CN115106617A - A scanning and tracking method for long welds in a narrow space - Google Patents

Abstract

The application relates to the technical field of robot welding, and discloses a method for scanning and tracking long welds in a narrow space. The method utilizes a robot control system and an image recognition system, manually places a calibration plate, and automatically recognizes the position of the welds of structural parts by three-dimensional vision. The method can automatically find the welding position, which can meet the needs of welding small-space components, and can weld the groove welds of multi-cavity steel components to realize robot automatic welding operations.

Description

A method for scanning and tracking long welds in narrow spaces

technical field

The present application relates to the technical field of robotic welding, in particular to a method for scanning and tracking long welds in a narrow space.

Background technique

With the increasing maturity of robot technology, welding robots are gradually being widely used in various fields of industrial manufacturing. Because the welding robot has multi-axis linkage and program intelligent control functions, it has the characteristics of high action repetition accuracy, stable product quality, and high production efficiency. It is especially suitable for application in assembly lines, such as automobile manufacturing, medical equipment, chemical production, packaging, etc. industry with the characteristics of chemicalization and standardization.

In water conservancy and hydropower steel structural parts, such as arc doors, racks, etc., hand arc welding is currently mainly used for welding. Manual arc welding has high labor intensity, unstable welding quality, and low production efficiency. Applying intelligent robot welding technology to the manufacture of such products can effectively solve the problems of quality and efficiency instead of manual arc welding. However, due to the complex structure of this type of welding parts, many welding direction changes, narrow space, the quality level of the weld is first-class, 100% UT and so on. Because the internal space of this type of structure is narrow, it is difficult to enter the cavity for welding. In order to realize welding in narrow space, a small space long cantilever internal welding device is used, which is installed on the robot arm, which can be used for multi-cavity steel welding. The groove welds of the components are welded. However, due to the narrow and closed space, the long welding seam and the straightness of the groove combined welding seam greater than 3mm, the real-time welding situation cannot be seen, and only relying on programming to realize welding cannot guarantee the quality of the welding seam.

SUMMARY OF THE INVENTION

In order to solve the above-mentioned problems and deficiencies in the prior art, the present application proposes a method for scanning and tracking long welds in a narrow space. The method of automatically identifying the welding position of structural parts, automatically finding the welding position, meeting the needs of welding small-space components, and welding the groove welds of multi-cavity steel components to realize robot automatic welding operations.

In order to realize the above-mentioned purpose of the invention, the technical scheme of the present application is as follows:

A method for scanning and tracking long welds in a narrow space, comprising:

Manually lay the calibration plate to the welding position of the workpiece, and the vision sensor collects the real-time image of the workpiece and transmits the image to the upper computer;

The host computer performs image processing on the collected workpiece image, identifies the calibration plate in the image, obtains the type of workpiece and the type of weld, and finally determines the motion trajectory of the welding robot;

Obtain the original 3D point cloud data of the workpiece, preprocess the original 3D point cloud data of the workpiece, and finally obtain the 3D model of the workpiece and the weld position of the workpiece;

The three-dimensional model of the workpiece is compared with the actual workpiece. If the two match or are within the error range, the upper computer transmits the motion trajectory command of the welding robot and the position information of the welding seam to the welding robot controller. The instruction and the position information of the welding seam control the execution structure of the motion axis of the welding robot, and finally realize the welding of the workpiece.

Further, the visual sensor includes a CCD camera, a laser and a narrow-band filter, the CCD camera and the laser are integrated in a housing, the housing is arranged on the shaft arm of the welding robot, and the narrow-band filter is arranged on the lens of the CCD camera. Front.

Further, the image processing includes image preprocessing, edge detection and feature extraction.

Further, the image preprocessing includes one or more of digitization, geometric transformation, normalization, smoothing, restoration and enhancement.

Further, the host computer performs image processing on the collected workpiece image, identifies the calibration plate in the image, obtains the type of the workpiece, the type of weld and the position of the weld, and finally determines the motion trajectory of the welding robot, including:

The host computer stores a basic database, which includes calibration plate type data, workpiece data and welding robot trajectory data, calibration plate type data, workpiece data and welding robot trajectory data correspondingly matched; when the workpiece weld position is identified, place it After calibrating the plate type, find the matching workpiece data in the basic database, and finally get the workpiece type, welding seam type and the corresponding welding robot motion trajectory.

Further, the preprocessing of the original 3D point cloud data of the workpiece includes:

Point cloud coordinate transformation: Solve the coordinate transformation matrix between the visual coordinate system and the robot coordinate system, and convert the acquired original 3D point cloud data from the coordinates of the scanning device to the coordinates of the robot;

Point cloud slicing: centrally slice the acquired 3D point cloud to obtain the main view slice point cloud, top view slice point cloud and left view slice point cloud of the workpiece;

Point cloud denoising: respectively denoising the main view slice point cloud, top view slice point cloud and left view slice point cloud of the above workpiece;

Point cloud segmentation: After obtaining the denoised point cloud slice, set a grayscale threshold, select the target area and background area, and extract the rough outline of the target area;

Point cloud slice reorganization: Reorganize the main view slice point cloud, top view slice point cloud and left view slice point cloud, and finally get the 3D model of the workpiece.

Beneficial effects of this application:

(1) This application uses the robot control system and image recognition system, adopts the manual placement of the guide plate, and the three-dimensional vision automatically identifies the welding position of the structural member, and automatically finds the welding position to meet the needs of welding of small space components. The groove welds of the body steel components are welded to realize the robot automatic welding operation.

(2) This application automatically guides the robot to align the welding torch at the designated position of the welding seam based on the laser scanning the position of the welding seam, which can ensure that the welding seam tracking accuracy meets the requirements of the welding process, and also reduces the difficulty and workload of online programming. After multiple programming, one program can be adapted to a variety of workpieces, and there is no need for teaching operations. You only need to compare the scanned workpiece model with the real object, and then find the weld according to the scanned workpiece. No matter what form of weld, the program Welding can be performed by automatically calling corresponding subprograms and corresponding parameters.

(3) The manual laying of the calibration plate mentioned in this application greatly simplifies the robot's large-scale search for workpieces and welding seams. The robot only needs to move to the position of the calibration plate to scan the workpiece and find the welding seam by itself.

Description of drawings

The foregoing and following detailed description of the present application will become more apparent when read in conjunction with the following accompanying drawings, in which:

FIG. 1 is a flow chart of the method of the application.

Detailed ways

In order to enable those skilled in the art to better understand the technical solutions in the present application, the following will further illustrate the technical solutions for realizing the purpose of the invention of the present application through several specific embodiments. It should be noted that the technical solutions claimed in the present application are Protocols include but are not limited to the following examples. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative work shall fall within the protection scope of this application.

At present, in the water conservancy and hydropower steel structure parts, such as arc doors, frames, etc., hand arc welding is mainly used for welding. Manual arc welding has high labor intensity, unstable welding quality, and low production efficiency. Applying intelligent robot welding technology to the manufacture of such products can effectively solve the problems of quality and efficiency instead of manual arc welding. However, due to the complex structure of this type of welding parts, many welding direction changes, narrow space, the quality level of the weld is first-class, 100% UT and so on. Especially the multi-cavity steel components, each cavity size is 175 mmx200mm, the depth is 1900 mm, the plate thickness is 50mm, and the groove is combined with welding seam. Because the internal space of this type of structure is narrow, it is difficult to enter the cavity for welding. In order to realize welding in narrow space, a small space long cantilever internal welding device is used, which is installed on the robot arm, which can be used for multi-cavity steel welding. The groove welds of the components are welded. However, due to the narrow and closed space, the long welding seam and the straightness of the groove combined welding seam greater than 3mm, the real-time welding situation cannot be seen, and only relying on programming to realize welding cannot guarantee the quality of the welding seam.

Based on this, in order to solve the problem of multi-cavity steel components, the size of each cavity is 175 mmx200 mm and the depth is 1900 mm, so it is difficult to observe the welding situation of the inner cavity weld. This embodiment provides a method for scanning and tracking long welding seams in a narrow space, using a robot control system and an image recognition system, manually placing a calibration plate, and automatically identifying the welding position of structural parts by 3D vision to automatically find the welding position. , to meet the needs of welding small space components, can weld the groove welds of multi-cavity steel components, and realize robot automatic welding operations.

Referring to accompanying drawing 1 of the description, the method specifically includes the following steps:

Step S1. First, manually lay the calibration plate at the position of the workpiece weld, and the vision sensor collects the real-time image of the workpiece and transmits the collected image to the host computer.

In this embodiment, the visual sensor is mainly composed of a CCD camera, a laser and a narrow-band filter. The CCD camera and the laser are integrated in a housing, the housing is set on the shaft arm of the welding robot, and the narrow-band filter is set in front of the CCD camera lens.

The CCD camera is vertically aligned with the workpiece, the laser is arranged obliquely, and the laser emits laser light, which is formed by a lens and finally irradiated onto the workpiece to form a narrow light band. When the light band is reflected or refracted by the workpiece, the light of a specific wavelength emitted by the laser is retained by the filter, and the light of other wavelengths is filtered out, and finally enters the lens of the CCD camera for imaging.

Step S2. The host computer performs image processing on the collected workpiece image, identifies the calibration plate in the image, obtains the type of workpiece and the type of welding seam, and finally determines the motion trajectory of the welding robot.

In this embodiment, the image processing method specifically includes image preprocessing, edge detection, and feature extraction.

Further, in this embodiment, the image preprocessing further includes one or more of digitization, geometric transformation, normalization, smoothing, restoration, and enhancement. The main purpose of image preprocessing is to eliminate irrelevant information in images, restore useful real information, enhance the detectability of relevant information and simplify data to the greatest extent, thereby improving the reliability of feature extraction, image segmentation, matching and recognition.

Further, in this embodiment, a basic database is stored inside the host computer, and the basic database includes calibration plate type data, workpiece data and welding robot motion trajectory data, calibration plate type data, workpiece data and welding robot trajectory data. Class data corresponds to matching.

Since a basic database is stored in the host computer in this application, when the host computer uses the above-mentioned image processing method to process the workpiece image and recognizes the type of the calibration plate at the position of the weld at this time, it will search for the same in the basic database. The matched workpiece data can finally get the workpiece type, weld type and the corresponding welding robot motion trajectory. That is to say, when the calibration plate placed at the weld position is identified, the corresponding matching can be found in the database. The workpiece type corresponding to the calibration plate, the welding seam type contained in the workpiece, and the motion trajectory of the welding robot corresponding to each welding seam. When ready to start welding, the host computer directly calls the motion trajectory of the welding robot corresponding to the welding seam of the workpiece in the basic database and combines the welding position information to control the welding robot to start welding directly.

In this embodiment, it should be noted that the methods used for image preprocessing, edge detection, and feature extraction are all conventional technical means known to those skilled in the art.

Step S3. Use the laser to scan the workpiece to obtain the original 3D point cloud of the workpiece. The 3D point cloud is three-dimensional. The original 3D point cloud obtained by scanning is preprocessed, and finally a 3D model of the workpiece is obtained to identify the workpiece and obtain the weld seam of the workpiece at the same time. location information.

In this embodiment, the original three-dimensional point cloud of the workpiece is preprocessed to finally obtain a three-dimensional model of the workpiece, thereby identifying the workpiece, which specifically includes the following processes and steps.

Cloud coordinate transformation: Solve the coordinate transformation matrix between the visual coordinate system and the robot coordinate system. Through matrix transformation, the acquired original 3D point cloud data is converted from the coordinate expression of the scanning device to the expression of the robot coordinate.

Point cloud slice: The acquired 3D point cloud is centrally sliced to obtain the main view slice point cloud, top view slice point cloud and left view slice point cloud of the workpiece.

Point cloud denoising: The above-mentioned main view slice point cloud, top view slice point cloud and left view slice point cloud are denoised respectively.

Point cloud segmentation: After obtaining the denoised point cloud slice, a gray threshold is set, and the target area and the background area are selected in the image according to the given threshold, so as to extract the rough outline of the target area.

In this embodiment, by dividing the threshold, the rough outline of the target area is extracted, and a clearer point cloud slice can be obtained.

Point cloud slice reorganization: Reorganize the main view slice point cloud, top view slice point cloud and left view slice point cloud, and finally get the 3D model of the workpiece.

In this embodiment, it should be noted that the above processing methods of the three-dimensional point cloud data are conventional technical means known to those skilled in the art.

In this embodiment, since the workpiece bevel is at different depths in the vertical direction of the workpiece, when viewed from the direction perpendicular to the workpiece, the reflected light is in the shape of a folded line, and the folded line reflects the three-dimensional positional relationship between the center of the light pattern and the middle of the weld groove, Finally, the image processing method is used to identify the features of the collected laser bar images.

In this embodiment, in image analysis, image threshold segmentation is commonly used, and it is also the simplest image segmentation method. The optimal threshold value is obtained by the segmentation method of the laser bar, and the edge of the weld image is obtained by filtering out interference signals such as electrical noise after smoothing filtering. A small amount of residual noise is filtered out, and the position of the weld seam in the image is found from the center line of the extracted laser band according to the characteristic of the largest change in the slope of the laser band at the weld seam, and finally the processed position information is sent to the welding robot.

Step S4. Compare the three-dimensional model of the workpiece with the actual workpiece. If the two are matched or within a certain error range, the upper computer transmits the motion trajectory instruction of the welding robot and the welding seam position information to the welding robot controller. , the welding robot controller controls the execution structure of the motion axis of the welding robot according to the above-mentioned instructions and position information, and finally realizes the welding of the workpiece.

In this embodiment, it should be noted that the specific error range is defined by the technicians themselves in the actual operation process.

In the description of this application, it should be understood that the terms "center", "portrait", "horizontal", "front", "rear", "left", "right", "vertical", "horizontal", The orientation or positional relationship indicated by "top", "bottom", "inner", "outer", etc. is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing the present application and simplifying the description, rather than indicating or implying The device or element referred to must have a specific orientation, be constructed and operate in a specific orientation, and therefore should not be construed as limiting the scope of protection of the present application.

In the description of this application, it should also be noted that, unless otherwise expressly specified and limited, the terms "arrangement", "installation" and "connection" should be understood in a broad sense, for example, it may be a fixed connection or a connectable connection. Detachable connection, or integral connection; may be mechanical connection or electrical connection; may be direct connection, or indirect connection through an intermediate medium, or internal communication between two components. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood in specific situations.

The above are only preferred embodiments of the present application, and are not intended to limit the present application in any form. Any simple modifications and equivalent changes made to the above embodiments according to the technical essence of the present application shall fall within the scope of the present application. within the scope of protection.

Claims (6)

1. a long welding seam scanning and tracking method in a narrow space, is characterized in that, comprising: Manually lay the calibration plate to the welding position of the workpiece, and the vision sensor collects the real-time image of the workpiece and transmits the image to the upper computer; The host computer performs image processing on the collected workpiece image, identifies the calibration plate in the image, obtains the type of workpiece and the type of weld, and finally determines the motion trajectory of the welding robot; Obtain the original 3D point cloud data of the workpiece, preprocess the original 3D point cloud data of the workpiece, and finally obtain the 3D model of the workpiece and the weld position of the workpiece; The three-dimensional model of the workpiece is compared with the actual workpiece. If the two match or are within the error range, the upper computer transmits the motion trajectory command of the welding robot and the position information of the welding seam to the welding robot controller. The instruction and the position information of the welding seam control the execution structure of the motion axis of the welding robot, and finally realize the welding of the workpiece. 2. the long welding seam scanning and tracking method in a kind of narrow space according to claim 1, is characterized in that, described vision sensor comprises CCD camera, laser and narrow-band filter, CCD camera and laser are integrated in a shell In the body, the shell is set on the shaft arm of the welding robot, and the narrow-band filter is set in front of the CCD camera lens. 3 . The method for scanning and tracking long welds in a narrow space according to claim 1 , wherein the image processing includes image preprocessing, edge detection and feature extraction. 4 . 4. The method for scanning and tracking long welds in a narrow space according to claim 3, wherein the image preprocessing comprises one of digitization, geometric transformation, normalization, smoothing, restoration and enhancement. one or more. 5. The method for scanning and tracking long welds in a narrow space according to claim 1, wherein the host computer performs image processing on the collected workpiece image, and identifies the calibration plate in the image, Obtain the type of workpiece and the type of welding seam, and finally determine the motion trajectory of the welding robot, including: The host computer stores a basic database, which includes calibration plate type data, workpiece data and welding robot trajectory data, calibration plate type data, workpiece data and welding robot trajectory data correspondingly matched; when the workpiece weld position is identified, place it After calibrating the plate type, find the matching workpiece data in the basic database, and finally get the workpiece type, welding seam type and the corresponding welding robot motion trajectory. 6. The method for scanning and tracking long welds in a narrow space according to claim 1, wherein the preprocessing of the original three-dimensional point cloud data of the workpiece comprises: Point cloud coordinate transformation: Solve the coordinate transformation matrix between the visual coordinate system and the robot coordinate system, and convert the acquired original 3D point cloud data from the coordinates of the scanning device to the coordinates of the robot; Point cloud slicing: centrally slice the acquired 3D point cloud to obtain the main view slice point cloud, top view slice point cloud and left view slice point cloud of the workpiece; Point cloud denoising: respectively denoising the main view slice point cloud, top view slice point cloud and left view slice point cloud of the above workpiece; Point cloud segmentation: After obtaining the denoised point cloud slice, set a grayscale threshold, select the target area and background area, and extract the rough outline of the target area; Point cloud slice reorganization: Reorganize the main view slice point cloud, top view slice point cloud and left view slice point cloud, and finally get the 3D model of the workpiece.

Images