Welding defect detection and automatic repair welding system

Abstract

The invention discloses a welding defect detection and automatic repair welding system, which comprises a repair welding execution module, a welding point cloud extraction module, a welding defect extraction module and a defect identification and classification module, and is characterized in that the welding point cloud extraction module is in signal connection with the repair welding execution module; complete point cloud data of a welding area are visually collected through the repair welding execution module, a target area is accurately extracted through the welding point cloud extraction module, and then efficient extraction and contour positioning of welding defects are completed through the welding defect extraction module based on point cloud feature analysis. Accurate identification and multi-type classification of defects are achieved through a defect identification and classification module, finally, six-dimensional coordinates of repair welding track points under a camera coordinate system are obtained through a repair welding instruction calculation module, and then a hand-eye calibration matrix, a tool calibration matrix and a transformation matrix between a wrist joint coordinate system and a base coordinate system are combined, so that the repair welding precision is improved. And through multi-coordinate system collaborative conversion, the six-dimensional pose of the repair welding track point of the mechanical arm welding gun is solved, and equipment is driven to complete repair welding operation.

Description

Technical Field

[0001] This invention belongs to the field of welding processing technology, specifically relating to a welding defect detection and automatic repair welding system. Background Technology

[0002] In the field of industrial welding, especially in the processing of automotive parts, traditional welding robots frequently encounter four defects during operation: burn-through, porosity, incomplete welding, and missed welding. These defects directly affect the welding quality of the workpiece and its subsequent performance, and have an adverse impact on product qualification rate and production efficiency.

[0003] Current industry practices for handling welding defects have significant limitations: after welding operations, workpieces must be manually transferred to a separate production line for visual inspection to detect defects; while some companies have built automated inspection platforms, these platforms require separate deployment and lack integration with existing welding production lines. Subsequent repair processes still heavily rely on manual operation after defect detection. Workers must manually repair problematic welds based on the inspection results. This process is highly dependent on worker skill levels and is inefficient, failing to meet the demands of modern production lines. It increases labor costs and extends production cycles. Therefore, we propose a welding defect detection and automated repair system. Summary of the Invention

[0004] This invention provides a welding defect detection and automatic repair welding system to solve the problems mentioned in the background art.

[0005] This invention provides the following technical solution: it includes a welding repair execution module, and further includes a welding point cloud extraction module, a welding defect extraction module, and a defect identification and classification module. The welding point cloud extraction module is signal-connected to the welding repair execution module, the welding defect extraction module is signal-connected to the welding point cloud extraction module, and the defect identification and classification module is signal-connected to the welding defect extraction module. The welding point cloud extraction module is used to preprocess the point cloud data collected after feature extraction by the welding repair module. The welding defect extraction module is used to separate the weld area and non-weld area obtained by the welding point cloud extraction module. The welding defect extraction module includes a defect segmentation processing unit, a filtering processing unit, a defect segmentation quality optimization unit, and a common defect input unit. The defect segmentation processing unit is used to initially extract the weld defect area after image acquisition by the welding point cloud extraction module. The filtering processing unit filters the weld defect area extracted by the defect segmentation processing unit to optimize the segmentation quality of the weld defect area. The defect segmentation quality optimization unit segments the extracted weld defect area data through a fusion clustering algorithm. The common defect input unit is used to input common defects to provide corresponding matching options when the defect segmentation quality optimization unit segments the weld defect area data. Common defects include burn-through, porosity, incomplete welding, and missing welding defects. The defect identification and classification module is used to read the point cloud data preprocessed by the welding defect extraction module and to expand the weld defect data using the point cloud data.

[0006] A further improvement of the present invention is that the welding execution module includes a vision sensor, an industrial computer, a control cabinet, and a welding robot. The vision sensor is installed on the working end of the welding robot and is connected to a DC power supply. The vision sensor is connected to the industrial computer via a network port and transmits welding images to the industrial computer in real time. The control cabinet is connected to the industrial computer via a network port.

[0007] A further improvement of this invention is that the welding point cloud extraction module includes a data preprocessing unit, a point cloud segmentation unit, a welding area separation unit, and a welding area point cloud denoising unit. The data preprocessing unit sets acquisition thresholds in the X, Y, and Z directions between the welding robot and the workpiece to crop out obviously irrelevant areas and reduce the range of the preprocessed image. The data preprocessing unit is also used to remove discrete noise points after resampling the point cloud. The point cloud segmentation unit separates the workpiece from non-workpieces in the point cloud preprocessed by the data preprocessing unit to obtain a point cloud containing only the workpiece to be welded. The welding area separation unit is used to extract a subset of point clouds containing the weld and its surrounding narrow area from the workpiece point cloud to obtain weld features. The welding area point cloud separated by the welding area point cloud denoising unit is finally purified to remove residual spatter attachment points and small holes in the area and smooth the surface, providing data for subsequent weld feature extraction.

[0008] A further improvement of the present invention is that the defect identification and classification module includes a point cloud data enhancement unit, a defect identification calculation unit, a machine learning model unit, and a defect result output unit. The point cloud data enhancement unit expands the weld defect dataset through cloud data enhancement. The defect identification calculation unit is used to convert the input data of the welding defect extraction module into output labels so as to correctly classify the new sample data. The machine learning model unit is used to obtain the feature values ​​of the sample data classification and feature vector formed by the machine learning model unit as input. Finally, the defect result output unit outputs the prediction results that meet the specific application requirements.

[0009] A further improvement of this invention lies in a welding defect detection system comprising a welding repair command calculation module, a coordinate calibration unit, a hand-eye calibration matrix processing unit, a tool calibration matrix processing unit, and a transformation matrix processing unit. The welding repair command calculation module includes a coordinate calibration unit for calibrating the six-dimensional coordinates of the welding repair trajectory points in the camera coordinate system. The hand-eye calibration matrix processing unit is used to perform hand-eye calibration matrix calculations by the coordinate calibration unit, and the tool calibration matrix processing unit is used to perform tool calibration matrix calculations by the coordinate calibration unit. The transformation matrix processing unit reads data acquired by the hand-eye calibration matrix processing unit and the tool calibration matrix processing unit, calculates the transformation matrix between the wrist joint coordinate system and the base coordinate system, and finally outputs the six-dimensional pose of the robotic arm welding torch welding trajectory points and implements the welding repair operation path control signal.

[0010] Compared with existing technologies, the beneficial effects of this invention are as follows: The welding execution module visually acquires complete point cloud data of the welding area, and the welding point cloud extraction module accurately extracts the target area. Subsequently, the welding defect extraction module efficiently extracts and locates welding defects based on point cloud feature analysis. The defect identification and classification module accurately identifies and classifies defects into multiple types. Finally, the welding instruction calculation module obtains the six-dimensional coordinates of the welding trajectory points in the camera coordinate system. Combined with the hand-eye calibration matrix, tool calibration matrix, and the transformation matrix between the wrist joint coordinate system and the base coordinate system, the six-dimensional pose of the robotic arm welding torch welding trajectory points is solved through multi-coordinate system collaborative calculation. This drives the equipment to complete the welding operation. Compared with existing technologies, this invention can efficiently and accurately complete weld inspection and welding repair, effectively solving the problem of disconnection between welding, inspection, and welding repair processes in existing technologies. It significantly improves welding quality while shortening the production cycle. Attached Figure Description

[0011] Figure 1 This is a composition diagram of the present invention; Figure 2 This is a diagram showing the composition of the welding repair module in this invention; Figure 3This is a flowchart of the operation process of the present invention; Figure 4 This is a flowchart of the operation of the welding instruction calculation module in this invention.

[0012] In the diagram: 1. Welding repair execution module; 11. Vision sensor; 12. Industrial computer; 13. Control cabinet; 14. Welding robot; 2. Welding point cloud extraction module; 21. Data preprocessing unit; 22. Point cloud segmentation processing unit; 23. Welding area separation unit; 24. Welding area point cloud denoising unit; 3. Welding defect extraction module; 31. Defect segmentation processing unit; 32. Filtering processing unit; 33. Defect segmentation quality optimization unit; 34. Common defect input unit; 4. Defect recognition and classification module; 41. Point cloud data enhancement unit; 42. Defect recognition calculation unit; 43. Machine learning model unit; 44. Defect result output unit; 5. Welding repair command calculation module; 51. Coordinate calibration unit; 52. Hand-eye calibration matrix processing unit; 53. Tool calibration matrix processing unit; 54. Transformation matrix processing unit. Detailed Implementation

[0013] The present invention will be further described below with reference to specific embodiments. The accompanying drawings are for illustrative purposes only and are schematic diagrams, not actual pictures. They should not be construed as limiting the present invention. In order to better illustrate the specific embodiments of the present invention, some parts in the drawings may be omitted, enlarged or reduced, and do not represent the actual product size. It is understandable for those skilled in the art that some well-known structures and their descriptions in the drawings may be omitted. All other specific embodiments obtained by those skilled in the art based on the specific embodiments of the present invention without creative effort are within the scope of protection of the present invention. Please see Figures 1-4 A welding defect detection and automatic repair welding system includes a repair welding execution module 1, a welding point cloud extraction module 2, a welding defect extraction module 3, and a defect identification and classification module 4. The welding point cloud extraction module 2 is signal-connected to the repair welding execution module 1, the welding defect extraction module 3 is signal-connected to the welding point cloud extraction module 2, and the defect identification and classification module 4 is signal-connected to the welding defect extraction module 3.

[0014] The welding point cloud extraction module 2 is used to preprocess the point cloud data collected after feature extraction by the welding repair module 1. The welding defect extraction module 3 is used to separate the weld area and non-weld area obtained by the welding point cloud extraction module 2. The welding defect extraction module 3 includes a defect segmentation processing unit 31, a filtering processing unit 32, a defect segmentation quality optimization unit 33, and a common defect input unit 34. The defect segmentation processing unit 31 is used to initially extract the weld defect area after image acquisition by the welding point cloud extraction module 2. The filtering processing unit 32 filters the weld defect area extracted by the defect segmentation processing unit 31 to optimize the segmentation quality of the weld defect area. The defect segmentation quality optimization unit 33 segments the extracted weld defect area data through a fusion clustering algorithm. The common defect input unit 34 is used to input common defects to provide corresponding matching options when the defect segmentation quality optimization unit 33 segments the weld defect area data. Common defects include burn-through, porosity, incomplete welding, and missing welding defects.

[0015] The defect identification and classification module 4 is used to read the point cloud data preprocessed by the welding defect extraction module 3, and to expand the weld defect data through the point cloud data.

[0016] In this embodiment, conventional technical solutions use a weld seam tracker for real-time path correction to address the issue of the welding torch misaligning with the weld seam. However, in actual operation, welding quality is the result of multiple factors. Besides the core issue of the positional deviation between the welding torch and the weld seam, there are also interfering factors such as fluctuations in robot welding voltage, poor consistency of part dimensions, and insufficient positioning accuracy of tooling fixtures. These factors make welding defects difficult to avoid, ultimately requiring extensive manual inspection and rework. In traditional operations, robotic arm inspection generally adopts a post-inspection mode, requiring the additional setup of special tooling to fix the workpiece for inspection after the welding robot 14 completes the welding operation. This not only increases the cost of tooling investment and production line space occupation, but also necessitates manual re-welding after defects are detected. Therefore, the above two processing methods are not ideal for the efficiency of weld seam inspection and re-welding operations. To solve the above problems, this invention proposes a welding point cloud extraction module 2, a welding defect extraction module 3, and a defect identification and classification module 4 for weld seam inspection, and a re-welding instruction calculation module 5 classifies defects according to their identification. Module 4 performs welding inspection data repair processing. The repair welding execution module visually acquires complete point cloud data of the welding area, and the welding point cloud extraction module 2 accurately extracts the target area. Then, the welding point cloud extraction module 2 completes efficient extraction and contour positioning of welding defects based on point cloud feature analysis. The defect recognition and classification module achieves accurate identification and multi-type classification of defects. Finally, the repair welding command calculation module obtains the six-dimensional coordinates of the repair welding trajectory points in the camera coordinate system. Combined with the hand-eye calibration matrix, tool calibration matrix, and transformation matrix between the wrist joint coordinate system and the base coordinate system, the six-dimensional pose of the robotic arm welding gun repair welding trajectory points is solved through multi-coordinate system collaborative calculation, and the equipment is driven to complete the repair welding operation.

[0017] In this embodiment, the welding execution module 1 includes a vision sensor 11, an industrial computer 12, a control cabinet 13, and a welding robot 14. The vision sensor 11 is installed on the working end of the welding robot 14 and is connected to a DC power supply. The vision sensor 11 is connected to the industrial computer 12 via a network port and transmits welding images to the industrial computer 12 in real time. The control cabinet 13 is connected to the industrial computer 12 via a network port. Please refer to the accompanying drawings for details. Figure 2 and Figure 3The welding repair module 1 uses existing welding equipment equipped with a vision sensor 11 (mounted on the output axis of the welding robot 14). During the preparation phase of the welding robot 14, the operator manually operates the robot using a teach pendant to record the workpiece welding path and the welding inspection path. This process is called path teaching and inspection teaching. After the welding robot 14 finishes its welding operation, the system automatically triggers an inspection command, driving the robot 14 to move the vision sensor 11 and perform a full scan of the workpiece welding area along a preset path. The scanned image data is transmitted in real time to the industrial computer 12. Data processing algorithms are used to perform noise reduction and enhancement preprocessing. Subsequently, feature extraction algorithms are applied to extract key features related to welding defects from the complex image information. Finally, a deeply trained intelligent algorithm model is used to identify and classify the extracted features and accurately locate the specific position of the defect. Based on the location and type of the defect, the trajectory points that need to be repaired (i.e., the key points that the robot's welding torch needs to pass through during repair welding) are extracted, and the repair welding trajectory points are converted into a six-dimensional pose in the robot's base coordinate system. The control cabinet 13 receives the pose data and guides the robotic arm to move along the trajectory points, thereby performing the welding repair work.

[0018] After the repair welding is completed, the repaired area is inspected again (using the same inspection method as the previous welded area scan) to check whether the repair welding was successful and whether there are any new defects or insufficient repair welding. If the repair welding quality re-inspection shows no defects, the repair welding is considered successful, and the robot returns to its initial position, completing the entire process.

[0019] The welding point cloud extraction module 2 includes a data preprocessing unit 21, a point cloud segmentation processing unit 22, a welding area separation unit 23, and a welding area point cloud denoising unit 24. The data preprocessing unit 21 sets acquisition thresholds in the X, Y, and Z directions between the welding robot 14 and the workpiece to crop out obviously irrelevant areas and reduce the range of the preprocessed image. The data preprocessing unit 21 is also used to remove discrete noise points after resampling the point cloud. The point cloud segmentation processing unit 22 separates the workpiece from non-workpieces in the point cloud preprocessed by the data preprocessing unit 21 to obtain a point cloud containing only the workpiece to be welded. The welding area separation unit 23 is used to extract a subset of point clouds containing the weld and its nearby narrow area from the workpiece point cloud to obtain weld features. The welding area point cloud denoising unit 24 performs final purification on the weld area point cloud separated from the weld area, removing residual spatter attachment points and small holes in the area, smoothing the surface, and providing data for subsequent weld feature extraction.

[0020] The defect identification and classification module 4 includes a point cloud data enhancement unit 41, a defect identification calculation unit 42, a machine learning model unit 43, and a defect result output unit 44. The point cloud data enhancement unit 41 expands the weld defect dataset through cloud data enhancement. The defect identification calculation unit 42 is used to convert the input data of the welding defect extraction module 3 into output labels so as to correctly classify the new sample data. The machine learning model unit 43 is used to obtain the feature values ​​of the sample data classification and feature vector formed by the machine learning model unit 43 as input. Finally, the defect result output unit 44 outputs the prediction results that meet the specific application requirements.

[0021] In this embodiment, in the field of industrial welding, especially in the automotive parts processing scenario, traditional welding robots 14 frequently encounter four types of defects during operation: burn-through, porosity, incomplete welding, and missed welding. These defects directly affect the welding quality and subsequent performance of the workpiece, adversely impacting product qualification rate and production efficiency. Furthermore, current industry procedures for handling welding defects have significant limitations. After welding, the workpieces must be manually transferred to another independent production line for visual inspection to detect defects. While existing technologies have automated inspection platforms, these platforms require separate deployment and lack integration with the existing welding production line. Therefore, after defect detection, subsequent repair still heavily relies on manual operation. Workers must manually repair problematic welds based on the inspection results. Not only is the repair accuracy highly dependent on the worker's skill level, but manual operation is also inefficient and difficult to adapt to the high-efficiency requirements of modern production lines. Therefore, to solve this problem, this solution combines the welding defect detection system with the defect repair welding system to perform repair welding on the acquired detection data, thus addressing the poor integration of the existing technology. Furthermore, this solution also proposes a welding defect repair welding system, which includes a repair welding command calculation module 5, a coordinate calibration unit 51, a hand-eye calibration matrix processing unit 52, a tool calibration matrix processing unit 53, and a transformation matrix processing unit 54. The repair welding command calculation module 5 includes a coordinate calibration unit 51, which is used to calibrate the six-dimensional coordinates of the repair welding trajectory points in the camera coordinate system.

[0022] Among them, the hand-eye calibration matrix processing unit 52 is used by the coordinate calibration unit 51 to perform hand-eye calibration matrix, the tool calibration matrix processing unit 53 is used by the coordinate calibration unit 51 to perform tool calibration matrix, and the transformation matrix processing unit 54 is used to read the data obtained by the hand-eye calibration matrix processing unit 52 and the tool calibration matrix processing unit 53, calculate the transformation matrix between the wrist joint coordinate system and the base coordinate system, and finally output the six-dimensional pose of the welding torch repair trajectory point of the robotic arm, and implement the repair welding operation path control signal.

[0023] In this embodiment, please refer to the accompanying drawings. Figure 4The welding repair instruction calculation module 5 includes the following welding repair processing steps: Step S1: The coordinate calibration unit 51 selects feature points that can represent the position and orientation of the defect from the defect point cloud segmented in the welding defect extraction module 3. For example, for porosity defects, the center point or key vertex of the boundary contour can be selected; for burn-through defects, the feature inflection point of the hole edge can be selected.

[0024] Step S2: The coordinate calibration unit 51 converts the pixel coordinates of the defect feature points into three-dimensional position coordinates in the sensor coordinate system based on the intrinsic parameters of the vision sensor 11. At the same time, through normal vector analysis or contour fitting of the defect point cloud, the three rotation angles of the defect in the camera coordinate system are calculated, ultimately forming a six-dimensional coordinate system.

[0025] Step S3: The hand-eye calibration matrix processing unit 52 solves the coordinate transformation relationship (i.e., the hand-eye relationship matrix) between the industrial robot base coordinate system or end-effector coordinate system and the sensor coordinate system. Through this matrix, the coordinates of the welding start point under the sensor coordinate system can be accurately transformed to the welding robot base coordinate system, thereby enabling the robot to accurately calculate the position information of the object identified by the sensor in its own base coordinate system during operation.

[0026] Step S4: Since the robotic arm in the welding robot 14 uses the wrist joint coordinate system as the reference for motion control by default, but the actual welding operation is performed by the end of the welding gun or other tools, the tool calibration matrix processing unit 53 calculates the translation vector and rotation matrix from the wrist joint coordinate system to the tool end coordinate system to establish a unique correspondence between the two.

[0027] Step S5: The transformation matrix between the wrist joint coordinate system and the base coordinate system in the transformation matrix processing unit 54 can be calculated using the existing robot kinematic model. This matrix is ​​used to establish a precise mapping between the position and posture of the robotic arm's wrist joint and the robot's fixed reference (base coordinate system). Then, through the following coordinate transformation steps, the welding trajectory point information obtained from the sensor is precisely transformed into the robot's base coordinate system, thereby guiding the robotic arm's welding torch to perform welding operations in the correct position and posture. The steps are as follows: First, using the hand-eye calibration matrix, the welding trajectory points are transformed from the sensor coordinate system to the wrist joint coordinate system: P_mid1=T_sc×P_c.

[0028] Next, the points in the wrist joint coordinate system are transformed to the welding torch tool end coordinate system using the tool calibration matrix: P_mid2=T_tw×P_mid1.

[0029] Finally, using the transformation matrix between the wrist joint and the base coordinate system, the points in the coordinate system of the welding torch tool end effector are transformed to the robot's base coordinate system, resulting in the six-dimensional pose of the welding torch re-welding trajectory points of the robotic arm: That is, P_rob=T_wb×P_mid2=T_sc×P_c×T_tw×T_wb.

[0030] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. A welding defect detection system, comprising a repair welding execution module (1), characterized in that, It also includes a welding point cloud extraction module (2), a welding defect extraction module (3), and a defect identification and classification module (4). The welding point cloud extraction module (2) is signal-connected to the welding repair execution module (1), the welding defect extraction module (3) is signal-connected to the welding point cloud extraction module (2), and the defect identification and classification module (4) is signal-connected to the welding defect extraction module (3). The welding point cloud extraction module (2) is used to preprocess the point cloud data collected after feature extraction by the welding repair module (1). The welding defect extraction module (3) is used to separate the weld area and non-weld area obtained by the welding point cloud extraction module (2). The welding defect extraction module (3) includes a defect segmentation processing unit (31), a filtering processing unit (32), a defect segmentation quality optimization unit (33), and a common defect recording unit (34). The defect segmentation processing unit (31) is used to preprocess the point cloud data collected by the welding point cloud extraction module (2). The weld defect area is initially extracted. The filtering processing unit (32) filters the weld defect area extracted by the defect segmentation processing unit (31) to optimize the segmentation quality of the weld defect area. The defect segmentation quality optimization unit (33) segments the extracted weld defect area data through a fusion clustering algorithm. The common defect input unit (34) is used to input common defects to provide corresponding matching options when the defect segmentation quality optimization unit (33) segments the weld defect area data. Common defects include burn-through, porosity, incomplete welding and missing welding defects. The defect identification and classification module (4) is used to read the preprocessed point cloud data of the welding defect extraction module (3) and to expand the weld defect data through the point cloud data.

2. The welding defect detection system according to claim 1, characterized in that: The welding execution module (1) includes a vision sensor (11), an industrial computer (12), a control cabinet (13), and a welding robot (14). The vision sensor (11) is installed on the working end of the welding robot (14). The vision sensor (11) is connected to a DC power supply. The vision sensor (11) is connected to the industrial computer (12) via a network port. The vision sensor (11) transmits welding images to the industrial computer (12) in real time. The control cabinet (13) is connected to the industrial computer (12) via a network port.

3. The welding defect detection system according to claim 1, characterized in that: The welding point cloud extraction module (2) includes a data preprocessing unit (21), a point cloud segmentation processing unit (22), a welding area separation unit (23), and a welding area point cloud denoising unit (24). The data preprocessing unit (21) sets acquisition thresholds in the X, Y, and Z directions between the welding robot (14) and the workpiece to cut off obviously irrelevant areas and reduce the range of the preprocessed image. The data preprocessing unit (21) is also used to remove discrete noise points after resampling the point cloud. The point cloud segmentation processing unit (22) separates the workpiece from the non-workpiece in the point cloud preprocessed by the data preprocessing unit (21) to obtain a point cloud containing only the workpiece to be welded. The welding area separation unit (23) is used to extract a subset of point clouds containing the weld and its surrounding narrow area from the workpiece point cloud to obtain weld features. The welding area point cloud denoising unit (24) performs final purification on the weld area point cloud separated from the workpiece, removing residual spatter attachment points and small holes in the area, smoothing the surface, and providing data for subsequent weld feature extraction.

4. The welding defect detection system according to claim 1, characterized in that: The defect identification and classification module (4) includes a point cloud data enhancement unit (41), a defect identification calculation unit (42), a machine learning model unit (43), and a defect result output unit (44). The point cloud data enhancement unit (41) expands the weld defect dataset through cloud data enhancement. The defect identification calculation unit (42) is used to convert the input data of the welding defect extraction module (3) into output labels so as to correctly classify the new sample data. The machine learning model unit (43) is used to obtain the feature values ​​of the sample data classification and feature vector formed by the machine learning model unit (43) as input. Finally, the defect result output unit (44) outputs the prediction results that meet the specific application requirements.

5. An automatic welding defect repair system, characterized in that: A welding defect detection system includes a welding repair command calculation module (5), a coordinate calibration unit (51), a hand-eye calibration matrix processing unit (52), a tool calibration matrix processing unit (53), and a transformation matrix processing unit (54). The welding repair command calculation module (5) includes a coordinate calibration unit (51), which is used to calibrate the six-dimensional coordinates of the welding repair trajectory points in the camera coordinate system. The hand-eye calibration matrix processing unit (52) is used to perform hand-eye calibration matrix in the coordinate calibration unit (51), and the tool calibration matrix processing unit (53) is used to perform tool calibration matrix in the coordinate calibration unit (51). The transformation matrix processing unit (54) is used to read the data obtained by the hand-eye calibration matrix processing unit (52) and the tool calibration matrix processing unit (53), calculate the transformation matrix between the wrist joint coordinate system and the base coordinate system, and finally output the six-dimensional pose of the welding repair trajectory points of the robotic arm welding gun and implement the welding repair operation path control signal.

Images