Robot intelligent welding method and system

Abstract

The invention relates to the technical field of welding automation and robot visual perception, and discloses a robot intelligent welding method and system.The method comprises the steps that a three-dimensional point cloud image is constructed through a laser triangulation method; filtering and enhancing the point cloud image, and extracting a weld point cloud set by adopting a U-Net semantic segmentation model; carrying out space path identification and attitude regression based on the set to obtain a track element of a position-attitude pair; fitting a path by adopting a cubic B-spline curve and interpolating to generate a continuous welding track; and in the welding process, visual feedback is combined, and proportional differential control logic is introduced for closed-loop track error compensation. Compared with the technical problem that in the prior art, weld joint path recognition and accurate tracking cannot be stably achieved under complex structures such as curved surfaces and variable cross sections, a perception control closed loop based on visual servo is constructed, high-precision automatic welding of the complex weld joint structures is achieved, and the track fitting precision is improved.

Description

Technical Field

[0001] The present invention relates to the field of welding automation and robot visual perception technology, and in particular to a robot intelligent welding method and system. Background Art

[0002] Currently, robotic automatic welding technology is widely used in the automotive, shipbuilding, and construction machinery industries, and is particularly well-suited for welds with high repeatability and regular structures. However, as industrial structural components grow larger, more complex, and feature multiple curvatures, traditional welding robots continue to face numerous technical bottlenecks when dealing with complex three-dimensional weld structures such as curved surfaces, stepped surfaces, pipe socket interfaces, and irregular cross-sections.

[0003] Existing technologies primarily rely on preset trajectories or offline programming for welding path planning. These methods rely heavily on workpiece positioning accuracy and lack the ability to perceive the actual weld geometry. This makes it difficult to adapt to weld position variations, workpiece assembly deviations, or weld drift caused by thermal deformation. Furthermore, some solutions use single-frame images or fixed-angle vision systems to extract weld edges, which cannot construct complete spatial information and makes it difficult to capture the weld tangent direction or posture change characteristics.

[0004] In terms of path control, common robot trajectory planning only considers the sequence of position points, while ignoring the dynamic changes in the end-point posture. This leads to inconsistencies between the welding gun angle and the weld tangent, resulting in uneven weld penetration and reduced weld quality. In terms of closed-loop control, existing methods mostly rely on position feedforward control, which lacks the error correction capability based on real-time visual feedback and cannot dynamically correct deviations during the welding process.

[0005] Therefore, existing technologies cannot fully meet the requirements for high-precision, adaptive, closed-loop control of complex three-dimensional welds. There is an urgent need for a robotic intelligent welding method that can achieve three-dimensional weld recognition, posture estimation, trajectory generation, and dynamic error compensation in complex weld structures, multiple posture changes, and real-time disturbance environments, thereby improving weld path fit, weld consistency, and finished product quality. Summary of the Invention

[0006] In response to the above-mentioned technical deficiencies, the purpose of the present invention is to propose a robot intelligent welding method, which aims to solve the technical problem in the existing technology that weld path recognition and precise tracking cannot be stably achieved under complex structures such as curved surfaces and variable cross-sections.

[0007] In order to solve the above technical problems, the present invention adopts the following technical solutions: The present invention provides a robot intelligent welding method,

[0008] The robot intelligent welding method comprises:

[0009] Step S10: Scanning the target welded workpiece with a laser transmitter and an industrial camera to obtain a weld area image, constructing a spatial mapping model based on the weld area image using a laser triangulation method, and outputting a three-dimensional point cloud image of the target welded workpiece;

[0010] Step S20: performing Gaussian filtering denoising and contrast-limited adaptive histogram equalization on the target welding workpiece 3D point cloud image to obtain a preprocessed point cloud image, and performing pixel-level weld segmentation using a pre-trained Python semantic segmentation model U-Net to generate a weld point cloud set;

[0011] Step S30: performing spatial path structure recognition and posture recognition based on the weld point cloud set to obtain trajectory primitives of position-posture pairs;

[0012] Step S40: Using a cubic B-spline curve to fit the path of the trajectory primitive based on the position-posture pair, and generating a continuous executable welding trajectory path and a welding instruction through posture interpolation, the robot reads the welding instruction and executes the welding operation;

[0013] Step S50: During the welding process, the robot performs closed-loop trajectory tracking control based on the welding trajectory path combined with the real-time weld detection results, and introduces proportional differential control logic for error compensation.

[0014] Preferably, in step S10, the steps of scanning the target weld workpiece by a laser transmitter and an industrial camera to obtain a weld area image, constructing a spatial mapping model based on the weld area image by using a laser triangulation method, and outputting a three-dimensional point cloud image of the target weld workpiece specifically include:

[0015] A laser line is formed on the target welding workpiece surface by a laser transmitter. The industrial camera captures the projection position of the laser line on the target welding workpiece surface at a preset fixed frame rate to obtain the weld area image and the position of the laser line pixel;

[0016] Laser triangulation is used to perform depth inverse calculation on the parallax value d of the position of the laser line pixel in the weld area image, and the depth z value of the position of the laser line pixel in the preset camera coordinate system is calculated;

[0017] Combined with the preset camera intrinsic parameter matrix K, the position of the laser line pixel is mapped to the camera coordinate system through back projection, and after the distortion of the camera coordinate system is corrected, the corresponding three-dimensional point cloud image of the target welding workpiece is obtained.

[0018] Preferably, in step S30, the step of performing spatial path structure recognition and posture recognition based on the weld point cloud set to obtain the trajectory primitive of the position-posture pair specifically includes: predicting the posture direction vector of the weld i using the posture regression model constructed based on the ResNet or MobileNet backbone network based on the weld point cloud set. And apply rigid body transformation to map the weld point cloud set to the preset robot base coordinate system to obtain the trajectory primitive of the position-posture pair Among them, p i is the three-dimensional coordinate position of weld i in the robot base coordinate system.

[0019] Preferably, in step S40, the step of fitting the path of the position-posture pair using a cubic B-spline curve is performed using the following formula: Among them, C(u) is the path fitting function, n is the number of welds, N i,k (u) is the B-spline basis function of order k, P i is the three-dimensional coordinate position of weld i in the robot base coordinate system.

[0020] Preferably, in step S40, the step of generating a continuous executable welding trajectory path by posture interpolation method specifically includes: the posture direction vector in the trajectory primitive of the position-posture pair Apply linear interpolation or Slerp spherical interpolation to generate path pose function Path fitting function C(u) and posture function Synchronous sampling is used to construct a trajectory point sequence as a continuously executable welding trajectory path.

[0021] Preferably, in step S50, during the welding process, closed-loop trajectory tracking control is performed based on the welding trajectory path combined with the real-time detection results of the weld, and proportional differential control logic is introduced to perform error compensation. Specifically, the steps include: during the welding process, the actual position of the weld j is obtained in real time through the camera and sensor. And calculate the current frame error vector e j ; The proportional differential control logic is used to calculate the trajectory correction value, and the welding trajectory path error is compensated according to the trajectory correction value to obtain the optimized welding trajectory path.

[0022] Preferably, in step S50, the proportional differential control logic adopts PD control method.

[0023] The present invention also provides a robot intelligent welding system comprising:

[0024] 3D mapping modeling module: used to scan the target welding workpiece with a laser transmitter and an industrial camera to obtain a weld area image, build a spatial mapping model based on the weld area image using laser triangulation, and output a 3D point cloud image of the target welding workpiece;

[0025] Weld point cloud segmentation module: This module performs Gaussian filtering denoising and contrast-limited adaptive histogram equalization on the 3D point cloud image of the target weld workpiece to obtain a preprocessed point cloud image. It then uses the pre-trained Python semantic segmentation model U-Net to perform pixel-level weld segmentation to generate a weld point cloud collection.

[0026] Posture recognition and trajectory primitive extraction module: used to perform spatial path structure recognition and posture recognition based on the weld point cloud set to obtain the trajectory primitives of the position-posture pair;

[0027] Welding trajectory fitting module: It is used to fit the path of the trajectory primitive based on the position-posture pair using a cubic B-spline curve, and generate a continuous and executable welding trajectory path and welding instructions through posture interpolation. The robot reads the welding instructions and performs the welding operation;

[0028] Closed-loop control and error compensation module: used for closed-loop trajectory tracking control of the robot during the welding process based on the welding trajectory path combined with the real-time weld detection results, and introduces proportional differential control logic for error compensation.

[0029] The present invention also provides a computer program product, comprising a robot intelligent welding program, which implements the robot intelligent welding method when executed by a processor.

[0030] The beneficial effects of the present invention are: by constructing a three-dimensional visual detection method based on laser triangulation and deep learning, the present invention realizes high-precision recognition and posture estimation of welds on complex surfaces, variable cross-sections and other structures, and can dynamically extract weld position and tangent direction information, making up for the shortcomings of traditional two-dimensional image detection methods in the spatial structure perception dimension, and improving the path recognition accuracy of robots in complex welding environments.

[0031] The present invention introduces real-time visual inspection and a proportional-differential control closed-loop mechanism during the welding process, and dynamically adjusts the robot welding gun trajectory based on the weld inspection results. This can effectively suppress weld drift problems caused by workpiece assembly errors, thermal deformation, or environmental disturbances, thereby improving the welding path fit and weld formation consistency. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] 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.

[0033] Figure 1 This is a flow chart of a first embodiment of a robot intelligent welding method of the present invention.

[0034] Figure 2 This is a schematic diagram of the structure of a robot intelligent welding system of the present invention

[0035] Figure 3 This is a schematic diagram of equipment for a robot intelligent welding method of the present invention. DETAILED DESCRIPTION

[0036] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments 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 efforts are within the scope of protection of the present invention.

[0037] like Figure 1 As shown, in one embodiment of the present invention, a schematic flow chart of a robot intelligent welding method includes the following steps:

[0038] Step S10: Scanning the target welded workpiece with a laser transmitter and an industrial camera to obtain a weld area image, constructing a spatial mapping model based on the weld area image using a laser triangulation method, and outputting a three-dimensional point cloud image of the target welded workpiece;

[0039] Specifically, in step S10, an industrial camera is used to scan the weld area image of the target welding workpiece, a spatial mapping model is constructed according to the weld area image by using the laser triangulation method, and a weld three-dimensional point cloud set is output. The steps specifically include: forming a laser line on the surface of the target welding workpiece by a laser emitter, and the industrial camera captures the projection position of the laser line on the workpiece surface at a preset fixed frame rate to obtain the weld area image and the position of the laser line pixel; using the laser triangulation method to perform depth inverse calculation on the parallax value d of the position of the laser line pixel in the weld area image, and calculating the depth z value of the position of the laser line pixel in the preset camera coordinate system; then, in combination with the preset camera intrinsic parameter matrix K, the position of the laser line pixel is mapped to the camera coordinate system by back projection, and after performing distortion correction on the camera coordinate system, the corresponding target welding workpiece three-dimensional point cloud image is obtained.

[0040] Specifically, the laser triangulation measurement system constructed by the laser emitter and the industrial camera in step S10 can obtain the three-dimensional depth information of the weld area in real time without contacting the workpiece, enabling the robot to have the spatial perception ability of the geometric contour of the weld. It is particularly suitable for complex structural scenes such as curved surfaces, variable cross-sections, and inclined welds, and improves the integrity and robustness of weld detection.

[0041] Specifically, in step S10, by back-projecting the laser line pixel points in combination with the camera intrinsic parameter matrix and performing distortion correction, the positioning deviation caused by image edge distortion can be effectively eliminated, ensuring the accuracy of the generated three-dimensional point cloud in geometric position, thereby providing high-precision visual inspection input for subsequent weld recognition, path fitting and robot end posture control, and improving the accuracy of the entire welding path planning and the consistency of welding formation.

[0042] Step S20: performing Gaussian filtering denoising and contrast-limited adaptive histogram equalization on the target welding workpiece 3D point cloud image to obtain a preprocessed point cloud image, and performing pixel-level weld segmentation using a pre-trained Python semantic segmentation model U-Net to generate a weld point cloud set;

[0043] Specifically, the above-mentioned image preprocessing operation significantly improves the grayscale distribution difference between the weld area and the background, and enhances the segmentation model's ability to discriminate the weld boundary. It is especially suitable for welding site image environments with low contrast, uneven lighting or strong reflection interference.

[0044] Specifically, the semantic segmentation model U-Net can learn the typical shape, texture and spatial distribution characteristics of welds from a large number of annotated weld images through end-to-end training, so that the weld area can still be accurately segmented on the surface of complex workpieces; combined with the point cloud back projection and mask mapping relationship, the weld pixel area is finally restored to a weld point cloud set, providing an accurate and stable input basis for subsequent path recognition and posture analysis.

[0045] Step S30: performing spatial path structure recognition and posture recognition based on the weld point cloud set to obtain trajectory primitives of position-posture pairs;

[0046] Specifically, in step S30, the spatial path structure recognition and posture recognition are performed based on the weld point cloud set to obtain the trajectory primitive of the position-posture pair, which specifically includes: predicting the posture direction vector of the weld i using the posture regression model built based on the ResNet or MobileNet backbone network based on the weld point cloud set. And apply rigid body transformation to map the weld point cloud set to the preset robot base coordinate system to obtain the trajectory primitive of the position-posture pair Among them, p iis the three-dimensional coordinate position of weld i in the robot base coordinate system.

[0047] The trajectory primitives of the position-attitude pair are obtained above The prerequisite is that by training on a large number of weld data sets, the local main direction of the weld path can be accurately identified, and stable direction estimation results can be provided even in areas with complex weld shapes or blurred boundaries; at the same time, a coordinate mapping method based on rigid body transformation is adopted to ensure that the weld establishes a unified spatial reference in the robot control system, avoiding problems such as path distortion or posture mismatch.

[0048] By obtaining the position-posture pair of the weld point, it not only provides a geometric continuity basis for subsequent path fitting, but also enables the posture of the welding gun end to be adjusted in real time according to the weld direction, achieving process control with uniform welding penetration and angle matching, thereby significantly improving welding quality and weld consistency. It is particularly suitable for complex weld structures with drastic spatial posture changes or multi-angle splicing.

[0049] Step S40: Using a cubic B-spline curve to fit the path of the trajectory primitive based on the position-posture pair, and generating a continuous executable welding trajectory path and a welding instruction through posture interpolation, the robot reads the welding instruction and executes the welding operation;

[0050] Specifically, in step S40, the path fitting step is performed using a cubic B-spline curve based on the trajectory primitives of the position-posture pair, and the formula used is: Among them, C(u) is the path fitting function, n is the number of welds, N i,k (u) is the B-spline basis function of order k, P i is the three-dimensional coordinate position of the weld i in the robot base coordinate system. And the steps of generating a continuous executable welding trajectory path by the posture interpolation method specifically include: the posture direction vector in the trajectory primitive of the position-posture pair Apply linear interpolation or Slerp spherical interpolation to generate path pose function Path fitting function C(u) and posture function Synchronous sampling is used to construct a trajectory point sequence as a continuously executable welding trajectory path.

[0051] The introduction of the posture interpolation function ensures that the welding trajectory remains not only continuous and smooth in position but also geometrically consistent in orientation, ensuring that the welding gun posture always aligns with the weld direction. This interpolated posture path effectively avoids the risk of welding gun angle jumps or posture instability, especially when the weld has sharp turns, spatial transitions, or height changes.

[0052] By jointly interpolating the path position and posture direction and synchronously sampling to form a trajectory point sequence, an execution path with controllable smoothness, continuity and spatial consistency is constructed, providing complete posture input for subsequent robot welding control, improving the weld trajectory tracking accuracy, posture response accuracy, and the stability and forming quality of the welding process.

[0053] Step S50: During the welding process, the robot performs closed-loop trajectory tracking control based on the welding trajectory path combined with the real-time weld detection results, and introduces proportional differential control logic for error compensation.

[0054] Specifically, the proportional differential control logic in step S50 adopts the PD control method.

[0055] By introducing real-time visual inspection feedback and PD control compensation strategy, the system can dynamically respond to slight deviations in the weld position during the welding process. Especially under non-ideal working conditions such as workpiece thermal deformation, assembly tolerance or posture disturbance, it can still keep the end of the welding gun running stably along the center line of the weld, effectively preventing the occurrence of welding defects such as weld deviation and undercut.

[0056] This closed-loop trajectory control mechanism establishes a real-time control chain of "weld seam visual inspection - error calculation - trajectory compensation," achieving perception-control synergy for robotic welding. Compared to traditional open-loop execution paths, this solution significantly improves weld path conformity, robot end-point tracking accuracy, and welding process robustness, ensuring consistent and stable welding quality even with complex weld structures.

[0057] In addition, if Figure 2 As shown, in one embodiment of the present invention, a robot intelligent welding system is proposed, and the robot intelligent welding system includes:

[0058] 3D mapping modeling module: This module is used to scan the target welding workpiece using a laser transmitter and an industrial camera mounted on the robot to obtain an image of the weld area. Based on the image of the weld area, a spatial mapping model is constructed using laser triangulation, and a 3D point cloud image of the target welding workpiece is output.

[0059] Weld point cloud segmentation module: This module performs Gaussian filtering denoising and contrast-limited adaptive histogram equalization on the 3D point cloud image of the target weld workpiece to obtain a preprocessed point cloud image. It then uses the pre-trained Python semantic segmentation model U-Net to perform pixel-level weld segmentation to generate a weld point cloud collection.

[0060] Posture recognition and trajectory primitive extraction module: used to perform spatial path structure recognition and posture recognition based on the weld point cloud set to obtain the trajectory primitives of the position-posture pair;

[0061] Welding trajectory fitting module: It is used to fit the path of the trajectory primitive based on the position-posture pair using a cubic B-spline curve, and generate a continuous and executable welding trajectory path through the posture interpolation method. The robot reads the welding trajectory path and performs the welding operation;

[0062] Closed-loop control and error compensation module: used for closed-loop trajectory tracking control of the robot during the welding process based on the welding trajectory path combined with the real-time weld detection results, and introduces proportional differential control logic for error compensation.

[0063] This application provides a robotic intelligent welding system that utilizes a robotic intelligent welding method described in the aforementioned embodiment to address the technical issues of robotic intelligent welding. Compared to the prior art, the beneficial effects of the robotic intelligent welding system provided by the present invention are the same as those of the robotic intelligent welding method described in the aforementioned embodiment. Other technical features of the robotic intelligent welding system are the same as those disclosed in the aforementioned embodiment and are not further elaborated upon here.

[0064] like Figure 3As shown, in one embodiment of the present invention, a schematic structural diagram of a robotic intelligent welding device suitable for implementing an embodiment of the present application is shown, comprising: at least one processor; and a memory in communication with the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can perform a robotic intelligent welding method in the above-mentioned embodiment 1. A robotic intelligent welding device in an embodiment of the present invention may include, but is not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), and fixed terminals such as digital TVs and desktop computers. A robotic intelligent welding device is merely an example and should not limit the functions and scope of use of the embodiments of the present invention. A robotic intelligent welding device may include a processing device 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes based on programs stored in a read-only memory (ROM) 1002 or programs loaded from a storage device 1003 into a random access memory (RAM) 1004. RAM 1004 also stores various programs and data required for the operation of the robotic intelligent welding device. The processing device 1001, ROM 1002, and RAM 1004 are interconnected via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, hard disk, etc.; and a communication device 1009. The communication device 1009 can allow a robotic intelligent welding device to communicate with other devices wirelessly or wired to exchange data. Although the figure shows a robotic intelligent welding device with various systems, it should be understood that it is not required to implement or have all of the systems shown. More or fewer systems may be implemented or have alternatively.

[0065] The present invention also provides a computer program product comprising a computer program. When executed by a processor, the computer program implements the steps of the aforementioned robotic intelligent welding method. The computer program product provided by the present invention can solve the technical problem of robotic intelligent welding. Compared with the prior art, the beneficial effects of the computer program product provided by the present invention are the same as those of the robotic intelligent welding method provided by the aforementioned embodiment, and are not further elaborated here.

[0066] In particular, according to the embodiments disclosed in the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, the embodiments disclosed in the present invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program comprising program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present invention are performed.

[0067] The various parts disclosed in the present invention may be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics may be combined in any appropriate manner in any one or more embodiments or examples.

[0068] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. A robot intelligent welding method, characterized in that: Methods include: Step S10: Scanning the target welded workpiece with a laser transmitter and an industrial camera to obtain a weld area image, constructing a spatial mapping model based on the weld area image using a laser triangulation method, and outputting a three-dimensional point cloud image of the target welded workpiece; Step S20: performing Gaussian filtering denoising and contrast-limited adaptive histogram equalization on the target welding workpiece 3D point cloud image to obtain a preprocessed point cloud image, and performing pixel-level weld segmentation using a pre-trained Python semantic segmentation model U-Net to generate a weld point cloud set; Step S30: performing spatial path structure recognition and posture recognition based on the weld point cloud set to obtain trajectory primitives of position-posture pairs; Step S40: Using a cubic B-spline curve to fit the path of the trajectory primitive based on the position-posture pair, and generating a continuous executable welding trajectory path and a welding instruction through posture interpolation, the robot reads the welding instruction and executes the welding operation; Step S50: During the welding process, the robot performs closed-loop trajectory tracking control based on the welding trajectory path combined with the real-time weld detection results, and introduces proportional differential control logic for error compensation.

2. A robot intelligent welding method according to claim 1, characterized in that: In step S10, the steps of scanning the target weld workpiece by a laser transmitter and an industrial camera to obtain a weld area image, constructing a spatial mapping model based on the weld area image using a laser triangulation method, and outputting a three-dimensional point cloud set of the target weld workpiece specifically include: A laser line is formed on the target welding workpiece surface by a laser transmitter. The industrial camera captures the projection position of the laser line on the workpiece surface at a preset fixed frame rate to obtain the weld area image and the position of the laser line pixel. Laser triangulation is used to perform depth inverse calculation on the parallax value d of the position of the laser line pixel in the weld area image, and the depth z value of the position of the laser line pixel in the preset camera coordinate system is calculated; Combined with the preset camera intrinsic parameter matrix K, the position of the laser line pixel is mapped to the camera coordinate system through back projection, and after the distortion of the camera coordinate system is corrected, the corresponding three-dimensional point cloud image of the target welding workpiece is obtained.

3. A robot intelligent welding method according to claim 1, characterized in that: In step S30, spatial path structure recognition and posture recognition are performed based on the weld point cloud set to obtain the trajectory primitive of the position-posture pair, specifically including: predicting the posture direction vector of weld i using the posture regression model built based on the ResNet or MobileNet backbone network based on the weld point cloud set And apply rigid body transformation to map the weld point cloud set to the preset robot base coordinate system to obtain the trajectory primitive of the position-posture pair Among them, p i is the three-dimensional coordinate position of weld i in the robot base coordinate system.

4. A robot intelligent welding method according to claim 1, characterized in that: In step S40, a cubic B-spline curve is used to fit the path of the position-posture pair trajectory primitives. The formula used is: Among them, C(u) is the path fitting function, n is the number of welds, N i,k (u) is the B-spline basis function of order k, P i is the three-dimensional coordinate position of weld i in the robot base coordinate system.

5. A robot intelligent welding method as claimed in claim 4, characterized in that: In step S40, the step of generating a continuous executable welding trajectory path by posture interpolation method specifically includes: the posture direction vector in the trajectory primitive of the position-posture pair Apply linear interpolation or Slerp spherical interpolation to generate path pose function Path fitting function C(u) and posture function Synchronous sampling is used to construct a trajectory point sequence as a continuously executable welding trajectory path.

6. A robot intelligent welding method according to claim 1, characterized in that: In step S50, during the welding process, closed-loop trajectory tracking control is performed based on the welding trajectory path combined with the real-time detection results of the weld, and proportional differential control logic is introduced to perform error compensation. Specifically, during the welding process, the actual position of the weld j is obtained in real time through the camera and sensor. And calculate the current frame error vector e j ; The proportional differential control logic is used to calculate the trajectory correction value, and the welding trajectory path error is compensated according to the trajectory correction value to obtain the optimized welding trajectory path.

7. A robot intelligent welding method according to claim 1, characterized in that: In step S50, the proportional differential control logic adopts the PD control method.

8. A robot intelligent welding system, applied to a robot intelligent welding method according to any one of claims 1 to 7, characterized in that: The robot intelligent welding system includes: 3D mapping modeling module: used to scan the target welding workpiece with a laser transmitter and an industrial camera to obtain a weld area image, build a spatial mapping model based on the weld area image using laser triangulation, and output a 3D point cloud image of the target welding workpiece; Weld point cloud segmentation module: This module performs Gaussian filtering denoising and contrast-limited adaptive histogram equalization on the 3D point cloud image of the target weld workpiece to obtain a preprocessed point cloud image. It then uses the pre-trained Python semantic segmentation model U-Net to perform pixel-level weld segmentation to generate a weld point cloud collection. Posture recognition and trajectory primitive extraction module: used to perform spatial path structure recognition and posture recognition based on the weld point cloud set to obtain the trajectory primitives of the position-posture pair; Welding trajectory fitting module: It is used to fit the path of the trajectory primitive based on the position-posture pair using a cubic B-spline curve, and generate a continuous and executable welding trajectory path and welding instructions through posture interpolation. The robot reads the welding instructions and performs the welding operation; Closed-loop control and error compensation module: used for closed-loop trajectory tracking control of the robot during the welding process based on the welding trajectory path combined with the real-time weld detection results, and introduces proportional differential control logic for error compensation.

9. A robot intelligent welding equipment, characterized in that, The robot intelligent welding equipment includes: a memory, a processor, and a robot intelligent welding program stored in the memory and executable on the processor. When the robot intelligent welding program is executed by the processor, a robot intelligent welding method according to any one of claims 1 to 7 is implemented.

10. A computer program product, characterized in that The computer program product includes a robot intelligent welding program, and when the robot intelligent welding program is executed by a processor, a robot intelligent welding method according to any one of claims 1 to 7 is implemented.

Images