CN117047237B - Intelligent flexible welding system and method for special-shaped parts - Google Patents
Intelligent flexible welding system and method for special-shaped parts Download PDFInfo
- Publication number
- CN117047237B CN117047237B CN202311307754.7A CN202311307754A CN117047237B CN 117047237 B CN117047237 B CN 117047237B CN 202311307754 A CN202311307754 A CN 202311307754A CN 117047237 B CN117047237 B CN 117047237B
- Authority
- CN
- China
- Prior art keywords
- welding
- point cloud
- weld
- welded
- special
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000003466 welding Methods 0.000 title claims abstract description 160
- 238000000034 method Methods 0.000 title claims abstract description 47
- 239000011159 matrix material Substances 0.000 claims abstract description 33
- 238000006243 chemical reaction Methods 0.000 claims abstract description 25
- 238000005070 sampling Methods 0.000 claims abstract description 12
- 238000001914 filtration Methods 0.000 claims abstract description 7
- 239000013598 vector Substances 0.000 claims description 14
- 238000004364 calculation method Methods 0.000 claims description 12
- 238000013135 deep learning Methods 0.000 claims description 9
- XKRFYHLGVUSROY-UHFFFAOYSA-N Argon Chemical compound [Ar] XKRFYHLGVUSROY-UHFFFAOYSA-N 0.000 claims description 6
- CURLTUGMZLYLDI-UHFFFAOYSA-N Carbon dioxide Chemical compound O=C=O CURLTUGMZLYLDI-UHFFFAOYSA-N 0.000 claims description 6
- 238000013136 deep learning model Methods 0.000 claims description 5
- 239000007789 gas Substances 0.000 claims description 4
- 229910052786 argon Inorganic materials 0.000 claims description 3
- 229910002092 carbon dioxide Inorganic materials 0.000 claims description 3
- 239000001569 carbon dioxide Substances 0.000 claims description 3
- 238000002372 labelling Methods 0.000 claims description 3
- 238000012216 screening Methods 0.000 claims description 2
- 230000007547 defect Effects 0.000 abstract description 2
- 238000004140 cleaning Methods 0.000 description 5
- 239000012636 effector Substances 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 238000004519 manufacturing process Methods 0.000 description 3
- 229910000831 Steel Inorganic materials 0.000 description 2
- 230000000295 complement effect Effects 0.000 description 2
- 238000000354 decomposition reaction Methods 0.000 description 2
- 239000010959 steel Substances 0.000 description 2
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000003754 machining Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 230000010355 oscillation Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B23—MACHINE TOOLS; METAL-WORKING NOT OTHERWISE PROVIDED FOR
- B23K—SOLDERING OR UNSOLDERING; WELDING; CLADDING OR PLATING BY SOLDERING OR WELDING; CUTTING BY APPLYING HEAT LOCALLY, e.g. FLAME CUTTING; WORKING BY LASER BEAM
- B23K9/00—Arc welding or cutting
- B23K9/16—Arc welding or cutting making use of shielding gas
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B23—MACHINE TOOLS; METAL-WORKING NOT OTHERWISE PROVIDED FOR
- B23K—SOLDERING OR UNSOLDERING; WELDING; CLADDING OR PLATING BY SOLDERING OR WELDING; CUTTING BY APPLYING HEAT LOCALLY, e.g. FLAME CUTTING; WORKING BY LASER BEAM
- B23K9/00—Arc welding or cutting
- B23K9/32—Accessories
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Plasma & Fusion (AREA)
- Mechanical Engineering (AREA)
- Numerical Control (AREA)
Abstract
The invention belongs to the technical field of robot welding, and particularly discloses an intelligent flexible welding system and method for special-shaped pieces, wherein the principle comprises the following steps: acquiring a three-dimensional point cloud model of the special-shaped piece to be welded; performing coordinate conversion and point cloud splicing on the three-dimensional point cloud model to obtain a plurality of complete weldment point clouds; acquiring a first weld joint position set of the special-shaped piece to be welded according to the point cloud of the complete welding piece by adopting a straight-through filtering method and a clustering algorithm, and acquiring a second weld joint position set according to the three-dimensional point cloud model after coordinate conversion by adopting a random sampling consistency fitting method; acquiring weld positions according to the first weld position set and the second weld position set; constructing a covariance matrix according to the weld point cloud to carry out path planning and pose planning; controlling a welding gun to perform welding operation according to the welding seam position, the welding path and the welding pose; the invention overcomes the defects of large workload and high labor intensity of the traditional manual welding, and is suitable for welding weldments with various specifications.
Description
Technical Field
The invention belongs to the technical field of robot welding, and particularly relates to an intelligent flexible welding system and method for special-shaped pieces.
Background
With the continuous breakthrough of intelligent manufacturing, robot welding has evolved rapidly. The welding robot has the advantages of high precision, good quality and high efficiency, and can greatly reduce labor cost, so that the welding robot becomes an unavoidable trend.
The traditional welding robot adopts a teaching programming mode, so that the welding seam track often deviates from the teaching track to influence the welding quality due to the influence of factors such as machining errors, clamping errors and the like; the existing vision guiding robot can accurately extract and track welding seams by additionally arranging a laser vision sensor; the latest 3D vision welding technology adopts a mode of matching a weldment point cloud and a template point cloud to realize welding, and for the current flexible and changeable weldment types, the two modes are relatively poor in welding speed and flexibility; the network frame needed by accurate positioning of the welding seam by deep learning is too complex, and the recognition time is too long, so that the production efficiency is affected.
Disclosure of Invention
The invention aims to provide an intelligent flexible welding system and method for special-shaped pieces, which are used for solving the problems in the prior art.
In order to achieve the above object, the present invention provides an intelligent flexible welding system for special-shaped pieces, comprising:
the 3D vision camera is used for acquiring a three-dimensional point cloud model of the special-shaped piece to be welded;
the weld joint identification positioning module is used for carrying out coordinate conversion and point cloud splicing on the three-dimensional point cloud model to obtain a plurality of complete weldment point clouds; acquiring a first weld joint position set of the special-shaped piece to be welded according to the complete weldment point cloud by adopting a straight-through filtering method and a clustering algorithm, and acquiring a second weld joint position set according to a three-dimensional point cloud model after coordinate conversion by adopting a random sampling consistency fitting method; acquiring weld positions according to the first weld position set and the second weld position set;
the weld path planning module is used for constructing a covariance matrix according to the weld point cloud and carrying out welding path planning according to the covariance matrix;
the welding pose planning module is used for planning welding poses according to the covariance matrix;
the robot is used for controlling the welding machine to perform welding operation according to the welding seam position, the welding path and the welding pose;
and the welding machine is used for carrying out welding operation on the special-shaped piece to be welded.
Optionally, a moving guide rail in sliding fit with the robot is further included.
Optionally, the shielding gas provided by the welder includes 80% carbon dioxide and 20% argon.
Optionally, the 3D vision camera is a structured light 3D camera, the light source is 450nm, the working distance is 350mm-700mm, the near-end view field is 285mm×192mm, the far-end view field is 515mm×373mm, the z-axis precision is 0.04-0.08mm, the resolution is 1280×1024, and soft triggering is performed through api.
On the other hand, in order to achieve the purpose, the invention provides an intelligent flexible welding method for special-shaped pieces, which comprises the following steps:
acquiring a three-dimensional point cloud model of the special-shaped piece to be welded;
performing coordinate conversion and point cloud splicing on the three-dimensional point cloud model to obtain a plurality of complete weldment point clouds;
screening the complete weldment point cloud based on the straight-through filtering, and extracting a weld point cloud based on a clustering algorithm;
acquiring a first weld joint position set of the special-shaped piece to be welded by adopting a deep learning method based on the complete weldment point cloud;
performing plane fitting on the three-dimensional point cloud model after coordinate conversion based on a random sampling consistency fitting method, obtaining a welding plane, obtaining two plane equations of the welding plane, solving the two plane equations, and obtaining a second weld joint position set;
acquiring a weld position based on the first weld position set and the second weld position set;
constructing a covariance matrix based on the weld point cloud and carrying out welding path planning and pose planning based on the covariance matrix;
and welding the special-shaped piece to be welded based on the welding path and the welding seam position.
Optionally, the process of obtaining the first weld joint position set of the special-shaped piece to be welded by adopting the deep learning method comprises the following steps:
performing format conversion and labeling on the weld point cloud;
inputting the marked weld point cloud into an OpenPCDet framework for training, and obtaining a deep learning model;
acquiring a recognition result of a part to be welded of the special-shaped piece to be welded based on the deep learning model;
and performing linear fitting on the recognition result of the part to be welded to obtain the first welding seam position set.
Optionally, the identifying result of the portion to be welded is a weld area point cloud approximate coordinate, and the process of performing straight line fitting on the identifying result of the portion to be welded includes:
selecting two groups of points from the recognition result of the part to be welded to calculate a linear equation, taking the rest points into the linear equation to calculate residual errors, constructing a residual error threshold, taking a point set with the residual errors smaller than the residual error threshold as internal points, recording, and repeatedly carrying out residual error calculation to obtain a plurality of groups of internal point sets;
and obtaining the points with the largest number of the inner points in the plurality of groups of inner point sets, and performing straight line fitting on the points with the largest number of the inner points based on a least square method to obtain the first welding seam position set.
Optionally, in the process of carrying out plane fitting on the three-dimensional point cloud model after coordinate conversion based on the random sampling consistency fitting method, marking a plane with the minimum plane centroid value and the upward normal as a first plane, intersecting other planes with the first plane to form a flat welding seam, and intersecting other planes to form a vertical welding seam.
Optionally, the process of acquiring the weld position based on the first set of weld positions and the second set of weld positions includes:
and performing cosine similarity calculation on the first welding seam position set and the second welding seam position set, performing distance calculation on a straight line with a cosine value of approximately 1 or-1, and performing matching based on a distance calculation result to obtain the welding seam position.
Optionally, the process of constructing a covariance matrix based on the weld point cloud and performing welding path planning and pose planning based on the covariance matrix includes:
performing point location selection on the weld point cloud, constructing a selected radius, performing decentration on all points in a spherical space formed by the selected points and the selected radius, and constructing the covariance matrix;
and calculating the covariance matrix, obtaining a characteristic value and a characteristic vector, taking the characteristic vector direction corresponding to the maximum characteristic value as a welding direction, and carrying out pose planning based on the characteristic vector corresponding to the minimum characteristic value.
The invention has the technical effects that:
the invention overcomes the defects of large workload and high labor intensity of the traditional manual welding, and the vision control robot is used for automatically completing the welding task, so that the cost is low and the efficiency is high.
The invention uses machine vision, solves the characteristics of low automation degree, poor flexibility and the like of the traditional teaching production, and effectively avoids the precision problem caused by workpiece placement errors, teaching errors and the like.
The invention avoids large errors caused by the fact that the pure deep learning network is used for recognizing the welding line, and simultaneously avoids the problem of overlong recognition time caused by the fact that the high-precision semantic division point cloud network is used.
The invention performs point cloud splicing through coordinate conversion, is suitable for identifying medium and large weldments, has higher flexibility compared with a point cloud template matching mode, and is suitable for welding weldments with various specifications.
Drawings
The accompanying drawings, which are included to provide a further understanding of the application, illustrate and explain the application and are not to be construed as limiting the application. In the drawings:
FIG. 1 is a block diagram of an intelligent welding system for special-shaped pieces in an embodiment of the invention;
fig. 2 is a schematic diagram of a system structure of intelligent welding according to an embodiment of the present invention, and reference numerals: 100-upper computer, 200-robot controller, 300-robot, 400-movable guide rail, 500-welder, 600-3D vision camera, 700-welding workbench and 800-automatic gun cleaning station;
FIG. 3 is a diagram of a weldment model in an embodiment of the invention;
FIG. 4 is a schematic diagram of a weld seam identification matching process in an embodiment of the invention;
fig. 5 is a schematic flow chart of an intelligent welding method in an embodiment of the invention.
Detailed Description
It should be noted that, in the case of no conflict, the embodiments and features in the embodiments may be combined with each other. The present application will be described in detail below with reference to the accompanying drawings in conjunction with embodiments.
It should be noted that the steps illustrated in the flowcharts of the figures may be performed in a computer system such as a set of computer executable instructions, and that although a logical order is illustrated in the flowcharts, in some cases the steps illustrated or described may be performed in an order other than that illustrated herein.
Example 1
As shown in fig. 1-4, the present embodiment provides a system and a method for intelligent flexible welding of a special-shaped piece.
The welding object comprises right-angle or right-angle-like welding seams such as multi-plane medium plate welding, square tube welding, H-shaped steel welding and the like; the upper computer 100 sends a welding instruction, the robot 300 controls the end effector to collect point clouds, the collected point clouds and the tail end pose of the robot 300 at the current moment are sent to the upper computer 100, and the upper computer 100 performs weld joint recognition positioning and plans a welding process, a welding path and a welding pose; then the robot 300 controls the end effector carrying the welding gun to weld, and after the welding is finished, the end effector is controlled to self-determined gun cleaning station to clean the gun.
As shown in fig. 2, the special-shaped part welding system provided by the invention comprises an upper computer 100, a robot controller 200, a robot 300, a movable guide rail 400, a welding machine 500, a 3D vision camera 600, a welding workbench 700 and an automatic gun cleaning station 800.
As shown in fig. 2, a moving rail 400 is used for the robot 300 to move on the moving rail 400. The robot controller 200 is configured to receive an instruction from the host computer 100 and control the robot 300 to execute the instruction. The robot 300 executes instructions of the host computer 100 under the control of the robot controller 200.
The upper computer 100 is provided with a welding seam identification positioning module, a welding process design module, a welding seam path planning module and a welding pose planning module.
The to-be-welded piece is placed on a welding workbench 700, a 3D vision camera 600 and a welding gun are arranged on an end effector, the end effector is arranged on a flange plate at the end of a robot 300, the welding machine 500 provides energy, shielding gas and welding materials, the upper computer 100 is connected with a robot controller 200 and the 3D vision camera 600 through an Ethernet, after a welding piece point cloud acquired by the 3D vision camera 600 is obtained, a welding process, a welding path and a welding pose are planned, obtained data are transmitted to the robot 300, and welding gun movement is realized.
In this embodiment, as shown in fig. 3, the workpiece to be welded is a multi-plane medium plate weldment, and has a length of 800mm, a width of 700mm, and a height of 500mm; the model of the welder 500 is an Artsen II PM/CM Myger Mi Tegong heavy-duty welder, the used shielding gas is carbon dioxide (80%) and argon (20%), and the welder 500 is connected with a welder power supply through a cable.
In the embodiment, the 3D vision camera 600 is a structured light 3D camera, the camera light source is 450nm, the working distance is 350mm-700mm, the near-end view field is 285mm×192mm, the far-end view field is 515mm×373mm, the z-axis precision is 0.04-0.08mm, the resolution is 1280×1024, and soft triggering is realized through api.
In the embodiment, the multi-plane medium plate weldment is placed on the welding table 700, and whether clamping by a clamp is needed or not is judged according to the weight of the weldment.
The specific implementation steps of the intelligent flexible welding method for the special-shaped piece in the embodiment are as follows:
step one: acquiring a multi-angle three-dimensional point cloud model of a weldment according to a preset angle, wherein the set robot angle can completely cover all view areas of the welding workbench 700, the acquired point cloud is in a pcd format, the point cloud is in an unordered point cloud type containing (X, Y and Z) coordinates, and then coordinate conversion is carried out to a base coordinate through a hand-eye matrix and a robot terminal coordinate system conversion matrix, wherein a point cloud coordinate conversion formula is as follows:
wherein (1)>The robot is a conversion matrix from the robot base coordinates to the tail end of the mechanical arm, the pose of the tail end of the robot is calculated by a Rodrigues formula, and the point cloud shot by each pose corresponds to a group of conversion matrices; />The hand-eye matrix from the tail end of the mechanical arm to the 3D camera is obtained through hand-eye calibration.
As shown in fig. 3, all weldments can be photographed and covered by adopting four directions of obliquely upward weldments.
And then the point clouds are spliced to form the complete weldment point clouds, and because the coordinate conversion is performed in advance, the plurality of weldment point clouds are all in a robot base coordinate system, and the precise registration such as ICP registration and the like is not needed, so that the welding is realized by simple superposition and splicing.
The weldment point cloud is roughly screened through the working area range set by the through filtering, the selection range is x-direction (-500, 500), y-direction (-400, 400), z-direction (-300, 300), the number of the point cloud is reduced through voxel downsampling, the voxel size is 1.2 multiplied by 1.2, and then the weldment point cloud is extracted through a clustering algorithm.
Converting the weld point cloud pcd format into a bin format, labeling by using a labelloud tool, inputting labeled point cloud data into a trained OpenPCDet framework, and training to obtain a deep learning model.
And acquiring the recognition result of the part to be welded of the weldment through the trained deep learning network, wherein the recognition result is the approximate coordinates of the point cloud of the welding line area. And determining the approximate position of the welding seam by carrying out straight line fitting on the point cloud of the welding seam area.
A straight line is fitted using a combination of random sample consensus (RANSAC) and least squares.
The specific steps of combining random sampling consistency (RANSAC) and a least square method are that firstly searching an interior point set, selecting two groups of points from point cloud data to calculate a linear equation, taking the rest points into the equation, calculating residual errors, taking the point set with the residual errors smaller than a threshold T as the interior points, recording the interior points, and repeating the steps to reach the cycle times to obtain a plurality of groups of interior point sets; and finding out the point with the largest number of the inner points in the step, and performing straight line fitting by a least square method to obtain a set A.
Step two: sequentially fitting planes to the integral point cloud model Q1 by using an improved random sampling consistency (RANSAC) plane fitting mode to obtain two plane equations, and setting a point cloud bounding box range according to the accuracy of the 3D camera shooting point cloud to ensure that the fitting plane cannot be over-fitted; because the weldment is placed on the horizontal workbench, the z value of the center of mass of the plane of the bottom medium plate is minimum, the point cloud normal of the bottom is upward and marked as a first plane, in addition, four planes are intersected with the first plane to form four flat welding seams, and the side planes form four vertical welding seams.
Then solving a two-to-two plane intersection equation and marking as a set B;
as shown in fig. 4, similarity matching is performed according to the above-mentioned weld line set a and set B, and the cosine similarity is specifically used for calculation, where the formula is:
and (3) performing distance calculation on two straight lines with cosine values of approximately 1 or-1, and completing matching when the distance is smaller than a fitting plane threshold value to obtain the accurate spatial position of the welding line straight line.
Step three: for the weld point cloud, first find all points V in the radius R sphere space around one of the points vi to calculate the mean, then de-center, and calculate the covariance as follows:
wherein->For the mean value, n is the number of samples, S 2 For sample variance, cov (V, V) is sample covariance.
Constructing a covariance matrix C according to the calculated covariance, and carrying out characteristic decomposition on the covariance matrix, wherein the formula is as follows:
wherein->，/>And P is a 3-order invertible matrix as a eigenvalue.
And calculating a covariance matrix of the point set, eigenvalues and eigenvectors, wherein one eigenvalue is obviously larger than the other two eigenvalues, and the corresponding eigenvector direction is the straight line direction.
As shown in fig. 3, a welding path is planned with the direction determined in step three as the welding direction.
And the feature vector corresponding to the minimum feature value is the normal vector of the point, the normal vector of the planes at the two sides of the welding line is calculated, and the complementary angle of the included angle between the welding gun and the first plane is determined to be kept between 10 and 15 degrees.
The welding machine parameters adopt 1.2mm solid welding wires, the welding current is selected to be 200A, the welding voltage is selected to be 20V, the welding speed is selected to be 80mm/min, the welding angle is selected to be 15 degrees, and sinusoidal oscillation is selected.
Step four: the robot 300 controls the tip welding gun to weld according to the planned welding path and pose. After welding, the moving guide rail 400 drives the robot 300 to move to the automatic gun cleaning station 800 for gun cleaning.
Example two
As shown in fig. 5, another implementation manner of the intelligent flexible welding method for the special-shaped piece is provided in the embodiment, which includes the following steps:
step 1: the weld joint is identified and roughly positioned through a deep learning network;
step 2: accurately positioning the welding line;
step 3: planning a welding process, a welding path and a welding pose;
step 4: the robot 300 controls the tip welding gun to weld according to the planned welding path and pose.
Step 1.1: the robot 300 controls the mechanical arm to shoot according to the set position and angle, the shooting range covers the whole welding area, multi-angle weldment point cloud data { P1, P2, & gt, pn } are collected, the weldment point cloud is converted into a robot base coordinate through coordinate conversion, and the whole weldment point cloud model Q1 is formed through splicing;
the point cloud coordinate conversion formula is:
wherein->The robot is a conversion matrix from the robot base coordinates to the tail end of the mechanical arm, the pose of the tail end of the robot is calculated by a Rodrigues formula, and the point cloud shot by each pose corresponds to a group of conversion matrices; />The hand-eye matrix from the tail end of the mechanical arm to the 3D camera is obtained through hand-eye calibration.
Step 1.2: and inputting the converted point cloud Q1 into an OpenPCDet of a point cloud deep learning frame trained in advance, and identifying a welding line area to obtain a point cloud set Q2.
Step 1.3: fitting a straight line to points in the weld point cloud Q2 by combining random sampling consistency (RANSAC) and a least square method to preliminarily obtain a weld straight line set A, and further:
the specific steps of combining random sampling consistency (RANSAC) and a least square method are that firstly searching an interior point set, selecting two groups of points from point cloud data to calculate a linear equation, taking the rest points into the equation, calculating residual errors, taking the point set with the residual errors smaller than a threshold T as the interior points, recording the interior points, and repeating the steps to reach the cycle times to obtain a plurality of groups of interior point sets; and finding out the point with the largest number of the inner points in the step, and performing straight line fitting by a least square method.
Step 2.1: sequentially fitting planes to the integral point cloud model Q1 by using an improved random sampling consistency (RANSAC) plane fitting mode, and setting a point cloud bounding box range according to the accuracy of the 3D camera shooting point cloud to ensure that the fitting plane cannot be over-fitted; because the weldment is placed on the horizontal workbench, the z value of the plane centroid of the bottom surface steel plate is minimum, the normal line of the bottom surface point cloud is upward and marked as a first plane, in addition, the plane and the first plane are intersected to form a flat welding seam, and the side plane forms a vertical welding seam.
Step 2.2: solving the intersecting line equation of every two planes and recording as a set B.
Step 2.3: according to the weld line set obtained in the step 1.3 and the step 1.5, the cosine similarity is adopted for calculation, and the formula is as follows:
wherein (1)>。
And (3) performing distance calculation on two straight lines with cosine values of approximately 1 or-1, and completing matching when the distance is smaller than a fitting plane threshold value to obtain the accurate spatial position of the welding line straight line.
Step 3.1: for the weld point cloud, first find all points V in the radius R sphere space around one of the points vi to calculate the mean, then de-center, and calculate the covariance as follows:
wherein->For the mean value, n is the number of samples, S 2 For sample variance, cov (V, V) is sample covariance.
Constructing a covariance matrix C according to the calculated covariance, and carrying out characteristic decomposition on the covariance matrix, wherein the formula is as follows:
wherein->，/>And P is a 3-order invertible matrix as a eigenvalue.
And calculating a covariance matrix of the point set, characteristic values and characteristic vectors, wherein one characteristic value is obviously larger than the other two characteristic values, and the corresponding characteristic vector direction is the straight line direction of the welding line.
Step 3.2: and (3) planning a welding path by taking the intersection points of the plurality of straight lines as starting points and taking the direction obtained in the step (3.1) as a welding direction.
Step 3.3: and 3.1, taking the feature vector corresponding to the minimum feature value in the step 3.1 as the normal vector of the point, solving the normal vector of the planes at two sides of the welding line, and determining that the complementary angle of the included angle between the welding gun and the ground is kept between 5 and 15 degrees.
The foregoing is merely a preferred embodiment of the present application, but the scope of the present application is not limited thereto, and any changes or substitutions easily contemplated by those skilled in the art within the technical scope of the present application should be covered by the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.
Claims (4)
1. An intelligent flexible welding system for special-shaped pieces, which is characterized by comprising:
a 3D vision camera (600) for acquiring a three-dimensional point cloud model of the special-shaped piece to be welded;
the weld joint identification positioning module is used for carrying out coordinate conversion and point cloud splicing on the three-dimensional point cloud model to obtain a plurality of complete weldment point clouds; acquiring a first weld joint position set of the special-shaped piece to be welded according to the complete weldment point cloud by adopting a straight-through filtering method and a clustering algorithm, and acquiring a second weld joint position set according to a three-dimensional point cloud model after coordinate conversion by adopting a random sampling consistency fitting method; acquiring weld positions according to the first weld position set and the second weld position set;
the welding line path planning module is used for constructing a covariance matrix according to the welding line point cloud and carrying out welding path planning according to the covariance matrix, wherein the welding line point cloud is generated based on weldment point cloud filtering and clustering;
the welding pose planning module is used for planning welding poses according to the covariance matrix;
a robot (300) for controlling a welder (500) to perform welding operation according to the weld position, the welding path and the welding pose;
the welding machine (500) is used for carrying out welding operation on the special-shaped piece to be welded;
the method comprises the following steps:
acquiring a three-dimensional point cloud model of the special-shaped piece to be welded;
performing coordinate conversion and point cloud splicing on the three-dimensional point cloud model to obtain a plurality of complete weldment point clouds;
screening the complete weldment point cloud based on the straight-through filtering, and extracting a weld point cloud based on a clustering algorithm;
acquiring a first weld joint position set of the special-shaped piece to be welded by adopting a deep learning method based on the complete weldment point cloud;
performing plane fitting on the three-dimensional point cloud model after coordinate conversion based on a random sampling consistency fitting method, obtaining a welding plane, obtaining two plane equations of the welding plane, solving the two plane equations, and obtaining a second weld joint position set;
acquiring a weld position based on the first weld position set and the second weld position set;
constructing a covariance matrix based on the weld point cloud and carrying out welding path planning and pose planning based on the covariance matrix;
welding the special-shaped piece to be welded based on the welding path and the welding seam position;
the process for obtaining the first weld joint position set of the special-shaped piece to be welded by adopting the deep learning method comprises the following steps:
performing format conversion and labeling on the weld point cloud;
inputting the marked weld point cloud into an OpenPCDet framework for training, and obtaining a deep learning model;
acquiring a recognition result of a part to be welded of the special-shaped piece to be welded based on the deep learning model;
performing straight line fitting on the recognition result of the part to be welded to obtain the first welding seam position set;
the identification result of the part to be welded is the approximate coordinates of the point cloud of the welding line area, and the process of carrying out straight line fitting on the identification result of the part to be welded comprises the following steps:
selecting two groups of points from the recognition result of the part to be welded to calculate a linear equation, taking the rest points into the linear equation to calculate residual errors, constructing a residual error threshold, taking a point set with the residual errors smaller than the residual error threshold as internal points, recording, and repeatedly carrying out residual error calculation to obtain a plurality of groups of internal point sets;
acquiring points with the largest number of internal points in a plurality of groups of internal point sets, and performing straight line fitting on the points with the largest number of internal points based on a least square method to acquire a first welding seam position set;
in the process of carrying out plane fitting on the three-dimensional point cloud model after coordinate conversion based on a random sampling consistency fitting method, marking a plane with the minimum plane centroid value and the upward normal as a first plane, intersecting other planes with the first plane to form a flat welding seam, and intersecting other planes to form a vertical welding seam;
the process of acquiring the weld position based on the first set of weld positions and the second set of weld positions includes:
performing cosine similarity calculation on the first welding seam position set and the second welding seam position set, performing distance calculation on a straight line with a cosine value of approximately 1 or-1, and performing matching based on a distance calculation result to obtain the welding seam position;
the process of constructing a covariance matrix based on the weld point cloud and performing welding path planning and pose planning based on the covariance matrix comprises the following steps:
performing point location selection on the weld point cloud, constructing a selected radius, performing decentration on all points in a spherical space formed by the selected points and the selected radius, and constructing the covariance matrix;
and calculating the covariance matrix, obtaining a characteristic value and a characteristic vector, taking the characteristic vector direction corresponding to the maximum characteristic value as a welding direction, and carrying out pose planning based on the characteristic vector corresponding to the minimum characteristic value.
2. The form intelligent flexible welding system of claim 1, further comprising a moving rail (400) in sliding engagement with the robot (300).
3. The form intelligent flexible welding system of claim 1, wherein the shielding gas provided by the welder (500) comprises 80% carbon dioxide and 20% argon.
4. The special-shaped piece intelligent flexible welding system according to claim 1, wherein the 3D vision camera (600) is a structured light 3D camera, the light source is 450nm, the working distance is 350mm-700mm, the near-end view field is 285mm x 192mm, the far-end view field is 515mm x 373mm, the z-axis precision is 0.04-0.08mm, the resolution is 1280 x 1024, and soft triggering is performed through api.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311307754.7A CN117047237B (en) | 2023-10-11 | 2023-10-11 | Intelligent flexible welding system and method for special-shaped parts |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311307754.7A CN117047237B (en) | 2023-10-11 | 2023-10-11 | Intelligent flexible welding system and method for special-shaped parts |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117047237A CN117047237A (en) | 2023-11-14 |
| CN117047237B true CN117047237B (en) | 2024-01-19 |
Family
ID=88666683
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311307754.7A Active CN117047237B (en) | 2023-10-11 | 2023-10-11 | Intelligent flexible welding system and method for special-shaped parts |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117047237B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117689678B (en) * | 2024-02-04 | 2024-05-28 | 法奥意威(苏州)机器人系统有限公司 | Workpiece weld joint identification method, device, equipment and storage medium |
| CN118513721A (en) * | 2024-07-24 | 2024-08-20 | 安徽工布智造工业科技有限公司 | Welding method, system and device for block-shaped member and storage medium |
Citations (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP0554888A1 (en) * | 1992-02-07 | 1993-08-11 | FIAT AUTO S.p.A. | Method for testing spot welds |
| CN103600167A (en) * | 2013-11-15 | 2014-02-26 | 太原科技大学 | Method for welding carbon steel and stainless steel composite plates |
| WO2016172078A1 (en) * | 2015-04-20 | 2016-10-27 | Edison Welding Institute, Inc. | Automated weld inspection system |
| CN110524581A (en) * | 2019-09-16 | 2019-12-03 | 西安中科光电精密工程有限公司 | A kind of flexible welding robot system and its welding method |
| WO2020205998A1 (en) * | 2019-04-05 | 2020-10-08 | Arconic Technologies Llc | Non-destructive evaluation and weld-to-weld adaptive control of metal resistance spot welds via topographical data collection and analysis |
| CN112958959A (en) * | 2021-02-08 | 2021-06-15 | 西安知象光电科技有限公司 | Automatic welding and detection method based on three-dimensional vision |
| CN113798634A (en) * | 2021-08-18 | 2021-12-17 | 中国科学院自动化研究所 | Method, system and equipment for teaching spatial circular weld and tracking weld |
| CN113920060A (en) * | 2021-09-09 | 2022-01-11 | 中国科学院自动化研究所 | Autonomous operation method and device for welding robot, electronic device, and storage medium |
| CN114515924A (en) * | 2022-03-24 | 2022-05-20 | 浙江大学 | Tower foot workpiece automatic welding system and method based on weld joint identification |
| CN114571160A (en) * | 2022-03-09 | 2022-06-03 | 华南理工大学 | Offline curved surface weld extraction and attitude estimation method |
| CN114571153A (en) * | 2022-04-07 | 2022-06-03 | 福州大学 | Weld joint identification and robot weld joint tracking method based on 3D point cloud |
| CN114800574A (en) * | 2022-05-27 | 2022-07-29 | 山东大学 | Robot automatic welding system and method based on double three-dimensional cameras |
| CN114888501A (en) * | 2022-05-20 | 2022-08-12 | 中国建筑第四工程局有限公司 | Teaching-free programming building component welding device and method based on three-dimensional reconstruction |
| CN115026470A (en) * | 2022-06-10 | 2022-09-09 | 重庆大学 | Robot welding system and method for reinforcing mesh |
| CN115229374A (en) * | 2022-07-07 | 2022-10-25 | 武汉理工大学 | Automobile body-in-white weld quality detection method and device based on deep learning |
| CN115330734A (en) * | 2022-08-18 | 2022-11-11 | 南京邮电大学 | Automatic robot repair welding system based on three-dimensional target detection and point cloud defect completion |
| CN218051002U (en) * | 2022-08-09 | 2022-12-16 | 深圳前海瑞集科技有限公司 | Cluster welding device |
| WO2023023243A1 (en) * | 2021-08-19 | 2023-02-23 | Magna International Inc. | Quality monitoring of welding process |
| CN115770988A (en) * | 2022-12-02 | 2023-03-10 | 河北工业大学 | Intelligent welding robot teaching method based on point cloud environment understanding |
| CN115965960A (en) * | 2023-01-31 | 2023-04-14 | 河北中电信普智能科技有限公司 | Weld joint identification method based on deep learning and 3D point cloud |
| CN116765569A (en) * | 2023-05-23 | 2023-09-19 | 江苏伍德精工科技有限公司 | Robot post nail roller surfacing path planning method based on point cloud |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020207512A1 (en) * | 2019-04-12 | 2020-10-15 | 北京城市网邻信息技术有限公司 | Three-dimensional object modeling method, image processing method, and image processing device |
| CN111936821A (en) * | 2019-07-12 | 2020-11-13 | 北京航迹科技有限公司 | System and method for positioning |
| US20220117218A1 (en) * | 2020-10-16 | 2022-04-21 | Verdant Robotics, Inc. | Autonomous system for light treatment of a plant |
-
2023
- 2023-10-11 CN CN202311307754.7A patent/CN117047237B/en active Active
Patent Citations (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP0554888A1 (en) * | 1992-02-07 | 1993-08-11 | FIAT AUTO S.p.A. | Method for testing spot welds |
| CN103600167A (en) * | 2013-11-15 | 2014-02-26 | 太原科技大学 | Method for welding carbon steel and stainless steel composite plates |
| WO2016172078A1 (en) * | 2015-04-20 | 2016-10-27 | Edison Welding Institute, Inc. | Automated weld inspection system |
| WO2020205998A1 (en) * | 2019-04-05 | 2020-10-08 | Arconic Technologies Llc | Non-destructive evaluation and weld-to-weld adaptive control of metal resistance spot welds via topographical data collection and analysis |
| CN110524581A (en) * | 2019-09-16 | 2019-12-03 | 西安中科光电精密工程有限公司 | A kind of flexible welding robot system and its welding method |
| CN112958959A (en) * | 2021-02-08 | 2021-06-15 | 西安知象光电科技有限公司 | Automatic welding and detection method based on three-dimensional vision |
| CN113798634A (en) * | 2021-08-18 | 2021-12-17 | 中国科学院自动化研究所 | Method, system and equipment for teaching spatial circular weld and tracking weld |
| WO2023023243A1 (en) * | 2021-08-19 | 2023-02-23 | Magna International Inc. | Quality monitoring of welding process |
| CN113920060A (en) * | 2021-09-09 | 2022-01-11 | 中国科学院自动化研究所 | Autonomous operation method and device for welding robot, electronic device, and storage medium |
| CN114571160A (en) * | 2022-03-09 | 2022-06-03 | 华南理工大学 | Offline curved surface weld extraction and attitude estimation method |
| CN114515924A (en) * | 2022-03-24 | 2022-05-20 | 浙江大学 | Tower foot workpiece automatic welding system and method based on weld joint identification |
| CN114571153A (en) * | 2022-04-07 | 2022-06-03 | 福州大学 | Weld joint identification and robot weld joint tracking method based on 3D point cloud |
| CN114888501A (en) * | 2022-05-20 | 2022-08-12 | 中国建筑第四工程局有限公司 | Teaching-free programming building component welding device and method based on three-dimensional reconstruction |
| CN114800574A (en) * | 2022-05-27 | 2022-07-29 | 山东大学 | Robot automatic welding system and method based on double three-dimensional cameras |
| CN115026470A (en) * | 2022-06-10 | 2022-09-09 | 重庆大学 | Robot welding system and method for reinforcing mesh |
| CN115229374A (en) * | 2022-07-07 | 2022-10-25 | 武汉理工大学 | Automobile body-in-white weld quality detection method and device based on deep learning |
| CN218051002U (en) * | 2022-08-09 | 2022-12-16 | 深圳前海瑞集科技有限公司 | Cluster welding device |
| CN115330734A (en) * | 2022-08-18 | 2022-11-11 | 南京邮电大学 | Automatic robot repair welding system based on three-dimensional target detection and point cloud defect completion |
| CN115770988A (en) * | 2022-12-02 | 2023-03-10 | 河北工业大学 | Intelligent welding robot teaching method based on point cloud environment understanding |
| CN115965960A (en) * | 2023-01-31 | 2023-04-14 | 河北中电信普智能科技有限公司 | Weld joint identification method based on deep learning and 3D point cloud |
| CN116765569A (en) * | 2023-05-23 | 2023-09-19 | 江苏伍德精工科技有限公司 | Robot post nail roller surfacing path planning method based on point cloud |
Non-Patent Citations (4)
| Title |
|---|
| 基于Matlab的ER50A型拆捆机器人运动特性分析;仝浩源, 马立东, 孟智娟;太原科技大学学报(第2期);全文 * |
| 基于深度学习的三维模型重建检索算法研究;冯智键;周燕;曾凡智;江展锋;;福建电脑(11);全文 * |
| 基于线结构光的型钢自动焊接位置检测技术;刘洪伟，马立东，马自勇，张富泉，杭嘉濠;焊接(第5期);全文 * |
| 红外激光遥感图像散斑特征提取;周贵华;周伟昌;;激光杂志(01);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117047237A (en) | 2023-11-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN117047237B (en) | Intelligent flexible welding system and method for special-shaped parts | |
| CN108747132B (en) | Autonomous mobile welding robot vision control system | |
| CN110524581A (en) | A kind of flexible welding robot system and its welding method | |
| CN110102855A (en) | A kind of robot automtion welding system, device and method | |
| CN114515924B (en) | Automatic welding system and method for tower foot workpiece based on weld joint identification | |
| Nele et al. | An image acquisition system for real-time seam tracking | |
| CN110524582A (en) | A kind of flexibility welding robot workstation | |
| CN113146620B (en) | Binocular vision-based double-arm cooperative robot system and control method | |
| CN112958959A (en) | Automatic welding and detection method based on three-dimensional vision | |
| CN105728972A (en) | Concave-convex angle-variable welding joint self-adaptive tracking control device and method | |
| CN106994684A (en) | The method of control machine people's instrument | |
| CN114434059A (en) | Automatic welding system and method for large structural part with combined robot and three-dimensional vision | |
| CN114289934B (en) | Automatic welding system and method for large structural part based on three-dimensional vision | |
| CN110170996B (en) | Robot rapid teaching system based on stereoscopic vision | |
| CN113223071B (en) | Workpiece weld joint positioning method based on point cloud reconstruction | |
| CN104400217A (en) | Full-automatic laser welding method and full-automatic laser welding device | |
| CN115464263A (en) | Automatic tracking method, detection method and device for laser welding seam | |
| CN111687515A (en) | Intelligent welding guide system for large steel structure | |
| CN110039520B (en) | Teaching and processing system based on image contrast | |
| CN112719702B (en) | Intelligent assembling and welding integrated device and method for complex pipeline | |
| CN217167072U (en) | Robot welding and real-time monitoring system based on line laser and binocular vision | |
| CN117381788A (en) | High-precision positioning and intelligent operation guiding method for composite robot | |
| CN117885096B (en) | Method and device for controlling welding operation of robot tail end welding gun | |
| CN114800574B (en) | Robot automatic welding system and method based on double three-dimensional cameras | |
| CN217618390U (en) | Laser welding system based on visual identification |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |