An online palletizing method with multiple error adjustments

Abstract

The present invention relates to the technical field of diaper production, and in particular to an online palletizing method with multiple error adjustments. An online palletizing method with multiple error adjustments includes S1, material preparation; S2, image acquisition and preliminary error adjustment; S3, clustering algorithm; S4, secondary error adjustment; S5, packaging data entry; S6, production line application. In the present invention, an industrial camera and a TOF depth camera are set on the production line for preliminary stacking of diapers to obtain real-time images of the current diaper stack. Through the target detection algorithm, the overly prominent diaper sheets are identified and removed. After new diaper sheets are replenished, the stacked diaper stack is pushed into the packaging production line. The packaging production line adjusts the mechanism data for secondary stacking alignment according to the identified diaper types, aligns the stacks, and retrieves the corresponding packaging data, enabling the subsequent packaging process to proceed programmatically, so that multiple types of diapers can be identified and packaged on the same diaper production line.

Description

Technical Field

[0001] The present invention relates to the technical field of diaper production, and in particular to an online palletizing method with multiple error adjustments. Background Art

[0002] Diapers, also known as disposable diapers, are an essential daily necessity in the process of raising infants and young children. In the prior art, diaper products usually adopt a bagged packaging method. Usually, the same packaging production line can only produce and package diaper stacks of the same specification size. If the production line needs to switch to package diaper stacks of another specification, the entire production line needs to be shut down. Producers need to re-adjust the corresponding packaging data according to production experience, and the production line cannot identify the type of diapers being packaged currently, so producers also need to manually adjust the production data of the production line. Summary of the Invention

[0003] Other features and advantages of the present invention will be described in the following specification, and will become partially apparent from the specification, or be understood by implementing the present invention. The objectives and other advantages of the present invention can be achieved and obtained through the structures specifically pointed out in the specification and other accompanying drawings of the specification.

[0004] The objective of the present invention is to overcome the above deficiencies and provide an online palletizing method with multiple error adjustments. An industrial camera and a TOF depth camera are set on the production line for preliminary stacking of diapers to obtain real-time images of the current diaper stack. Diaper sheets that are too prominent are identified through a target detection algorithm and removed. After new diaper sheets are replenished, the stacked diaper stack is pushed into the packaging production line. The packaging production line adjusts the mechanism data for secondary stacking alignment according to the identified diaper type, aligns the stack, and retrieves the corresponding packaging data, enabling the subsequent packaging process to proceed in a programmed manner, enabling multiple types of diapers to be identified and packaged on the same diaper production line, and making the stack neatly arranged while the packaging data corresponds.

[0005] The present invention provides an online palletizing method with multiple error adjustments, including:

[0006] S1. Material preparation: Input multiple groups of diaper stacks of different types on the production line for preliminary stacking of diapers, and the number of groups of each type of diaper stack is at least three groups;

[0007] S2. Image acquisition and preliminary error adjustment: Set an industrial camera and a TOF depth camera on the production line for preliminary stacking of diapers, collect real-time images of each group of diaper stacks and input them into the control center A. Establish a target detection model in the control center A for preliminary stacking adjustment to remove diaper sheets that are too prominent;

[0008] S3. Clustering algorithm: Collect the real-time image of the diaper stack after preliminary stacking adjustment and input it into the control center B. The control center B performs the K-means clustering algorithm on the real-time image, classifies the diaper stack according to the clustering center, and records the volume and stacking layers corresponding to each group of clustering centers;

[0009] S4. Secondary error adjustment: Push out multiple groups of diaper stacks of different types from the preliminary stacking production line one by one and enter the packaging production line. Control the palletizing robotic arm to perform secondary stacking adjustment on the diaper stack through the fuzzy control algorithm to make the diaper pieces in a group of diaper stacks arranged neatly;

[0010] S5. Packaging data entry: According to production experience, set the corresponding packaging data for multiple groups of diaper stacks of different types, feedback and input the packaging data into the control center B, and combine it with the clustering result to form a diaper stack data set group;

[0011] S6. Production line application: Input the diaper stack data set group established in steps S3 - S5 into the control center B. When there are multiple types of diaper stacks in the preliminary stacking production line, the target detection algorithm in the control center A performs online monitoring on the diaper stack and collects real-time images, identifies the corresponding diaper stack type, removes the overly prominent diaper pieces and automatically replenishes them. After the diaper stack is pushed into the packaging production line, call the diaper stack data set group in the control center B to perform secondary stacking adjustment and packaging according to the diaper stack type.

[0012] In some embodiments, in this S2 step, the specific steps of collecting the real-time image by the industrial camera are as follows: Collect the RGB image of the current diaper stack on the production line through the industrial camera, input the RGB image in the loaded state into the control center A, add rectangular frames and labels to the diaper pieces of different sizes through the labelme software, define the diaper type and identify the corresponding quantity, input the processed image into the Mask RCNN instance segmentation network, learn the image features and establish a target detection model, mark the overly prominent diaper pieces, and control the removal robotic arm to remove the prominent diaper pieces from the production line.

[0013] In some embodiments, during the process of establishing the target detection model and marking the overly prominent diaper pieces, the alignment algorithm is involved. After the target detection model detects each independent diaper piece in the current diaper stack, add rectangular frame marks to the diaper pieces, calculate the center points of each rectangular frame and the slope between the center points. If the slope exceeds the set threshold, it is determined as an overly prominent diaper piece, and control the removal robotic arm to remove the prominent diaper piece from the production line. Combine the number of removed diaper pieces and production requirements, replenish the missing diaper pieces in the current diaper stack, recalculate the alignment until the alignment is within the threshold range, and then enter step S3.

[0014] In some embodiments, in step S3, the specific steps of performing a clustering algorithm operation on the diaper stack after preliminary stacking adjustment are as follows:

[0015] S31. An industrial camera collects an RGB image of the preliminary stacking production line in the no-load state, and a TOF depth camera collects a depth image of the preliminary stacking production line in the no-load state. Align the RGB image and the depth image, and record the real-time image sample in the no-load state;

[0016] S32. An industrial camera collects an RGB image of the current diaper stack on the preliminary stacking production line, and transmits the RGB image in the load state to the control center B. The TOF depth camera synchronously collects the depth image of the diaper stack in the load state, performs correction processing on the depth image, retains the outer frame, and obtains the area of the current diaper stack;

[0017] S33. Align the RGB image and the depth image, obtain a comprehensive image of the arrangement state between each diaper sheet in the diaper stack in the current state, repeatedly obtain comprehensive images of multiple groups of diaper stacks of different types, perform the K-means clustering algorithm, classify the diaper stacks according to the cluster centers, and obtain multiple cluster centers;

[0018] S34. Combine the depth image in the no-load state with the depth image in the load state, calculate the height and volume of the current diaper stack, and synchronously input the volume information of the current diaper stack and the comprehensive image into the K-means clustering algorithm, so that each data point carries volume information;

[0019] S35. In the control center B, combine the obtained cluster centers with the actual production information, add type, stacking layer number, and volume information labels to the diaper stacks according to the cluster centers, and at the same time import the information into the control center A with a target detection model, so that the classification in the target detection model corresponds to the classification of the cluster centers.

[0020] In some embodiments, in step S4, after the current diaper stack is pushed into the packaging production line, the specific steps of performing secondary stacking adjustment on the diaper stack are as follows:

[0021] S41. The diaper stack slides into the front end of the packaging production line, and the left clamping arm, the right clamping arm, and the rear clamping arm extend to form a U-shaped frame, so that the diaper stack is located within the U-shaped frame. The left clamping arm, the right clamping arm, and the rear clamping arm are all plate-like mechanisms controlled by hydraulic telescopic rods and are used to limit the diaper stack;

[0022] S42. The front side of the diaper stack extends out of the palletizing robotic arm. A limiting plate is connected to the end of the palletizing robotic arm through a rotating shaft, and this rotating shaft is connected to the control center A. In the control center A, according to the target detection algorithm, the center points of each diaper sheet in the current diaper stack are stored, and the least squares method is used to calculate the center points to obtain the slope of the current diaper stack.

[0023] S43. A fuzzy control algorithm module is also set in the control center A. The rotating shaft is connected to the fuzzy control algorithm module. After inputting the slope of the current diaper stack, the corresponding initial rotation angle and rotation rate of the rotating shaft are output through the fuzzy control algorithm. The limiting plate rotates in the initial state to adapt to the slope of the current diaper stack. During the process of gradually pushing and adjusting the diaper stack to align it, the rotating shaft rotates gradually until it is parallel to the rear clamping arm, thereby realizing the secondary palletizing adjustment of the diaper stack.

[0024] In some embodiments, in step S4, the elongation amounts of the left clamping arm, the right clamping arm, the rear clamping arm, and the palletizing robotic arm are set by the production personnel according to the current diaper stack type, and the set elongation amount data is synchronized to the control center A and the control center B.

[0025] In some embodiments, after the secondary palletizing adjustment of the diaper stack is completed, the rear clamping arm and the palletizing robotic arm are withdrawn. The left clamping arm and the right clamping arm press the diaper stack inward. The vacuum pump arranged at the rear of the packaging production line unfolds the corresponding type of diaper packaging bag. The left clamping arm and the right clamping arm move backward. After sending the diaper stack into the packaging bag, it is taken out, and the sealing machine seals the opening of the packaging bag; the clamping distance and pressure of the left clamping arm and the right clamping arm, the moving distance of the left clamping arm and the right clamping arm, the operating data of the vacuum pump, and the operating data of the sealing machine are all packaging data, which are set by the production personnel according to the current diaper stack type, and the set packaging data is synchronously input into the control center B and combined with the clustering result to form a diaper palletizing data set group.

[0026] In some embodiments, in step S6, the specific operation of automatic filling is as follows: after removing the overly protruding diaper sheets, the control of the preliminary stacking production line makes the filling robotic arm push the rear diaper sheets forward to fill. When running to the last stack of this type of diaper stack, if there is a shortage of diapers, the control center A issues an alarm, and the producer can choose to remove this stack of diaper stacks or add new diapers to this stack of diaper stacks.

[0027] By adopting the above technical solutions, the beneficial effects of the present invention are:

[0028] In the present invention, an industrial camera and a TOF depth camera are arranged on the production line for the preliminary stacking of diapers to obtain a real-time image of the current diaper stack. Through a target detection algorithm, the overly protruding diaper sheets are identified and removed. After new diaper sheets are replenished, the stacked diaper stack is pushed into the packaging production line. The packaging production line adjusts the mechanism data for secondary stacking alignment according to the identified diaper types, aligns the stacks, and retrieves the corresponding packaging data, enabling the subsequent packaging process to proceed programmatically. This allows multiple types of diapers to be identified and packaged on the same diaper production line, while ensuring that the stacks are neatly arranged and the packaging data is corresponding.

[0029] It should be understood that the above general description and the following detailed description are merely exemplary and explanatory, and do not limit the present disclosure.

[0030] Undoubtedly, such objects of the present invention and other objects will become more apparent after the details of the preferred embodiments described in the following with multiple drawings and illustrations.

[0031] To make the above and other objects, features, and advantages of the present invention more obvious and understandable, one or more preferred embodiments are specifically exemplified below and described in detail in conjunction with the accompanying drawings as follows. Description of the Drawings

[0032] 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, but do not constitute a limitation to the present invention.

[0033] In the drawings, the same components are denoted by the same reference numerals, and the drawings are schematic and not necessarily drawn to actual scale.

[0034] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are only one or several embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on such drawings.

[0035] Figure 1 Schematic diagram of the overall process of the online stacking method in some embodiments of the present invention;

[0036] Figure 2 Schematic diagram of the recognition of the center point and slope by the target detection model in some embodiments of the present invention;

[0037] Figure 3 Schematic diagram of the diaper classification obtained by the clustering algorithm in some embodiments of the present invention;

[0038] Figure 4Schematic diagram of the packaging production line structure in some embodiments of the present invention.

[0039] Description of the main reference numerals:

[0040] 1, left clamping arm; 2, right clamping arm; 3, rear clamping arm; 4, palletizing robotic arm. Detailed implementation manners

[0041] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below in conjunction with specific implementation manners. It should be understood that the specific implementation manners described herein are only used to explain the present invention, but not to limit the present invention.

[0042] In addition, in the description of the present invention, it should be understood that the orientation or positional relationship indicated by terms such as "center", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "axial", "radial", "circumferential", etc. is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus cannot be understood as a limitation of the present invention.

[0043] In the present invention, unless otherwise clearly specified and defined, terms such as "installation", "connection", "connection", "fixation", etc. should be understood in a broad sense. For example, it may be a fixed connection, a detachable connection, or integrated; it may be directly connected, or indirectly connected through an intermediate medium, and may be the communication inside two elements or the interaction relationship between two elements. However, indicating a direct connection means that there is no connection relationship constructed through a transition structure between the two connected main bodies, and they are only connected through the connection structure to form a whole. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific situations.

[0044] In the present invention, unless otherwise clearly specified and defined, the first feature being "above" or "below" the second feature may be that the first and second features are in direct contact, or the first and second features are in indirect contact through an intermediate medium. In the description of this specification, the description with reference to terms such as "one embodiment", "some embodiments", "example", "specific example", or "some examples", etc. means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in a suitable manner in any one or more embodiments or examples.

[0045] Refer to 1-Figure 4 , Figure 1 is a schematic diagram of the overall process of the online palletizing method in some embodiments of the present invention; Figure 2 is a schematic diagram of the center point and slope recognized by the target detection model in some embodiments of the present invention; Figure 3 is a schematic diagram of the classification of diapers obtained by the clustering algorithm in some embodiments of the present invention; Figure 4 is a schematic diagram of the structure of the packaging production line in some embodiments of the present invention.

[0046] According to some embodiments of the present invention, the present invention provides an online palletizing method with multiple error adjustments, including:

[0047] S1. Material preparation: Input multiple groups of diaper stacks of different types on the production line for preliminary stacking of diapers. The number of groups of each type of diaper stack is at least three;

[0048] S2. Image acquisition and preliminary error adjustment: Set an industrial camera and a TOF depth camera on the production line for preliminary stacking of diapers, collect real-time images of each group of diaper stacks and input them into the control center A. Establish a target detection model in the control center A for preliminary stacking adjustment, and remove the overly protruding diaper pieces;

[0049] The specific steps for collecting real-time images by the industrial camera are as follows: Use the industrial camera to collect RGB images of the current diaper stack on the production line, input the RGB images in the loaded state into the control center A, add rectangular frames and labels to diaper pieces of different sizes through the labelme software, define the diaper types and identify the corresponding quantities, input the processed images into the Mask RCNN instance segmentation network, learn the image features and establish a target detection model, mark the overly protruding diaper pieces, and control the removal robotic arm to remove the protruding diaper pieces from the production line;

[0050] As Figure 2 shown, in the process of establishing the target detection model and marking the overly protruding diaper pieces, the alignment algorithm is involved. After the target detection model detects each independent diaper piece in the current diaper stack, add rectangular frame marks to the diaper pieces, calculate the center points of each rectangular frame and the slope between the center points. If the slope exceeds the set threshold, it is determined as an overly protruding diaper piece, control the removal robotic arm to remove the protruding diaper pieces from the production line, and supplement the missing diaper pieces in the current diaper stack according to the number of removed diaper pieces and production needs, recalculate the alignment until the alignment is within the threshold range, and enter step S3.

[0051] S3. Clustering algorithm: Collect the real-time image of the diaper stack after preliminary stacking adjustment and input it into the control center B. The control center B performs the K-means clustering algorithm on the real-time image, classifies the diaper stack according to the clustering centers, and records the volume and stacking layers corresponding to each group of clustering centers;

[0052] The specific steps for performing the clustering algorithm operation on the diaper stack after preliminary stacking adjustment are as follows:

[0053] S31. The industrial camera collects the RGB image of the preliminary stacking pipeline in the no-load state, and the TOF depth camera collects the depth image of the preliminary stacking pipeline in the no-load state. Align the RGB image and the depth image, and record the real-time image sample in the no-load state;

[0054] S32. The industrial camera collects the RGB image of the current diaper stack on the preliminary stacking pipeline, transmits the RGB image in the load state to the control center B, and the TOF depth camera synchronously collects the depth image of the diaper stack in the load state. Perform correction processing on the depth image, retain the outer frame, and obtain the area of the current diaper stack;

[0055] S33. Align the RGB image and the depth image, obtain the comprehensive image of the arrangement state between each diaper sheet in the diaper stack in the current state, repeat to obtain multiple groups of comprehensive images of different types of diaper stacks, perform the K-means clustering algorithm, and classify the diaper stack according to the clustering centers. As Figure 3 shown, obtain multiple clustering centers;

[0056] S34. Combine the depth image in the no-load state and the depth image in the load state, calculate the height and volume of the current diaper stack, and synchronously input the volume information of the current diaper stack and the comprehensive image into the K-means clustering algorithm, so that each data point carries volume information;

[0057] S35. In the control center B, combine the obtained clustering centers with the actual production information, add the type, stacking layers, and volume information labels of the diaper stack according to the clustering centers, and at the same time import the information into the control center A with the target detection model, so that the classification in the target detection model corresponds to the classification of the clustering centers.

[0058] S4. Secondary error adjustment: Push out multiple groups of diaper stacks of different types from the preliminary stacking pipeline one by one and enter the packaging pipeline. Control the palletizing robot arm 4 through the fuzzy control algorithm to perform secondary stacking adjustment on the diaper stack, so that the diaper sheets in a group of diaper stacks are neatly arranged;

[0059] After the current diaper stack is pushed into the packaging pipeline, the specific steps for performing secondary stacking adjustment on the diaper stack are as follows:

[0060] S41. The diaper stack slides into the front end of the packaging production line. As shown in Figure 4 , the left clamping arm 1, the right clamping arm 2, and the rear clamping arm 3 extend to form a U-shaped frame, with the diaper stack positioned within the U-shaped frame. The left clamping arm 1, the right clamping arm 2, and the rear clamping arm 3 are all plate-like mechanisms controlled by hydraulic telescopic rods, used to limit the diaper stack.

[0061] S42. The front side of the diaper stack extends the palletizing robotic arm 4. The end of the palletizing robotic arm 4 is connected with a limiting plate through a rotating shaft, and this rotating shaft is connected to the control center A. In the control center A, according to the target detection algorithm, the center points of each diaper sheet in the current diaper stack are stored, and the least squares method is used to calculate the slope of the current diaper stack.

[0062] S43. The control center A is also provided with a fuzzy control algorithm module. Connect the rotating shaft to the fuzzy control algorithm module. After inputting the slope of the current diaper stack, the corresponding initial rotation angle and rotation rate of the rotating shaft are output through the fuzzy control algorithm. The limiting plate rotates a certain angle in the initial state to adapt to the slope of the current diaper stack. During the process of gradually pushing and adjusting the diaper stack to align, the rotating shaft rotates gradually until it is parallel to the rear clamping arm 3, thereby realizing the secondary palletizing adjustment of the diaper stack.

[0063] Among them, the elongation amounts of the left clamping arm 1, the right clamping arm 2, the rear clamping arm 3, and the palletizing robotic arm 4 are set by the production personnel according to the current diaper stack type, and the set elongation amount data is synchronized to the control center A and the control center B.

[0064] S5. Packaging data entry: According to production experience, corresponding packaging data is set for multiple groups of diaper stacks of different types, and the packaging data is fed back and input into the control center B, combined with the clustering result to form a diaper stack data set group.

[0065] After the secondary palletizing adjustment of the diaper stack is completed, the rear clamping arm 3 and the palletizing robotic arm 4 are withdrawn. The left clamping arm 1 and the right clamping arm 2 press the diaper stack inward. The vacuum air pump arranged at the rear side of the packaging production line unfolds the corresponding type of diaper packaging bag. The left clamping arm 1 and the right clamping arm 2 move backward. After sending the diaper stack into the packaging bag and then taking it out, the sealing machine seals the opening of the packaging bag. The clamping distance and pressure of the left clamping arm 1 and the right clamping arm 2, the moving distance of the left clamping arm 1 and the right clamping arm 2, the operating data of the vacuum air pump, and the operating data of the sealing machine are all packaging data, set by the production personnel according to the current diaper stack type, and the set packaging data is synchronously input into the control center B, corresponding to the clustering result and combined to form a diaper stack data set group.

[0066] S6. Production line application: Input the established disposable diaper palletizing data set group in steps S3 - S5 into control center B. When there are multiple types of disposable diaper pallets in the preliminary stacking pipeline, the target detection algorithm in control center A monitors the disposable diaper pallets online, collects real-time images, identifies the corresponding types of disposable diaper pallets, removes overly protruding disposable diaper sheets and automatically replenishes them. After the disposable diaper pallets are pushed into the packaging pipeline, call the disposable diaper palletizing data set group in control center B and perform secondary stacking adjustment and packaging according to the types of disposable diaper pallets;

[0067] The specific operation of automatic replenishment is as follows: After removing the overly protruding disposable diaper sheets, the control of the preliminary stacking pipeline makes the replenishing robotic arm push the disposable diaper sheets at the back forward to complete the replenishment. When running to the last stack of this type of disposable diaper pallet and there is a shortage of disposable diapers, control center A issues an alarm, and the producer can choose to remove this stack of disposable diaper pallets or replenish new disposable diapers in this stack of disposable diaper pallets. Embodiment

[0068] Refer to 1 - Figure 4 , Figure 1 is a schematic diagram of the overall process of the online palletizing method in some embodiments of the present invention; Figure 2 is a schematic diagram of the recognition of the center point and slope of the target detection model in some embodiments of the present invention; Figure 3 is a schematic diagram of the classification of disposable diapers obtained by the clustering algorithm in some embodiments of the present invention; Figure 4 is a schematic diagram of the structure of the packaging pipeline in some embodiments of the present invention.

[0069] According to some embodiments of the present invention, the present invention provides an online palletizing method with multiple error adjustments, including:

[0070] S1. Material preparation: Input multiple groups of disposable diaper pallets of different types on the pipeline for preliminary stacking of disposable diapers, and there are three groups for each type of disposable diaper pallet; specifically, there are three categories of disposable diapers produced on this production line, namely A, B, and C. Under each of the three categories A, B, and C of disposable diapers, there are three stacking specifications, namely 10 pieces, 20 pieces, and 30 pieces. Therefore, there are a total of 9 types of disposable diaper pallets running on this production line, and each type of disposable diaper pallet is repeated three times and sequentially input under the industrial camera and the TOF depth camera;

[0071] S2. Image acquisition and preliminary error adjustment: Set an industrial camera and a TOF depth camera on the pipeline for preliminary stacking of disposable diapers, collect real-time images of each group of disposable diaper pallets and input them into control center A. Establish a target detection model in control center A, perform preliminary stacking adjustment, and remove overly protruding disposable diaper sheets;

[0072] The specific steps for collecting real-time images through an industrial camera are as follows: Use the industrial camera to collect RGB images of the current stack of diapers on the production line, input the RGB images in the loaded state into Control Center A, add rectangular frames and labels to diaper sheets of different sizes through the labelme software, define the diaper types and identify the corresponding quantities, input the processed images into the Mask RCNN instance segmentation network, learn the image features and establish an object detection model, mark the overly prominent diaper sheets, and control the rejection robotic arm to remove the prominent diaper sheets from the production line;

[0073] Add labels to diaper sheets of three categories A, B, and C, input them into the Mask RCNN instance segmentation network, learn the image features and establish an object detection model. Through the object detection model, Control Center A can correspondingly identify the current diaper category and its corresponding stacking quantity;

[0074] As Figure 2 shown, during the process of establishing the object detection model and marking overly prominent diaper sheets, an alignment algorithm is involved. After the object detection model detects each independent diaper sheet in the current diaper stack, add a rectangular frame label to the diaper sheet, calculate the center points of each rectangular frame and the slope between the center points. If the slope exceeds 30 degrees, it is determined as an overly prominent diaper sheet, and control the rejection robotic arm to remove the prominent diaper sheet from the production line. Combine the number of rejected diaper sheets and production requirements, supplement the missing diaper sheets in the current diaper stack, recalculate the alignment until the alignment is within the threshold range, and then enter Step S3. For example, if two diaper sheets are removed from a 10-piece diaper stack, two diaper sheets are replenished from the back, and the 10-piece stack is pushed into the packaging production line.

[0075] S3. Clustering algorithm: Collect the real-time image of the diaper stack after preliminary stacking adjustment and input it into Control Center B. Control Center B performs the K-means clustering algorithm on the real-time image, classifies the diaper stack according to the clustering centers, and records the volume and stacking layers corresponding to each clustering center;

[0076] The specific steps for performing the clustering algorithm operation on the diaper stack after preliminary stacking adjustment are as follows:

[0077] S31. Use the industrial camera to collect the RGB image of the preliminary stacking production line in the unloaded state, and use the TOF depth camera to collect the depth image of the preliminary stacking production line in the unloaded state. Align the RGB image and the depth image, and record the real-time image sample in the unloaded state;

[0078] S32. The industrial camera captures the RGB image of the current diaper stack on the preliminary stacking pipeline, transmits the RGB image in the loaded state to the control center B. The TOF depth camera synchronously captures the depth image of the diaper stack in the loaded state, corrects the depth image, retains the outer frame, and obtains the area of the current diaper stack.

[0079] The specific steps of the correction process are as follows: extract the hot spot area from the depth image, screen and correct the data in the hot spot area, retain the rectangular hot spot area of the outer frame, delete the irrelevant internal data, perform noise processing on the rectangular hot spot area of the outer frame, and finally calculate and obtain the outer frame area of the current diaper stack.

[0080] S33. Align the RGB image with the depth image to obtain the comprehensive image of the arrangement state between each diaper sheet in the diaper stack in the current state. Repeat to obtain multiple groups of comprehensive images of different types of diaper stacks, perform the K-means clustering algorithm, and classify the diaper stacks according to the clustering centers. Figure 3 As shown, set 9 clustering centers in the K-means clustering algorithm. After clustering, 9 classifications are formed. Add classification labels to the currently obtained clustering centers, such as A1, 10-piece pack; A2, 20-piece pack; A3, 30-piece pack; B1, 10-piece pack, etc.

[0081] S34. Combine the depth image in the unloaded state with the depth image in the loaded state to calculate the height and volume of the current diaper stack. Synchronously input the volume information of the current diaper stack and the comprehensive image into the K-means clustering algorithm so that each data point carries volume information.

[0082] The ranging principle of the TOF depth camera is as follows: continuously send light pulses to the target, then use the sensor to receive the light returned from the object, and obtain the distance of the target object by detecting the round-trip time of the light pulse. The TOF camera thus obtains the depth information of the entire image. The relevant calculation method for the TOF camera to calculate the height can refer to the existing technology and will not be elaborated here. Combining the height with the detected outer frame area of the diaper stack can obtain the volume information of the diaper stack.

[0083] S35. In the control center B, combine the obtained clustering centers with the actual production information, add labels for the type, stacking layer number, and volume information of the diaper stack according to the clustering centers, and at the same time import the information into the control center A with the target detection model so that the classification in the target detection model corresponds to the classification of the clustering centers.

[0084] The obtained cluster centers are manually identified, and the cluster centers and labels that conform to the actual production information are retained. The erroneous cluster center information is re-clustered and re-imported. In this embodiment, 9 cluster centers are obtained. The relevant information of the 9 cluster centers is imported into the control center A, so that when the early target detection model detects the corresponding diaper category, the control center B jumps to the cluster center under the corresponding category and calls the corresponding production information. For example, the target detection model detects that the type of the current diaper stack on the assembly line is A1, 10 pieces, and calls the corresponding cluster center volume label to select the packaging bag corresponding to the type.

[0085] S4, secondary error adjustment: push different types of multiple groups of diaper stacks out of the preliminary stacking line one by one, and enter the packaging line. The fuzzy control algorithm is used to control the stacking robot 4 to perform secondary stacking adjustment on the diaper stacks, so that the diaper pieces in a group of diaper stacks are arranged neatly;

[0086] After the current diaper stack is pushed into the packaging line, the specific steps for secondary stacking adjustment of the diaper stack are as follows:

[0087] S41, the diaper stack slides into the front end of the packaging line, such as Figure 4 As shown, the left clamping arm 1, the right clamping arm 2 and the rear clamping arm 3 extend to form a U-shaped frame, so that the diaper stack is located in the U-shaped frame, and the left clamping arm 1, the right clamping arm 2 and the rear clamping arm 3 are all plate-shaped mechanisms controlled by hydraulic telescopic rods, which are used to limit the diaper stack;

[0088] S42, a stacking robot arm 4 extends from the front side of the diaper stack, and the end of the stacking robot arm 4 is connected to a limit plate through a rotating shaft, and the rotating shaft is connected to the control center A. The control center A stores the center points of each diaper piece in the current diaper stack according to the target detection algorithm, and performs a least squares calculation on the center points to obtain the slope of the current diaper stack;

[0089] S43, the control center A is also provided with a fuzzy control algorithm module, the rotating shaft is connected to the fuzzy control algorithm module, after the slope of the current diaper stack is input, the corresponding initial rotation angle and rotation rate of the rotating shaft are output through the fuzzy control algorithm, the limit plate rotates a certain angle in the initial state to adapt to the slope of the current diaper stack, and in the process of gradually pushing and adjusting the diaper stack, the rotating shaft gradually rotates until it is parallel to the rear clamping arm 3, thereby realizing the secondary stacking adjustment of the diaper stack;

[0090] Among them, the elongation amounts of the left clamping arm 1, the right clamping arm 2, the rear clamping arm 3 and the palletizing robotic arm 4 are set by the production personnel according to the current type of diaper stack, and the set elongation amount data is synchronized to the control centers A and B. It should be understood that the specific structures of the left clamping arm 1, the right clamping arm 2, the rear clamping arm 3 and the palletizing robotic arm 4 can refer to the prior art. The component structures in the drawings are only for illustration to assist understanding and do not represent their actual structures as shown. Their actual structures can be adaptively changed according to the combination of the realized functions and the prior art.

[0091] S5. Packaging data entry: According to production experience, corresponding packaging data is set for multiple groups of diaper stacks of different types, and the packaging data is fed back and input into the control center B, and combined with the clustering result to form a diaper stack data set group.

[0092] After the secondary stacking adjustment of the diaper stack is completed, the rear clamping arm 3 and the palletizing robotic arm 4 are withdrawn. The left clamping arm 1 and the right clamping arm 2 press the diaper stack inward. The vacuum pump arranged at the rear of the packaging assembly line unfolds the corresponding type of diaper packaging bag. The left clamping arm 1 and the right clamping arm 2 move backward, send the diaper stack into the packaging bag and then take it out, and the sealing machine seals the opening of the packaging bag; the clamping distance and pressure of the left clamping arm 1 and the right clamping arm 2, the moving distance of the left clamping arm 1 and the right clamping arm 2, the operating data of the vacuum pump and the operating data of the sealing machine are all packaging data, which are set by the production personnel according to the current type of diaper stack. The set packaging data is synchronously input into the control center B and combined with the clustering result to form a diaper stack data set group.

[0093] S6. Production line application: Input the diaper stack data set group established in steps S3 - S5 into the control center B. When there are multiple types of diaper stacks in the initially stacked assembly line, the target detection algorithm in the control center A monitors the diaper stacks online and collects real-time images, identifies the corresponding diaper stack types, removes the overly protruding diaper sheets and automatically replenishes them. After the diaper stack is pushed into the packaging assembly line, the diaper stack data set group in the control center B is called, and secondary stacking adjustment and packaging are performed according to the diaper stack type.

[0094] The specific operation of automatic replenishment is as follows: After removing the overly protruding diaper sheets, the initially stacked assembly line controls the replenishment robotic arm to push the rear diaper sheets forward to replenish. When running to the last stack of this type of diaper stack and there is a shortage of diapers, the control center A issues an alarm, and the producer can choose to remove this stack of diaper stacks or replenish new diapers in this stack of diaper stacks.

[0095] During the actual operation of the industrial camera and the TOF depth camera in the production line, while performing the target detection algorithm, real-time images of the diaper stacks after preliminary error adjustment are acquired. The real-time images are supplemented as new data points into the K-means clustering algorithm. When enough data points are obtained, the clustering center of the current type of diaper stack will shift. When the shift path exceeds the threshold, an alarm is issued to remind the producer to adjust the production data on the packaging production line in a timely manner.

[0096] It should be understood that the embodiments disclosed in the present invention are not limited to the specific processing steps or materials disclosed herein, but should extend to equivalent alternatives of such features understood by those of ordinary skill in the relevant art. It should also be understood that the terms used herein are for the purpose of describing specific embodiments only and do not imply limitation.

[0097] The "embodiments" mentioned in the specification mean that the specific features or characteristics described in connection with the embodiments are included in at least one embodiment of the present invention. Therefore, the phrase "embodiments" that appears throughout the specification does not necessarily refer to the same embodiment.

[0098] In addition, the described features or characteristics can be combined in any other suitable manner into one or more embodiments. In the above description, some specific details, such as thickness, quantity, etc., are provided to provide a comprehensive understanding of the embodiments of the present invention. However, those skilled in the relevant art will understand that the present invention can be implemented without one or more of the above specific details or can also be implemented using other methods, components, materials, etc.

Claims

1. An online palletizing method with multiple error adjustments, characterized in that: include: S1. Material preparation: input multiple groups of diaper stacks of different types into the production line for preliminary stacking of diapers, and the number of groups of each type of diaper stack is at least three; S2. Image acquisition and preliminary error adjustment: An industrial camera and a TOF depth camera are set up on the production line for preliminary stacking of diapers to collect real-time images of each group of diaper stacks and input them to the control center A. A target detection model is established in the control center A to perform preliminary stacking adjustments and remove diaper pieces that are too protruding; S3, clustering algorithm: collect the real-time image of the diaper stack after preliminary stacking adjustment and input it to the control center B, which performs K-means clustering algorithm on the real-time image, classifies the diaper stack according to the cluster center, and records the volume and number of stacking layers corresponding to each group of cluster centers; S4, secondary error adjustment: Multiple groups of diaper stacks of different types are pushed out of the initial stacking line one by one and enter the packaging line. The stacking robot arm is controlled by the fuzzy control algorithm to perform secondary stacking adjustment on the diaper stacks, so that the diaper pieces in a group of diaper stacks are arranged neatly; S5. Packaging data entry: according to production experience, corresponding packaging data is set for different types of multiple groups of diaper stacks, and the packaging data is fed back to the control center B, and combined with the clustering results to form a diaper stacking data set group; S6. Production line application: Input the diaper stacking data set group established in steps S3-S5 into control center B. When there are multiple types of diaper stacks in the initial stacking line, the target detection algorithm in control center A monitors the diaper stacks online and collects real-time images, identifies the corresponding diaper stack types, removes overly protruding diaper pieces and automatically fills them. After the diaper stacks are pushed into the packaging line, the diaper stacking data set group in control center B is called to perform secondary stacking adjustments and packaging according to the diaper stack types.

2. The online palletizing method with multiple error adjustments according to claim 1, characterized in that: In the S2 step, the specific steps of collecting real-time images through the industrial camera are as follows: collecting RGB images of the current diaper stack on the assembly line through the industrial camera, inputting the RGB image in the loading state to the control center A, adding rectangular frames and labels to diaper pieces of different sizes through the labelme software, defining the diaper type and identifying the corresponding quantity, inputting the processed image into the Mask RCNN instance segmentation network, learning image features and establishing a target detection model, marking diaper pieces that are too protruding, and controlling the rejection robot arm to remove the protruding diaper pieces from the assembly line.

3. The online palletizing method with multiple error adjustments according to claim 2, characterized in that: In the process of establishing the target detection model and marking the diaper pieces that are too protruding, an alignment algorithm is involved. After the target detection model detects the independent diaper pieces in the current diaper stack, a rectangular frame mark is added to the diaper piece, and the center point of each rectangular frame and the slope between the center points are calculated. If the slope exceeds the set threshold, it is judged as a diaper piece that is too protruding, and the rejection robot arm is controlled to remove the protruding diaper piece from the assembly line. Combined with the number of diaper pieces rejected and production needs, the missing diaper pieces in the current diaper stack are supplemented, and the alignment is recalculated until the alignment is within the threshold range, and then the S3 step is entered.

4. The online palletizing method with multiple error adjustments according to claim 3 is characterized in that: In step S3, the specific steps of performing clustering algorithm operation on the diaper stack after preliminary stacking adjustment are as follows: S31, the industrial camera collects the RGB image of the preliminary stacking pipeline in the no-load state, the TOF depth camera collects the depth image of the preliminary stacking pipeline in the no-load state, aligns the RGB image and the depth image, and records the real-time image samples in the no-load state; S32, the industrial camera collects the RGB image of the current diaper stack on the preliminary stacking line, transmits the RGB image in the loading state to the control center B, and the TOF depth camera synchronously collects the depth image of the diaper stack in the loading state, corrects the depth image, retains the outer frame and obtains the area of ​​the current diaper stack; S33, aligning the RGB image with the depth image, obtaining a comprehensive image of the arrangement state of the diaper pieces in the diaper stack in the current state, repeatedly obtaining comprehensive images of multiple groups of diaper stacks of different types, performing a K-means clustering algorithm, classifying the diaper stacks according to the cluster centers, and obtaining multiple cluster centers; S34, combining the depth image in the empty state with the depth image in the loaded state, calculating the height and volume of the current diaper stack, and synchronously inputting the volume information of the current diaper stack and the comprehensive image into the K-means clustering algorithm, so that each data point carries the volume information; S35. In control center B, the obtained cluster center is combined with the actual production information, and the type, number of stacking layers and volume information labels of the diaper stack are added according to the cluster center. At the same time, the information is imported into control center A with a target detection model, so that the classification in the target detection model corresponds to the classification of the cluster center.

5. The online palletizing method with multiple error adjustments according to claim 4, characterized in that: In step S4, after the current diaper stack is pushed into the packaging line, the specific steps of performing secondary stacking adjustment on the diaper stack are as follows: S41, the stack of diapers slides into the front end of the packaging assembly line, and the left clamping arm, the right clamping arm and the rear clamping arm extend to form a U-shaped frame, so that the stack of diapers is located in the U-shaped frame, and the left clamping arm, the right clamping arm and the rear clamping arm are all plate-shaped mechanisms controlled by hydraulic telescopic rods, which are used to limit the stack of diapers; S42, a stacking robot arm extends from the front side of the diaper stack, and the end of the stacking robot arm is connected to a limit plate through a rotating shaft, and the rotating shaft is connected to the control center A. The control center A stores the center points of each diaper piece in the current diaper stack according to the target detection algorithm, and performs a least squares calculation on the center points to obtain the slope of the current diaper stack; S43. A fuzzy control algorithm module is also provided in the control center A. The rotating shaft is connected to the fuzzy control algorithm module. After the slope of the current diaper stack is input, the corresponding initial rotation angle and rotation rate of the rotating shaft are output through the fuzzy control algorithm. The limit plate rotates in the initial state to adapt to the slope of the current diaper stack. In the process of gradually pushing, adjusting and aligning the diaper stack, the rotating shaft gradually rotates until it is parallel to the rear clamping arm, thereby realizing the secondary stacking adjustment of the diaper stack.

6. The online palletizing method with multiple error adjustments according to claim 5, characterized in that: In step S4, the extension amounts of the left clamping arm, the right clamping arm, the rear clamping arm and the palletizing robot arm are set by the production personnel according to the current diaper stack type, and the set extension amount data is synchronized to control center A and control center B.

7. The online palletizing method with multiple error adjustments according to claim 6, characterized in that: After the secondary stacking adjustment of the diaper stack is completed, the rear clamping arm and the stacking robot arm are withdrawn, the left clamping arm and the right clamping arm press the diaper stack inward, and the vacuum pump set at the rear side of the packaging line unfolds the corresponding type of diaper packaging bag, and the left clamping arm and the right clamping arm move backward to send the diaper stack into the packaging bag and then pull it out, and the plastic sealing machine plastic seals the opening of the packaging bag; the clamping distance and pressure of the left clamping arm and the right clamping arm, the moving distance of the left clamping arm and the right clamping arm, the operating data of the vacuum pump and the operating data of the plastic sealing machine are all packaging data, which are set by the production personnel according to the current diaper stack type, and the set packaging data are synchronously input into the control center B, and combined with the corresponding clustering results to form a diaper stacking data set group.

8. The online palletizing method with multiple error adjustments according to claim 7, characterized in that: In step S6, the specific operation of automatic replenishment is that after removing the diaper pieces that are too protruding, the initial stacking production line controls the replenishment robot arm to push the diaper pieces on the rear side forward to replenish. When it runs to the last stack of diapers of this type, there is a shortage of diapers. The control center A issues an alarm, and the producer chooses to remove the stack of diapers or add new diapers to the stack of diapers.

Images