Large-format board pose recognition linear mechanical arm carrying system and method

Abstract

The application provides a large-format board pose recognition linear mechanical arm carrying system and method, which comprises a stacking lifting module, a mechanical arm module, a data detection module, a data processing module and a control module, the mechanical arm module comprises a mechanical arm track and a mechanical arm, the data detection module comprises a first detection radar and a second detection radar, the first detection radar is mounted on the mechanical arm track, the second detection radar is mounted on the mechanical arm, the first detection radar is used for detecting the corner point of the first board on the top layer, and the second detection radar is used for acquiring the data information of the board on the top layer; the detected data information is transmitted to the data processing module, the board pose and size are acquired through point cloud geometry mapping calculation, and the control module controls the mechanical arm module to grasp and feed the board according to the processing result; through the first detection radar coarse positioning and the second detection radar fine scanning, the board pose and size are calculated through point cloud geometry mapping, and the low-cost, high-precision pose calculation and rapid feeding of the board are realized.

Description

Technical Field

[0001] This invention relates to the field of sheet metal processing technology, and in particular to a linear robotic arm handling system and method for large-format sheet metal position recognition. Background Technology

[0002] In the process of unstacking and loading non-standard boards, obtaining the size and position of the top layer board is the most critical step. For example, custom furniture boards include a variety of materials, colors, and lengths (up to 3 meters). Existing technologies and implementation plans can only use expensive large-field-of-view 3D vision cameras for detection and large-arm heavy-duty industrial multi-joint robotic arms for handling and loading the boards. This results in high project costs, large space requirements, slow recognition speed, and compromised safety.

[0003] Therefore, a low-cost, high-precision, and fast large-format sheet metal pose recognition linear robotic arm handling system and method is developed. Summary of the Invention

[0004] The main objective of this invention is to overcome the above-mentioned shortcomings and deficiencies of the prior art and to provide a linear robotic arm handling system and method for large-format sheet metal pose recognition.

[0005] A linear robotic arm handling system for large-format sheet metal posture recognition includes a worktable, a loading platform, a stacking and lifting module, a robotic arm module, a data detection module, a data processing module, and a control module. The stacking and lifting module, robotic arm module, data detection module, data processing module, and control module are connected. The stacking and lifting module is mounted on the worktable, which is arranged parallel to the loading platform. The robotic arm module includes a robotic arm track and at least two robotic arms mounted parallel to the track. The robotic arms can move horizontally along the track. The data detection module includes at least one first detection radar and at least one second detection radar. The first detection radar is horizontally mounted below the robotic arm track and in front of the stacking side. The second detection radar is mounted on one of the robotic arms. The measurement radius of the first detection radar covers the sheet metal stacking area on the stacking and lifting module, used to detect the corner point of the first sheet metal at the top layer. The measurement radius of the second detection radar covers the width of the sheet metal stack, used to acquire the top surface contour, height, top surface width, and side data information of the top sheet metal.

[0006] The data detection module transmits the detected data information to the data processing module. The data processing module performs point cloud geometric mapping calculations on the received data information to obtain the pose and dimensions of the board. The control module controls the robotic arm module to grip and load the board according to the processing results.

[0007] Compared to the expensive large field-of-view 3D vision cameras and long-reach industrial robotic arms of existing technologies, this invention uses a horizontal first detection linear array radar for coarse positioning, and a downwardly tilted second detection linear array radar for fine feature scanning to obtain the height, top width, and side features of the board. Then, the point cloud geometric mapping of the data processing module is used to calculate the pose and size of the board, thereby achieving low-cost, high-precision pose calculation and rapid correction and loading of the board.

[0008] In one embodiment, the control module controls the first detection radar to perform a horizontal scan of the plate stacking area to acquire a mixed point cloud containing the side of the plate and the background; the point cloud of the side of the plate is extracted using a Euclidean clustering segmentation algorithm, and a straight line of the side edge of the plate is fitted to obtain the straight line. and corner point coarse coordinates ;

[0009] The control module controls the robotic arm to move towards the corner coarse coordinates. Move in the X direction to control the second detection radar to move to the center of the robotic arm;

[0010] The control module controls the robotic arm to move along the robotic arm track toward the sheet metal, and the second detection radar continuously collects point clouds. Due to the tilted installation of the second detection radar, its scanning line forms multiple feature line segments on the top surface, side edges and sides of the sheet metal. The tilted installation of the second detection radar amplifies the change in point cloud distance after the scanning line of the thin sheet metal goes out of bounds.

[0011] The point cloud is segmented using a greedy region growing algorithm (a region growing algorithm based on seed points), and the top surface scan line segments are extracted respectively. and side scan line segments ;

[0012] Based on the installation tilt angle θ and scanning distance of the second detection radar, the height H of the top surface of the plate relative to the second detection radar is calculated in real time, and combined with... The length variation precisely locks onto the top edge of the board.

[0013] Pose fusion and size calculation: The sequence of top edge points extracted by the second detection radar is fitted with a straight line using the least squares method to obtain the precise straight line of the top edge of the plate. ;

[0014] Combined with the side straight line obtained by the first detection radar The straight line of the top edge obtained by the second detection radar Construct a rectangular geometric model of the board material;

[0015] Through calculation The slope of the straight line equation is used to obtain the deflection angle α of the plate; according to Calculate the center coordinates and length and width of the plate by taking the intercept and scan length.

[0016] In one embodiment, the robotic arm is provided with a suction cup module that can move along the extension direction of the robotic arm, and the second detection radar is mounted on the suction cup module.

[0017] In one embodiment, the suction cup module is a single suction cup or a rotatably connected composite suction cup.

[0018] In one embodiment, the suction cup module includes a cylinder and a suction cup connected to the cylinder, the cylinder driving the suction cup to rise and fall.

[0019] A method for handling large-format sheet metal using a linear robotic arm with pose recognition, employing any of the aforementioned large-format sheet metal pose recognition linear robotic arm handling systems, includes the following steps:

[0020] Step 1: Use the first detection radar to perform coarse positioning and corner detection on the plates on the stacking and lifting module;

[0021] Step 2: The robotic arm approaches the plate and works in conjunction with the second detection radar to scan the top surface, side edges, and sides of the plate on the stacking and lifting module;

[0022] Step 3: Extract point cloud features based on the greedy growing region algorithm to obtain the top surface outline, height, width and side data of the top panel;

[0023] Step 4: The data processing module calculates the center coordinates and length and width dimensions of the board by solving the board pose and fitting a rectangle.

[0024] Step 5: Based on the board information from Step 4, the control module controls the robotic arm to precisely grab and adsorb the board and move it to the upper material platform. The two robotic arms work together to rotate and adjust the position of the board.

[0025] Step 6: After reaching the loading platform, the control module controls the robotic arm to release the sheet metal. If the stacking and lifting module has not reached the limit, it will gradually raise the unit height. The first detection radar will start to detect the sheet metal again, preparing for the next loading.

[0026] In one embodiment, step 1 includes the following steps:

[0027] Step 1.1: The first detection radar is installed horizontally on the workbench, located below the rail and in front of the stacking side, with a field of view covering the entire side area of ​​the stacked sheet material of the stacking lifting module;

[0028] Step 1.2: Acquire point cloud data Filtering removes environmental noise;

[0029] Step 1.3: Set the sliding window: with the current point Centered on a point, select k points before and after it as a local window;

[0030]

[0031] Step 1.4: Calculate the distance difference: Calculate the sum of the distances between the front and back points within the window, and subtract the multiple of the distance to the center point. The formula can be simplified to:

[0032] in It is the distance from the point to the radar, or you can directly calculate the local variance of the point in the plane coordinate system;

[0033] Step 1.5: Set the threshold for determining corner points: In a straight area, Approaching 0; when a right-angle corner is scanned, The absolute value of will increase significantly when When the value exceeds the set threshold, that point becomes the corner position. .

[0034] If there are multiple corner points, select the one closest to the radar.

[0035] In one embodiment, step 2 includes the following steps:

[0036] Step 2.1: According to The position is determined by driving the robotic arm close to the board to move to that position, and the suction cup module reaches the center of the robotic arm;

[0037] Step 2.2: The second detection radar is installed at an angle downwards, with its laser plane at a certain angle or perpendicular to the direction of movement of the suction cup module;

[0038] Step 2.3: When the top surface of the board scanned by the second detection radar is tilted or warped, resulting in slight fluctuations in height (such as ±10mm), the scanning lines of the top or side surface of the board can still be obtained.

[0039] In one embodiment, step 3 includes the following steps:

[0040] Step 3.1: Point cloud acquired by the second detection radar Includes: aerial points (invalid), top surface points of the board, and side edge points (corner points) of the board;

[0041] Step 3.2: Greedy growth fitting:

[0042] Select The point with the closest intermediate distance is used as the seed point;

[0043] Set distance threshold and angle threshold ;

[0044] Search the neighborhood of the seed point and classify the points that satisfy the smoothness constraint as the top surface scan line segment;

[0045] This algorithm is used to obtain the top surface scan line segment cluster;

[0046] Step 3.3: Edge contour acquisition: Connect the end of the top surface cluster line segment (edge ​​point) to obtain the side edge of the board.

[0047] In one embodiment, step 4 includes the following steps:

[0048] Step 4.1: Rectangle Fitting: Collect edge points of all top surface line segment clusters during the robotic arm's propulsion process. ;

[0049] These points are fitted using the overall least squares method to obtain a high-precision equation for the straight line at the edge of the sheet metal. ;

[0050] Construct the complete rectangular outline of the sheet material using its two sides;

[0051] Step 4.2: Output the center coordinates of the board material Deflection angle θ, plate length L, and width W.

[0052] The beneficial effects of this invention are as follows:

[0053] 1. The first detection linear array radar on the horizontal line is used for coarse positioning, and the second detection linear array radar installed downwardly is used for fine feature scanning to obtain the height, top width and side features of the plate. Then, the point cloud geometric mapping of the data processing module is used to calculate the pose and size of the plate, so as to realize low-cost, high-precision pose calculation and rapid correction and feeding of the plate.

[0054] 2. This system is simple to implement and significantly reduces costs. By using the tilted installation of the second detection radar, the changes in point cloud distance after the scanning line of the board material goes out of bounds are amplified, realizing simultaneous detection of height and edge, reducing hardware costs, and improving the recognition rate and stability of the board material pose.

[0055] 3. The greedy growing region algorithm is used to segment the point cloud, which can effectively handle noise and breakpoints in the point cloud and can still accurately extract features even when the edges of the board are not clear.

[0056] 4. It occupies little space and consumes little energy. It only requires a linear track two-dimensional robotic arm and a stacker lift to move the sheet material quickly in a small space.

[0057] 5. Synchronous dynamic alignment: The scanning and recognition are completed while the robotic arm is moving towards the board. After recognition, the board can be immediately adsorbed, which improves the work cycle.

[0058] 6. By adding a detection radar and modifying the control logic, the device of the present invention can also be applied to the stacking of sheet metal, that is, stacking the sheet metal pieces on the loading platform one by one. Attached Figure Description

[0059] Figure 1 This is a schematic diagram of the linear robotic arm handling system for large-format sheet metal pose recognition according to the present invention.

[0060] Figure 2 This is a structural schematic diagram of the linear robotic arm handling system for large-format sheet metal pose recognition from another angle.

[0061] Figure 3 This is a schematic diagram illustrating the usage process of the large-format sheet metal pose recognition linear robotic arm handling system of the present invention;

[0062] Figure 4 This is a schematic diagram of the linear robotic arm handling method for large-format sheet metal pose recognition according to the present invention. Detailed Implementation

[0063] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0064] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," "counterclockwise," "axial," "radial," and "circumferential" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0065] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "multiple" means two or more, unless otherwise explicitly specified.

[0066] 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 or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0067] In this invention, unless otherwise explicitly specified and limited, "above" or "below" the second feature can mean that the first feature is in direct contact with the second feature, or that the first feature is in indirect contact with the second feature through an intermediate medium. Furthermore, "above," "over," and "on top" of the second feature can mean that the first feature is directly above or diagonally above the second feature, or simply that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" the second feature can mean that the first feature is directly below or diagonally below the second feature, or simply that the first feature is at a lower horizontal level than the second feature.

[0068] This invention discloses a linear robotic arm handling system for large-format sheet metal posture recognition, comprising a worktable, a loading platform, a stacking and lifting module, a robotic arm module, a data detection module, a data processing module, and a control module. The stacking and lifting module, robotic arm module, data detection module, data processing module, and control module are connected. The stacking and lifting module is mounted on the worktable, which is arranged parallel to the loading platform. The robotic arm module includes a robotic arm track and at least two robotic arms mounted parallel to the track, allowing the robotic arms to move horizontally along the track's extension direction. The data detection module includes at least one first detection radar and at least one second detection radar. The first detection radar is mounted below the robotic arm track, and the second detection radar is mounted on one of the robotic arms. The measurement radius of the first detection radar covers the sheet metal stacking area on the stacking and lifting module, used to detect the corner point of the first sheet metal at the top layer. The measurement radius of the second detection radar covers the width of the sheet metal stack, used to acquire the top surface contour, height, top surface width, and side surface data of the top sheet metal.

[0069] The data detection module transmits the detected data information to the data processing module. The data processing module performs point cloud geometric mapping calculations on the received data information to obtain the pose and dimensions of the board. The control module controls the robotic arm module to grip and load the board according to the processing results.

[0070] This invention uses a first detection linear array radar on a horizontal line for coarse positioning, and a second detection linear array radar installed at a downward tilt for fine feature scanning to obtain the height, top width, and side features of the board. Then, the point cloud geometric mapping of the data processing module is used to calculate the pose and size of the board, thereby achieving low-cost, high-precision pose calculation and rapid correction and loading of the board.

[0071] Example 1

[0072] Please see Figures 1 to 3 This invention provides a linear robotic arm handling system for large-format sheet metal posture recognition, including a workbench 1, a loading platform 2, a stacking and lifting module 3, a robotic arm module 4, a data detection module, a data processing module, and a control module. The stacking and lifting module 3, robotic arm module 4, data detection module, data processing module, and control module are connected. The stacking and lifting module 3 is mounted on the workbench 1, and a track or rollers are installed below the workbench 1 to transport the stack to a position parallel to the loading platform 2. The robotic arm module 4 includes a robotic arm track 41 and two robotic arms 42 mounted parallel to the robotic arm track 41. The robotic arms 42 can move horizontally along the extension direction of the robotic arm track 41. The data detection module includes a first detection radar 5 and a second detection radar 5. Radar 6, the first detection radar 5 is horizontally mounted below the robotic arm track 41, and the second detection radar 6 is mounted on one of the robotic arms 42. Both the first detection radar 5 and the second detection radar 6 are linear array lidars. The first detection radar 5 is horizontally installed at a 45-degree angle along the stacking direction of the sheet metal facing the stacking lifting module 3. The installation height of the first detection radar 5 is slightly higher than the table surface of the workbench 1. The second detection radar 6 is installed at a downward tilt of 45 degrees (the preferred angle range is 30-60 degrees; the smaller the angle, the greater the displacement difference of the laser line in the thickness direction of the sheet metal, thus amplifying the signal of the change in the height of the thin sheet metal; the larger the angle, the more focused the laser line, and the more accurate the measured distance. The preferred 45-degree angle allows for small changes in the height of the sheet metal to be effectively measured). Generates distance measurement data from lidar The projection change is multiplied by a factor of two, thus amplifying the measurement signal and improving the signal-to-noise ratio. The measurement radius of the first detection radar 5 covers the stacking area of ​​the plates on the stacking lifting module 3, and is used to detect the corner point of the first plate on the top layer. The measurement radius of the second detection radar 6 covers the width of the plate stack, and is used to obtain the top surface outline, height, top surface width and side data information of the top plate.

[0073] The data detection module transmits the detected data to the data processing module. The data processing module performs point cloud geometric mapping calculations on the received data to obtain the pose and dimensions of the board. The control module controls the robotic arm module 4 to grab and load the board according to the processing results.

[0074] Compared to the expensive large field-of-view 3D vision cameras and long-reach industrial robotic arms of existing technologies, this invention uses a horizontal first detection linear array radar for coarse positioning, and a downwardly tilted second detection linear array radar for fine feature scanning to obtain the height, top width, and side features of the board. Then, the point cloud geometric mapping of the data processing module is used to calculate the pose and size of the board, thereby achieving low-cost, high-precision pose calculation and rapid correction and loading of the board.

[0075] For more details, please refer to Figure 3 The control module controls the first detection radar 5 to perform a horizontal scan of the plate stacking area, acquiring a mixed point cloud containing the sides of the plates and the background; the Euclidean clustering segmentation algorithm is used to extract the point cloud of the side of the plates, and the straight line of the side of the plates is fitted to obtain the straight line of the side of the plates. and corner point coarse coordinates ;

[0076] The control module controls the coarse coordinates of the 42-axis corner points of the robotic arm. Move in the X direction to control the second detection radar 6 to move to the center position of the robotic arm 42;

[0077] The control module controls the robotic arm 42 to move along the robotic arm track 41 toward the plate, and the second detection radar 6 continuously collects point clouds. Because the second detection radar 6 is installed at an angle, its scanning line forms multiple feature line segments on the top surface, side edges and sides of the plate. The angled installation of the second detection radar 6 amplifies the change in point cloud distance after the scanning line of the thin plate goes out of bounds.

[0078] The point cloud is segmented using a greedy region growing algorithm (a region growing algorithm based on seed points), and the top surface scan line segments are extracted separately. and side scan line segments ;

[0079] Based on the installation tilt angle θ and scanning distance of the second detection radar 6, the height H of the top surface of the plate relative to the second detection radar 6 is calculated in real time, and combined with... The length variation precisely locks onto the top edge of the board.

[0080] Pose fusion and size calculation: The sequence of top edge points extracted by the second detection radar 6 is fitted with a straight line using the least squares method to obtain the precise straight line of the top edge of the plate. ;

[0081] Combined with the side straight line obtained by the first detection radar 5 The straight line of the top edge acquired by the second detection radar 6 Construct a rectangular geometric model of the board material;

[0082] Through calculation The slope of the straight line equation is used to obtain the deflection angle α of the plate; according to Calculate the center coordinates and length and width of the plate by taking the intercept and scan length.

[0083] For more details, please refer to Figures 1 to 3 The robotic arm 42 is equipped with a suction cup module 7 that can move along the extension direction of the robotic arm 42, and a second detection radar 6 is mounted on the suction cup module 7. The suction cup module 7 is a single-axis suction cup or a rotatable composite suction cup, with the rotatable suction cup allowing the two arms to work together to straighten the material during handling. Preferably, the suction cup module 7 includes a cylinder 71 and a suction cup 72 connected to the cylinder 71, with the cylinder 71 driving the suction cup 72 to rise and fall.

[0084] The more specific implementation process of this invention is as follows:

[0085] 1. Initialization:

[0086] Set a deviation threshold;

[0087] Set the maximum value for the consecutive failure counter to max_fail_count = 3;

[0088] Arrange the point cloud set in order (e.g., sort by X-axis coordinate from smallest to largest, or according to the order of radar scans).

[0089] Choose the first point in the set as the starting point of the line segment. .

[0090] 2. Traversal and Dynamic Fitting:

[0091] Starting from the next point after the starting point, traverse the point cloud sequentially.

[0092] For each new point Use the current starting point And the previous point that was successfully incorporated into the line To construct a temporary straight line and calculate... Distance to the line .

[0093] judge:

[0094] if Deviation from threshold: This indicates that the point is on a straight line. Incorporate into the current line segment and update And reset the continuous failure counter to zero;

[0095] if Deviation from threshold: This indicates that the point deviates from the straight line, and the continuous failure counter is incremented by 1.

[0096] 3. Termination of judgment:

[0097] As long as the continuous failure counter has not reached 3, continue to check the next point (1-2 noise points or glitch in the middle are allowed).

[0098] Once the consecutive failure counter reaches 3, the traversal stops immediately. At this point, the last point successfully included becomes the end point of the line segment.

[0099] 4. Final line segment generation:

[0100] Extract all points marked "on a straight line" between the starting point and the ending point;

[0101] The least squares method is used to perform a final straight line fit on these points to obtain the most accurate equation for the line segment.

[0102] 5. Gripping, aligning, and feeding the sheet material.

[0103] After the data processing module of this invention obtains the size and orientation of the board, it can control the robotic arm module 4 and the suction cup module 7 to pick up, move and straighten the board. For long boards, two or more robotic arms 42 are used to suction and move them in coordination, while for short boards, only a single robotic arm 42 is used to suction and move them.

[0104] Example 2

[0105] Please see Figure 4 This invention provides a method for handling large-format sheet metal using a linear robotic arm with pose recognition. The method, using the aforementioned large-format sheet metal pose recognition linear robotic arm handling system, includes the following steps:

[0106] A method for handling large-format sheet metal using a linear robotic arm with pose recognition, employing any of the aforementioned large-format sheet metal pose recognition linear robotic arm handling systems, includes the following steps:

[0107] Step S1: The first detection radar 5 is used to perform coarse positioning and corner detection on the plate material on the stacking lifting module 3;

[0108] Step S2: The robotic arm 42 approaches the plate and works in conjunction with the second detection radar 6 to scan the top surface, side edges and sides of the plate on the stacking lifting module 3;

[0109] Step S3: Extract point cloud features based on the greedy growth region algorithm to obtain the top surface outline, height, width and side data of the top panel;

[0110] Step S4: The data processing module calculates the center coordinates and length and width dimensions of the board by solving the board pose and fitting a rectangle.

[0111] Step S5: Based on the board information from Step 4, the control module controls the robotic arm 42 to precisely grasp and adsorb the board and move it to the upper material platform 2, and rotate and adjust the posture of the board.

[0112] Step S6: After reaching the loading platform 2, the control module controls the robotic arm 42 to release the board. If the stacking lifting module does not reach the limit, it will gradually raise the unit height. The first detection radar 5 will start to detect the board again, preparing for the next board to be picked up and loaded.

[0113] More specifically, step S1 includes the following steps:

[0114] Step 1.1: The first detection radar 5 is installed horizontally on the workbench 1, and its field of view covers the entire side area of ​​the stacked plates on the stacking lifting module 3;

[0115] Step S1.2: Acquire point cloud data Filtering removes environmental noise;

[0116] Step S1.3: Set the sliding window: with the current point Centered on a point, select k points before and after it as a local window;

[0117] Step S1.4: Calculate the distance difference: Calculate the sum of the distances between the front and back points within the window, and subtract the multiple of the distance to the center point. The formula can be simplified to:

[0118]

[0119] in It is the distance from the point to the radar, or you can directly calculate the local variance of the point in the plane coordinate system;

[0120] Step S1.5: Set threshold for determining corner points: in straight areas, Approaching 0; when a right-angle corner is scanned, The absolute value of will increase significantly when When the value exceeds the set threshold, that point becomes the corner position. .

[0121] If there are multiple corner points, select the one closest to the radar.

[0122] More specifically, step S2 includes the following steps:

[0123] Step S2.1: According to The position is determined by driving the robotic arm 42, which is close to the board, to move towards that position, and the suction cup module 7 reaches the center of the robotic arm 42;

[0124] Step S2.2: The second detection radar 6 is installed at an angle downwards, and its laser plane is at a certain angle or perpendicular to the movement direction of the suction cup module 7;

[0125] Step S2.3: When the top surface of the board scanned by the second detection radar 6 is tilted or warped, causing slight fluctuations in height (such as ±10mm), the scanning lines of the top or side surface of the board can still be obtained.

[0126] More specifically, step S3 includes the following steps:

[0127] Step S3.1: Point cloud acquired by the second detection radar 6 Includes: aerial points (invalid), top surface points of the board, and side edge points (corner points) of the board;

[0128] Step S3.2: Greedy growth fitting:

[0129] Select The point with the closest intermediate distance is used as the seed point;

[0130] Set distance threshold and angle threshold ;

[0131] Search the neighborhood of the seed point and classify the points that satisfy the smoothness constraint as the top surface scan line segment;

[0132] This algorithm is used to obtain the top surface scan line segment cluster;

[0133] Step S3.3: Edge contour acquisition: Connect the end of the top surface cluster line segment (edge ​​point), which is the side edge of the board.

[0134] More specifically, step S4 includes the following steps:

[0135] Step S4.1: Rectangle Fitting: Collect edge points of all top surface line segment clusters during the advancement of the robotic arm 42. ;

[0136] These points are fitted using the overall least squares method to obtain a high-precision equation for the straight line at the edge of the sheet metal. ;

[0137] Construct the complete rectangular outline of the sheet material using its two sides;

[0138] Step S4.2: Output the center coordinates of the board material Deflection angle θ, plate length L, and width W.

[0139] Among them, the greedy growth region algorithm is the "greedy growth region line extraction based on seed point" algorithm. Its logic is: starting from the starting point, like a greedy snake, it continuously swallows points that conform to the characteristics of a straight line. Once a certain number (3) of "bad points" are encountered in a row, the current straight line segment is considered to have ended.

[0140] Assume the current line segment starts from the starting point. and current endpoint Sure.

[0141] Coefficients of a straight line equation: The general form of a straight line determined by two points is... ,in:

[0142]

[0143]

[0144]

[0145] (The sign definitions of coefficients A, B, and C do not affect the calculation of the absolute value of the distance.)

[0146] Distance from a point to a line: for the next point to be detected. Its perpendicular distance to the current line for:

[0147]

[0148] if If a point deviates from the threshold, it belongs to the current line.

[0149] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0150] The above embodiments merely illustrate several implementation methods of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. A linear robotic arm handling system and method for large-format sheet metal posture recognition, characterized in that: The system includes a workbench, a loading platform, a stacking and lifting module, a robotic arm module, a data detection module, a data processing module, and a control module. These modules are connected. The stacking and lifting module is mounted on the workbench, which is arranged parallel to the loading platform. The robotic arm module includes a robotic arm track and at least two robotic arms mounted parallel to the track. The robotic arms can move horizontally along the track's extension direction. The data detection module includes at least one first detection radar and at least one second detection radar. The first detection radar is horizontally mounted below the robotic arm track, and the second detection radar is mounted on one of the robotic arms. The measurement radius of the first detection radar covers the stacking area of ​​the sheet metal on the stacking and lifting module, used to detect the corner of the first sheet metal at the top layer. The measurement radius of the second detection radar covers the width of the stacked sheet metal, used to acquire the top surface contour, height, width, and side data of the top sheet metal. The data detection module transmits the detected data information to the data processing module. The data processing module performs point cloud geometric mapping calculations on the received data information to obtain the pose and dimensions of the board. The control module controls the robotic arm module to grip and load the board according to the processing results.

2. The large-format sheet metal pose recognition linear robotic arm handling system according to claim 1, characterized in that: The control module controls the first detection radar to perform a horizontal scan of the plate stacking area to acquire a mixed point cloud containing the side of the plate and the background; the Euclidean clustering segmentation algorithm is used to extract the point cloud of the side of the plate, and the straight line of the side edge of the plate is fitted to obtain the straight line of the side of the plate. and corner point coarse coordinates ; The control module controls the robotic arm to move towards the corner coarse coordinates. Move in the X direction to control the second detection radar to move to the center of the robotic arm; The control module controls the robotic arm to move along the robotic arm track toward the plate, and the second detection radar continuously collects point clouds. The scanning lines of the second detection radar form multiple feature line segments on the top surface, side edges and sides of the plate. The point cloud is segmented using a greedy region growing algorithm (a region growing algorithm based on seed points), and the top surface scan line segments are extracted respectively. ; Based on the installation tilt angle θ and scanning distance of the second detection radar, the height H of the top surface of the plate relative to the second detection radar is calculated in real time, and combined with... The length variation precisely locks onto the top edge of the board. Pose fusion and size calculation: The sequence of top edge points extracted by the second detection radar is fitted with a straight line using the least squares method to obtain the precise straight line of the top edge of the plate. ; Combined with the side straight line obtained by the first detection radar The straight line of the top edge obtained by the second detection radar Construct a rectangular geometric model of the board material; Through calculation The slope of the straight line equation is used to obtain the deflection angle α of the plate; according to Calculate the center coordinates and length and width of the plate by taking the intercept and scan length.

3. The large-format sheet metal pose recognition linear robotic arm handling system according to claim 1, characterized in that: The robotic arm is equipped with a suction cup module that can move along the extension direction of the robotic arm, and the second detection radar is mounted on the suction cup module.

4. The large-format sheet metal pose recognition linear robotic arm handling system according to claim 3, characterized in that: The suction cup module is a single-axis suction cup or a rotatable composite suction cup.

5. The large-format sheet metal pose recognition linear robotic arm handling system according to claim 4, characterized in that: The suction cup module includes a cylinder and a suction cup connected to the cylinder, and the cylinder drives the suction cup to rise and fall.

6. A method for handling large-format sheet metal using a linear robotic arm with pose recognition, characterized in that: The large-format sheet metal pose recognition linear robotic arm handling system according to any one of claims 1-5 includes the following steps: Step 1: Use the first detection radar to perform coarse positioning and corner detection on the plates on the stacking and lifting module; Step 2: The robotic arm approaches the plate and works in conjunction with the second detection radar to scan the top surface, side edges, and sides of the plate on the stacking and lifting module; Step 3: Extract point cloud features based on the greedy growing region algorithm to obtain the top surface outline, height, width and side data of the top panel; Step 4: The data processing module calculates the center coordinates and length and width dimensions of the board by solving the board pose and fitting a rectangle. Step 5: Based on the board material information from Step 4, the control module controls the robotic arm to precisely grasp and adsorb the board material, move it to the upper material platform, and rotate and adjust the posture of the board material. Step 6: After reaching the loading platform, the control module controls the robotic arm to release the sheet metal. If the stacking and lifting module has not reached the limit, it will gradually raise the unit height. The first detection radar will start to detect the sheet metal again, preparing for the next loading.

7. The method for handling large-format sheet metal using a linear robotic arm with pose recognition according to claim 6, characterized in that: Step 1 includes the following steps: Step 1.1: The first detection radar is installed horizontally on the workbench, with its field of view covering the entire side area of ​​the stacked sheet metal on the stacking lifting module; Step 1.2: Acquire point cloud data Filtering removes environmental noise; Step 1.3: Set the sliding window: with the current point Centered on a point, select k points before and after it as a local window; Step 1.4: Calculate the distance difference: Calculate the sum of the distances between the front and back points within the window, and subtract the multiple of the distance to the center point. The formula can be simplified to: in It is the distance from the point to the radar, or you can directly calculate the local variance of the point in the plane coordinate system; Step 1.5: Set the threshold for determining corner points: In a straight area, Approaching 0; when a right-angle corner is scanned, The absolute value of will increase significantly when When the value exceeds the set threshold, that point becomes the corner position. . If there are multiple corner points, select the one closest to the radar.

8. The method for handling large-format sheet metal using a linear robotic arm with pose recognition according to claim 6, characterized in that: Step 2 includes the following steps: Step 2.1: According to The position is determined by driving the robotic arm close to the board to move to that position, and the suction cup module reaches the center of the robotic arm; Step 2.2: The second detection radar is installed at an angle downwards, with its laser plane at a certain angle or perpendicular to the direction of movement of the suction cup module; Step 2.3: Even when the top surface of the board being scanned by the second detection radar experiences slight fluctuations in height due to tilting or warping, the scanning lines on the top or side surfaces of the board can still be obtained.

9. The method for handling large-format sheet metal using a linear robotic arm with pose recognition according to claim 6, characterized in that: Step 3 includes the following steps: Step 3.1: Point cloud acquired by the second detection radar Includes: aerial points (invalid), top surface points of the board, and side edge points (corner points) of the board; Step 3.2: Greedy growth fitting: Select The point closest to the medium-range radar is used as the seed point; Set distance threshold and angle threshold ; Search the neighborhood of the seed point and classify the points that satisfy the smoothness constraint as the top surface scan line segment; This algorithm is used to obtain the top surface scan line segment cluster; Step 3.3: Edge contour acquisition: Connect the end of the top surface cluster line segment (edge ​​point) to obtain the side edge of the board.

10. The method for handling large-format sheet metal using a linear robotic arm with pose recognition according to claim 6, characterized in that: Step 4 includes the following steps: Step 4.1: Rectangle Fitting: Collect edge points of all top surface line segment clusters during the robotic arm's propulsion process. ; These points are fitted using the overall least squares method to obtain a high-precision equation for the straight line at the edge of the sheet metal. ; Construct the complete rectangular outline of the sheet material using its two sides; Step 4.2: Output the center coordinates of the board material Deflection angle θ, plate length L, and width W.

Images