Pre-weld workpiece one-way position deviation detection and welding path pre-correction method and system

Abstract

The present application relates to the technical fields of industrial welding automation and machine vision detection, and aims to provide a pre-welding workpiece one-way position offset detection and welding path pre-correction method and system, which detects the actual assembly position of the workpiece to be welded before the welding process starts, and pre-corrects the welding path based on the detection result, so that the welding path can match the actual assembly state of each workpiece, thereby improving the one-time success rate of welding operation.

Description

Technical Field

[0001] This invention relates to the field of industrial welding automation and machine vision inspection technology, specifically to a method and system for detecting the assembly position offset of the workpiece to be welded using machine vision before the start of the welding process, and for performing pre-welding workpiece unidirectional position offset detection and welding path pre-correction based on the detection results. Background Technology

[0002] In the field of industrial automated welding, especially in robotic welding operations of metal components such as automotive parts and engineering machinery parts, common welding objects include workpieces with various structural forms such as steel bars, profiles, and stampings. During the forming, handling, and clamping process, these workpieces are prone to varying degrees of deformation due to factors such as material elastic rebound, release of processing stress, or assembly errors, resulting in deviations between the actual relative positions of the workpieces to be welded and their designed positions.

[0003] In actual production, such as Figure 1 As shown, when a welding robot welds a workpiece with assembly deviations according to a preset welding path, the welding torch may not match the actual welding position, which may lead to welding quality problems such as incomplete fusion of the weld, burn-through, or unstable weld formation. In severe cases, the welded workpiece may need to be reworked or scrapped, affecting production efficiency and product consistency.

[0004] To address welding quality issues caused by workpiece assembly deviations during the welding process, the following technical methods are mainly employed in existing technologies.

[0005] One approach is manual teaching or teaching-and-reproduction. This involves manually operating the welding robot's teaching pendant to adjust the welding path point by point to match the actual assembly state of the workpiece. This method relies heavily on the operator's experience, is time-consuming to debug, and struggles to adapt to random variations in workpiece assembly deviations between different batches or workstations.

[0006] Another approach is offline programming and simulation correction. This method involves introducing a 3D model of the workpiece and planning and correcting the welding path in offline programming software. However, this method relies on a high-precision workpiece model and assembly error parameters, making it difficult to reflect random deviations caused by workpiece deformation, fixture errors, and other factors in actual production. In practice, multiple trial welds and manual intervention are still required.

[0007] With the development of sensing technology, welding path correction technologies based on sensor feedback have emerged, among which online weld seam tracking technology is a typical example. This type of technology typically uses vision sensors, arc sensors, or contact sensors to detect changes in the weld seam position in real time during welding and dynamically adjust the welding path. While this technology can improve welding accuracy to some extent, it usually requires continuous signal acquisition and real-time control during the welding process, resulting in a complex system structure, high adaptability to the welding environment, and susceptibility to interference from welding arcs, spatter, and fumes.

[0008] In addition, some existing technologies propose using machine vision to detect the workpiece position before welding in order to achieve workpiece positioning or welding start point correction. However, such solutions mostly focus on overall workpiece positioning or multi-dimensional position detection, which usually requires the introduction of a three-dimensional vision system or complex spatial reconstruction algorithm, resulting in high system costs. Furthermore, in practical applications, they do not address the specific technical design for situations where assembly deviations of welded workpieces are dominant in a particular direction.

[0009] Therefore, existing technologies still lack a solution that can effectively detect the positional deviation of the dominant direction of the workpiece during assembly before the welding process begins, and achieve pre-correction of the welding path without introducing real-time tracking of the welding process. How to simplify the system structure, reduce control complexity, and improve the first-time success rate of welding operations while ensuring welding quality remains a pressing technical problem for those skilled in the art. Summary of the Invention

[0010] To address the problem of inconsistent welding quality caused by mismatch between the welding path and the actual welding position due to workpiece assembly deformation or clamping errors in existing welding operations, the present invention aims to provide a method and system for detecting unidirectional workpiece position offset and pre-correcting the welding path before welding. By detecting the actual assembly position of the workpiece to be welded before the start of the welding process and pre-correcting the welding path based on the detection results, the welding path can be matched with the actual assembly state of the workpiece each time, thereby improving the first-time success rate of welding operations.

[0011] The technical solution of this invention is as follows: A method for detecting unidirectional workpiece position offset before welding and pre-correcting welding path includes the following steps: 1) Before the welding process begins, clamp the first and second workpieces to be welded at the welding station and position the welding robot in the initial position to be welded. 2) Before the welding process is performed, a machine vision acquisition device set on one side of the welding station is used to acquire static images of the target detection areas of the first workpiece and the second workpiece. 3) Perform image preprocessing and feature extraction on the acquired workpiece images to obtain feature point information that characterizes the relative assembly relationship between the first workpiece and the second workpiece; 4) Based on the camera calibration parameters, convert the pixel coordinates of the feature points into physical coordinates, and calculate the linear position offset between the first workpiece and the second workpiece in the preset detection direction; 5) Convert the linear position offset into welding path correction parameters, and before the welding process is executed, superimpose the welding path correction parameters onto the preset welding path of the welding robot to generate the corrected welding path. 6) Control the welding robot to complete the welding operation according to the modified welding path.

[0012] Preferably, the preset detection direction is a single linear direction along the welding path normal or the main offset direction of the workpiece assembly.

[0013] Preferably, the calculation of the position offset does not involve the reconstruction of the three-dimensional shape of the workpiece, but is obtained based on the coordinate difference of feature points in the two-dimensional image in the preset detection direction.

[0014] Preferably, the image preprocessing includes grayscale processing, filtering and noise reduction processing, and edge detection processing, and the feature extraction includes extracting at least one of workpiece edge feature points, center feature points, or extreme feature points from the processed image.

[0015] Preferably, the linear position offset is obtained by converting the pixel coordinates of the feature point into physical coordinates based on pre-completed camera calibration parameters, and calculating the difference between the physical coordinates and the preset detection direction.

[0016] Preferably, the welding path correction parameter is the coordinate offset along the preset detection direction, and the coordinate offset is uniformly superimposed on the coordinates of each path point in the preset welding path of the welding robot before the welding process is executed.

[0017] Preferably, the image acquisition and position offset detection are not performed before welding begins or during welding.

[0018] A pre-welding workpiece unidirectional position offset detection and welding path pre-correction system for implementing any of the above methods, comprising: Machine vision acquisition devices are used to acquire image information of the workpiece to be welded before the welding process begins; An industrial computing unit, connected to the machine vision acquisition device, is used to process the acquired images, calculate the linear position offset of the workpiece in the preset detection direction, and generate welding path correction parameters. The communication interface module is used to send the welding path correction parameters to the welding robot control system; A welding robot control system is used to superimpose the welding path correction parameters onto a preset welding path before the welding process is executed, and to control the welding robot to complete the welding operation according to the corrected welding path.

[0019] Preferably, the machine vision acquisition device is equipped with an openable and closable protective mechanism to protect the acquisition device when it is not in the image acquisition state.

[0020] Preferably, the machine vision acquisition device is any one of a two-dimensional area array camera, a 2.5D camera, or a three-dimensional vision camera.

[0021] Compared with the prior art, the present invention has at least the following technical effects: By inspecting the assembly state of the workpiece to be welded before the welding process begins and pre-correcting the welding path before welding, the welding path can be matched with the actual assembly state of the workpiece, avoiding the problem of welding position mismatch caused by directly using the preset welding path due to assembly deviation.

[0022] Since both position offset detection and welding path correction are completed before the welding process begins, this invention avoids the system complexity caused by real-time detection and dynamic control during the welding process, reduces the dependence on real-time sensing, control response speed and anti-interference ability, and helps to improve the overall stability of the welding system.

[0023] This invention addresses the characteristic that assembly deviations in welded workpieces are dominant in a specific direction. It detects and corrects linear positional offsets only in a preset detection direction. While ensuring welding quality, it reduces the amount of computation required for multi-dimensional position detection and spatial reconstruction, making the welding path correction process simpler and more efficient.

[0024] By uniformly mapping the detected position offset to welding path correction parameters and superimposing them on the preset welding path before the welding process is executed, the overall consistency correction of the welding path is achieved, avoiding the path discontinuity problem that may be caused by point-by-point correction or multiple corrections, which is conducive to improving the smoothness and controllability of the welding trajectory.

[0025] Since the position offset detection is performed in a static state after the workpiece is clamped and before welding has started, it effectively avoids the influence of factors such as arc light, spatter and smoke during welding on the detection accuracy. In complex welding environments, it actually improves the reliability of the position offset detection results.

[0026] This invention decouples welding path correction from specific welding process parameters, correcting only the welding path without relying on real-time adjustments to process parameters such as welding current, voltage, or welding speed. This allows the technical solution of this invention to adapt to different welding process conditions and production scenarios, and has good versatility.

[0027] Furthermore, by performing pre-welding detection and correction only on positional offsets in the assembly-dominant direction, this invention achieves effective control over welding quality without introducing multi-dimensional detection or real-time tracking of the welding process. This technical effect is not a simple replacement or superposition of existing technologies, but rather a comprehensive technical effect brought about by a technical strategy that combines pre-welding detection with path pre-correction.

[0028] In addition, by uniformly pre-correcting the welding path before the welding process begins, the welding torch movement trajectory remains continuous and consistent during the welding process, avoiding potential path abrupt changes or control instability during online correction, thus improving the controllability of the welding process from the system control level.

[0029] The technical solution of the present invention can be implemented without changing the original welding process, is easy to integrate with existing welding robot systems, and has good engineering applicability. Attached Figure Description

[0030] Figure 1 This is a schematic diagram of non-fusion welding in existing technologies; Figure 2 This is a schematic diagram of the overall structure of the pre-welding unidirectional position offset detection and welding path pre-correction system; Figure 3 This is a schematic diagram illustrating the effect of extracting features from the workpiece and calculating the unidirectional position offset using image processing in Example 5. Figure 4 This is a schematic diagram of the coordinate system used in Example 5 to illustrate the transformation relationship between pixel coordinates and physical coordinates. In the figure: 6: first workpiece, 61: clamping device for the first workpiece, 7: second workpiece, 71: clamping device for the second workpiece; Detailed Implementation

[0031] Example 1: This embodiment provides a method for detecting unidirectional positional offset of workpiece before welding and pre-correcting welding path, which is applicable to robot welding operation scenarios, especially to situations where the position of the welding workpiece in the dominant direction is offset due to deformation or clamping error during the assembly process.

[0032] In this embodiment, the method includes the following steps, and its overall process is shown in Figure 3.

[0033] Step S1: Workpiece clamping and pre-welding preparation The first and second workpieces to be welded are clamped on the fixtures at the welding station, so that the two workpieces are in a state to be welded; the welding robot is in the initial position before the start of the welding process and has not yet performed the welding action.

[0034] In this state, the workpiece is clamped and remains stationary, and welding has not yet begun.

[0035] Step S2: Pre-welding image acquisition Before the welding process begins, a machine vision acquisition device set up on one side of the welding station is used to acquire images of the target detection areas of the first and second workpieces.

[0036] The image acquisition is completed in a static state before welding, and the welding robot does not perform welding actions during the acquisition process.

[0037] Step S3: Workpiece position offset detection The acquired workpiece images are processed to obtain feature information that characterizes the relative assembly position relationship between the first workpiece and the second workpiece, and the linear position offset of the two workpieces in a preset detection direction is calculated based on the feature information.

[0038] The preset detection direction is either the welding path normal or the dominant direction of positional offset during workpiece assembly.

[0039] The positional offset between the first workpiece and the second workpiece in the preset detection direction is obtained through the above method.

[0040] Step S4: Generation of welding path correction parameters The position offset is converted into a welding path correction parameter, which is used to characterize the overall offset correction amount of the welding path in the preset detection direction.

[0041] Step S5: Pre-correction of welding path Before the welding process begins, the welding path correction parameters are superimposed onto the welding path preset by the welding robot to uniformly correct the welding path and generate the corrected welding path.

[0042] The welding path correction is completed once before welding begins, and no position detection or path correction is performed during the welding process.

[0043] Step S6: Welding execution The welding robot is controlled to perform welding operations according to the modified welding path, so that the welding path matches the actual assembly state of the workpiece.

[0044] Explanation of the implementation effect of this embodiment: Under the conditions of this embodiment, by detecting the assembly state of the workpiece before the welding process begins, and pre-correcting the welding path based on the detected position offset, the welding torch movement trajectory during the welding process can be kept consistent with the actual assembly position of the workpiece, thereby reducing the welding position mismatch problem caused by assembly deviation.

[0045] Example 2: Based on the method for pre-welding unidirectional position offset detection and welding path pre-correction described in Example 1, this example further explains the specific methods for pre-welding image acquisition, position offset detection, and welding path correction.

[0046] Furthermore, in step S2, the pre-welding image acquisition includes: before the welding process begins, taking one or more images of the welding area using an industrial camera. During the acquisition process, the camera's shooting angle and shooting position are set according to the structural characteristics of the welding station and the location of the inspection area to ensure that the target area of ​​the workpiece to be inspected appears completely in the acquired image.

[0047] Further, in step S3, the workpiece position offset detection includes: performing image preprocessing on the acquired workpiece image, extracting feature point information from the image to characterize the relative assembly position relationship between the first workpiece and the second workpiece, and calculating the linear position offset of the two workpieces in a preset detection direction based on the feature point information.

[0048] The image preprocessing may include one or more of grayscale processing, filtering and noise reduction processing, and edge detection processing; the feature points may be feature points on the edge of the workpiece, contour center points, or other feature points that can characterize the relative positional relationship of the workpiece.

[0049] Further, in step S3, the calculation of the position offset includes: based on the pre-completed camera calibration parameters, converting the pixel coordinates of feature points in the image into corresponding physical coordinates, and obtaining the linear position offset between the first workpiece and the second workpiece according to the coordinate difference of the physical coordinates in the preset detection direction.

[0050] Further, in step S4, the welding path correction parameter is: the overall offset of the welding path along the preset detection direction, which is used to uniformly correct each path point in the preset welding path of the welding robot.

[0051] Further, in step S5, the pre-correction of the welding path includes: before the welding process begins, superimposing the welding path correction parameters onto the coordinates of each path point in the preset welding path of the welding robot, so that the welding path is translated as a whole in the preset detection direction, thereby generating the corrected welding path.

[0052] Additional notes for this embodiment: It should be noted that all the further limitations described in this embodiment are optional implementations of the method described in Embodiment 1. This invention is not limited to the specific limitations described above. Equivalent substitutions of image processing methods, feature point selection methods, and path correction methods without departing from the technical concept of this invention should all fall within the protection scope of this invention.

[0053] Example 3: This embodiment provides a pre-welding unidirectional position offset detection and welding path pre-correction system, which is used to detect the assembly position offset of the workpiece to be welded before the start of the welding process, and pre-correct the welding path based on the detection results, so as to execute the methods described in Embodiments 1 and 2.

[0054] The system includes: Machine vision acquisition device Industrial computing units Communication interface module, Welding robot control system.

[0055] Machine vision acquisition device: The machine vision acquisition device is installed on one side of the welding station and is used to acquire images of the target detection areas of the first and second workpieces before the welding process begins. The machine vision acquisition device operates in a static state before welding and does not participate in real-time detection during the welding process after image acquisition is completed.

[0056] Industrial computing unit: The industrial computing unit is connected to the machine vision acquisition device and is used to receive the acquired workpiece image and process the workpiece image to obtain the linear position offset of the first workpiece and the second workpiece in a preset detection direction.

[0057] In addition, the industrial computing unit is also used to convert the position offset into welding path correction parameters.

[0058] Communication interface module: The communication interface module is connected to the industrial computing unit and the welding robot control system respectively, and is used to send the welding path correction parameters to the welding robot control system.

[0059] The communication interface module can be implemented using Ethernet, fieldbus, or other industrial communication methods.

[0060] Welding robot control system: The welding robot control system is used to receive the welding path correction parameters, and before the welding process begins, superimpose the welding path correction parameters onto the welding path preset by the welding robot to generate a corrected welding path; and control the welding robot to perform welding operations according to the corrected welding path.

[0061] System Operation Description: In this embodiment, the detection of workpiece position offset before welding, the generation of welding path correction parameters, and the pre-correction of the welding path are all completed before the welding process begins; during the welding process, the system no longer performs position detection or welding path correction.

[0062] The implementation effect of this embodiment is explained as follows: Through the above system structure settings, the detection of workpiece position offset and the pre-correction of welding path can be completed before welding begins, avoiding the system complexity caused by real-time detection and control during welding, and improving the stability of system operation.

[0063] Example 4: Based on the pre-welding unidirectional position offset detection and welding path pre-correction system described in Example 3, this example further explains the specific configuration of each component unit in the system.

[0064] Furthermore, the machine vision acquisition device includes an industrial camera and lens assembly, used to acquire images of the target inspection area of ​​the workpiece before the welding process begins. The industrial camera can be any one of a two-dimensional area scan camera, a 2.5D camera, or a three-dimensional vision camera.

[0065] Furthermore, the machine vision acquisition device is equipped with a protective mechanism, which is used to protect the industrial camera and lens assembly in non-image acquisition state to avoid the impact of arc light, spatter or dust on the acquisition device during the welding process.

[0066] Furthermore, the industrial computing unit is an industrial computer used to run image processing programs and position offset calculation programs, and to generate welding path correction parameters.

[0067] Furthermore, the communication interface module adopts Ethernet communication or industrial fieldbus communication. This enables data transmission between the industrial computing unit and the welding robot control system.

[0068] Furthermore, the welding robot control system includes a main controller and a servo drive unit. The main controller is used to parse the welding path correction parameters and convert the corrected welding path into servo drive commands. The servo drive unit is used to drive the welding robot to perform welding actions.

[0069] Additional notes on this embodiment: It should be understood that the specific configuration of each component of the system in this embodiment is only an illustrative example. Equivalent substitutions for system hardware type, communication method or protection structure without departing from the technical concept of this invention should fall within the protection scope of this invention.

[0070] Example 5: This embodiment is used to illustrate an exemplary implementation of the workpiece position offset detection step in the pre-welding unidirectional position offset detection method described in Embodiment 1. It does not constitute an independent technical solution, and the present invention is not limited to the specific algorithm or calculation formula described in this embodiment.

[0071] As shown in Figure 2, the first workpiece 6 and the second workpiece 7 are fixed on the welding station by clamping devices 61 and 71, respectively. Due to assembly errors or deformation, the second workpiece 7 has a unidirectional position offset Δd relative to the first workpiece 6 in the preset detection direction.

[0072] (1) Image acquisition and grayscale processing (example formula) Before the welding process begins, the original image I(x,y) of the welding area is acquired through a machine vision acquisition device.

[0073] The original image can be converted to grayscale, for example, by converting a color image to a grayscale image G(x,y) in the following manner: in: R(x,y), G(x,y), and B(x,y) represent the pixel values ​​of the image in the red, green, and blue channels, respectively. α, β, and γ are weighting coefficients, and their values ​​can be set according to actual needs.

[0074] Note: This formula is only used to illustrate the grayscale processing process and does not limit the specific weights or implementation methods.

[0075] (2) Edge detection and grayscale change rate calculation (example formula) After grayscale processing is completed, the changes in pixel grayscale in the image can be analyzed to help determine the edge position of the workpiece.

[0076] For example, edge features can be characterized by calculating the rate of change in grayscale between adjacent pixels: or: Where ∇G represents the magnitude of pixel grayscale change.

[0077] Note: The above expressions are only used to illustrate the grayscale change calculation process and do not limit the specific operators or edge detection algorithms.

[0078] (3) Feature point selection and center position calculation (example formula) After extracting the workpiece contour, feature points that can characterize the workpiece position can be selected from the contour point set.

[0079] For example, the center position of a feature point can be obtained by averaging the coordinates of the contour points along a preset detection direction: in: This represents the coordinate value of the i-th contour point in the preset detection direction; N represents the number of contour points involved in the calculation; This indicates the characteristic position of the workpiece in the detection direction.

[0080] Note: This formula is only an example of how to calculate feature locations.

[0081] (4) Conversion from pixel coordinates to physical coordinates (example formula) like Figure 4 As shown, the proportional relationship between pixel coordinates and physical coordinates is established through camera calibration.

[0082] For example, pixel coordinates can be converted to physical coordinates according to the following linear relationship: in: x is the pixel coordinate; X represents the corresponding physical coordinates; k is the proportionality coefficient; b represents the offset compensation amount.

[0083] Note: The above formula is only used to illustrate the mapping relationship between pixels and physical coordinates, and does not limit the specific calibration model.

[0084] (5) Calculation of position offset (core formula, required) After completing the physical coordinate transformation, the linear position offset between the first workpiece and the second workpiece is calculated only in the preset detection direction described in Example 1.

[0085] For example, the linear position offset Δd can be expressed as: or: in: This represents the physical coordinates of the first workpiece in the preset detection direction; This indicates the physical coordinates of the second workpiece in the preset detection direction.

[0086] (6) Offset output and path correction call like Figure 4 As shown, the calculated linear position offset Δd is output to generate welding path correction parameters, and is used by the method described in Example 1 to pre-correct the welding path before the welding process begins.

[0087] It should be understood that the grayscale calculation method, grayscale change rate calculation method, feature point position calculation method, coordinate transformation formula and offset calculation formula listed in this embodiment are all exemplary implementation methods for implementing the method described in Embodiment 1; Without departing from the technical concept of this invention, any other equivalent calculation method used to obtain the linear position offset of the workpiece in the preset detection direction should fall within the protection scope of this invention.

Claims

1. A method for detecting unidirectional workpiece position offset before welding and pre-correcting the welding path, characterized in that, Includes the following steps: 1) Before the welding process begins, clamp the first and second workpieces to be welded at the welding station and position the welding robot in the initial position to be welded. 2) Before the welding process is performed, a machine vision acquisition device set on one side of the welding station is used to acquire static images of the target detection areas of the first workpiece and the second workpiece. 3) Perform image preprocessing and feature extraction on the acquired workpiece images to obtain feature point information that characterizes the relative assembly relationship between the first workpiece and the second workpiece; 4) Based on the camera calibration parameters, convert the pixel coordinates of the feature points into physical coordinates, and calculate the linear position offset between the first workpiece and the second workpiece in the preset detection direction; 5) Convert the linear position offset into welding path correction parameters, and before the welding process is executed, superimpose the welding path correction parameters onto the preset welding path of the welding robot to generate the corrected welding path. 6) Control the welding robot to complete the welding operation according to the modified welding path.

2. The method according to claim 1, characterized in that, The preset detection direction is a single linear direction along the welding path normal or the main offset direction of the workpiece assembly.

3. The method for detecting unidirectional workpiece position offset and pre-correcting welding path before welding according to claim 1 or 2, characterized in that, The calculation of the position offset does not involve the reconstruction of the three-dimensional shape of the workpiece, but is obtained based on the coordinate difference of feature points in the two-dimensional image in the preset detection direction.

4. The method for detecting unidirectional workpiece position offset and pre-correcting welding path before welding according to claim 1, characterized in that, The image preprocessing includes grayscale processing, filtering and noise reduction processing, and edge detection processing. The feature extraction includes extracting at least one of the following from the processed image: workpiece edge feature points, center feature points, or extreme feature points.

5. The method for detecting unidirectional workpiece position offset and pre-correcting welding path before welding according to claim 1, characterized in that, The linear position offset is obtained in the following way: Based on pre-completed camera calibration parameters, the pixel coordinates of the feature points are converted into physical coordinates, and the difference between the physical coordinates and the preset detection direction is calculated.

6. The method for detecting unidirectional workpiece position offset and pre-correcting welding path before welding according to claim 1, characterized in that, The welding path correction parameter is the coordinate offset along the preset detection direction. The coordinate offset is uniformly superimposed on the coordinates of each path point in the preset welding path of the welding robot before the welding process is executed.

7. The method for detecting unidirectional workpiece position offset and pre-correcting welding path before welding according to claim 1, characterized in that, The image acquisition and position offset detection will not be performed before welding begins or during welding.

8. A pre-welding workpiece unidirectional position offset detection and welding path pre-correction system for implementing the method according to any one of claims 1 to 7, characterized in that, include: Machine vision acquisition devices are used to acquire image information of the workpiece to be welded before the welding process begins; An industrial computing unit, connected to the machine vision acquisition device, is used to process the acquired images, calculate the linear position offset of the workpiece in the preset detection direction, and generate welding path correction parameters. The communication interface module is used to send the welding path correction parameters to the welding robot control system; A welding robot control system is used to superimpose the welding path correction parameters onto a preset welding path before the welding process is executed, and to control the welding robot to complete the welding operation according to the corrected welding path.

9. The pre-welding workpiece unidirectional position offset detection and welding path pre-correction system according to claim 8, characterized in that, The machine vision acquisition device is equipped with an openable and closable protective mechanism to protect the acquisition device when it is not in the image acquisition state.

10. The pre-welding workpiece unidirectional position offset detection and welding path pre-correction system according to claim 8, characterized in that, The machine vision acquisition device is any one of a two-dimensional area array camera, a 2.5D camera, or a three-dimensional vision camera.

Images