Planar short weld seam recognition method, welding method and device

Abstract

The present application relates to a kind of plane short weld seam identification method, welding method and device, belong to the field of automated welding;Through the two-step photographing method of single camera autonomous welding model, accurately identify the three-dimensional coordinates of spatial point under the robot base coordinate system, give the autonomy of robot welding non-standard plane weld or batch plane weld;YOLOv5 neural network is used to locate the arc starting point, arc extinguishing point, workpiece corner point and weld area, improve the perception ability of robot to space target;Through brightness and contrast enhancement algorithm, edge detection algorithm, morphological processing algorithm, accurately extract weld point cloud from weld area;Through stereo matching algorithm and B spline curve fitting algorithm, identify and fit welding trajectory from irresistible, dynamic, irregular interference information;The fitted welding trajectory is transmitted to robot system to carry out autonomous welding on plane weld.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of automated welding, in particular to a planar short weld seam recognition method, a welding method and a device. BACKGROUND

[0002] The manual teaching welding form cannot adapt to the changing external environment, and it is impossible to automatically correct the welding trajectory error that may occur under non-ideal conditions. It is rapidly changing towards autonomous welding form. Autonomous welding is difficult to achieve because it is difficult for robots to abandon the identification, programming and other processes of the teaching trajectory, and it is difficult for robots to dynamically and flexibly perceive weld seam information like humans, such as arc starting point position, arc extinguishing point position and weld seam area position. For this reason, mechanical sensors, arc sensors, infrared sensors, ultrasonic sensors, electromagnetic sensors and optical sensors are widely used in robot welding systems. At present, the main forms of intelligent welding are sensor-assisted real-time tracking welding, pre-weld scanning trajectory recognition welding and pre-weld photographing trajectory recognition welding.

[0003] Sensor-assisted real-time tracking welding form has the advantage of efficient welding for long straight welds and large quantities of standard welds. However, it has the following technical deficiencies for planar short welds: 1. As the weld length becomes shorter, the efficiency of this welding method becomes lower, because when the weld length is less than the pre-positioning distance of the sensor, the tracking welding function will fail; 2. Tracking welding is not completely autonomous operation that completely separates from the teaching trajectory, it needs the guidance of the teaching trajectory, so when the workpiece placement error is too large, tracking welding cannot be effectively carried out; 3. Tracking welding mainly controls the trajectory of the welding torch during welding, but it lacks the identification and positioning of the arc starting point and the arc extinguishing point.

[0004] Sensor-assisted pre-weld scanning trajectory recognition welding form can enable the robot to autonomously identify the welding trajectory before welding to some extent, without relying on the teaching playback method, but it needs to edit the sensor scanning trajectory in advance, which is very inconvenient and inefficient for the robot welding system. Sensor-assisted pre-weld photographing trajectory recognition welding form mainly uses binocular vision assistance, which improves the degree of autonomy of planar short weld welding to some extent, but still has problems: 1. Binocular vision sensor is a complex composite sensor with a very complex internal structure; it is also very difficult to determine the spatial relationship between the two cameras inside the sensor; 2. The lack of a highly adaptive weld seam stripe extraction algorithm under complex environmental light is the main reason why binocular vision cannot be widely used in actual welding; 3. There are a lot of noise, reflection, bright spots and dark spots at the arc starting point and the arc extinguishing point, which makes it easy for conventional weld seam stripe extraction algorithms to get incomplete weld seam stripes; the traditional binocular vision-assisted welding method does not determine the positioning method of the arc starting point and the arc extinguishing point, making it difficult to meet the requirements of high-precision welding.

[0005] Therefore, planning the welding track of the flat weld is always the key of the autonomous welding of the robot, however, the recognition and positioning of the weld, the starting and extinguishing points and the programming of the welding track are often realized by the inefficient manual teaching, the starting and extinguishing points of the flat short weld cannot be simply and efficiently recognized and positioned, the welding track cannot be simply and efficiently fitted, and the intelligent and efficient welding is insufficient. SUMMARY

[0006] The present application aims to provide a flat short weld recognition method, which is realized by two-step photographing, and is especially suitable for the robot welding of the flat short weld, and can realize efficient welding, such as the flat short butt weld with a groove, the flat short butt weld without a groove, the flat short lap weld or the flat short fillet weld.

[0007] The above object of the present application is realized by the following technical scheme:

[0008] A flat short weld recognition method, comprising the following steps:

[0009] Step one, the global photographing of the workpiece in two random postures is realized by using a single camera installed on the robot, the camera coordinates of a space point in the first photographing posture and the camera coordinates of the point in the second photographing posture are substituted into the system calibration equation, and the three-dimensional coordinates of the point in the robot base coordinate system are solved;

[0010] The system calibration equation is as follows:

[0011]

[0012] In the formula, (X B ,Y B ,Z B ) are the coordinates of a space point in the robot base coordinate system O B X B Y B Z B under the first photographing posture and the second photographing posture; is the conversion matrix of the tool coordinate system O H1 X H1 Y H1 Z H1 to the robot base coordinate system O B X B Y B Z B under the first photographing posture; is the conversion matrix of the tool coordinate system O H2 X H2 Y H2 Z H2 to the robot base coordinate system O B XB Y B Z B The transformation matrix; (X C1 ,Y C1 Z C1 Let O be a spatial point in the camera coordinate system under the first shooting posture. C1 X C1 Y C1 Z C1 The coordinates below; Camera coordinate system O C1 X C1 Y C1 Z C1 To tool coordinate system O H1 X H1 Y H1 Z H1 The transformation matrix; (X C2 ,Y C2 Z C2 () is a spatial point in the camera coordinate system O under the second shooting posture. C2 X C2 Y C2 Z C2 The coordinates below; Camera coordinate system O C2 X C2 Y C2 Z C2 To tool coordinate system O H2 X H2 Y H2 Z H2 The transformation matrix.

[0013] Step 2: Identify and fit the welding trajectory:

[0014] Extract the arc initiation point, arc extinguishing point, weld area, and four corner points of the workpiece; and assign the coordinates of the corner points to the camera coordinate system under the first shooting posture. Coordinates in the camera coordinate system under the second shooting posture Substitute the stereo matching relation to obtain the stereo matching relation T between the two images. R ;

[0015] The stereo matching relationship is as follows:

[0016]

[0017] After extracting the weld stripe image, a morphological processing method involving erosion followed by dilation is applied to the weld stripe image. The weld stripe point cloud in the first captured image after processing is then described as follows: The weld seam stripe point cloud in the image taken in the second shooting posture is Set the camera coordinate system O C1 XC1 Y C1 Z C1 Down The coordinates of each point are obtained through the 3D matching relationship T R Find the corresponding stereo matching point cloud. Set the camera coordinate system O C2 X C2 Y C2 Z C2 Down The coordinates of each point are obtained through the 3D matching relationship T R Find the corresponding stereo matching point cloud.

[0018] Step 3: Obtain the welding trajectory;

[0019] Will Each point in camera coordinate system O C1 X C1 Y C1 Z C1 The coordinates below and Each point in camera coordinate system O C2 X C2 Y C2 Z C2 Substitute the coordinates below into the system calibration equation to obtain the welding trajectory F1; Each point in camera coordinate system O C2 X C2 Y C2 Z C2 The coordinates below and Each point in camera coordinate system O C1 X C1 Y C1 Z C1 Substitute the coordinates below into the system calibration equation to obtain the welding trajectory F2; obtain the center curve F of the welding trajectory F1 and welding trajectory F2. 中 and the central curve F 中 Perform B-spline curve fitting; the trajectory curve F after B-spline curve fitting 焊 It was identified as the final welding trajectory.

[0020] As a preferred technical solution of the present invention, the process of solving the system calibration equation is as follows:

[0021] The calibration equation of the step system is rewritten as Ax = b; the coefficient matrix A of Ax = b is sampled to construct a system of 6 standard linear equations A i x = b i (i = 1, 2, 3, 4, 5, 6); Solve each system of linear equations A i x = b i The standard solution x of (i = 1, 2, 3, 4, 5, 6)i (i = 1, 2, 3, 4, 5, 6); calculate Chebyshev residual r i (x) =‖Ax i -b‖ ∞ ; identify the solution corresponding to the minimum value of r i (x) absolute value as the system calibration equation solution.

[0022] As a more optimal technical solution of the application, the extraction process of the weld stripe image is as follows:

[0023] The weld area is subjected to brightness and contrast enhancement, and the brightness and contrast enhancement function is:

[0024] g(u,v) = α·f(u,v) + β

[0025] In the formula, α is a gain, α > 0; β is a pixel deviation parameter; f(u,v) is a pixel value of an input image; and g(u,v) is a pixel value of an output image.

[0026] The weld stripe is detected as:

[0027] s(u,v) = |Δ x f| + |Δ y f|

[0028] In the formula, s(u,v) is a gradient of a pixel point; Δ x f is a horizontal edge pixel gradient; and Δ y f is a vertical edge pixel gradient.

[0029] As a more optimal technical solution of the application, the establishment process of the system calibration equation is as follows:

[0030] The camera coordinate system is converted to the tool coordinate system;

[0031]

[0032] In the formula, (X C1 ,Y C1 ,Z C1 ) is a coordinate of a space point in the camera coordinate system O C1 X C1 Y C1 Z C1 under the first shooting posture; (X H1 ,Y H1 ,Z H1 ) is a coordinate of the point in the tool coordinate system O H1 X H1 Y H1 Z H1 under the first shooting posture; is a coordinate of the point in the camera coordinate system OC1 X C1 Y C1 Z C1 to tool coordinate system O H1 X H1 Y H1 Z H1 .

[0033]

[0034] where (X C2 , Y C2 , Z C2 ) are the coordinates of a space point in the camera coordinate system O C2 X C2 Y C2 Z C2 at the second photographing posture; (X H2 , Y H2 , Z H2 ) are the coordinates of the point in the tool coordinate system O H2 X H2 Y H2 Z H2 at the second photographing posture; is the conversion matrix from the camera coordinate system O C2 X C2 Y C2 Z C2 to the tool coordinate system O H2 X H2 Y H2 Z H2 .

[0035] The tool coordinate system is converted to the robot base coordinate system;

[0036]

[0037] where (X B , Y B , Z B ) are the coordinates of a space point in the robot base coordinate system O B X B Y B Z B at the first photographing posture and at the second photographing posture; is the conversion matrix from the tool coordinate system O H1 X H1 Y H1 Z H1 at the first photographing posture to the robot base coordinate system O B X B Y B Z B . The transformation matrix of the tool coordinate system O H2 X H2 Y H2 Z H2 to the robot base coordinate system O B X B Y B Z B .

[0038] The system calibration equation is obtained according to the above two transformations.

[0039] The application also aims to provide a planar short weld seam welding device, comprising a robot 3, a welding gun 5 fixed on the robot 3, a camera 4 fixed on the welding gun 5 and a workpiece 7 to be welded placed on a welding platform 6.

[0040] The welding process of the planar short weld seam welding device is as follows:

[0041] The camera can take pictures of the whole workpiece in each shooting posture, the robot 3 is set in a first shooting posture to take pictures of the workpiece 7, and the robot 3 is set in a second shooting posture to take pictures of the workpiece 7, so as to locate the arc starting point and the arc extinguishing point, identify and fit the welding trajectory; the final welding trajectory is obtained through the above planar short weld seam identification method, and finally the final welding trajectory, the arc starting point position and the arc extinguishing point position are transmitted to the robot system, and the robot drives the welding gun to weld.

[0042] The application also aims to provide a planar short weld seam welding method, wherein the final welding trajectory is obtained through the above planar short weld seam identification method, and then the final welding trajectory, the arc starting point position and the arc extinguishing point position are transmitted to the robot system, and the robot drives the welding gun to weld.

[0043] The beneficial effects are as follows:

[0044] The application realizes autonomous welding through the two-step shooting method of a single camera, so that the robot can autonomously locate and identify the welding trajectory of the planar weld seam, improve the autonomy and flexibility of the robot welding, and the assembly position and assembly precision of the workpiece are not strictly constrained, and the welding trajectory can be identified only by a single and simple camera without the aid of complex sensors, thereby assisting the robot to realize autonomous welding. BRIEF DESCRIPTION OF DRAWINGS

[0045] The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this application, illustrate embodiments of the application and serve to explain the principles of the application, and do not limit the application in any way.

[0046] Figure 1For the welding device of the application, wherein: 1, the first shooting posture; 2, the second shooting posture; 3, the robot; 4, the camera; 5, the welding torch; 6, the welding platform; 7, the workpiece to be welded; 8, the weld.

[0047] Figure 2 For the workpiece information recognition result in embodiment 1 of the application; from left to right, the photo taken in the first shooting posture, the photo taken in the second shooting posture, A is the corner point, B is the row weld area, C is the arc starting point, and D is the arc extinguishing point.

[0048] Figure 3 For the schematic diagram of extracting the weld stripe in embodiment 1 of the application; from left to right, the weld area, the brightness and contrast enhancement, the weld stripe detection, and the morphological processing.

[0049] Figure 4 For the weld stereo matching effect diagram in embodiment 1 of the application.

[0050] Figure 5 For the welding trajectory calculation effect diagram of the plane short weld in embodiment 1 of the application. DETAILED DESCRIPTION

[0051] The technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative work fall within the protection scope of the application. In order to make the above-mentioned purposes, features and advantages of the application more obvious and easy to understand, the application will be further described in detail below with reference to the drawings and specific embodiments.

[0052] The application first proposes a two-step shooting method autonomous welding model of a single camera, which can accurately identify the three-dimensional coordinates of a space point in the robot base coordinate system, and gives the robot the autonomy to weld non-standard plane welds or batch plane welds. Secondly, the YOLOv5 neural network is used to locate the arc starting point, arc extinguishing point, workpiece corner point and weld area, greatly improving the robot's perception ability of space targets. Then, the brightness and contrast enhancement algorithm, edge detection algorithm and morphological processing algorithm are used to accurately extract the weld point cloud from the weld area. Through the stereo matching algorithm and B-spline curve fitting algorithm, the welding trajectory is identified and fitted from the irresistible, dynamic and irregular interference information. Finally, the fitted welding trajectory is transmitted to the robot system for autonomous welding of the plane weld.

[0053] The welding method provided by the application includes three modules: S1, establishing a two-step shooting method autonomous welding model; S2, identifying and fitting the welding trajectory; S3, autonomous welding.

[0054] S1, establishing a two-step photographing method autonomous welding model;

[0055] S1.1, model establishment;

[0056] The first photographing posture and the second photographing posture are arbitrarily set, and it is ensured that the camera can photograph the whole global of the workpiece under each photographing posture. The robot 3 photographs the workpiece under the first photographing posture and under the second photographing posture, and a two-step photographing method autonomous welding model is established.

[0057] S1.2, system calibration equation establishment;

[0058] The purpose of establishing the calibration equation is to convert the coordinates of a space point in the camera coordinate system under two postures into three-dimensional coordinates in the base coordinate system which can be recognized by the robot.

[0059] S1.2.1, conversion from the camera coordinate system to the tool coordinate system;

[0060]

[0061] In the formula: (X C1 ,Y C1 ,Z C1 ) is the coordinate of a space point in the camera coordinate system O C1 X C1 Y C1 Z C1 under the first photographing posture; (X H1 ,Y H1 ,Z H1 ) is the coordinate of the point in the tool coordinate system O H1 X H1 Y H1 Z H1 under the first photographing posture; is the conversion matrix from the camera coordinate system O C1 X C1 Y C1 Z C1 to the tool coordinate system O H1 X H1 Y H1 Z H1 .

[0062]

[0063] In the formula: (X C2 ,Y C2 ,Z C2 ) is the coordinate of a space point in the camera coordinate system O C2 X C2 Y C2 Z C2The coordinates below; (X H2 ,Y H2 Z H2 () represents the tool coordinate system O of this point in the second shooting posture. H2 X H2 Y H2 Z H2 The coordinates below; Camera coordinate system O C2 X C2 Y C2 Z C2 To tool coordinate system O H2 X H2 Y H2 Z H2 The transformation matrix.

[0064] S1.2.2 Transformation from tool coordinate system to robot base coordinate system

[0065]

[0066] In the formula: (X B ,Y B Z B Let O be the robot's base coordinate system under the first and second photographing postures. B X B Y B Z B The coordinates below; Tool coordinate system O in the first shooting posture H1 X H1 Y H1 Z H1 To the robot's base coordinate system O B X B Y B Z B The transformation matrix; Tool coordinate system O in the second shooting posture H2 X H2 Y H2 Z H2 To the robot's base coordinate system O B X B Y B Z B The transformation matrix.

[0067] S1.2.3 Solving the system calibration equations;

[0068] From steps S1.2.1 and S1.2.2, the system calibration equations are as follows:

[0069]

[0070] S1.3 Solving the system calibration equations

[0071] The system calibration equation is an over-determined equation, and cannot be solved by a conventional equation solving method (X B ,Y B ,Z B ), therefore, the present application sets an over-determined equation solving algorithm based on Chebyshev thought, which is divided into 5 steps:

[0072] S1.3.1, rewriting the system calibration equation obtained in step S1.2.3 as Ax=b;

[0073] S1.3.2, sampling the coefficient matrix A of Ax=b to construct 6 standard linear equations

[0074] A i x=b i (i=1, 2, 3, 4, 5, 6);

[0075] S1.3.3, solving the standard solution of each linear equation A i x=b i (i=1, 2, 3, 4, 5, 6)

[0076] x i (i=1, 2, 3, 4, 5, 6);

[0077] S1.3.4, calculating the Chebyshev residual r i (x)=‖Ax i -b‖ ∞ ;

[0078] S1.3.5, identifying the solution corresponding to the minimum value of the absolute value of r i (x) as the solution of the system calibration equation.

[0079] Therefore, by substituting the coordinates (X C1 ,Y C1 ,Z C1 ) of a space point in the first photographing posture and the coordinates (X C2 ,Y C2 ,Z C2 ) of the point in the second photographing posture into the system calibration equation, the three-dimensional coordinates of the point in the robot base coordinate system O B X B Y B Z B can be obtained through the over-determined equation solving algorithm based on Chebyshev thought.

[0080] S2, identifying and fitting a welding trajectory;

[0081] S2.1, extracting information of interest of a workpiece;

[0082] YOLOv5 is used to recognize the workpiece information of interest on the two photographed images; the workpiece information of interest includes four workpiece corner points, an arc starting point, an arc extinguishing point, and a weld seam region, as shown in Figure 2 .

[0083] S2.2, extract the weld seam stripes, as shown in Figure 3 ;

[0084] S2.2.1, perform brightness and contrast enhancement on the weld seam region;

[0085] The brightness and contrast enhancement function is:

[0086] g(u,v) = a f(u,v) + b

[0087] In the formula, a is the gain, a > 0; b is the pixel deviation parameter; f(u,v) is the pixel value of the input image; g(u,v) is the pixel value of the output image.

[0088] S2.2.2, weld seam stripe detection;

[0089] Sobel algorithm is used to detect the weld seam stripes:

[0090] s(u,v) = |D x x f| + |D y y f|

[0091] In the formula, s(u,v) is the gradient of the pixel point; D x x f is the horizontal edge pixel gradient; D y y f is the vertical edge pixel gradient.

[0092] S2.2.3, morphological processing of weld seam stripes;

[0093] The weld seam stripe image is subjected to morphological processing of erosion first and then dilation.

[0094] S2.3, weld seam stereo matching

[0095] S2.3.1, obtaining stereo matching relationship

[0096] The stereo matching relationship between the two images is as follows:

[0097]

[0098] The coordinates of the four workpiece corner points obtained in step S2.1 in the camera coordinate system under the first photographing posture and the coordinates in the camera coordinate system under the second photographing posture are substituted into the above equation to obtain the stereo matching relationship T R between the two images.

[0099] S2.3.2, Weld seam point cloud 3D matching, see [link / reference] Figure 4 As shown;

[0100] Let the weld seam stripe point cloud in the image under the first photographic posture after processing in step S2.2.3 be... The weld seam stripe point cloud in the image taken in the second shooting posture is

[0101] Set the camera coordinate system O C1 X C1 Y C1 Z C1 Down The coordinates of each point are obtained through the 3D matching relationship T R Find the corresponding stereo matching point cloud. Set the camera coordinate system O C2 X C2 Y C2 Z C2 Down The coordinates of each point are obtained through the 3D matching relationship T R Find the corresponding stereo matching point cloud.

[0102] S2.4 Welding trajectory determination;

[0103] S2.4.1, Preliminary welding trajectory determination;

[0104] Will Each point in camera coordinate system O C1 X C1 Y C1 Z C1 The coordinates below and Each point in the camera coordinate system O C2 X C2 Y C2 Z C2 Substituting the coordinates below into the system calibration equations, the welding trajectory F1 can be obtained;

[0105] Will Each point in the camera coordinate system O C2 X C2 Y C2 Z C2 The coordinates below and Each point in the camera coordinate system O C1 X C1 Y C1 Z C1 The welding trajectory F2 can be obtained by substituting the coordinates below into the system calibration equation.

[0106] S2.4.2 Calculation of the final welding trajectory;

[0107] The center curve F of the welding trajectory F1 and the welding trajectory F2 is obtained 中 , and B-spline curve fitting is performed on the center curve F 中 . The trajectory curve F 焊 after B-spline curve fitting is identified as the final welding trajectory, as Figure 5 shown.

[0108] S3, autonomous welding;

[0109] The final welding trajectory F 焊 is transmitted to the robot system together with the arc striking point position and the arc extinguishing point position to realize autonomous welding of the robot planar short weld.

[0110] As can be seen from the above embodiments, the welding method provided by the present application realizes simple and efficient identification and positioning of the arc striking point and the arc extinguishing point of the planar short weld and fitting of the welding trajectory, and is extremely suitable for welding applications of the planar short weld in the welding field.

[0111] In addition to the above method, the present application also proposes a planar short weld welding device, as shown in Figure 1 , comprising a robot 3, a camera 4, a welding platform 6 and a piece to be welded 7; the camera 4 is fixed on the welding torch 5 and integrated with the welding torch 5; the welding torch 5 is at the end of the robot 3; the piece to be welded 7 is placed on the welding platform 6 at will; the weld 8 on the piece to be welded 7 can be a beveled planar short butt weld, a non-beveled planar short butt weld, a planar short lap weld, a planar short fillet weld, etc.; the robot 3 takes pictures of the piece to be welded 7 on the welding platform 6 through the first and second photographing postures set at will to locate the arc striking point, locate the arc extinguishing point and identify and fit the welding trajectory.

[0112] It should be understood that the specific embodiments of the present disclosure are described with reference to the accompanying drawings; however, they are presented for illustrative purposes only, and the present disclosure is not limited thereto. As Figure 1 shown, the piece to be welded 7 is only for illustration; in actual applications, such as shipbuilding applications, the structure of the welding object can be quite complex; however, based on similar principles and operations, the welds of complex structures can also be identified. In addition, the welding object is illustrated as a combination of cuboids; however, the present disclosure is not limited thereto, and the welding object can include various combinations of various objects, such as cylinders, cones, spheres, hemispheres or any other shapes or combinations thereof, and thus those skilled in the art can also identify the welds thereof from the teachings provided herein.

[0113] Those skilled in the art will further appreciate that the solution provided herein can be implemented in software, hardware, firmware or any combination thereof. As such, the term "implementation" (or the like, such as "implementation(s)" or "implementation(s)") as used herein is intended to encompass a microprocessor, a digital signal processor, a simple chip machine, a programmed processor, etc., in combination with appropriate software, etc.

[0114] The foregoing detailed description of implementations has been presented for purposes of illustration and description. It is understood that the description is not intended to limit the inventions to the forms disclosed herein. Many modifications and variations (including, without limitation, those substituting equivalents to keep within the scope of the inventions or the like) will be apparent to those skilled in the art. It is therefore understood that other implementations can be used and / or practiced the teachings described herein. Accordingly, the disclosure is not intended to be limited to the examples described herein; rather, the true scope is defined by the appended claims along with their full scope of equivalents.

[0115] Various modifications to the foregoing exemplary implementations can and will be apparent to those skilled in the art in view of the foregoing description, wherein like numerals indicate like elements, and wherein but a few implementations have been described. Any and all modifications or variations that may depend from the above will still fall within the scope of the non-limiting and exemplary implementations of the disclosure. Further, any features described herein can be implemented in hardware, software, firmware or any combination thereof. Moreover, any features described herein can be implemented as a system, method, apparatus or article of manufacture using standard devices.

Claims

1. A method for flat short weld recognition, characterized in that It comprises the following steps: Step one, using a single camera mounted on the robot to take pictures of the workpiece in two random poses, and substituting the camera coordinates of a space point in the first photographing pose and the camera coordinates of the point in the second photographing pose into the system calibration equation to obtain the three-dimensional coordinates of the point in the robot base coordinate system; The system calibration equation is as follows: wherein: (X B ,Y B ,Z B ) are coordinates of a space point in the first photographing posture and the second photographing posture in the robot base coordinate system O B X B Y B Z B ; is a transformation matrix of the tool coordinate system O H1 X H1 Y H1 Z H1 to the robot base coordinate system O B X B Y B Z B in the first photographing posture; is a transformation matrix of the tool coordinate system O H2 X H2 Y H2 Z H2 to the robot base coordinate system O B X B Y B Z B in the second photographing posture;(X C1 ,Y C1 ,Z C1 ) are coordinates of a space point in the first photographing posture in the camera coordinate system O C1 X C1 Y C1 Z C1 ; is a transformation matrix of the camera coordinate system O C1 X C1 Y C1 Z C1 to the tool coordinate system O H1 X H1 Y H1 Z H1 ;(X C2 ,Y C2 ,Z C2 ) are coordinates of a space point in the second photographing posture in the camera coordinate system O C2 X C2 Y C2 Z C2 ; is a transformation matrix of the camera coordinate system O C2 X C2 Y C2 Z C2 to the tool coordinate system O H2 X H2 Y H2 Z H2 . Step two, identifying and fitting the welding trajectory; Extract the arc starting point, arc extinguishing point weld area and four corner point information of the workpiece; the coordinates of the corner points in the camera coordinate system under the first shooting posture and the coordinates in the camera coordinate system under the second shooting posture Substitute the stereo matching relationship to obtain the stereo matching relationship T between the two images R ; The stereo matching relationship is as follows: After the weld seam stripe image is extracted, morphological processing of erosion first and then dilation is performed on the weld seam stripe image, so that the weld seam stripe point cloud on the image in the first photographing posture after processing is The weld seam stripe point cloud on the image in the second photographing posture is The camera coordinate system O C1 X C1 Y C1 Z C1 is established The coordinates of each point are obtained through the stereo matching relationship T R The corresponding stereo matching point cloud is obtained The camera coordinate system O C2 X C2 Y C2 Z C2 is established The coordinates of each point are obtained through the stereo matching relationship T R The corresponding stereo matching point cloud is obtained Step three, obtaining the welding trajectory; Will The coordinates of each point in the camera coordinate system O C1 X C1 Y C1 Z C1 The coordinates of each point in the camera coordinate system O The coordinates of each point in the camera coordinate system O C2 X C2 Y C2 Z C2 The coordinates of each point in the camera coordinate system O The coordinates of each point in the camera coordinate system O C2 X C2 Y C2 Z C2 The coordinates of each point in the camera coordinate system O The coordinates of each point in the camera coordinate system O C1 X C1 Y C1 Z C1 The coordinates of each point in the camera coordinate system O 中 The center curve F 中 of the welding track F1 and the welding track F2 is obtained, and the B-spline curve fitting is performed on the center curve F 焊 The track curve F 焊 after B-spline curve fitting is identified as the final welding track. The system calibration equation solving process is as follows: Rewrite the step system calibration equation as Ax = b; sample the coefficient matrix A of Ax = b to construct 6 standard linear equations A i x = b i (i = 1, 2, 3, 4, 5, 6); solve each linear equation group A i x = b i (i = 1, 2, 3, 4, 5, 6) standard solution x i (i = 1, 2, 3, 4, 5, 6); calculate Chebyshev residual r i (x) = ‖Ax i -b‖ ∞ ; the solution corresponding to the minimum value of r i (x) absolute value is identified as the system calibration equation solution; The extraction process of the weld seam stripe image is as follows: The brightness and contrast of the weld seam area are enhanced, and the brightness and contrast enhancement function is: g(u,v)=α·f(u,v)+β In the formula: alpha is the gain, alpha>0; beta is the pixel deviation parameter; f(u,v) is the pixel value of the input image; g(u,v) is the pixel value of the output image; The weld seam stripe is detected as: s(u,v) = |Δ x f|+|Δ y f| where s(u, v) is the gradient of the pixel; Δ x f is the lateral edge pixel gradient; Δ y f is the longitudinal edge pixel gradient.

2. The planar short weld bead recognition method of claim 1, wherein, The system calibration equation solving process is as follows: Convert the camera coordinate system to the tool coordinate system; wherein: (X H1 ,Y H1 ,Z H1 ) are the coordinates of the point in the tool coordinate system O H1 X H1 Y H1 Z H1 under the first photographing attitude; wherein: (X H2 ,Y H2 ,Z H2 ) is the coordinate of the point in the tool coordinate system O H2 X H2 Y H2 Z H2 under the second photographing posture; Convert the tool coordinate system to the robot base coordinate system; The system calibration equation is obtained according to the above two conversions.

3. A flat short weld welding apparatus, characterized by: It comprises a robot, a welding gun fixed on the robot, a camera fixed on the welding gun, and a workpiece to be welded placed on a welding platform; The welding process of the flat short weld seam welding device is as follows: Ensure that the camera can take pictures of the workpiece in each photographing pose, set the first photographing pose of the robot randomly, take pictures of the workpiece to be welded, set the second photographing pose of the robot randomly, take pictures of the workpiece to be welded, locate the arc starting point and arc extinguishing point, identify and fit the welding trajectory, obtain the final welding trajectory through the flat short weld seam identification method of claim 1, and finally transmit the final welding trajectory, the arc starting point position and the arc extinguishing point position to the robot system, and the robot drives the welding gun to weld.

4. A method of welding a flat short weld, characterized by: First, the final welding trajectory is obtained through the flat short weld seam identification method of claim 1, and then the final welding trajectory, the arc starting point position and the arc extinguishing point position are transmitted to the robot system, and the robot drives the welding gun to weld.

Images