A plate flexible fixed-point machining method based on visual virtual reference reconstruction

Abstract

The application discloses a plate flexible fixed-point processing method based on visual virtual reference reconstruction and relates to the technical field of plate processing. The method comprises the following steps: by analyzing the processing data record associated with the processed piece, the target processing feature orthogonal coordinate sequence and the reference feature definition are obtained with a single global origin as the absolute coordinate origin; the current pose image of the processed piece is collected, the contour pixel distribution of at least two intersecting edges is extracted, and the analytical reference straight line is generated by fitting; the virtual reference origin is reconstructed, the analytical reference straight line is compared with the processing execution end coordinate system, and the pose deflection relationship is determined; based on the translation and rotation constraints, the coordinate affine transformation model is constructed, the target processing feature coordinates are mapped, the actual execution coordinates are generated, and the processing execution end is driven to complete the processing. The application effectively avoids the positioning error caused by the plate edge defects, the edge sealing chamfer and the like, eliminates the cumulative dimensional tolerance, supports the random placement of the plate, and realizes the high-precision and high-flexibility fixed-point processing.

Description

Technical Field

[0001] This invention relates to the field of sheet metal processing technology, specifically to a flexible fixed-point processing method for sheet metal based on visual virtual reference reconstruction. Background Technology

[0002] In sheet metal processing, such as customized furniture, decorative panels, and electronic packaging substrates, the precision and efficiency of point-to-point processing, such as drilling, gluing, cutting, and milling, directly determine product quality and production costs. Traditional sheet metal processing methods mainly rely on the following two types of technologies:

[0003] Firstly, mechanical templates or physical stops are used for positioning. Fixed pneumatic stops or clamps are installed on the processing equipment, and the sheet material is manually pushed to a reference surface to achieve positioning. This method has significant drawbacks: mechanical contact easily damages the surface or edges of the sheet material, especially causing irreversible indentations to sheets with sealed edges or films; furthermore, different sheet material specifications require manual adjustment of the stop positions or replacement of clamps, resulting in extremely poor flexibility and making it unsuitable for flexible production needs involving small batches and multiple varieties.

[0004] Secondly, visual positioning directly captures physical corner points. Industrial cameras capture the corner areas of the sheet metal, directly identifying the intersections of the physical edges as processing benchmarks. However, during actual sheet metal cutting, edge banding, chamfering, or handling, the physical corners often suffer from microscopic geometric distortions such as edge banding thickness interference, rounded corners, localized chipping, or burrs. Directly capturing pixels in this area can cause the benchmark origin to jitter or shift, introducing millimeter-level or even sub-millimeter-level errors into subsequent coordinate mapping, severely affecting the positional accuracy of features such as hinge holes and mounting holes.

[0005] Furthermore, existing technologies often employ multiple reference systems or polar coordinate systems in constructing the machining coordinate system. For example, local references are set for different edges of the sheet metal, and machining features are described by polar diameter and polar angle. This approach limits the relative position of the final machining features to the dimensional tolerances of the sheet metal: if a dimensional deviation occurs in the preceding cutting process (such as a length deviation exceeding 1mm), this deviation will be introduced into the conversion between different references, causing the relative distance between machining features to drift cumulatively, ultimately resulting in serious quality problems such as mismatch between hinge hole spacing and hardware, and assembly interference.

[0006] In summary, existing sheet metal fixed-point processing technologies have significant shortcomings in terms of robustness of positioning datum, decoupling ability of dimensional tolerances, and adaptability to arbitrary placement postures.

[0007] Therefore, there is an urgent need for a processing method that can be immune to physical edge defects, eliminate cumulative dimensional tolerances, and support highly flexible blind discharge. Summary of the Invention

[0008] The purpose of this invention is to provide a flexible fixed-point machining method for sheet metal based on visual virtual reference reconstruction. It aims to provide a high-precision fixed-point machining technology that is immune to sheet metal placement posture, unaffected by physical edge defects of the sheet metal, and can completely eliminate dimensional cumulative tolerances by constructing a single fixed-point orthogonal coordinate system and a virtual reference reconstruction algorithm, thereby solving the problems mentioned in the background art.

[0009] To achieve the above objectives, the present invention provides the following technical solution: a method for flexible fixed-point processing of sheet metal based on visual virtual reference reconstruction, comprising the following steps:

[0010] The processing data records associated with the workpiece are parsed to obtain the orthogonal coordinate sequence of the target processing features with a single global origin as the absolute coordinate origin, and the baseline feature definition that has a spatial mapping relationship with the single global origin is extracted.

[0011] Acquire the current pose image of the workpiece, and under the guidance of the defined reference features, locate and extract the contour pixel distribution of at least two intersecting edges of the workpiece in the current pose image;

[0012] Based on the contour pixel distribution, at least two analytical reference lines corresponding to the intersecting edges are generated by fitting.

[0013] The intersection of at least two analytical reference lines is calculated to generate theoretical intersection point coordinates. The theoretical intersection point coordinates are then anchored to a single global origin in the processing data record to reconstruct a virtual reference origin aligned with the digital drawing in the physical processing space.

[0014] The analytical reference line is compared with the preset physical reference axis of the machining execution end coordinate system to determine the pose deflection relationship of the workpiece relative to the machining execution end coordinate system; based on the translation constraint established by the virtual reference origin and the rotation constraint established by the pose deflection relationship, a coordinate affine transformation model is constructed, and the orthogonal coordinate sequence of the target machining feature is mapped by the coordinate affine transformation model to generate the actual execution coordinates, and the machining execution end is driven to perform machining according to the actual execution coordinates.

[0015] Preferably, the relative distance between any two target processing features in the orthogonal coordinate sequence of the target processing features is exclusively determined by the coordinate difference of the target processing feature relative to the single global origin, and is decoupled from the dimensional tolerance of the workpiece.

[0016] Preferably, the step of parsing the processing data record associated with the workpiece includes: decoding the data carrier attached to the workpiece and extracting the structured data encapsulated therein; the structured data includes at least topological constraint information for anchoring the single global origin, and the orthogonal coordinate sequence of the target processing feature composed of discrete two-dimensional coordinate points, wherein the topological constraint information constitutes the reference feature definition.

[0017] Preferably, the step of extracting the contour pixel distribution of at least two intersecting edges of the workpiece in the current pose image includes: using an edge detection operator to identify edge candidate pixels and removing nonlinear outlier pixels located in the physical intersection area of ​​the intersecting edges due to deformation or chamfering.

[0018] Preferably, the edge detection operator includes a Canny operator or a Sobel operator configured to calculate the spatial gradient of image pixels to extract contours; the step of fitting based on the contour pixel distribution includes: performing convergent fitting of the contour pixel distribution using the least squares method or the Hough transform algorithm.

[0019] Preferably, the step of comparing the analytical reference line with the preset physical reference axis of the machining execution end coordinate system to determine the pose deflection relationship includes: selecting the first analytical reference line representing the main extension direction of the workpiece among the at least two analytical reference lines, calculating the spatial angle between the first analytical reference line and the specified coordinate axis of the machining execution end coordinate system, and configuring it as the deflection angle in the pose deflection relationship.

[0020] Preferably, the step of constructing a coordinate affine transformation model for coordinate mapping includes: using the physical coordinates of the virtual reference origin in the coordinate system of the machining execution end as the translation compensation parameter corresponding to the translation constraint, and using the deflection angle as the rotation compensation parameter corresponding to the rotation constraint, to generate a two-dimensional transformation matrix containing translation and rotation attributes; substituting each target machining feature in the orthogonal coordinate sequence into the two-dimensional transformation matrix for solution, and outputting the actual execution coordinates relative to the coordinate system of the machining execution end.

[0021] Preferably, the solution relationship of the actual execution coordinates satisfies the following equation:

[0022] ;

[0023] ;

[0024] in, The original drawing coordinates of the target machining feature in the orthogonal coordinate sequence. The deflection angle is... The physical coordinates of the virtual reference origin.

[0025] Preferably, the step of decoding the data carrier attached to the workpiece includes: dynamically establishing a processing task association under the current placement posture in response to an external scanning trigger signal of the data carrier, and pushing the structured data into the motion control unit.

[0026] Preferably, the workpiece is a customized sheet metal component, the processing execution end is configured as a CNC drilling mechanism, and the target processing feature is characterized by preset assembly holes distributed on the customized sheet metal component.

[0027] In summary, the beneficial effects of this invention are:

[0028] Abandoning the traditional method of directly identifying physical corner pixels, the algorithm pioneered a virtual origin reconstruction logic of "edge grabbing and intersection finding". Even if the workpiece (such as sheet metal) has distortions in the physical intersection area such as edge banding thickness interference, R-corner chamfering, or local edge chipping, the algorithm can automatically remove noise and deduce an absolutely accurate mathematical virtual intersection point, ensuring the absolute accuracy of machine tool coordinate mapping from the source of physical perception.

[0029] This model forcibly transforms a multi-reference system into a Cartesian coordinate system with a single global origin. Mathematically, this model locks in the internal topological relationships of all machining features, ensuring that the distance between machining holes is determined solely by the difference in orthogonal coordinates. Regardless of the dimensional errors caused by previous processes, the final hole spacing perfectly matches the standard assembly dimensions, completely eliminating cumulative hole drift.

[0030] By introducing a two-dimensional coordinate affine transformation model, real-time dynamic mapping between digital machining drawings and physical placement postures is achieved. The system not only eliminates the need for easily worn mechanical structures such as pneumatic stops, but also allows the workpiece to be placed at any tilt on the worktable. The control algorithm can correct coordinate offsets based on the fitted deflection angle within milliseconds, significantly reducing the alignment difficulty of manual loading.

[0031] A closed loop was creatively constructed, linking data-defined benchmarks, visual verification benchmarks, and algorithm-driven machinery. Relying on the scanning trigger of data carriers (such as QR codes), the same production line can achieve seamless mixed-line processing of plates with different sizes and hole designs, truly breaking the limitation of traditional machinery requiring machine stoppage and parameter adjustment based on molds, and providing underlying technical support for large-scale flexible customized production starting from a single piece. Attached Figure Description

[0032] To more clearly illustrate the technical solutions in the embodiments of the invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0033] Figure 1 This is a schematic diagram of the process framework of a flexible fixed-point processing method for sheet metal based on visual virtual reference reconstruction according to the present invention.

[0034] Figure 2 This is a schematic diagram illustrating the virtual points in a flexible fixed-point processing method for sheet metal based on visual virtual reference reconstruction according to the present invention.

[0035] Figure 3 This is a schematic diagram of coordinate affine transformation for a flexible fixed-point processing method for sheet metal based on visual virtual reference reconstruction according to the present invention.

[0036] Figure 4 This is a schematic diagram illustrating the visual acquisition and edge extraction of the sheet material in a flexible fixed-point processing method based on visual virtual reference reconstruction according to the present invention.

[0037] Figure 5 This is a schematic diagram illustrating the virtual reconstruction of a sheet material in a flexible fixed-point processing method based on visual virtual reference reconstruction according to the present invention. Detailed Implementation

[0038] The present invention will now be described in further detail with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention. These drawings are simplified schematic diagrams, which are only used to illustrate the basic structure of the present invention in a schematic manner, and therefore only show the components related to the present invention.

[0039] To facilitate understanding of the present invention, a more complete description of the invention will be given below with reference to the accompanying drawings, which illustrate several embodiments of the invention. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided so that the disclosure of the invention will be more thorough and complete.

[0040] All features disclosed in this specification, or all steps in all disclosed methods or processes, may be combined in any way, except for mutually exclusive features and / or steps.

[0041] Any feature disclosed in this specification (including any appended claims, abstract, and drawings) may be replaced by other equivalent or similar features for a similar purpose, unless specifically stated otherwise. That is, unless specifically stated otherwise, each feature is merely one example of a series of equivalent or similar features.

[0042] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection, a direct connection, or an indirect connection through an intermediate medium; they can refer to the internal communication of at least two elements or the interaction relationship of at least two elements, unless otherwise explicitly limited. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0043] Please see Figures 1-5 This invention provides an embodiment of a flexible fixed-point processing method for sheet metal based on visual virtual reference reconstruction. This method breaks through the limitations of traditional mechanical templates and direct visual corner grasping. By coupling a single fixed-point orthogonal model in digital space with a virtual origin reconstruction algorithm in physical space, it achieves high-precision and high-tolerance blind placement processing. In a preferred embodiment of this invention, the main execution body of this method includes:

[0044] Data parsing modules, such as barcode scanners or industrial gateways;

[0045] Vision acquisition modules, such as industrial cameras mounted above machine tools;

[0046] Computation and control units, such as industrial computers and motion control cards;

[0047] Machining execution end, such as multi-axis CNC drilling machine.

[0048] Taking flexible drilling of customized furniture (such as bathroom vanities) panels as an example, this method specifically includes the following steps:

[0049] Step S1: Parse the processing data records to obtain the global orthogonal coordinates and datum feature definitions;

[0050] During the sheet material loading stage, operators or automated robotic arms place the sheet material on the processing table. Subsequently, the system responds to external scanning trigger signals, such as a barcode scanner reading the QR code or RFID tag attached to the surface of the sheet material, dynamically establishing the association of the current processing task.

[0051] The computation and control unit parses the decoded structured data and extracts two core pieces of information:

[0052] First, quasi-feature definition: clearly define the specific location on the drawing as the absolute coordinate origin, for example, specify the lower left corner of the front of the board as a single global origin;

[0053] Secondly, the orthogonal coordinate sequence of target processing features: all holes on the plate that need to be processed, no matter how far away from the origin, are expressed in a rectangular coordinate system relative to the global origin, that is, a discrete two-dimensional coordinate point sequence set.

[0054] In traditional multi-reference polar coordinate positioning, if a 1000mm designed sheet material is actually cut into 1001mm pieces, and the hinge holes on the left and right sides are referenced to the top left and right corners respectively, this 1mm cutting error will be carried over into the hole spacing, making it impossible to install the hinges. However, in this embodiment, the relative distance between the holes is exclusively determined by the coordinate difference relative to a single global origin, completely decoupled from the dimensional tolerances of the sheet material. This means that regardless of fluctuations in the length and width of the sheet material, the hole spacing will always perfectly match the dimensions of standard hardware.

[0055] Step S2: Image acquisition and baseline-guided edge contour extraction;

[0056] An industrial camera captures an image of the sheet metal in its current orientation. Guided by the baseline feature definition extracted in step S1, the system defines a region of interest (ROI) in the image, locates and extracts the pixel distribution of the two edges intersecting at that baseline corner, such as a long side and an adjacent short side, referencing... Figure 4 .

[0057] During this process, the system utilizes edge detection operators, preferentially employing either the Canny or Sobel operator to calculate the spatial gradient of image pixels to identify candidate edge pixels. To avoid interference from the thickness of the board's edge banding, chamfers (R-angles), or chipped edges due to local handling, the algorithm automatically performs filtering operations to eliminate nonlinear outlier pixels located within the physical intersection area of ​​intersecting edges, retaining only the straight line segments that truly represent the macroscopic extension direction of the board.

[0058] Step S3: Line fitting and virtual reference origin reconstruction, refer to Figure 2 and Figure 5 ;

[0059] For the extracted and filtered contour pixel distribution, the operation and control unit uses the least squares method or the Hough transform algorithm to perform convergence fitting, generating two continuous mathematical straight line equations corresponding to the long side and the short side, respectively, which are the analytical reference lines.

[0060] Subsequently, the system performs intersection calculations on these two analytical reference lines to obtain the coordinates of the theoretical intersection point. This theoretical intersection point is not necessarily physically located on the actual pixel edge of the board material, as the actual physical corners may have been rounded or chipped; however, it is the mathematical extension and intersection point of two perfect boundaries. The system anchors this theoretical intersection point to the single global origin in the drawing data from step S1, thereby successfully reconstructing the virtual reference origin in the physical processing space. This step bypasses the defects of the actual corner points of the board material from the source of physical perception, ensuring absolute positioning accuracy.

[0061] Step S4: Pose deflection calculation and coordinate affine transformation, refer to Figure 3 ;

[0062] Since this invention supports arbitrary blind placement of the sheet material, such as placing it at a 15-degree angle on the worktable, the system needs to determine the current position and orientation of the sheet material. The computation and control unit selects the first analytical reference line representing the main extension direction of the sheet material, such as the line containing the long side, calculates the spatial angle between it and the machine coordinate system of the machining execution end, and configures this angle as the deflection angle. .

[0063] At this point, the system has obtained the physical coordinates (i.e., the virtual reference origin) corresponding to the translation constraint and the parameters (i.e., the deflection angle) corresponding to the rotation constraint. Based on this, the system constructs a two-dimensional coordinate affine transformation model, whose solution relationship satisfies the following equation:

[0064] ;

[0065] ;

[0066] In the above equation, The original drawing coordinates of the target assembly hole positions in the orthogonal coordinate sequence. For the above deflection angle, The absolute coordinates of the virtual reference origin in the machine tool physical coordinate system.

[0067] Step S5: Drive the actuator to complete flexible machining;

[0068] The computation and control unit substitutes the drawing coordinates of all assembly holes into the aforementioned affine transformation matrix one by one for millisecond-level calculation, and outputs the actual mechanical coordinates of each hole in the current tilt posture. Subsequently, the coordinate sequence is compiled into machine tool executable code, which drives the servo motor and CNC drilling mechanism to complete precise machining along the tilt angle of the sheet metal.

[0069] The integrated processing flow described above greatly shortens the cycle time, enabling the continuous and seamless feeding of plates with different lengths, widths, and hole designs onto the same production line. This truly achieves highly flexible production without the need for manual alignment or mechanical template shutdowns for parameter adjustment.

[0070] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0071] In another embodiment, for step S3, in order to cope with extremely harsh physical edges, such as severe edge chipping or wide chamfers, this embodiment introduces a joint optimization mechanism of random sampling consensus algorithm combined with least squares method in the straight line fitting stage. The specific process is as follows:

[0072] First, the RANSAC algorithm is used to randomly select a subset from the extracted candidate edge pixels to construct an initial straight line model, and the normal distance from the remaining pixels to this initial straight line model is calculated. Pixels with a normal distance less than a preset tolerance threshold, such as 0.1 mm, are classified as inliers, while the rest are classified as outliers, representing noise points or impurities. After multiple iterations, the model with the most inliers is selected as the effective edge dataset.

[0073] Subsequently, the least squares method was used to perform a global convergence fitting on the selected effective edge dataset, generating two continuous analytical reference line equations corresponding to the long and short sides of the board, respectively. In the two-dimensional image coordinate system, their standard general form equations are expressed as:

[0074] First analytical reference line (long side): ;

[0075] Second analytical reference line (short side): ;

[0076] Finally, the system solves the problem by simultaneously solving the equations of the two analytical reference lines. To avoid no solution or infinite solutions caused by parallel or coincident lines, the system pre-calculates the Jacobian determinant. Ensure that the two adjacent edges of the board are nearly perpendicular (i.e., Under the premise that the values ​​have significant absolute values, the theoretical intersection coordinates can be directly calculated using Cramer's rule. :

[0077]

[0078]

[0079] The coordinates of this theoretical intersection point are the reconstructed virtual reference origin. Since this origin is obtained by the intersection of two macroscopic reference lines that have been fitted with global noise reduction, it is completely immune to local microscopic defects at the physical intersection of the sheet metal. The system then uses camera calibration parameters, such as the pixel physical equivalent ratio K, to convert the pixel-level coordinates into absolute mechanical coordinates in the physical space of the processing execution end.

[0080] In another embodiment, the specific method for the pose deflection calculation and coordinate affine transformation in step S4 is as follows:

[0081] After obtaining the virtual reference origin in the physical space, the system needs to establish the dynamic pose mapping relationship between the digital drawings and the physical board material, so as to support the high tolerance of "blind placement" of the board material.

[0082] First, the pose deflection relationship is calculated. The computational control unit extracts the first analytical reference line (i.e., the line representing the main extension direction of the plate) that characterizes the main extension direction of the plate. ), calculate the slope of its straight line:

[0083]

[0084] Subsequently, based on the preset physical reference axis of the machining execution end coordinate system, typically set as the machine tool's mechanical X-axis, the absolute deflection angle of the sheet metal placement direction relative to the machine tool's mechanical coordinate system is calculated using the arctangent function. :

[0085]

[0086] The system will determine the quadrant interval where the line lies. Perform full quadrant normalization from 0 to 360 degrees.

[0087] Subsequently, an affine transformation model is constructed. To achieve efficient batch coordinate system mapping, this embodiment adopts the standard homogeneous coordinate representation method in robotics and computer vision. The system uses the physical coordinates of the virtual reference origin. As a translation compensation parameter, the deflection angle As a rotation compensation parameter, construct a third-order two-dimensional affine transformation matrix T:

[0088]

[0089] For the orthogonal coordinate sequence of the target machining feature obtained in step S1, the original two-dimensional coordinates of any hole position on the drawing are... Expanded into homogeneous coordinate vectors The system solves by substituting all target processing features one by one into the affine transformation model through matrix multiplication:

[0090]

[0091] Expanding this matrix multiplication yields the final actual mechanical coordinates used to drive the servo motor.

[0092]

[0093]

[0094] Through the aforementioned extremely rigorous matrix affine transformation, the coordinates of each hole, originally dependent on a single fixed-point drawing, are instantly "projected" onto a physical plate with random tilt angles and random placement positions. The computation and control unit sends these actual mechanical coordinates to the multi-axis CNC mechanism, allowing the machine tool to directly address and drill without any mechanical alignment adjustments, thus completely realizing flexible fixed-point machining based on digital twin mapping.

[0095] In addition, the present invention also provides another embodiment of high-precision flexible dispensing processing for PCB circuit boards, the details of which are as follows:

[0096] The workpiece being processed is a printed circuit board (PCB), the processing execution end is configured as a multi-axis automatic dispensing machine, and the target processing feature is characterized by preset component dispensing positions distributed on the PCB, such as chip bottom filler glue and solder paste application points.

[0097] In SMT (Surface Mount Technology) production lines for electronics manufacturing, after PCB boards undergo front-end depaneling processes such as V-Cut cutting or stamp hole breaking, irregular burrs or fiberglass protrusions often remain on their physical edges. If traditional visual positioning directly uses the physical corners of the PCB board as the origin, these microscopic burrs can cause severe pixel-level misalignment, leading to dispensing needle misalignment and potentially causing short circuits in components.

[0098] The technical solution of this invention is as follows:

[0099] Data and benchmark anchoring: The scanning component reads the DataMatrix laser-engraved QR code attached to the PCB board or its carrier and parses out the dispensing coordinate array with a theoretical corner point in the drawing as the global origin.

[0100] Anti-burr edge extraction: Industrial cameras acquire PCB board pose images. When extracting edges, the vision system uses the RANSAC algorithm to filter out abrupt burrs caused by the V-Cut section, i.e., nonlinear outlier pixels, and only retains the inliers representing the true contour of the PCB board for least squares fitting to obtain a smooth analytical reference line.

[0101] Virtual intersection point and dynamic dispensing: A virtual reference origin is determined by solving the simultaneous linear equations. This origin mathematically eliminates the interference from PCB deflection burrs. Subsequently, the system calculates the deflection angle of the PCB board on the dispensing table. By substituting all the coordinates of the dispensing drawing into a two-dimensional affine transformation matrix, the actual mechanical coordinates of the dispensing process can be instantly calculated. The dispensing head can directly press down to perform high-precision dispensing without waiting for the mechanical limit cylinder to align.

[0102] This embodiment fully demonstrates that the virtual benchmark reconstruction algorithm of the present invention also has excellent correction and anti-interference capabilities in precision machining scenarios with edge geometric distortions at the microscale.

[0103] It is worth mentioning that, in another preferred embodiment of the present invention, fixture-free laser marking / cutting processing is performed on heavy sheet metal parts;

[0104] The workpiece is a heavy sheet metal part, the processing execution end is configured as a CNC laser cutting machine or a laser galvanometer marking machine, and the target processing feature is characterized as the processing trajectory that requires secondary drilling or engraving of traceability information on this sheet metal part.

[0105] In traditional heavy-duty sheet metal processing, the shape of sheet metal parts is usually formed by plasma cutting or flame cutting at the front end, resulting in rough edges and severe slag buildup (large deformation in the heat-affected zone). At the same time, it is extremely difficult to manually load heavy metal sheets, making it hard to accurately push them onto the zero-point stop of the machine tool, and they usually exhibit a random tilted posture.

[0106] The technical solution of this invention is as follows:

[0107] Coordinate and pose decoupling: The loading gantry crane randomly places the sheet metal parts on the laser processing machine. By reading the RFID tag accompanying the work order, the control unit obtains the secondary processing drawing features based on a single global origin.

[0108] Visual slag removal: A large-view industrial camera mounted above the machine bed acquires images, and the edge extraction algorithm automatically ignores the jagged slag edges left by plasma cutting, fitting two analytical reference lines representing the macroscopic dimensions of the sheet metal part.

[0109] Fixtureless attitude mapping: After calculating a virtual reference intersection point unaffected by slag, the system aligns the drawing origin with this virtual intersection point. For sheet metal parts arbitrarily positioned on the machine bed, the system uses a coordinate affine transformation model to directly perform virtual rotation of the digital drawing within the control system. The laser cutting head or galvanometer then directly initiates processing based on the mapped actual coordinates.

[0110] This embodiment completely eliminates the reliance on expensive mechanical fixtures and manual physical alignment in heavy metal processing, realizing fixture-free, extremely flexible processing with "place and cut as needed," which greatly improves the processing cycle and safety in the heavy industry field.

[0111] To further verify the accuracy of the two-dimensional coordinate affine transformation model of this invention in practical engineering, this embodiment provides a set of specific operating examples:

[0112] (1) Data parsing stage:

[0113] The current workpiece is a rectangular bathroom vanity side panel. After scanning the QR code, the system obtains and establishes a single global origin with the "bottom left corner of the front" of the panel. An orthogonal coordinate system.

[0114] The drawing data defines two hinge base mounting holes that need to be machined:

[0115] The original drawing for hole position 1 (P1) has the following orthogonal coordinates: (100, 50), in millimeters.

[0116] The original drawing for hole position 2 (P2) has the following orthogonal coordinates: (700, 50), in millimeters.

[0117] Note: In digital logic, the absolute distance between these two holes is... Millimeters, this relative topological relationship has been locked by the system.

[0118] (2) Material loading and vision grasping stage:

[0119] The operator placed the sheet material randomly on the worktable of the CNC drilling machine. Due to the lack of any mechanical alignment, the sheet material shifted and tilted.

[0120] The image is acquired by the vision system above the machine tool. After deburring using the RANSAC algorithm and fitting using the least squares method, two analytical reference lines are obtained.

[0121] Through simultaneous calculation, the coordinates of the virtual reference origin at the "lower left corner of the front" of the sheet metal in the absolute mechanical coordinate system of the machine tool are obtained as follows: .

[0122] Meanwhile, by calculating the angle between the long side analytical reference line and the machine tool's mechanical X-axis, the actual deflection angle of the sheet metal is obtained. .

[0123] (3) Coordinate affine transformation stage:

[0124] The computation control unit retrieves the homogeneous coordinate affine transformation equation from Example 1. It is known that... .

[0125] Mapping for hole position 1 (P1):

[0126] ;

[0127] ;

[0128] ;

[0129] ;

[0130] That is, the actual drilling coordinates of hole position 1 on the machine tool are (119.80, 106.60).

[0131] Mapping for hole position 2 (P2):

[0132] ;

[0133] ;

[0134] ;

[0135] ;

[0136] That is, the actual drilling coordinates of hole position 2 on the machine tool are (710.68, 210.76).

[0137] (4) Result verification:

[0138] The calculation control unit sends the calculated actual execution coordinates (119.80, 106.60) and (710.68, 210.76) to the CNC execution end, and the machine tool drill bit drills two holes on the inclined plate.

[0139] Physical distance verification was performed on the two drilled holes:

[0140] ;

[0141] .

[0142] The calculation results show that although the sheet metal was tilted on the machine tool Although it deviates from the origin, the system, through dynamic mapping between the virtual origin and the deflection angle, still perfectly maintains the absolute spacing of 600 mm defined in the drawing in the physical space. The slight error comes from the preservation of the accuracy of the trigonometric function. This conclusively proves, from both mathematical and engineering perspectives, that the present invention completely eliminates the cumulative dimensional tolerances caused by traditional positioning and fully realizes high-precision flexible machining that is not affected by the placement posture.

[0143] The above description is merely a specific embodiment of the invention, but the scope of protection of the invention is not limited thereto. Any variations or substitutions conceived without inventive effort should be included within the scope of protection of the invention. Therefore, the scope of protection of the invention should be determined by the scope defined in the claims.

Claims

1. A flexible point-to-point processing method for sheet metal based on visual virtual reference reconstruction, characterized in that: Includes the following steps: The processing data records associated with the workpiece are parsed to obtain the orthogonal coordinate sequence of the target processing features with a single global origin as the absolute coordinate origin, and the baseline feature definition that has a spatial mapping relationship with the single global origin is extracted. Acquire the current pose image of the workpiece, and under the guidance of the reference feature definition, locate and extract the contour pixel distribution of at least two intersecting edges of the workpiece in the current pose image; Based on the contour pixel distribution, at least two analytical reference lines corresponding to the intersecting edges are generated by fitting. The intersection of at least two analytical reference lines is calculated to generate theoretical intersection point coordinates. The theoretical intersection point coordinates are then anchored to a single global origin in the processing data record to reconstruct a virtual reference origin aligned with the digital drawing in the physical processing space. The analytical reference line is compared with the preset physical reference axis of the machining execution end coordinate system to determine the pose deflection relationship of the workpiece relative to the machining execution end coordinate system. Specifically, the following steps are included: selecting the first analytical reference line representing the main extension direction of the workpiece among the at least two analytical reference lines, calculating the spatial angle between the first analytical reference line and the specified coordinate axis of the machining execution end coordinate system, and configuring it as the deflection angle in the pose deflection relationship; Based on the translation constraints established by the virtual reference origin and the rotation constraints established by the pose deflection relationship, a coordinate affine transformation model is constructed. The orthogonal coordinate sequence of the target processing features is mapped using this model to generate actual execution coordinates, and the processing execution end is driven to process according to these actual execution coordinates. The step of constructing the coordinate affine transformation model and performing coordinate mapping includes: using the physical coordinates of the virtual reference origin in the coordinate system of the processing execution end as the translation compensation parameter corresponding to the translation constraint, and using the deflection angle as the rotation compensation parameter corresponding to the rotation constraint, to generate a two-dimensional transformation matrix containing translation and rotation attributes; substituting each target processing feature in the orthogonal coordinate sequence into the two-dimensional transformation matrix for solution, and outputting the actual execution coordinates relative to the coordinate system of the processing execution end.

2. The method for flexible fixed-point processing of sheet metal based on visual virtual reference reconstruction according to claim 1, characterized in that: The relative distance between any two target processing features in the orthogonal coordinate sequence of the target processing features is exclusively determined by the coordinate difference of the target processing feature relative to the single global origin, and is decoupled from the dimensional tolerance of the workpiece.

3. The method for flexible fixed-point processing of sheet metal based on visual virtual reference reconstruction according to claim 2, characterized in that: The step of parsing the processing data records associated with the workpiece includes: decoding the data carrier attached to the workpiece and extracting the structured data encapsulated therein; the structured data includes at least topological constraint information for anchoring the single global origin, and the orthogonal coordinate sequence of the target processing feature composed of discrete two-dimensional coordinate points, wherein the topological constraint information constitutes the baseline feature definition.

4. The method for flexible fixed-point processing of sheet metal based on visual virtual reference reconstruction according to claim 3, characterized in that: The step of extracting the contour pixel distribution of at least two intersecting edges of the workpiece in the current pose image includes: using an edge detection operator to identify edge candidate pixels and removing nonlinear outlier pixels located in the physical intersection area of ​​the intersecting edges due to deformation or chamfering.

5. The method for flexible fixed-point processing of sheet metal based on visual virtual reference reconstruction according to claim 4, characterized in that: The edge detection operator includes a Canny operator or a Sobel operator configured to calculate the spatial gradient of image pixels to extract contours; the step of fitting based on the contour pixel distribution includes: using the least squares method or the Hough transform algorithm to perform convergent fitting on the contour pixel distribution.

6. The method for flexible fixed-point processing of sheet metal based on visual virtual reference reconstruction according to claim 1, characterized in that: The solution relationship of the actual executed coordinates satisfies the following equation: ; ; in, The original drawing coordinates of the target machining feature in the orthogonal coordinate sequence. The deflection angle is... The physical coordinates of the virtual reference origin.

7. The method for flexible fixed-point processing of sheet metal based on visual virtual reference reconstruction according to claim 3, characterized in that: The step of decoding the data carrier attached to the workpiece includes: dynamically establishing a processing task association under the current placement posture in response to an external scan trigger signal of the data carrier, and pushing the structured data into the motion control unit.

8. The method for flexible fixed-point processing of sheet metal based on visual virtual reference reconstruction according to claim 1, characterized in that: The workpiece is a customized sheet metal component, the processing execution end is configured as a CNC drilling mechanism, and the target processing feature is characterized by preset assembly holes distributed on the customized sheet metal component.

Images