is filtered, and high-frequency vibration noise is eliminated through the sliding window mean filtering method to generate a denoised pose time series sequence; A linear motion prediction equation is constructed based on the conveyor line speed parameters, and a friction coefficient compensation factor μ between the workpiece and the conveyor belt is introduced into the equation to calculate the predicted trajectory within the future Δt time window; The predicted trajectory is corrected through the Kalman filter, and the covariance matrix Q value is iteratively updated using the first three historical pose data to output the corrected trajectory with a confidence interval. The corrected trajectory is converted into labeling head motion control parameters, and combined with the acceleration limit of the robot arm joint to generate an adjustment instruction set with a timestamp. The adjustment instruction set includes the displacement Δx, the compensation angle θ and the dynamic response coefficient k.

[0010] Optionally, the motion trajectory of the multi-axis robotic arm is controlled based on the adjustment instruction set, and the fitting pressure of the label contact surface is adjusted by the pressure feedback module to complete the labeling operation on the workpiece curved surface. Specifically, the following steps are performed: Receive the displacement Δx, compensation angle θ and dynamic response coefficient k in the adjustment instruction set, generate a corresponding multi-axis interpolation path through the motion controller, and drive the labeling head to a pre-bonding point that maintains a preset safety distance from the target labeling position; After completing the pre-bonding positioning, the labeling head enters the slow approach mode. The built-in distance sensor module monitors the distance between the labeling head and the workpiece surface. When the distance is less than the set threshold d0, the pressure control unit is triggered to slowly apply the bonding force at the preset target pressure. After the labeling is completed, the labeling head is controlled to move at a constant speed along the labeling track, and at the end, a slight pressure is applied to retract the label to automatically detach the end of the label, completing the labeling.

[0011] Optionally, the method of scanning the geometric position of the labeled object by a line scan camera module and comparing the result with a preset quality standard to generate a real-time detection result, triggering a compensation signal if there is a position deviation, and performing a compensation operation by a compensation head, specifically includes: After the label application is completed, the line scan camera module is controlled to synchronously start scanning along the axial direction of the workpiece, to collect a detection image of the label application area, and to divide the detection image into multiple detection segments with a fixed step size; Preprocess the divided detection segments to identify the edge contour of the label, and compare it with the standard template through the feature matching algorithm to extract the label relative offset Δd and deflection angle Δθ; Based on the comparison results, it is determined whether the labeling accuracy meets the preset quality standards. If either Δd or Δθ exceeds the corresponding allowable threshold, the NG signal is automatically triggered and compensation operation is performed; Call the compensation head action module, perform vector differential calculation on the current position coordinates and the detection offset Δd, generate compensation motion instructions, and control the compensation head to perform a directional secondary pressing operation on the offset area with a constant pressure value; After the compensation operation is completed, the quality re-inspection sub-process is called to upload the defect image and detection parameters to the database for archiving.

[0012] Optionally, the comparison generates a real-time detection result by comparing with a preset quality standard, triggers a compensation signal if there is a position deviation, and performs a compensation operation through a compensation pressure head, and then further includes: Integrate historical labeling data and equipment operating parameters to build a process optimization model, and output optimized process parameters through the process optimization model.

[0013] The present invention also provides a visual labeling mechanism device, an automatic labeling method using the visual labeling mechanism device described above, wherein the visual labeling mechanism device includes a multi-axis robotic arm, and a conveyor belt for conveying workpieces is provided below the multi-axis robotic arm; The driving end of the multi-axis robotic arm is provided with a mounting plate, on which a pressing assembly, a line scan camera and a compensation assembly are sequentially arranged along the conveying direction of the conveyor belt. A label belt assembly is arranged opposite to the pressing assembly, and the conveying direction of the label belt assembly is perpendicular to the conveying direction of the conveyor belt.

[0014] Optionally, the pressing assembly includes a first linear module, a driving end of the first linear module is provided with a connecting block, the connecting block is provided with a rotating driving member, the driving end of the rotating driving member is provided with a labeling head, and the labeling head includes a plurality of press head units arranged in parallel; The compensation component includes a compensation pressure head, which is one of the plurality of pressure head units.

[0015] Optionally, the label tape assembly includes a first reel and a second reel that are spaced apart, and a conveying passage for conveying the label tape is formed between the first reel and the second reel; A plurality of tensioning wheels are provided along the conveying path, wherein a driving roller is provided at a preset position of the conveying path, and the driving roller is connected to a driving motor for driving the driving roller to rotate; A passive roller is provided on one side of the active roller, and a third linear module is provided on one side of the passive roller. The third linear module is used to drive the passive roller to move toward or away from the active roller.

[0016] Compared with the prior art, the present invention has the following beneficial effects: first, the surface image of the workpiece is collected in real time by a visual component arranged in front of the conveyor line, and its three-dimensional posture parameters are extracted using an image processing algorithm to generate a posture data set including spatial coordinates and deflection angles; the data set is input into a motion prediction model, and combined with the real-time speed of the conveyor line, the predicted motion trajectory of the workpiece in the short time domain is calculated, and an adjustment instruction set for multi-axis robotic arm control is generated; further, the multi-axis robotic arm on which the labeling head is installed is controlled to perform precise following motion according to the instructions, and at the same time, the fitting pressure output by the labeling head is adjusted in combination with the pressure feedback module to achieve positional adaptive fitting of the label to the curved surface of the workpiece; finally, the line scan camera module arranged at the end of the labeling head is used to perform geometric position scanning and detection of the labeling result. If a labeling deviation is found, the compensation pressure head is automatically triggered to perform a secondary pressing operation to ensure that the labeling quality meets the preset standard; this method effectively improves the stability and fitting accuracy of dynamic labeling through the coordinated control of visual recognition, trajectory prediction and compensation. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0018] The structures, proportions, sizes, etc. depicted in the drawings of this specification are only used to match the contents disclosed in the specification so as to facilitate understanding and reading by persons familiar with this technology. They are not intended to limit the conditions under which the present invention can be implemented and therefore have no substantive technical significance. Any structural modifications, changes in proportional relationships, or adjustments in size should still fall within the scope of the technical contents disclosed in the present invention without affecting the effects and objectives that can be achieved by the present invention.

[0019] Figure 1 Schematic diagram of the process of the automatic labeling method of the visual labeling mechanism device of the first embodiment; Figure 2 This is a structural diagram of the labeling part of the visual labeling mechanism device of the second embodiment; Figure 3 This is a structural diagram of the label belt assembly of the visual labeling mechanism device of the second embodiment.

[0020] Illustrations: multi-axis robotic arm 10, conveyor belt 20, mounting plate 11, pressing assembly 30, line scan camera 40, compensation assembly 50, label tape assembly 60, first linear module 31, rotary drive member 32, labeling head 33, compensation pressure head 51, first reel 61, second reel 62, tensioning wheel 63, active roller 64, drive motor 65, passive roller 66, third linear module 67. DETAILED DESCRIPTION

[0021] In order to make the purpose, features, and advantages of the present invention more obvious and easy to understand, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described below are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0022] In the description of the present invention, it should be understood that the terms "upper," "lower," "top," "bottom," "inner," "outer," etc., indicating orientations or positional relationships, are based on the orientations or positional relationships shown in the accompanying drawings and are intended solely to facilitate the description of the present invention and simplify the description. They are not intended to indicate or imply that the devices or elements referred to must have a specific orientation, be constructed, or operate in a specific orientation. Therefore, they should not be construed as limitations on the present invention. It should be noted that when a component is considered to be "connected" to another component, it may be directly connected to the other component or there may be a centrally located component.

[0023] The technical solution of the present invention will be further described below with reference to the accompanying drawings and through specific implementation methods.

[0024] Example 1: Combine Figure 1 As shown, an embodiment of the present invention provides an automatic labeling method of a visual labeling mechanism device, comprising: S1, uses the visual component to obtain the surface image of the moving workpiece in real time, combines the image processing algorithm to extract the posture parameters of the workpiece, and generates a posture data set containing coordinate parameters and deflection angles.

[0025] In step S2, the pose data set is input into the motion prediction model. Combined with the conveyor line speed parameters, the future trajectory of the workpiece is calculated, generating an adjustment instruction set for the labeling head 33. By introducing speed compensation and filtering correction mechanisms, trajectory deviations caused by factors such as conveyor belt 20 fluctuations and workpiece slippage can be reduced, thereby enabling advance control of the labeling path and timing, improving the stability and response speed of the labeling operation.

[0026] S3, based on the adjustment instruction set, the motion trajectory of the multi-axis robot arm 10 is controlled, and the fitting pressure of the label contact surface is adjusted through the pressure feedback module to complete the labeling operation on the workpiece curved surface.

[0027] The multi-axis robotic arm 10 is driven to perform labeling along a predicted trajectory, and the built-in pressure feedback module adjusts the pressure between the label and the workpiece surface to accommodate varying curvatures or materials. This process emphasizes dynamic alignment of the label path with the workpiece surface normal, while incorporating pressure feedback for adaptive adjustment. This helps improve label adhesion integrity and bond strength, while reducing the likelihood of defects such as bubbles and wrinkles.

[0028] S4, scan the geometric position of the labeled object through the line scan camera 40 module, compare it with the preset quality standard to generate real-time detection results, and trigger a compensation signal if there is a position deviation, and perform compensation operation through the compensation head 51.

[0029] The setting provides a fast deviation correction mechanism without affecting the main labeling process, which can effectively improve labeling consistency and product appearance quality. It is especially suitable for labeling tasks of workpieces with complex shapes or large curved surface changes.

[0030] S5, integrates historical labeling data and equipment operating parameters to build a process optimization model, and outputs optimized process parameters through the process optimization model.

[0031] By continuously collecting and integrating historical labeling data and equipment operating parameters, an optimization model for the labeling process is constructed, and then updated process parameters (such as bonding pressure, motion path fine-tuning amount, compensation strategy, etc.) are output in the subsequent labeling process.

[0032] The working principle of the present invention is as follows: first, the surface image of the workpiece is collected in real time by a visual component arranged in front of the conveyor line, and its three-dimensional posture parameters are extracted using an image processing algorithm to generate a posture data set including spatial coordinates and deflection angles; the data set is input into a motion prediction model, and combined with the real-time speed of the conveyor line, the predicted motion trajectory of the workpiece in the short time domain is calculated, and an adjustment instruction set for controlling the multi-axis robot 10 is generated; further, the multi-axis robot 10 installed with the labeling head 33 is controlled to perform precise following movement according to the instruction, and at the same time, the fitting pressure output by the labeling head 33 is adjusted in combination with the pressure feedback module to achieve positional adaptive fitting of the label to the workpiece curved surface; finally, the labeling result is subjected to geometric position scanning and detection by a line scan camera 40 module arranged at the end of the labeling head 33. If a labeling deviation is found, the compensation pressure head 51 is automatically triggered to perform a secondary pressing operation to ensure that the labeling quality meets the preset standard; this method effectively improves the stability and fitting accuracy of dynamic labeling through the coordinated control of visual recognition, trajectory prediction and compensation.

[0033] In this embodiment, it is specifically explained that step S1 includes the following steps: S11, when the workpiece enters the visual capture area of ​​the visual component, the shooting is triggered, and multi-view images of the front, side and top of the workpiece surface are collected according to the preset time sequence.

[0034] When a workpiece enters the vision capture area of ​​the vision module, the system immediately triggers a capture action and captures multi-view images from the front, side, and top of the workpiece according to the preset image acquisition sequence. This is done to fully capture the complete surface information of the workpiece in its current conveying state, especially any areas with potential obstructions or curved structures.

[0035] S12 uses an adaptive polarization light source module to adjust the polarization direction of the incident light in real time, eliminate the interference of metal reflections on the workpiece surface, and generate original image data; To address image interference caused by metallic reflections, highlights, or material variations on the workpiece surface, the system employs an adaptive polarization light source module, adjusting the polarization direction of the incident light in real time during capture. This technology effectively suppresses reflective noise through polarization filtering, enhancing image edge and texture clarity, and ensuring high contrast and geometric legibility in the original image data.

[0036] S13, performing alignment processing on the original image data, and generating a ring projection view with complete geometric features through an image stitching algorithm based on the optical flow method.

[0037] To address the spatial misalignment between multi-view images, the system incorporates an optical flow-based image stitching algorithm to align and fuse the raw image data. Optical flow enables continuous stitching of images from different viewpoints based on pixel motion trends between images, generating a circular projection view with complete geometric representation. This projection view effectively "unfolds" the workpiece surface onto a two-dimensional plane, facilitating subsequent edge analysis, contour extraction, and labeling positioning, serving as the geometric foundation for high-precision labeling.

[0038] S14, loading a pre-installed feature extraction module on the annular projection view, synchronously identifying the edge contour of the workpiece and positioning the reference mark, and outputting a geometric parameter matrix containing the coordinates of the contour vertices and the reference angle.

[0039] After acquiring the annular projection view, the system loads a pre-set image feature extraction module and performs simultaneous recognition of the workpiece's contours, edge corners, and fiducial markers (such as locating holes and structural corners). By extracting these key features, the system outputs a geometric parameter matrix reflecting the workpiece's external structure and labeling benchmarks, including contour vertex coordinates, normal directions, and local angle information.

[0040] S15, the geometric parameter matrix is ​​input into the pose solving module, and the spatial coordinates (X, Y, Z), the deflection angle θ around the axis and the radial offset Δr of the workpiece are calculated by combining the binocular vision coordinate system transformation relationship, and a standardized pose data set is generated and transmitted to the motion controller.

[0041] The system inputs the geometric parameter matrix into the pose calculation module. Combined with the binocular vision system's coordinate transformation model, it calculates the workpiece's spatial position (X, Y, Z) in the conveyor line coordinate system, the deflection angle θ about the main axis, and the radial offset Δr from the labeling reference plane. The resulting pose dataset is in a standardized format, highly versatile, and real-time, and can be directly transmitted to the motion controller for labeling path planning and mechanism drive control.

[0042] In this embodiment, it is specifically explained that step S2 includes the following steps: S21, filtering the input posture data set, eliminating high-frequency vibration noise through sliding window mean filtering method, and generating a noise-reduced posture time series sequence.

[0043] Due to mechanical interference such as vibration and vibration of the conveyor belt 20 in actual production line environments, the pose data generated by the workpiece during movement is often accompanied by high-frequency noise, which can affect labeling accuracy if used directly. Therefore, this step first filters the input pose data set and uses a sliding window mean filter to perform a local average operation on each dimension of data to eliminate short-term disturbance signals. This method is simple to implement, has strong real-time performance, and can effectively smooth time series data, providing a more stable basic input for subsequent trajectory prediction.

[0044] S22, construct a linear motion prediction equation based on the conveyor line speed parameters, introduce the friction coefficient compensation factor μ (μ = 0.2-0.5 adjustable) between the workpiece and the conveyor belt 20 into the equation, and calculate the predicted trajectory within the future Δt time window.

[0045] After obtaining the de-noised pose time series, the system constructs a linear motion prediction equation based on the conveyor line's current speed parameters to estimate the workpiece's displacement trend within a future time window of Δt. Considering the friction and slip between the workpiece and the conveyor belt 20 during actual conveyance, a friction coefficient compensation factor μ (μ=0.2–0.5) is introduced into the equation to improve the prediction's accuracy. This correction factor compensates for actual displacement offsets.

[0046] S23, correct the error of the predicted trajectory through the Kalman filter, iteratively update the covariance matrix Q value using the first three historical pose data, and output the corrected trajectory with a confidence interval.

[0047] Considering that linear prediction may still deviate in real-world scenarios due to factors such as system errors and measurement delays, this step introduces a Kalman filter to correct the predicted trajectory. The system uses the first three historical pose data as the observation sequence and iteratively updates the Kalman covariance matrix Q, thereby dynamically adjusting the filter's response sensitivity.

[0048] S24, converting the corrected trajectory into motion control parameters of the labeling head 33, and generating an adjustment instruction set with a timestamp in combination with the acceleration limit of the robot arm joint. The adjustment instruction set includes the displacement Δx, the compensation angle θ and the dynamic response coefficient k.

[0049] After obtaining the corrected trajectory information, the system must convert it into a set of actual control instructions for the labeling head 33. This step calculates and encodes control parameters such as displacement Δx, compensation angle θ, and dynamic response coefficient k based on the acceleration limits and dynamic response characteristics of the robot arm's joint structure, generating a set of adjustment instructions with precise timestamps. These instructions are used to drive the labeling head 33 to align with the target labeling position along the workpiece's motion path in real time, achieving synchronous compensation and flexible matching.

[0050] In this embodiment, it is specifically explained that step S3 includes the following steps: S31, receiving the displacement Δx, compensation angle θ and dynamic response coefficient k in the adjustment instruction set, generating a corresponding multi-axis interpolation path through the motion controller, and driving the labeling head 33 to a pre-bonding point that maintains a preset safety distance from the target labeling position.

[0051] The multi-axis interpolation path is solved in real time based on the inverse kinematics model of the robot arm, and a timestamp parameter is introduced to achieve synchronous execution of the path over time; when executing path planning, a surface fitting compensation algorithm is introduced to dynamically correct the end posture of the labeling head according to the curvature change at the historical fitting position, ensuring that the labeling head 33 is consistent with the normal direction of its local surface before approaching the workpiece, avoiding relative slippage in the initial stage of fitting.

[0052] S32, after completing the pre-bonding positioning, the labeling head 33 enters the slow approach mode, and monitors the distance between the labeling head 33 and the workpiece surface through the built-in distance sensor module. When the distance is less than the set threshold d0, the pressure control unit is triggered to slowly apply the bonding force at the preset target pressure.

[0053] The pressure control unit is composed of a miniature strain force sensor embedded in the nozzle of the labeling head 33 and connected to the real-time controller to achieve force value refresh and feedback correction during the bonding process; at the same time, the bonding speed is fine-tuned in combination with the dynamic response coefficient k to ensure that slow start control of speed-pressure is applied in areas where the workpiece surface changes drastically to prevent the label from warping or deformation.

[0054] S33, after the labeling is completed, the labeling head 33 is controlled to move along the labeling track at a constant speed, and a slight pressure is withdrawn at the end to achieve automatic detachment of the label end, thereby completing the labeling.

[0055] During this process, the control system reads the error ΔF between the pressure sensor feedback value and the expected value in real time. If ΔF exceeds ±10%, the end path posture is automatically corrected to maintain a constant pressure bonding state. At the same time, the angular deflection θ of the labeling head 33 is compensated by the synchronous rotation of the label roll drive shaft to ensure that the label is evenly attached along the circumferential direction of the workpiece. After bonding is completed, the labeling head 33 automatically retracts to the standby position to prepare for the next labeling cycle.

[0056] In this embodiment, it is specifically explained that step S4 includes the following steps: S41, after the label pasting action is completed, the line scan camera 40 module is controlled to start scanning synchronously along the axial direction of the workpiece, collect the detection image of the label pasting area, and divide the detection image into multiple detection segments with a fixed step size.

[0057] The line scan camera 40 adopts a pixel array structure and combines the linear array motion encoder data for synchronous sampling to ensure that the image is consistent with the actual fitting trajectory; during the image acquisition process, the dynamic exposure adjustment module is called to automatically optimize the exposure time according to the current workpiece surface brightness to enhance contrast and reduce reflection artifact interference.

[0058] S42 , pre-processing the divided detection segments, identifying the edge contour of the tag, and comparing it with the standard template through a feature matching algorithm to extract the tag relative offset Δd and the deflection angle Δθ.

[0059] The standard template is a grayscale reference model constructed based on qualified attached samples during the initial labeling phase and contains a preset tolerance range (e.g., Δd threshold < 0.3mm). When the test results contain multiple characteristic error points, the average offset value is calculated and a test report is generated, including offset information for the label's starting edge, end edge, and center area. S43, based on the comparison results, determine whether the labeling accuracy meets the preset quality standard. If either Δd or Δθ exceeds the corresponding allowable threshold, an NG signal is automatically triggered and compensation is performed; At the same time, a defect type mark is generated and associated with the current product barcode as a basis for subsequent production line rejection or rework. This step also activates the compensation control process within the controller and suspends the labeling instructions for the next cycle to prevent batch mislabeling.

[0060] S44, calling the action module of the compensation pressure head 51, performing vector differential calculation between the current position coordinates and the detected offset Δd, generating a compensation movement instruction, and controlling the compensation pressure head 51 to perform a directional secondary pressing operation on the offset area with a constant pressure value; The internal flexible limiting structure of the compensation pressure head 51 can keep the fitting posture stable during small position adjustments to avoid displacement of the entire label; the fitting pressure is adjusted to 15~18N through a proportional valve closed loop, and the compensation duration is limited to no more than 0.5s to prevent material fatigue or secondary displacement.

[0061] S45, after the compensation operation is completed, the quality re-inspection sub-process is called to upload the defect image and detection parameters to the database for archiving.

[0062] During the re-inspection process, only a local window detection area of ​​5mm×20mm is set for the label offset, which improves detection efficiency and avoids production line delays. At the same time, the system synchronizes the re-inspection results to the visual recognition module for dynamic correction of the labeling path and offset tolerance parameters for the next cycle, realizing closed-loop process optimization.

[0063] Example 2: Combine Figure 2 and Figure 3 As shown, the present invention also provides a visual labeling mechanism device, which adopts the automatic labeling method of the visual labeling mechanism device as in Example 1, and the visual labeling mechanism device includes a multi-axis robot arm 10, and a conveyor belt 20 for conveying workpieces is arranged below the multi-axis robot arm 10; a mounting plate 11 is provided at the driving end of the multi-axis robot arm 10, and a pressing assembly 30, a line scan camera 40 and a compensation assembly 50 are sequentially arranged on the mounting plate 11 along the conveying direction of the conveyor belt 20, and a label belt assembly 60 is arranged opposite to the pressing assembly 30, and the conveying direction of the label belt assembly 60 is arranged perpendicular to the conveying direction of the conveyor belt 20.

[0064] It should be noted that during the operation of the visual labeling mechanism device described in this embodiment, the conveyor belt 20 is responsible for stably conveying the workpiece to be labeled to the labeling area; when the workpiece enters the operating range of the multi-axis robot arm 10, the system starts the visual guidance process, and the visual component installed on the mounting plate 11 collects the workpiece image and calculates its posture information. Based on the acquired posture data and conveying speed parameters, the multi-axis robot arm 10 performs path planning and dynamic adjustment to quickly and accurately guide the labeling head 33 installed at its end to the preset fitting position. After the labeling action is completed, the line scan camera 40 arranged on the mounting plate 11 immediately scans the labeling area and transmits the detected image to the image processing unit to evaluate the fitting quality. If a position deviation is detected, the compensation component 50 set at the tail end of the mounting plate 11 immediately responds to the compensation signal and starts the selected pressure head unit to complete the local re-pressing operation.

[0065] In this embodiment, the pressing assembly 30 includes a first linear module 31, the driving end of the first linear module 31 is provided with a connecting block, the connecting block is provided with a rotating driving member 32, the driving end of the rotating driving member 32 is provided with a labeling head 33, and the labeling head 33 includes a plurality of pressure head units arranged in parallel; the compensation assembly 50 includes a compensation pressure head 51, and the compensation pressure head 51 is set as one of the multiple pressure head units.

[0066] In the early stage of labeling by the labeling head 33, the rotary drive member 32 adjusts the pressure head posture according to the target angle so that it maintains normal contact with the workpiece surface as much as possible, thereby improving the uniformity of label contact. The multiple pressure head units in the labeling head 33 act synchronously on the label area, which not only ensures the uniform pressure distribution of the overall label paste, but also improves the adaptability to non-planar surfaces. After labeling is completed, if the line scan camera 40 detects that there is an attachment offset in a certain area, the compensation component 50 located at the rear end of the mounting plate 11 will be activated, and the corresponding pressure head unit will be independently activated as the compensation pressure head 51, and will perform secondary directional pressing on the target area under the guidance of the controller, ensuring that the labeling quality meets the preset standards while avoiding repeated whole-sticking, thereby improving the system response efficiency and flexible processing capabilities.

[0067] In this embodiment, it is specifically described that the label tape assembly 60 includes a first reel 61 and a second reel 62 arranged at an interval, and a conveying path for conveying the label tape is formed between the first reel 61 and the second reel 62; a plurality of tensioning rollers 63 are arranged along the conveying path, wherein an active roller 64 is provided at a preset position of the conveying path, and the active roller 64 is connected to a driving motor 65 for driving it to rotate; a passive roller 66 is provided on one side of the active roller 64, and a third linear module 67 is provided on one side of the passive roller 66, and the third linear module 67 is used to drive the passive roller 66 to move toward or away from the active roller 64.

[0068] In the clamping section formed by the active roller 64 and the passive roller 66, the active roller 64 is driven by a drive motor 65 to continuously rotate, providing a constant feed force. Driven by the third linear module 67, the passive roller 66 dynamically adjusts its position based on real-time tension conditions, achieving closed-loop control of the tape path tension. If the system detects tension fluctuations (such as a loose or stuck tape), the third linear module 67 immediately responds by fine-tuning the position of the passive roller 66 to compensate for the tension change, maintaining constant tape tension.

[0069] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments can still be modified, or some of the technical features thereof can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An automatic labeling method for a visual labeling mechanism, characterized in that: include: The surface image of the moving workpiece is acquired in real time through the visual component, and the pose parameters of the workpiece are extracted by combining the image processing algorithm to generate a pose data set containing coordinate parameters and deflection angles. The pose data set is input into the motion prediction model, and the future motion trajectory of the workpiece is calculated in combination with the conveyor line speed parameters to generate an adjustment instruction set for the labeling head; Based on the adjustment instruction set, the motion trajectory of the multi-axis robotic arm is controlled, and the fitting pressure of the label contact surface is adjusted through the pressure feedback module to complete the labeling operation on the workpiece surface; The line scan camera module scans the geometric position of the labeled part and compares it with the preset quality standards to generate real-time detection results. If there is a position deviation, a compensation signal is triggered and compensation is performed through the compensation head.

2. The automatic labeling method of the visual labeling mechanism according to claim 1, characterized in that: The method of acquiring a surface image of a moving workpiece in real time through a visual component, extracting the workpiece's posture parameters in combination with an image processing algorithm, and generating a posture data set including coordinate parameters and deflection angles includes the following steps: When the workpiece enters the visual capture area of ​​the visual component, the shooting is triggered, and multi-view images of the front, side and top of the workpiece surface are collected according to the preset time sequence; Adopting the adaptive polarization light source module to adjust the polarization direction of the incident light in real time, eliminating the interference of metal reflection on the workpiece surface and generating original image data; The original image data is aligned and fused using an image stitching algorithm based on the optical flow method to generate a ring projection view with complete geometric features.

3. The automatic labeling method of the visual labeling mechanism according to claim 2, characterized in that: The method further includes performing alignment processing on the original image data and fusing the image data using an optical flow-based image stitching algorithm to generate a ring-shaped projection view with complete geometric features. Loading a pre-installed feature extraction module on the annular projection view, synchronously identifying the edge contour of the workpiece and positioning the reference mark, and outputting a geometric parameter matrix including the coordinates of the contour vertices and the reference angle; The geometric parameter matrix is ​​input into the pose calculation module, and the spatial coordinates (X, Y, Z), the deflection angle θ around the axis, and the radial offset Δr of the workpiece are calculated in combination with the binocular vision coordinate system conversion relationship to generate a standardized pose data set and transmit it to the motion controller.

4. The automatic labeling method of the visual labeling mechanism according to claim 1, characterized in that: The pose data set is input into the motion prediction model, and the future motion trajectory of the workpiece is calculated in combination with the conveyor line speed parameters to generate an adjustment instruction set for the labeling head, including the following steps: The input pose data set is filtered, and high-frequency vibration noise is eliminated through the sliding window mean filtering method to generate a denoised pose time series sequence; A linear motion prediction equation is constructed based on the conveyor line speed parameters, and a friction coefficient compensation factor μ between the workpiece and the conveyor belt is introduced into the equation to calculate the predicted trajectory within the future Δt time window; The predicted trajectory is corrected through the Kalman filter, and the covariance matrix Q value is iteratively updated using the first three historical pose data to output the corrected trajectory with a confidence interval. The corrected trajectory is converted into labeling head motion control parameters, and combined with the acceleration limit of the robot arm joint to generate an adjustment instruction set with a timestamp. The adjustment instruction set includes the displacement Δx, the compensation angle θ and the dynamic response coefficient k.

5. The automatic labeling method of the visual labeling mechanism according to claim 1, characterized in that: Based on the adjustment instruction set, the motion trajectory of the multi-axis robot arm is controlled, and the fitting pressure of the label contact surface is adjusted through the pressure feedback module to complete the labeling operation on the workpiece curved surface. The specific steps are as follows: Receive the displacement Δx, compensation angle θ and dynamic response coefficient k in the adjustment instruction set, generate a corresponding multi-axis interpolation path through the motion controller, and drive the labeling head to a pre-bonding point that maintains a preset safety distance from the target labeling position; After completing the pre-bonding positioning, the labeling head enters the slow approach mode and monitors the distance between the labeling head and the workpiece surface through the built-in distance sensor module. When the distance is less than the set threshold d0, the pressure control unit is triggered to slowly apply the bonding force at the preset target pressure. After the labeling is completed, the labeling head is controlled to move at a constant speed along the labeling track, and at the end, a slight pressure is applied to retract the label to automatically detach the end of the label, completing the labeling.

6. The automatic labeling method of the visual labeling mechanism according to claim 1, characterized in that: The line scan camera module scans the geometric position of the label, compares it with the preset quality standard to generate real-time detection results, and triggers a compensation signal if there is a position deviation, and performs compensation operations through the compensation head, specifically including: After the label application is completed, the line scan camera module is controlled to synchronously start scanning along the axial direction of the workpiece to collect a detection image of the label application area, and the detection image is divided into multiple detection segments with a fixed step size; Preprocess the divided detection segments to identify the edge contour of the label, and compare it with the standard template through the feature matching algorithm to extract the label relative offset Δd and deflection angle Δθ; Based on the comparison results, it is determined whether the labeling accuracy meets the preset quality standards. If either Δd or Δθ exceeds the corresponding allowable threshold, the NG signal is automatically triggered and compensation operation is performed; Call the compensation pressure head action module, perform vector differential calculation on the current position coordinates and the detection offset Δd, generate compensation movement instructions, and control the compensation pressure head to perform a directional secondary pressing operation on the offset area with a constant pressure value; After the compensation operation is completed, the quality re-inspection sub-process is called to upload the defect image and detection parameters to the database for archiving.

7. The automatic labeling method of the visual labeling mechanism according to claim 1, characterized in that: The comparison generates a real-time detection result by comparing the preset quality standard, triggers a compensation signal if there is a position deviation, and performs a compensation operation through the compensation head, and then further includes: Integrate historical labeling data and equipment operating parameters to build a process optimization model, and output optimized process parameters through the process optimization model.

8. A visual labeling mechanism, characterized in that: An automatic labeling method using a visual labeling mechanism according to any one of claims 1 to 7, wherein the visual labeling mechanism comprises a multi-axis robotic arm, and a conveyor belt for conveying workpieces is provided below the multi-axis robotic arm; The driving end of the multi-axis robotic arm is provided with a mounting plate, on which a pressing assembly, a line scan camera and a compensation assembly are sequentially arranged along the conveying direction of the conveyor belt. A label belt assembly is arranged opposite to the pressing assembly, and the conveying direction of the label belt assembly is perpendicular to the conveying direction of the conveyor belt.

9. The visual labeling mechanism according to claim 8, characterized in that: The pressing assembly includes a first linear module, a driving end of the first linear module is provided with a connecting block, the connecting block is provided with a rotating driving member, the driving end of the rotating driving member is provided with a labeling head, and the labeling head includes a plurality of press head units arranged in parallel; The compensation component includes a compensation pressure head, which is one of the plurality of pressure head units.

10. The visual labeling mechanism according to claim 8, characterized in that: The label tape assembly includes a first reel and a second reel that are spaced apart, and a conveying passage for conveying the label tape is formed between the first reel and the second reel; A plurality of tensioning wheels are provided along the conveying path, wherein a driving roller is provided at a preset position of the conveying path, and the driving roller is connected to a driving motor for driving the driving roller to rotate; A passive roller is provided on one side of the active roller, and a third linear module is provided on one side of the passive roller. The third linear module is used to drive the passive roller to move toward or away from the active roller.

Images