An intelligent control method and system for an automatic hemmer

Abstract

The present invention provides an intelligent control method for an automatic hemmer, including: grasping and feeding the fabric to be overlocked and adjusting its placement position; performing image recognition on the placed fabric and determining the corresponding overlocking process according to the recognition result; and completing the discharging work after performing the overlocking work on the fabric based on the overlocking process. By means of the present invention, the hemmer can automatically handle the upper and lower fabrics and automatically complete the overlocking work.

Description

Technical Field

[0001] The present invention relates to the field of automatic control of AI intelligence, and particularly to an intelligent control method and system for an automatic lower swing machine. Background Art

[0002] The lower swing machine, also known as an overlock sewing machine, is a special sewing device with a stitch former having more than two straight needles and a curved hook. The formed overlock stitch is flat, and the strength and elasticity of the stitch are relatively good. It is suitable for sewing pajamas, underwear, trousers, various vests, and the rolling collar, hemming, flanging, overlocking, splicing seam, and edging of knitted garments. Since the majority of the sewing positions are at the lower hem of the clothing, it is called the lower swing machine.

[0003] The traditional lower swing machine uses manual feeding. After manually placing the clothing on the front and rear groups of feed dogs, the lower swing machine is started to work, and after the work is completed, the lower swing machine is manually controlled to stop. This process is very wasteful of human resources. Therefore, an intelligent control method and system for an automatic lower swing machine that can automatically perform loading and unloading and automatically perform overlock sewing work are needed. Summary of the Invention

[0004] The present invention provides an intelligent control method for an automatic lower swing machine to automatically perform upper and lower cloth feeding and automatically complete overlock sewing work.

[0005] An intelligent control method for an automatic lower swing machine provided by the present invention includes:

[0006] Step S1: Grab and feed the cloth to be overlocked and adjust the placement position;

[0007] Step S2: Perform image recognition on the placed cloth, and determine the corresponding overlock sewing process according to the recognition result;

[0008] Step S3: Perform overlock sewing work on the cloth based on the overlock sewing process and complete the unloading work.

[0009] Preferably, the grabbing and feeding of the cloth to be overlocked and adjusting the placement position include:

[0010] Shoot the cloth on the conveyor belt through a preset first camera to obtain a first image;

[0011] Analyze the first image to determine the style type corresponding to the cloth in the first image;

[0012] Determine the preset grabbing point position and the position where overlock sewing work needs to be performed corresponding to the style type;

[0013] Determine the corresponding preset placement method according to the position where overlock sewing work needs to be performed on the cloth;

[0014] The fabric is grasped by the robotic arm based on the position of the grasping point, and the fabric is placed on the overlock platform based on the placement method.

[0015] Preferably, the analysis of the first image to determine the style type corresponding to the fabric in the first image includes:

[0016] Step 1: Grayscale the first image to obtain a first grayscale image, and use a Gaussian image blur filter to smooth and filter the first grayscale image to remove image noise;

[0017] Step 2: Determine whether the probability distribution state of the pixel values in the first grayscale image satisfies the normal distribution. If it satisfies the normal distribution, adjust the lighting device to supplement light for the fabric and then acquire the first image again, and loop through steps 1 and 2;

[0018] Step 3: Until the probability distribution state of the pixel values in the obtained first grayscale image does not satisfy the normal distribution, calculate the first-order derivatives of each pixel point in the first grayscale image within a preset first range in the x-axis and y-axis directions, so as to obtain the gradient value and gradient direction of each pixel point position in the first grayscale image;

[0019] Step 4: For all pixel points in the first grayscale image, determine the local maximum gradient value among the multiple gradient values corresponding to the multiple pixel points within a preset second range in the x-axis and y-axis directions for each pixel point, and use the pixel points corresponding to all the local maximum gradient values in the first grayscale image to construct a set of first image contour points;

[0020] Step 5: Determine the average gray value of all pixel points in the set of first image contour points, preset a gray value upper threshold and a gray value lower threshold according to the average gray value, and screen all pixel points in the set of first image contour points according to the gray value upper threshold and the gray value lower threshold to obtain a set of second image contour points;

[0021] Step 6: According to the coordinates of each pixel point in the set of second image contour points in the first image, determine the distance between any pixel point in the set of second image contour points and other pixel points, and screen out the two pixel points with the closest distance as the adjacent points of the pixel point and connect them with a line;

[0022] Step 7: After connecting each pixel point in the set of second image contour points with a line, obtain the contour feature of the fabric in the first image;

[0023] Step 8: Match the contour feature with the samples in the preset style type - contour feature comparison library, and determine the style type corresponding to the contour feature with the highest matching degree as the style type corresponding to the fabric in the first image.

[0024] Preferably, the grasping of the fabric by the robotic arm based on the position of the grasping point includes:

[0025] Taking first and second visual images of the fabric by a binocular vision camera set on the robotic arm;

[0026] Analyzing the first and second visual images respectively to determine the first position information of the grasping point position in the first visual image and the second position information of the grasping point position in the second visual image;

[0027] Determining the first shooting distance between the first camera and the fabric and the second shooting distance between the second camera and the fabric in the binocular vision camera, and determining the relative position relationship of the grasping point position on the fabric relative to the binocular vision camera according to the first shooting distance, the second shooting distance, the preset relative distance and the shooting angle between the first camera and the second camera;

[0028] Determining the relative position relationship between the robotic arm gripper and the grasping point position according to the relative position relationship of the grasping point position on the fabric relative to the binocular vision camera, and automatically controlling the robotic arm to approach the grasping point position to grasp the fabric.

[0029] Preferably, the image recognition of the placed fabric and determining the corresponding overlock process according to the recognition result includes:

[0030] Taking a second image of the fabric placed on the workbench by a preset second camera;

[0031] Recognizing the second image to determine whether the style type corresponding to the fabric is the same as the recognition result corresponding to the first image;

[0032] If not, directly controlling the robotic arm to put the fabric on the conveyor belt for manual inspection;

[0033] If the same, determining the position where the overlock work needs to be carried out and determining the preset overlock rules corresponding to the fabric of this style type, and determining the needle dropping position when the fabric is overlocked at the hem according to the overlock rules;

[0034] Analyzing the second image, determining the specific model of the style type corresponding to the fabric according to the size of the fabric in the second image and the shooting distance between the second camera and the fabric, and further determining the overlock stroke corresponding to this model;

[0035] Determining the overlock process according to the needle dropping position and the overlock stroke when the fabric is overlocked at the hem.

[0036] Preferably, the completion of the blanking work after the overlock work on the fabric based on the overlock process includes:

[0037] Control the overlock sewing machine head to move to the needle dropping position and start overlock sewing work;

[0038] Move the fabric and capture the position image of the fabric during movement through the second camera;

[0039] Identify the position image and extract the edge contour of the fabric, and at the same time determine the included angle between the edge contour and the scale line on the overlock sewing platform;

[0040] Judge whether the included angle meets the preset overlock sewing standard. If not, control the overlock sewing machine head to adjust the angle for compensation and / or adjust the movement angle of the fabric until the overlock sewing work is completed.

[0041] Grab the fabric after the overlock sewing work is completed with a robotic arm and place it on a predetermined conveyor belt to complete the blanking work.

[0042] Preferably, during the overlock sewing work, the remaining length of the thread reel is detected in real time and the staff is reminded to replace it. Specifically, it includes:

[0043] Obtain the image information of the thread reel and determine the remaining radius of the thread reel in the image;

[0044] Determine the starting radius of the thread reel according to the preset thread reel information, determine the radius difference between the remaining radius and the starting radius, and remind the staff to replace the thread reel when the radius difference is lower than the preset difference threshold;

[0045] Or, preset a rotating fan blade on the thread reel, and pass the rotating fan blade through a photoelectric sensor. Determine the number of rotations of the thread reel through the counting of the photoelectric sensor and the number of rotating fan blades;

[0046] Determine the remaining number of turns according to the number of rotations of the thread reel and the preset number of turns of the thread reel corresponding to it. When the remaining number of turns is lower than the preset number of turns threshold, remind the staff to replace the thread reel, and reset the number of rotations of the thread reel after the thread reel is replaced.

[0047] Preferably, during the overlock sewing work, the state of the sewing needle is detected. Specifically, it includes:

[0048] Set laser generating devices above and below the overlock sewing platform respectively. Among them, the laser generating device above the overlock sewing platform is aligned with the highest needle tip position when the sewing needle lifts up during overlock sewing, and the laser generating device below the overlock sewing platform is aligned with the lowest needle tip position when the sewing needle presses down during overlock sewing;

[0049] Capture the highest needle tip position and the lowest needle tip position of the sewing needle through a preset third camera to obtain a third image;

[0050] Analyze the third image to determine whether there is a laser flashing point in the third image. If there are two laser flashing points at the same time, it is determined that the sewing needle is in a normal state;

[0051] If there is no laser flashing point, perform a self-check on the laser generating device to determine whether the laser generating device has a fault. If there is no fault, it is determined that the sewing needle has broken and the staff is reminded to replace it.

[0052] To achieve the above object, an embodiment of the present invention further provides an intelligent control system for an automatic bottom hemming machine, including:

[0053] A loading module for grasping and loading the fabric to be overedge stitched and adjusting the placement position;

[0054] A process determination module for performing image recognition on the placed fabric and determining the corresponding overedge stitching process according to the recognition result;

[0055] An overedge stitching module for performing overedge stitching on the fabric based on the overedge stitching process and then completing the unloading work.

[0056] Preferably, the loading module includes:

[0057] A first image acquisition unit for photographing the fabric on the conveyor belt through a preset first camera to obtain a first image;

[0058] An image analysis unit for analyzing the first image to determine the style type corresponding to the fabric in the first image;

[0059] A position determination unit for determining the position of the preset grasping point corresponding to the style type and the position where the overedge stitching work needs to be performed;

[0060] A placement determination unit for determining the corresponding preset placement method according to the position where the overedge stitching work needs to be performed on the fabric;

[0061] A grasping control unit for grasping the fabric by the robotic arm based on the grasping point position and placing the fabric on the overedge stitching platform based on the placement method.

[0062] Next, through the drawings and embodiments, the technical solutions of the present invention will be further described in detail. Description of the Drawings

[0063] The drawings are used to provide a further understanding of the present invention and constitute a part of the specification. They are used together with the embodiments of the present invention to explain the present invention and do not constitute a limitation to the present invention. In the drawings:

[0064] Figure 1 is a step flow chart of an intelligent control method for an automatic bottom hemming machine in an embodiment of the present invention;

[0065] Figure 2 This is a flowchart of the steps for grasping and feeding the fabric to be overlocked and adjusting its placement position in the embodiments of the present invention;

[0066] Figure 3 This is a schematic structural diagram of an intelligent control system for an automatic hemmer in the embodiments of the present invention. Specific Embodiments

[0067] The following describes the preferred embodiments of the present invention with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are only for the purpose of illustrating and explaining the present invention, and are not used to limit the present invention.

[0068] The embodiments of the present invention provide an intelligent control method for an automatic hemmer, as Figure 1 , including:

[0069] Step S1: Grasp and feed the fabric to be overlocked and adjust its placement position;

[0070] Step S2: Perform image recognition on the placed fabric, and determine the corresponding overlock process according to the recognition result;

[0071] Step S3: After performing the overlock work on the fabric based on the overlock process, complete the discharging work.

[0072] The working principle and beneficial effects of the above technical solution are as follows: The manipulator grasps and feeds the fabric to be overlocked and adjusts its placement position. By performing image recognition on the placed fabric, relevant information such as the style type, size model, and overlock thread pattern of the fabric is determined. According to the recognition result, the corresponding overlock process is determined. Finally, after performing the overlock work on the fabric based on the overlock process, the discharging work is completed. Based on the working process of manipulator recognition and grasping - fabric information recognition - automatic determination of overlock process - overlock work - discharging integration, through AI intelligent recognition and decision-making of images, the hemmer automatically performs the upper and lower fabrics and automatically completes the overlock work, reducing manual operation behavior and improving the overlock work efficiency of fabric hemming.

[0073] In a preferred embodiment, as Figure 2 , grasping and feeding the fabric to be overlocked and adjusting its placement position includes:

[0074] Step S11: Take a first image of the fabric on the conveyor belt through a preset first camera;

[0075] Step S12: Analyze the first image to determine the style type of the fabric in the first image;

[0076] Step S13: Determine the preset grasping point position corresponding to the style type and the position where the overlock work needs to be performed;

[0077] Step S14. Determine the corresponding preset placement method according to the position on the fabric where overedge sewing work needs to be performed.

[0078] Step S15. Grasp the fabric by the robotic arm based on the grasping point position, and place the fabric on the overedge sewing platform based on the placement method.

[0079] The working principle and beneficial effects of the above technical solution are as follows: When grasping and loading the fabric that needs to be overedge sewn, the fabric on the conveyor belt is photographed by a preset first camera to obtain a first image; the first image is analyzed to determine the style type corresponding to the fabric in the first image; determine the preset grasping point position and the position where overedge sewing work needs to be performed corresponding to the style type; determine the corresponding preset placement method according to the position on the fabric where overedge sewing work needs to be performed; grasp the fabric by the robotic arm based on the grasping point position, and place the fabric on the overedge sewing platform based on the placement method. Through the above method, the intelligent recognition and grasping of the fabric on the conveyor belt are realized, and the fabric is reasonably placed according to the overedge sewing requirements to improve the overedge sewing efficiency.

[0080] In a preferred embodiment, analyzing the first image to determine the style type corresponding to the fabric in the first image includes:

[0081] Step 1. Perform grayscale processing on the first image to obtain a first grayscale image, and use a Gaussian image blur filter to smooth and filter the first grayscale image to remove image noise.

[0082] Step 2. Determine whether the probability distribution state of the pixel values in the first grayscale image satisfies the normal distribution. If it satisfies the normal distribution, adjust the lighting device to supplement light to the fabric and then obtain the first image again, and loop to execute Step 1 and Step 2.

[0083] Step 3. Until the probability distribution state of the pixel values in the obtained first grayscale image does not satisfy the normal distribution, calculate the first-order derivatives of each pixel point in the first grayscale image within a preset first range in the x-axis and y-axis directions, so as to obtain the gradient value and gradient direction of each pixel point position in the first grayscale image.

[0084] Step 4. For all pixel points in the first grayscale image, determine the local maximum gradient value among the multiple gradient values corresponding to the multiple pixel points within a preset second range in the x-axis and y-axis directions for each pixel point, and use the pixel points corresponding to all the local maximum gradient values in the first grayscale image to construct a first image contour point set.

[0085] Step 5: Determine the average gray value of all pixel points in the first image contour point set. Preset the upper threshold and lower threshold of the gray value according to the average gray value. Screen all pixel points in the first image contour point set according to the upper threshold and lower threshold of the gray value to obtain the second image contour point set;

[0086] Step 6: According to the coordinates of each pixel point in the second image contour point set in the first image, determine the distance between any pixel point in the second image contour point set and other pixel points, and screen out the two pixel points with the closest distance as the adjacent points of this pixel point and perform line connection;

[0087] Step 7: After connecting each pixel point in the second image contour point set with a line, obtain the contour feature of the fabric in the first image;

[0088] Step 8: Match the contour feature with the samples in the preset style type - contour feature comparison library, and determine the style type corresponding to the contour feature with the highest matching degree as the style type corresponding to the fabric in the first image.

[0089] The working principle and beneficial effects of the above technical solution are as follows: By grayscaling the first image to obtain the first grayscale image, using a Gaussian image blur filter to smooth and filter the first grayscale image to remove image noise and reduce the impact of salt-and-pepper noise on the recognition result, determining whether the probability distribution state of the pixel values in the first grayscale image satisfies the normal distribution. If it satisfies the normal distribution, it is determined that the contrast of the pixel points in the image is not high enough. After adjusting the lighting device to supplement light to the fabric to increase the contrast, the first image is acquired again, and the above steps are cyclically executed until the probability distribution state of the pixel values in the obtained first grayscale image does not satisfy the normal distribution. Then, according to the first-order derivatives of each pixel point in the first grayscale image within a preset first range in the x-axis and y-axis directions, the gradient value and gradient direction of each pixel point position in the first grayscale image are obtained. The gradient value can reflect the contrast of pixel values in adjacent ranges in the image. For all pixel points in the first grayscale image, determine the local maximum gradient value among the multiple gradient values corresponding to multiple pixel points within a preset second range in the x-axis and y-axis directions for each pixel point. By taking the pixel point corresponding to the local maximum gradient value as the representative point of all pixel points within a range, it is used as the basic point of the image object contour for subsequent work, which can effectively identify all contour feature points in the image. Use the pixel points corresponding to all local maximum gradient values in the first grayscale image to construct the first image contour point set. Determine the average grayscale value of all pixel points in the first image contour point set, preset the upper grayscale value threshold and the lower grayscale value threshold according to the average grayscale value, and screen all pixel points in the first image contour point set according to the upper grayscale value threshold and the lower grayscale value threshold to obtain the second image contour point set. By setting the upper grayscale value threshold and the lower grayscale value threshold, the gross error points are screened out to achieve the refinement of the image contour. According to the coordinates of each pixel point in the second image contour point set in the first image, determine the distance between any pixel point in the second image contour point set and other pixel points, and screen out the two pixel points with the closest distance as the adjacent points of the pixel point and connect them with a line, thereby realizing the transformation of the contour from points to lines. During the connection process, the pixel points passed by the connection line are assigned values according to the average grayscale value of all pixel points in the first image contour point set to further refine the image contour. After connecting each pixel point in the second image contour point set with a line, the contour feature of the fabric in the first image is obtained. Compare the contour feature with the samples in the preset style type - contour feature comparison library, and determine the style type corresponding to the contour feature with the highest matching degree as the style type corresponding to the fabric in the first image. Thus, the intelligent processing and recognition of the image are realized, the accuracy of contour recognition in the image is improved, and finally the accurate search for the style type corresponding to the fabric is realized.

[0090] In a preferred embodiment, the grasping of the fabric by the robotic arm based on the grasping point position includes:

[0091] Taking pictures of the fabric through a binocular vision camera arranged on the robotic arm to obtain a first visual image and a second visual image respectively;

[0092] Analyzing the first visual image and the second visual image respectively to determine the first position information of the grasping point position in the first visual image and the second position information of the grasping point position in the second visual image;

[0093] Determining the first shooting distance between the first camera in the binocular vision camera and the fabric and the second shooting distance between the second camera and the fabric, and determining the relative position relationship of the grasping point position on the fabric relative to the binocular vision camera according to the first shooting distance, the second shooting distance, the preset relative distance between the first camera and the second camera, and the shooting angle;

[0094] Determining the relative position relationship between the robotic arm gripper and the grasping point position according to the relative position relationship of the grasping point position on the fabric relative to the binocular vision camera, and automatically controlling the robotic arm to approach the grasping point position to grasp the fabric.

[0095] The working principle and beneficial effects of the above technical solution are as follows: When the fabric is grasped by the robotic arm based on the position of the grasping point, the binocular vision camera installed on the robotic arm takes pictures of the fabric to obtain the first visual image and the second visual image respectively, and uses the accuracy of position recognition by binocular vision detection to locate the specific position on the fabric through the images. Analyze the first visual image and the second visual image respectively to determine the first position information of the grasping point position in the first visual image and the second position information of the grasping point position in the second visual image. Since the shooting points of the first visual image and the second visual image are different, the positions of the grasping point position in the first visual image and the second visual image are also different; determine the first shooting distance between the first camera and the fabric and the second shooting distance between the second camera and the fabric in the binocular vision camera, and determine the relative position relationship of the grasping point position on the fabric relative to the binocular vision camera according to the first shooting distance, the second shooting distance, the preset relative distance between the first camera and the second camera, and the shooting angle, so as to unify the corresponding grasping point positions in the first visual image and the second visual image, and determine the relative position relationship of the grasping point position on the fabric relative to the binocular vision camera according to the parameters of the binocular vision camera, such as the camera shooting line-of-sight angle, the relative position distance of the camera, the zoom ratio of the binocular vision camera, etc.; determine the relative position relationship between the robotic arm gripper and the grasping point position according to the relative position relationship of the grasping point position on the fabric relative to the binocular vision camera, and automatically control the robotic arm to approach the grasping point position to grasp the fabric. It realizes the intelligent control of the robotic arm and provides a basis for automatic grasping and feeding, thereby improving the automation degree of production operations.

[0096] In a preferred embodiment, image recognition is performed on the placed fabric, and determining the corresponding overlock process according to the recognition result includes:

[0097] The second image is obtained by taking pictures of the fabric placed on the workbench through a preset second camera;

[0098] Perform recognition on the second image to determine whether the style type corresponding to the fabric is the same as the recognition result corresponding to the first image;

[0099] If they are not the same, directly control the robotic arm to put the fabric on the conveyor belt for manual inspection;

[0100] If they are the same, determine the position where the overlock work needs to be performed and determine the preset overlock rule corresponding to the fabric of this style type, and determine the needle dropping position when the fabric is overlocked at the lower hem according to the overlock rule;

[0101] Analyze the second image, determine the specific model of the pattern type corresponding to the fabric according to the size of the fabric in the second image and the shooting distance between the second camera and the fabric, and further determine the corresponding overlock stitch travel under this model;

[0102] Determine the overlock stitch process according to the needle drop position and overlock stitch travel when overlocking the hem of the fabric.

[0103] The working principle and beneficial effects of the above technical solution are as follows: During the process of determining the overlock stitch process, the fabric placed on the workbench is photographed by a preset second camera to obtain a second image. By identifying the second image, it is determined whether the pattern type corresponding to the fabric is the same as the recognition result corresponding to the first image, so as to further judge whether there are situations such as inaccurate placement position or fabric folding and wrinkling during the placement of the robotic arm. After such situations occur, it is not conducive to the subsequent overlock stitch work and will cause irreparable situations such as overlock stitch errors in the overlock stitch work. When it is determined that the pattern type corresponding to the fabric is different from the recognition result corresponding to the first image, the robotic arm is directly controlled to put the fabric onto the conveyor belt for manual inspection, and the fabric is grabbed and loaded by the operator for overlock stitch work. If they are the same, it is determined that accurate overlock stitch work can be carried out. After determining the position where the overlock stitch work needs to be carried out, determine the preset overlock stitch rules corresponding to the fabric of this pattern type. The overlock stitch rules include the pattern of the overlock stitch line corresponding to the fabric of this pattern type during the overlock stitch work, and the overlock stitch methods such as three-thread, four-thread or five-thread overlock stitch edge-locking, the needle drop point, the selection of overlock stitch line color and model material, the deviation of the line during the overlock stitch process, etc. Determine the needle drop position when overlocking the hem of the fabric according to the overlock stitch rules; Analyze the second image, determine the specific model of the pattern type corresponding to the fabric according to the size of the fabric in the second image and the shooting distance between the second camera and the fabric, and further determine the corresponding overlock stitch travel under this model. The overlock stitch travel may include information such as the overlock stitch path, overlock stitch span, overlock stitch frequency or speed, etc. Since the overlock stitch travels required for fabrics of different model pattern types are inconsistent, model determination is required. Determine the overlock stitch process according to the needle drop position and overlock stitch travel when overlocking the hem of the fabric. Thus, by determining the corresponding overlock stitch process according to the fabric pattern type, the overlock stitch machine can automatically complete the overlock stitch work.

[0104] In a preferred embodiment, after the overlock stitch work on the fabric is completed based on the overlock stitch process, the material unloading work includes:

[0105] Control the overlock stitch head to move to the needle drop position and start the overlock stitch work;

[0106] Move the fabric and take a position image of the fabric during the movement through the second camera.

[0107] Identify the image at this position, extract the edge contour of the fabric, and determine the angle between the edge contour and the scale line on the overlock platform;

[0108] Determine whether the angle meets the preset overlock standard. If not, control the overlock head to adjust the angle for compensation and / or adjust the moving angle of the fabric until the overlock work is completed;

[0109] After the overlock work is completed, grab the fabric with a robotic arm and place it on a predetermined conveyor belt to complete the blanking work.

[0110] The working principle and beneficial effects of the above technical solution are as follows: Control the overlock head to move to the needle dropping position and start the overlock work; During the process of moving the fabric, use the second camera to capture the position image of the fabric during movement; Identify the image at this position, extract the edge contour of the fabric, and determine the angle between the edge contour and the scale line drawn on the overlock platform; Then determine whether the angle meets the preset overlock standard. If not, it is determined that the fabric has a position deflection during movement and needs to be corrected in a timely manner. It is necessary to control the overlock head to adjust the angle for compensation and / or adjust the moving angle of the fabric until the overlock work is completed. After the overlock work is completed, grab the fabric with a robotic arm and place it on a predetermined conveyor belt to complete the blanking work. Thus, automatic correction of the overlock line during the overlock process is achieved, reducing the error rate during the overlock process, correcting in a timely manner in case of deviation, and preventing material waste caused by incorrect production.

[0111] In a preferred embodiment, during the overlock work, the remaining length of the thread roll is detected in real time and the staff is reminded to replace it. Specifically, it includes:

[0112] Obtain the image information of the thread roll and determine the remaining radius of the thread roll in the image;

[0113] Determine the starting radius of the thread roll according to the preset thread roll information, determine the radius difference between the remaining radius and the starting radius, and remind the staff to replace the thread roll when the radius difference is lower than the preset difference threshold;

[0114] Alternatively, a rotating fan blade is preset on the thread roll, and the rotating fan blade is passed through a photoelectric sensor. The number of rotations of the thread roll is determined by counting through the photoelectric sensor and the number of rotating fan blades;

[0115] Determine the remaining number of turns according to the number of rotations of the thread roll and the preset number of turns of the thread roll corresponding to it. When the remaining number of turns is lower than the preset number of turns threshold, remind the staff to replace the thread roll, and reset the number of rotations of the thread roll after the thread roll is replaced.

[0116] The working principle and beneficial effects of the above technical solution are as follows: By detecting the remaining length of the thread roll in real time and reminding the staff to replace it, the production efficiency of the hemming machine is improved. During the detection process, the image information of the thread roll can be obtained by setting up a camera, and the remaining radius of the thread roll in the image can be determined; According to the preset thread roll information, the starting radius of the currently used thread roll is determined, and the radius difference between the remaining radius and the starting radius is determined. When the radius difference is lower than the preset difference threshold, the staff is reminded to replace the thread roll; In the further inspection process, information such as the length of the thread roll and the diameter of the thread can be determined through the recognition result of the image to calculate the remaining length of the overedge sewing thread. The calculation process is as follows: Based on the spiral radius r = i + jθ in the polar coordinate system, where i is the starting radius of the spiral, j is the radius increase rate, and θ is a preset correction constant. The radius increase rate can be determined by the diameter of the overedge sewing thread. If the diameter of the overedge sewing thread is S, the radius of the thread roll will increase by S for each turn of the spiral. So j = S / 2π. Considering that the thread roll has a drum, the starting radius i of the thread roll = r0 + S / 4, where r0 is the radius of the drum. So the radius of the thread roll is r = r0 + S / 4 + Sθ / 2π. Therefore, in the case of the same plane, the length of the tightly wound spiral overedge sewing thread satisfies the formula d l = rd θ = (r0 + S / 4 + Sθ / 2π)d θ , after integration and multiplying by the length L of the thread roll, the length l of the overedge sewing thread on the thread roll is obtained as l = L[(r0 + S / 4)θ + Sθ 2 / 4π]. It is also possible to preset a rotating fan blade on the thread roll, and connect the rotating fan blade to a photoelectric sensor. The number of rotations of the thread roll is determined by counting through the photoelectric sensor and the number of rotating fan blades; According to the number of rotations of the thread roll and the preset number of turns of the thread roll corresponding to it, the remaining number of turns is determined. When the remaining number of turns is lower than the preset number of turns threshold, the staff is reminded to replace the thread roll, and the number of rotations of the thread roll is reset after the thread roll is replaced. The length of the overedge sewing thread is calculated in a relatively cheap and accurate way. Through the above method, the monitoring of the remaining length of the overedge sewing thread is realized, which enables the staff to understand the remaining situation of the thread roll in advance, so as to replace the thread roll seamlessly in time and improve the production efficiency.

[0117] In a preferred embodiment, during the overedge sewing process, the state of the sewing needle is detected, specifically including:

[0118] Laser generating devices are respectively arranged above and below the overedge sewing platform. Among them, the laser generating device above the overedge sewing platform is aligned with the highest tip position of the sewing needle when it moves upward and lifts the needle during the overedge sewing process, and the laser generating device below the overedge sewing platform is aligned with the lowest tip position of the sewing needle when it moves downward and presses the needle during the overedge sewing process;

[0119] The highest tip position and the lowest tip position of the sewing needle are photographed by a preset third camera to obtain a third image;

[0120] Analyze the third image to determine whether there are laser flashing points in the third image. If there are two laser flashing points at the same time, it is determined that the sewing needle is in a normal state;

[0121] If there are no laser flashing points, self-check the laser generating device to determine whether the laser generating device fails. If there is no failure, it is determined that the sewing needle is broken and the staff is reminded to replace it.

[0122] The working principle and beneficial effects of the above technical solution are as follows: Laser generating devices are respectively arranged above and below the overlock platform. Among them, the laser generating device above the overlock platform is aligned with the highest tip position of the sewing needle when it lifts the needle upward during the overlock process, and the laser generating device below the overlock platform is aligned with the lowest tip position of the sewing needle when it presses the needle downward during the overlock process; Subsequently, the highest tip position and the lowest tip position of the sewing needle are photographed by a preset third camera to obtain a third image; Further analyze the third image to determine whether there are laser flashing points in the third image. If there are two laser flashing points at the same time, it is determined that the sewing needle is in a normal state; If there are no laser flashing points, self-check the laser generating device to determine whether the laser generating device fails. If there is no failure, it is determined that the sewing needle is broken and the staff is reminded to replace it. Thus, the detection of the working state of the sewing needle is realized, and the efficiency loss caused by the inability to resume production in time after the sewing needle is bent or broken is prevented.

[0123] To achieve the above objectives, an intelligent control system for an automatic bottom hemmer is further provided in an embodiment of the present invention, as Figure 3 , including:

[0124] A feeding module for grasping and feeding the fabric to be overlocked and adjusting its placement position;

[0125] A process determination module for performing image recognition on the placed fabric and determining the corresponding overlock process according to the recognition result;

[0126] An overlock module for performing overlock work on the fabric based on the overlock process and then completing the discharging work.

[0127] The working principle and beneficial effects of the above technical solution are as follows: The feeding module uses a robotic arm to grasp and feed the fabric to be overlocked and adjust its placement position. By performing image recognition on the placed fabric, relevant information such as the style type, size model, and overlock thread pattern corresponding to the fabric is determined. The process determination module determines the corresponding overlock process based on the recognition result. Finally, the overlock module performs the overlock work on the fabric based on the overlock process and completes the discharging work. Based on the working process of robotic arm recognition and grasping - fabric information recognition - automatic overlock process determination - overlock work - discharging integration, intelligent recognition and decision-making of images are performed by AI, reducing manual operation behaviors and improving the overlock work efficiency of the fabric hem.

[0128] In a preferred embodiment, the feeding module includes:

[0129] The first image acquisition unit is used to capture a first image of the fabric on the conveyor belt through a preset first camera;

[0130] The image analysis unit is used to analyze the first image and determine the style type corresponding to the fabric in the first image;

[0131] The position determination unit is used to determine the preset grasping point position corresponding to the style type and the position where the overlock work needs to be performed;

[0132] The placement determination unit is used to determine the corresponding preset placement method according to the position where the overlock work needs to be performed on the fabric;

[0133] The grasping control unit is used to grasp the fabric by the robotic arm based on the grasping point position and place the fabric on the overlock platform based on the placement method.

[0134] The working principle and beneficial effects of the above technical solution are as follows: The first image acquisition unit captures a first image of the fabric on the conveyor belt through a preset first camera; the image analysis unit analyzes the first image and determines the style type corresponding to the fabric in the first image; the position determination unit determines the preset grasping point position corresponding to the style type and the position where the overlock work needs to be performed; the placement determination unit determines the corresponding preset placement method according to the position where the overlock work needs to be performed on the fabric; the grasping control unit grasps the fabric by the robotic arm based on the grasping point position and places the fabric on the overlock platform based on the placement method. Through the above method, intelligent recognition and grasping of the fabric on the conveyor belt are realized, and the fabric is reasonably placed according to the overlock requirements to improve the overlock efficiency.

[0135] Obviously, those skilled in the art can make various modifications and variations to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention also intends to include these modifications and variations.

Claims

1. An intelligent control method for an automatic hemmer, characterized in that, Including: Grasping and loading the fabric to be overlock stitched and adjusting its placement position; Performing image recognition on the placed fabric and determining the corresponding overlock stitching process according to the recognition result; Completing the unloading work after performing the overlock stitching work on the fabric based on the overlock stitching process; Among them, the grasping and loading of the fabric to be overlock stitched and adjusting its placement position includes: Taking a first image of the fabric on the conveyor belt through a preset first camera; Analyzing the first image to determine the style type corresponding to the fabric in the first image; Determining the preset grasping point position corresponding to the style type and the position where the overlock stitching work needs to be performed; Determining the corresponding preset placement method according to the position on the fabric where the overlock stitching work needs to be performed; Grasping the fabric based on the grasping point position by the robotic arm, and placing the fabric on the overlock stitching platform based on the placement method. When grasping the fabric based on the grasping point position by the robotic arm, taking a first visual image and a second visual image of the fabric respectively through the binocular vision camera set on the robotic arm; analyzing the first visual image and the second visual image respectively to determine the first position information of the grasping point position in the first visual image and the second position information of the grasping point position in the second visual image; determining the first shooting distance between the first camera and the fabric and the second shooting distance between the second camera and the fabric in the binocular vision camera, and determining the relative position relationship of the grasping point position on the fabric relative to the binocular vision camera according to the first shooting distance, the second shooting distance, the preset relative distance and the shooting angle between the first camera and the second camera; determining the relative position relationship between the robotic arm gripper and the grasping point position according to the relative position relationship of the grasping point position on the fabric relative to the binocular vision camera, and automatically controlling the robotic arm to approach the grasping point position to grasp the fabric.

2. The intelligent control method of an automatic hemmer according to claim 1, wherein, The analyzing the first image to determine the style type corresponding to the fabric in the first image includes: Step 1: Performing grayscale processing on the first image to obtain a first grayscale image, and using a Gaussian image blur filter to perform smoothing filtering on the first grayscale image to remove image noise; Step 2: Determining whether the probability distribution state of the pixel values in the first grayscale image satisfies the normal distribution. If it satisfies the normal distribution, adjusting the lighting device to supplement light to the fabric and then obtaining the first image again, and looping to execute Step 1 and Step 2; Step 3: Until the probability distribution state of the pixel values in the obtained first grayscale image does not satisfy the normal distribution, then obtaining the gradient value and gradient direction of each pixel point position in the first grayscale image according to the first-order derivative within a preset first range in the x-axis and y-axis directions of each pixel point in the first grayscale image; Step 4: For all pixel points in the first grayscale image, determining the local maximum gradient value among the multiple gradient values corresponding to the multiple pixel points within a preset second range in the x-axis and y-axis directions of each pixel point, and constructing a first image contour point set using the pixel points corresponding to all the local maximum gradient values in the first grayscale image; Step 5: Determine the average gray value of all pixel points in the first image contour point set. According to the average gray value, preset the upper gray value threshold and the lower gray value threshold, and screen all pixel points in the first image contour point set according to the upper gray value threshold and the lower gray value threshold to obtain the second image contour point set; Step 6: According to the coordinates of each pixel point in the second image contour point set in the first image, determine the distance between any pixel point in the second image contour point set and other pixel points, and screen out the two pixel points with the closest distance as the adjacent points of this pixel point and connect them with a line; Step 7: After connecting each pixel point in the second image contour point set with a line, obtain the contour feature of the fabric in the first image; Step 8: Match the contour feature with the samples in the preset style type - contour feature comparison library, and determine the style type corresponding to the contour feature with the highest matching degree as the style type corresponding to the fabric in the first image.

3. The intelligent control method of an automatic hemmer according to claim 1, characterized in that, The image recognition of the placed fabric and determining the corresponding overlock sewing process according to the recognition result includes: Shoot the fabric placed on the workbench through a preset second camera to obtain a second image; Recognize the second image to determine whether the style type corresponding to the fabric is the same as the recognition result corresponding to the first image; If they are not the same, directly control the robotic arm to place the fabric on the conveyor belt for manual inspection; If they are the same, determine the position where the overlock sewing work needs to be carried out and determine the preset overlock sewing rule corresponding to the fabric of this style type, and determine the needle dropping position when the fabric is sewn at the hem according to the overlock sewing rule; Analyze the second image, determine the specific model of the style type corresponding to the fabric according to the size of the fabric in the second image and the shooting distance between the second camera and the fabric, and further determine the corresponding overlock sewing stroke under this model; Determine the overlock sewing process according to the needle dropping position and the overlock sewing stroke when the fabric is sewn at the hem.

4. The intelligent control method of an automatic lower hem machine according to claim 1, characterized in that, The blanking work is completed after the fabric is sewn based on the overlock sewing process, including: Control the overlock sewing head to move to the needle dropping position and start the overlock sewing work; Move the fabric and shoot the position image of the fabric during the movement through the second camera; Recognize the position image and extract the edge contour of the fabric, and at the same time determine the angle between the edge contour and the scale line on the overlock sewing platform; Judge whether the angle meets the preset overlock sewing standard. If it does not meet, control the overlock sewing head to adjust the angle for compensation and / or adjust the movement angle of the fabric until the overlock sewing work is completed; Grab the sewn fabric with the robotic arm and place it on the predetermined conveyor belt to complete the blanking work.

5. The intelligent control method of an automatic lower hem machine according to claim 1, wherein, During the overlock sewing process, the remaining length of the thread reel is detected in real time and the staff is reminded to replace it. Specifically, it includes: Obtain the image information of the thread reel and determine the remaining radius of the thread reel in the image; Determine the starting radius of the thread reel according to the preset thread reel information, determine the radius difference between the remaining radius and the starting radius, and remind the staff to replace the thread reel when the radius difference is lower than the preset difference threshold; Alternatively, a rotating fan blade is preset on the online coil, and the rotating fan blade is counted by a photoelectric sensor, and the number of rotations of the coil is determined by the number of rotations of the rotating fan blade and the number of the rotating fan blades; The remaining number of turns is determined according to the number of rotations of the coil and the preset number of turns of the coil corresponding to the coil. When the remaining number of turns is lower than the preset turn threshold, the staff is reminded to replace the coil, and the number of rotations of the coil is reset after the coil is replaced.

6. The intelligent control method of an automatic hemmer according to claim 1, wherein During the overlock sewing process, the state of the sewing needle is detected, specifically including: Laser generating devices are respectively arranged above and below the overlock sewing platform. Among them, the laser generating device above the overlock sewing platform is aligned with the highest needle tip position of the sewing needle when it is lifted upward during overlock sewing, and the laser generating device below the overlock sewing platform is aligned with the lowest needle tip position of the sewing needle when it is pressed downward during overlock sewing; The highest needle tip position and the lowest needle tip position of the sewing needle are photographed by a preset third camera to obtain a third image; The third image is analyzed to determine whether there are laser flashing points in the third image. If there are two laser flashing points at the same time, it is determined that the state of the sewing needle is normal; If there are no laser flashing points, the laser generating device is self-checked to determine whether the laser generating device fails. If no failure occurs, it is determined that the sewing needle is broken and the staff is reminded to replace it.

7. An intelligent control system for an automatic hemmer, characterized in that, Including: A feeding module for grasping and feeding the fabric to be overlock sewn and adjusting the placement position; A process determination module for performing image recognition on the placed fabric and determining the corresponding overlock sewing process according to the recognition result; An overlock sewing module for performing overlock sewing on the fabric based on the overlock sewing process and then completing the discharging work; Among them, the feeding module includes: A first image acquisition unit for photographing the fabric on the conveyor belt through a preset first camera to obtain a first image; An image analysis unit for analyzing the first image to determine the style type corresponding to the fabric in the first image; A position determination unit for determining the preset grasping point position corresponding to the style type and the position where overlock sewing work needs to be performed; A placement determination unit for determining the corresponding preset placement method according to the position where overlock sewing work needs to be performed on the fabric; The grasping control unit is used to grasp the fabric by the robotic arm based on the grasping point position and place the fabric on the overlock platform based on the placement method. When the grasping control unit grasps the fabric by the robotic arm based on the grasping point position, the binocular vision cameras arranged on the robotic arm are used to take pictures of the fabric to obtain a first visual image and a second visual image respectively. The first visual image and the second visual image are analyzed respectively to determine the first position information of the grasping point position in the first visual image and the second position information of the grasping point position in the second visual image. The first shooting distance between the first camera and the fabric and the second shooting distance between the second camera and the fabric in the binocular vision cameras are determined, and the relative position relationship of the grasping point position on the fabric relative to the binocular vision cameras is determined according to the first shooting distance, the second shooting distance, the preset relative distance between the first camera and the second camera, and the shooting angle. The relative position relationship between the robotic arm gripper and the grasping point position is determined according to the relative position relationship of the grasping point position on the fabric relative to the binocular vision cameras, and the robotic arm is automatically controlled to approach the grasping point position to grasp the fabric.

Images