A hard carton automatic feeding system for a tobacco packaging machine

Abstract

This invention relates to an automatic feeding system for rigid carton boxes in tobacco packaging machines, comprising a material transfer module, a material storage module, and a material conveying module. The material transfer module includes a robot mounting cabinet, a robot lifting column module, a robot module, suction cups, and a 3D vision module. The robot mounting cabinet is used to mount and fix the robot lifting column module, the robot module is mounted on the lifting column module, and the suction cups and 3D vision module are mounted at the end of the robot module. The material storage module is located on both sides of the robot module and includes a rigid carton material storage tray and a partition paper material storage tray. The material conveying module includes a material conveyor belt and a blocking and positioning device. This invention enables automatic feeding of rigid carton materials, replacing manual feeding of tobacco packaging machines, saving labor costs, improving feeding efficiency, and achieving a high degree of automation.

Description

Technical Field

[0001] This invention belongs to the field of tobacco packaging technology, specifically relating to an automatic feeding system for rigid carton boxes used in tobacco packaging machines. Background Technology

[0002] Currently, the feeding of rigid carton boxes is mostly done manually. Operators manually open the cartons containing the finished rigid carton boxes according to the machine's operating status, remove the moisture-proof plastic film lining the cartons, and ensure that the front and bottom surfaces of each stack are correctly oriented. Finally, each rigid carton box is removed and stacked in sequence on the filling machine's conveyor belt. When the tobacco packaging machine's speed and capacity are not high, 1-2 operators continuously feeding the boxes can meet production needs.

[0003] Traditional rigid carton feeding methods require 1-2 operators to continuously feed the cartons, resulting in low efficiency, high labor intensity, and a chaotic feeding environment. During on-site work, it has been observed that some cartons are incorrectly loaded (left-right or right-side up), and operators easily overlook these errors, severely impacting the efficiency of subsequent automated packaging. Rigid cartons possess distinctive features that can be extracted using deep learning methods, employing convolutional neural networks to learn features and distinguish directions—a highly efficient solution. Furthermore, the feeding speed of rigid cartons needs to match the feeding speed of the main machine. Manual feeding leads to problems such as untimely feeding, heavy personnel burden, and significant errors in feeding position, causing equipment downtime. Moreover, the quantity of rigid cartons fed relies primarily on timed recording by on-site workers, resulting in a large workload and poor accuracy. In addition, in the past, each machine required 1-2 dedicated workers for feeding, leading to high learning and operating costs. The manual feeding method for rigid cartons has hindered the automation and intelligent development of tobacco packaging machines. Summary of the Invention

[0004] The purpose of this invention is to provide an automatic feeding system for rigid cartons in tobacco packaging machines, which replaces manual labor in judging the front and bottom surfaces and orientation of rigid carton stacks and counting quantities to reduce labor costs; solves the problems of on-site chaos and disordered material placement during manual feeding; adapts to various production speed requirements of tobacco packaging machines; and assists in the automation and intelligent upgrading of tobacco packaging equipment.

[0005] To achieve the above-mentioned objectives, the following technical solution is employed:

[0006] An automatic feeding system for rigid cartons in a tobacco packaging machine includes a material transfer module, a material storage module, and a material conveying module. The material transfer module is located at the front end of the material conveying module and is used to transfer the rigid cartons from the material storage module to the material conveying module.

[0007] The material transfer module includes a robot mounting cabinet, a robot lifting column module, a robot module, a suction cup, and a 3D vision module; the robot mounting cabinet is used to install and fix the robot lifting column module, the robot module is installed on the lifting column module, and the suction cup and 3D vision module are installed at the end of the robot module.

[0008] The material storage module is located on both sides of the robot module and includes a rigid box material storage tray and a partition paper material storage tray.

[0009] The material conveying module includes a material conveyor belt and a blocking and positioning device. The blocking and positioning device is installed at the front end of the material conveyor belt and includes a pair of guide rod cylinders and a baffle connected to the guide rod cylinders.

[0010] Furthermore, the material transfer module also includes a waste removal module, an electrical control cabinet, and connectors. The waste removal module is fixed on top of the electrical control cabinet, and the electrical control cabinet is connected to the robot mounting cabinet via connectors.

[0011] Furthermore, the partition paper is stored in the partition paper material storage tray. The partition paper includes 6 cells. The rigid strip boxes are arranged in groups of 15 in the 6 cells of the partition paper. Each rigid strip box material storage tray holds 10 layers of rigid strip box material.

[0012] Furthermore, the deep learning-based visual detection and localization system for hard strip boxes includes the following steps:

[0013] S1. Use a depth camera to capture images of the actual production workshop where rigid strip boxes are stacked and loaded;

[0014] S2. Divide the images collected in step S1 into training dataset, validation dataset and test dataset in a ratio of 8:1:1.

[0015] S3. Training dataset data annotation: Connect the four corners of the front face (containing the cigarette brand, barcode, QR code, and place of origin information) and label it as "box"; connect the four corners of the bottom face and label it as "reverse box"; connect the outer frame of the outer hard box containing the cigarette brand, barcode, and QR code as a whole and label it as "box head".

[0016] S4. Establish a hard box recognition network model based on Mask R-CNN network, using ResNet50+FPN as the backbone neural network and Region Proposal Network (RPN) as the feature map for convolution. Finally, the bounding box, category, and mask are used as outputs to generate the hard box recognition network model.

[0017] S5. Implement the orientation determination of a single front-facing hard box. For a front-facing hard box, after using a deep learning hard box recognition network, a single box will have two recognition labels, box and box head. At the same time, the position coordinates of their respective center points in the image are obtained. The direction angle from box head to box is calculated to obtain the orientation determination and store it in the angle attribute of box.

[0018] S6. Implement hard box stack counting and judgment. Take the image origin as the starting point and the center point of the box closest to the origin as the starting point of the first group. Then, similarly calculate the center point of the box closest to the starting point of the first group. Calculate the distance between the two center points to judge whether they meet the threshold. Record the adjacent boxes as the new starting point until each hard box in the image is assigned to each group.

[0019] S7. Determine the consistency of orientation for each group and the final grasping posture for each pile. Judge the angle and direction of each hard pack in each group. If the angle deviation exceeds the threshold, the orientation is judged to be inconsistent and recorded as a rejected group. Record it in the group's attributes. When the orientation is consistent, take the median angle of each pack as the overall suction angle.

[0020] S8. Determine the capture center of each group, and calculate the average x and y values ​​of each hard box in the image as the center position of each group.

[0021] Furthermore, a pair of blocking and positioning devices are set on both sides of the material conveyor belt. When the robot module places a set of rigid strip box materials onto the material conveyor belt, the blocking and positioning devices close, and the blocking devices provide a supporting force along the X direction to the set of rigid strip box materials placed on the conveyor belt.

[0022] Further, the process flow is as follows: The tobacco packaging machine sends a material request signal to the system, the robot lifting column module rises to the designated photo position, the 3D vision module takes a photo of a layer of hard carton material on a certain side of the hard carton material storage tray, and removes materials that cannot be identified; then, it grabs the materials that meet the requirements from the remaining cells and moves them above the conveyor belt of the material conveying module, the blocking positioning device closes to block a group of hard carton materials, so that the hard carton materials stand stably on the conveyor belt; then, the suction cup releases this group of hard carton materials, the blocking positioning device opens, and this group of hard carton materials is then conveyed into the tobacco packaging machine along with the conveyor belt;

[0023] After all six cells of material in a layer of rigid box material have been grabbed, the 3D vision module takes a picture of the partition paper and grabs the partition paper into the partition paper material storage tray. After the rigid box material on one side of the tray is grabbed, the robot immediately grabs the rigid box material on the other side of the tray. At the same time, the system sends a material request signal, and a person or AGV cart drags away the empty rigid box material storage tray and replaces it with a full stack of rigid box material storage trays.

[0024] Compared with the prior art, the beneficial effects of the present invention are:

[0025] 1) This invention enables automatic feeding of materials for hard pack boxes, replacing manual feeding of tobacco packaging machines, saving labor costs, improving feeding efficiency, and achieving a high degree of automation.

[0026] 2) This invention can ensure the cleanliness and order of the feeding site, and realize the stable, efficient and rapid feeding of materials. It can place materials neatly and accurately on the conveyor belt. Compared with manual feeding, it has high efficiency and continuity, avoids the downtime of packaging machine caused by feeding problems, and improves the stability and safety of equipment operation.

[0027] 3) The automatic feeding system for rigid box robots designed in this invention can adjust the feeding rhythm according to the needs of the packaging machine equipment, and can intelligently adapt to different feeding speed requirements of the equipment. It has the characteristics of high flexibility and wide applicability. Attached Figure Description

[0028] Figure 1 This invention relates to an automatic feeding system for rigid carton robots.

[0029] Figure 2 This is a schematic diagram of the material transfer module structure of the present invention.

[0030] Figure 3 This is a schematic diagram of the material storage module structure of the present invention.

[0031] Figure 4 This is a schematic diagram of the rigid strip box structure.

[0032] Figure 5 This is a schematic diagram of the partition paper structure.

[0033] Figure 6 This is a schematic diagram of a single layer of rigid strip box material.

[0034] Figure 7 This is a schematic diagram of the material conveying module.

[0035] Figure 8a and Figure 8b These are schematic diagrams showing the blocking device in both closed and open states.

[0036] Figure 9 This is a schematic diagram of a deep learning-based visual detection and localization system for rigid boxes.

[0037] Figure 10 This is a schematic diagram of the force exerted by the blocking device.

[0038] Explanation of reference numerals in the attached figures:

[0039] 1. Material transfer module; 2. Material storage module; 3. Material conveying module; 11. Robot mounting cabinet; 12. Robot lifting column module; 13. Robot module; 14. Suction cup and 3D vision module; 15. Waste removal module; 16. Electrical control cabinet; 17. Connector; 21. Rigid strip box material storage tray; 22. Partition paper material storage tray; 31. Material conveyor belt; 32. Blocking and positioning device; 321. Guide rod cylinder; 322. Baffle; 4. Rigid strip box; 5. Partition paper. Detailed Implementation

[0040] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings. The following embodiments are merely exemplary and can only be used to explain and illustrate the technical solution of the present invention, and should not be construed as limiting the technical solution of the present invention.

[0041] like Figure 1 As shown, this application provides an automatic feeding system for rigid carton robots, used to transport rigid carton materials to the conveyor belt of a tobacco packaging machine. It includes a material transfer module 1, a material storage module 2, and a material conveying module 3. The material transfer module 1 is located at the front end of the material conveying module 3 and is used to transfer the rigid carton 4 from the material storage module 2 to the material conveying module 3.

[0042] like Figure 2 As shown, the material transfer module 1 includes a robot mounting cabinet 11, a robot lifting column module 12, a robot module 13, a suction cup, and a 3D vision module 14. The robot mounting cabinet 11 is used to install and fix the robot lifting column module 12. The robot module 13 is installed on the robot lifting column module 12 and can move up and down along the Z direction. The suction cup and 3D vision module 14 are installed at the end of the robot module 13. It also includes a waste rejection module 15 fixed on the top of the electrical control cabinet 16 for storing unidentified materials. The robot mounting cabinet 11 and the electrical control cabinet 16 are connected by a connector 17 to ensure the stability of the overall module.

[0043] like Figures 3 to 6 As shown, the material storage module 2 is placed on both sides of the material transfer module 1, and consists of a rigid box material storage tray 21 and a partition paper material storage tray 22, wherein the rigid box 4 is as follows Figure 4 As shown, the partition paper 5 is as follows Figure 5As shown, the partition paper 5 is divided into 6 cells; the rigid boxes 4 are arranged in groups of 15 within the 6 cells of the partition paper, forming a layer of rigid box material, as shown. Figure 6 As shown, each hard box material storage tray 21 can hold 10 layers of hard box materials. After the robot module removes the hard box materials from 6 cells of a layer, it places the partition paper 5 on the partition paper material storage tray 22.

[0044] like Figure 7 , Figure 8a and Figure 8b As shown, the material conveying module 3, used to convey the rigid carton material picked up and placed by the robot module 13, consists of a material conveyor belt 31 and a blocking and positioning device 32, as follows. Figure 7 As shown. The material conveyor belt 31 can transport a set of rigid carton materials grasped by the robot module 13 into the tobacco packaging machine. The set of rigid carton materials is arranged as follows: Figure 7 The arrangement shown is on the material conveyor belt and moves along the X direction; the material conveyor belt 31 is a component of the tobacco packaging machine, and its specific connection method and structure are existing technologies, which will not be described in detail here; the blocking and positioning device 32 is installed at the front end of the material conveyor belt 31 along the X direction, and the extension and retraction of the baffle 322 are controlled by a pair of guide rod cylinders 321, and the two states are as follows Figure 8a and Figure 8b As shown.

[0045] The system in this application is based on a deep learning-based hard box visual detection and localization system, and includes the following steps:

[0046] S1. Use a depth camera to capture images of the actual production workshop where rigid carton stacks are being loaded. Images of the rigid carton stacks need to be captured from different shooting heights, at different time periods, from different front and bottom views, and from different orientations.

[0047] S2. Divide the images collected in step S1 into training dataset, validation dataset and test dataset in a ratio of 8:1:1.

[0048] S3. Training Set Data Labeling. Connect the four corners of the front side (containing the cigarette brand, barcode, QR code, and country of origin information) and label it "box". Connect the four corners of the yellow, top-facing side and label it "reverse box". Connect the outer frame of the outer carton containing the cigarette brand, barcode, and QR code as a whole and label it "box head".

[0049] S4. Establish a hard box recognition network model based on Mask R-CNN. Since the detection target is singular, ResNet50+FPN is preferred as the backbone neural network, and Region Proposal Network (RPN) is used as the feature map for convolution. Finally, three branches output bounding boxes, categories, and masks to generate the hard box recognition network model. After testing on the test set, the recognition success rate reached 99% with a confidence level greater than 0.8.

[0050] S5. Determine the orientation of a single, face-up rigid box. For a face-up rigid box, after using a deep learning rigid box recognition network, a single box will have two recognition labels: "box" and "box head," and the position coordinates of their respective center points in the image will be obtained. The orientation angle from the "box head" to the "box" will be calculated, and the result will be stored in the "box's" angle attribute.

[0051] S6. Implement counting and judgment of hard-pack cartons. Based on the arrangement characteristics of the hard-pack cartons on site, they are grouped into small squares of 15 cartons each, separated by partitions slightly lower than the height of the cigarette pack. Due to the thinness of the partitions and the small gaps between each group of cartons, the pickup pose is prone to deviation, thus requiring counting and judgment. Starting from the image origin, the center point of the box closest to the origin is taken as the starting point of the first group. Similarly, the center point of the box closest to the starting point of the first group is calculated, and the distance between the two center points is used to determine if a threshold is met. If it is met, the count is incremented by 1 until the number reaches 15. If the threshold is not met, it indicates that there is a group of cartons with less than the specified number, and it is marked as a discard group. The adjacent box is marked as the new starting point, until all the cartons in the image are assigned to each group.

[0052] S7. Determine the consistency of orientation for each group and the final grasping posture for each pile. The angle and orientation of each hard pack in each group are judged. If the angle deviation exceeds a threshold, the orientation is considered inconsistent, and the group is removed and the result is recorded in the group's attributes. When the orientation is consistent, the median angle of each pack is taken as the overall grasping angle.

[0053] S8. Determine the gripping center of each group by calculating the average x and y values ​​of each hard box in the image as the center position of each group. From this, the center gripping pose of each group of hard boxes in the image can be determined. Finally, the suction cup gripping pose is determined by transforming the image coordinate system to the camera coordinate system, then to the robot coordinate system, and finally to the tool coordinate system. When there are problematic materials in a group of hard boxes, such as inconsistent left-right orientation or reversed orientation, the robot module will grip the problematic materials and move them to the waste removal module. The suction cup and 3D vision module are integrated at the very end of the robot module, improving the aesthetics and safety of the equipment. The material movement module adopts a modular design, facilitating maintenance, transportation, and disassembly.

[0054] The material storage module is arranged on both sides of the material moving module, which can make full use of the robot's workspace and the idle space next to the tobacco packaging machine, without occupying the operator's workspace, thus improving space utilization. In addition, the hard carton material storage trays are arranged on both sides of the robot. After the robot finishes grabbing the hard carton material on one side of the tray, it can immediately grab the hard carton material on the other side of the tray, ensuring the continuity of material feeding and improving efficiency.

[0055] A pair of blocking and positioning devices are added to both sides of the material conveyor belt. When the robot module places a set of rigid strip boxes onto the material conveyor belt, the blocking and positioning devices close, and the blocking devices provide a supporting force along the X direction to the set of rigid strip boxes placed on the conveyor belt. Figure 10 As shown, the stability of the material is ensured when it is placed on the material conveyor belt by the robot module; after a set of hard carton materials is stably placed on the material conveyor belt, the blocking and positioning device opens, and the set of hard carton materials is transported to the inside of the tobacco packaging machine along with the material conveyor belt.

[0056] The process flow of the automatic feeding system for rigid carton robots is as follows:

[0057] The tobacco packaging machine sends a material request signal to the automatic feeding system of the rigid carton robot. The robot's lifting column module rises to the designated camera position, and the 3D vision module photographs a layer of rigid carton materials on one side of the material storage tray, rejecting unidentifiable materials. Next, it grabs the qualified materials from the remaining cells and moves them above the conveyor belt of the material conveying module. The blocking positioning device closes, blocking a group of rigid carton materials, ensuring the materials stand stably on the conveyor belt. Then, the suction cup releases this group of materials, the blocking positioning device opens, and this group of materials is conveyed into the tobacco packaging machine along with the conveyor belt. After all six cells of material in one layer of rigid carton materials have been grabbed, the 3D vision module photographs the partition paper and grabs it into the partition paper material storage tray. Once the robot has finished grabbing the hard box material from one pallet, it immediately moves on to grab the hard box material from the other pallet. At the same time, the hard box robot automatic feeding system sends a material request signal, and a person or AGV cart removes the empty hard box material storage pallet and replaces it with a full stack of hard box material storage pallets.

[0058] The feeding speed of the robot module can be adjusted according to the operating speed of the tobacco packaging machine to adapt to various production speed requirements of the tobacco packaging machine; in addition, the robot module can accurately record and calculate the number of hard cartons fed in each shift and the number of defective hard cartons rejected, which facilitates production management.

[0059] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. An automatic feeding system for hard boxes in a tobacco packaging machine, characterized in that, It includes a material transfer module, a material storage module, and a material conveying module. The material transfer module is located at the front end of the material conveying module and is used to transfer the hard box from the material storage module to the material conveying module. The material transfer module includes a robot mounting cabinet, a robot lifting column module, a robot module, a suction cup, and a 3D vision module; the robot mounting cabinet is used to install and fix the robot lifting column module, the robot module is installed on the robot lifting column module, and the suction cup and 3D vision module are installed at the end of the robot module. The material storage module is located on both sides of the robot module and includes a rigid box material storage tray and a partition paper material storage tray. The material conveying module includes a material conveyor belt and a blocking and positioning device. The blocking and positioning device is installed at the front end of the material conveyor belt and includes a pair of guide rod cylinders and a baffle connected to the guide rod cylinders. A deep learning-based visual detection and localization system for rigid boxes includes the following steps: S1. Use a depth camera to capture images of the actual production workshop where rigid strip boxes are stacked and loaded; S2. Divide the images collected in step S1 into training dataset, validation dataset and test dataset in a ratio of 8:1:

1. S3. Training dataset data annotation: Connect the four corners of the front face (containing the cigarette brand, barcode, QR code, and place of origin information) and label it as "box"; connect the four corners of the bottom face and label it as "reverse box"; connect the outer frame of the outer hard box containing the cigarette brand, barcode, and QR code as a whole and label it as "box head". S4. Establish a hard box recognition network model based on Mask R-CNN network, using ResNet50+FPN as the backbone neural network and Region Proposal Network (RPN) as the feature map for convolution. Finally, the bounding box, category, and mask are used as outputs to generate the hard box recognition network model. S5. Implement the orientation determination of a single front-facing hard box. For a front-facing hard box, after using a deep learning hard box recognition network, a single box will have two recognition labels, box and box head. At the same time, the position coordinates of their respective center points in the image are obtained. The direction angle from box head to box is calculated to obtain the orientation determination and store it in the angle attribute of box. S6. Implement hard box stack counting and judgment. Take the image origin as the starting point and the center point of the box closest to the origin as the starting point of the first group. Then, similarly calculate the center point of the box closest to the starting point of the first group. Calculate the distance between the two center points to judge whether they meet the threshold. Record the adjacent boxes as the new starting point until each hard box in the image is assigned to each group. S7. Determine the consistency of orientation for each group and the final grasping posture for each pile. Judge the angle and direction of each hard pack in each group. If the angle deviation exceeds the threshold, the orientation is judged to be inconsistent and recorded as a rejected group. Record it in the group's attributes. When the orientation is consistent, take the median angle of each pack as the overall suction angle. S8. Determine the capture center of each group, and calculate the average x and y values ​​of each hard box in the image as the center position of each group.

2. The automatic hard-pack feeding system for a tobacco packaging machine according to claim 1, characterized in that, The material transfer module also includes a waste removal module, an electrical control cabinet, and connectors. The waste removal module is fixed on top of the electrical control cabinet, and the electrical control cabinet is connected to the robot mounting cabinet via connectors.

3. The automatic hard-pack feeding system for a tobacco packaging machine according to claim 1, characterized in that, The partition paper is stored in the partition paper material storage tray. The partition paper includes 6 cells. The rigid strip boxes are arranged in groups of 15 in the 6 cells of the partition paper. Each rigid strip box material storage tray holds 10 layers of rigid strip box material.

4. The automatic hard-pack feeding system for a tobacco packaging machine according to claim 1, characterized in that, A pair of blocking and positioning devices are set on both sides of the material conveyor belt. When the robot module places a set of rigid strip box materials onto the material conveyor belt, the blocking and positioning devices close, and the blocking devices provide a supporting force along the X direction to the set of rigid strip box materials placed on the conveyor belt.

5. The automatic hard-pack feeding system for a tobacco packaging machine according to claim 4, characterized in that, The process flow is as follows: The tobacco packaging machine sends a material request signal to the system. The robot lifting column module rises to the designated photo position. The 3D vision module takes a photo of a layer of hard carton material on a certain side of the hard carton material storage tray and removes materials that cannot be identified. Then, it grabs the materials that meet the requirements from the remaining cells and moves them above the conveyor belt of the material conveying module. The blocking and positioning device closes to block a group of hard carton materials, so that the hard carton materials stand stably on the conveyor belt. Then, the suction cup releases this group of hard carton materials, the blocking and positioning device opens, and this group of hard carton materials is conveyed into the tobacco packaging machine along with the conveyor belt. After all six cells of material in a layer of rigid box material have been grabbed, the 3D vision module takes a picture of the partition paper and grabs the partition paper into the partition paper material storage tray. After the rigid box material on one side of the tray is grabbed, the robot immediately grabs the rigid box material on the other side of the tray. At the same time, the system sends a material request signal, and a person or AGV cart drags away the empty rigid box material storage tray and replaces it with a full stack of rigid box material storage trays.

Images