Method of controlling a stack separation apparatus, control device and stack separation apparatus

Abstract

The application provides a control method and device of a stack separation device and the stack separation device, and relates to the technical field of logistics equipment. The control method of the stack separation device comprises the following steps: continuously acquiring multiple package image data on an inclined conveying device by a camera assembly; determining the three-dimensional position, size and contour feature of each package in each package image data, and generating package features corresponding to each package; calculating the spatial overlap and contact area between each package feature, and determining the adhesion degree and stacking relationship between adjacent packages; determining the optimal separation order and corresponding motion path of each package according to the adhesion degree and stacking relationship; and adjusting the running speed of each inclined conveying device according to the optimal separation order, so that each package is sequentially separated into a single package along the motion path. The application realizes precise separation control of stacked packages, and has the advantages of improving the separation accuracy and efficiency of the packages and reducing the package damage rate.

Description

Technical Field

[0001] This application relates to the field of logistics equipment technology, and in particular to a control method, control device and stacking separation equipment for stacking parts. Background Technology

[0002] Stacking and separating equipment is a key front-end device in modern automated sorting systems, mainly used to solve the problem of low sorting efficiency caused by the stacking and adhesion of packages or parts during transportation. Existing methods are mostly based on static image analysis, lacking the ability to dynamically evaluate and adaptively adjust the adhesion relationship of packages in motion, resulting in decreased separation efficiency, increased error rate, and seriously affecting the overall throughput and reliability of the system. Summary of the Invention

[0003] The main objective of this invention is to provide a control method for a stacked package separation device, which aims to improve the accuracy and efficiency of package separation and reduce the package damage rate.

[0004] To achieve the above objectives, the present invention provides a control method for a stacked component separation device, the stacked component separation device comprising a plurality of inclined transport devices arranged sequentially and a plurality of camera components disposed above the inclined transport devices; the control method for the stacked component separation device includes: The camera assembly is controlled to continuously acquire image data of multiple packages on the tilting transport device; The three-dimensional position, size, and contour features of each package in each package image data are determined, and package features corresponding to each package are generated; Calculate the spatial overlap and contact area between the features of each package to determine the degree of adhesion and stacking relationship between adjacent packages; The optimal separation order and corresponding movement path for each package are determined according to the degree of adhesion and the layering relationship. The operating speed of each of the tilting transport devices is adjusted according to the optimal separation sequence so that each package is sequentially separated into individual packages along the movement path.

[0005] Optionally, determining the three-dimensional position, size, and contour features of each package in each package image data, and generating package features corresponding to each package, includes: Each package image data is preprocessed sequentially according to time. Based on the image segmentation algorithm, the package edges of each package image data are identified, and the package position and background position in the package image data are determined; Generate a point cloud model for each package corresponding to the package location and the background location, and calculate the three-dimensional coordinates of each package; The size parameters of each package are determined based on the three-dimensional coordinates, and the size parameters include length, width, and height values; Generate contour features corresponding to the wrapping edge, the contour features being generated through curvature calculation and shape descriptor; The wrapping feature is generated from the three-dimensional coordinates, size parameters, and contour features.

[0006] Optionally, the step of identifying the package edges of each package image data based on the image segmentation algorithm, and determining the package location and background location in the package image data, includes: The preprocessed package image data is analyzed based on a deep learning segmentation model, and pixel-level labels for the package region and background region are output. Based on the pixel-level labels, an edge detection algorithm is used to extract the outline of the package edge; The location of the broken edge in the outline of the package is determined, and the broken edge is reconstructed continuously using morphological repair and curve interpolation algorithms to generate complete edge outline data; Determine the package location and background location based on complete edge contour data; The process of generating a point cloud model for each package corresponding to the package location and the background location, and calculating the three-dimensional coordinates of each package, includes: Based on the package location and background location corresponding to each package, the depth image data of the corresponding package area is extracted and three-dimensionally mapped to generate point cloud data containing three-dimensional spatial coordinates. Generate a point cloud model corresponding to the point cloud data; The centroid coordinates of the point cloud model are calculated as the three-dimensional position of the package, and the three-dimensional coordinates of each package are determined.

[0007] Optionally, generating the contour features corresponding to the wrapping edge includes: Extract the sequence of contour points corresponding to the edge of the package; Calculate the curvature value corresponding to the contour point sequence and identify significant geometric feature points in the contour; The frequency domain characteristic coefficient sequence of the contour is generated by performing frequency domain transformation on the contour point sequence using Fourier descriptors. The frequency domain feature coefficient sequence is subjected to dimensionality reduction processing to construct a low-dimensional contour feature vector corresponding to the dimensionality-reduced frequency domain feature coefficient sequence; The low-dimensional contour feature vector is fused with curvature values ​​and geometric feature point information to generate the contour feature.

[0008] Optionally, calculating the spatial overlap and contact area between the features of each package, and determining the degree of adhesion and stacking relationship between adjacent packages, includes: Based on the three-dimensional position of each package, calculate the intersection area and union area of ​​the package bounding boxes, and generate the corresponding spatial overlap value. The point cloud model is used to analyze the contact area on the surface of the package, measure the point density distribution, and determine the actual contact area. Obtain the friction coefficient and surface characteristic parameters from the packaging material database, and establish an adhesive force calculation model; Based on the spatial overlap value and the actual contact area, the adhesion strength level between adjacent packages is evaluated, and the degree of adhesion is determined. Construct a topological diagram of the stacked packages and analyze the vertical support relationships and stacking order of each package; The stacking relationship is generated based on the support relationship and the stacking order.

[0009] Optionally, the step of evaluating the adhesion strength level between adjacent packages based on the spatial overlap value and the actual contact area, and determining the degree of adhesion, includes: Obtain the friction coefficient, surface roughness, and elastic modulus parameters for each package; Based on the actual contact area and friction coefficient, calculate the estimated static adhesion force. Based on the surface roughness and elastic modulus parameters, the estimated static adhesion force is corrected to generate the corrected adhesion force value. The corrected adhesion force value is weighted and calculated with the spatial overlap value to evaluate the adhesion strength level between adjacent packages and determine the degree of adhesion.

[0010] Optionally, determining the optimal separation order and corresponding movement path for each package based on the degree of adhesion and the stacking relationship includes: The degree of adhesion is determined to separate and sort each package, and the sorting status of each package is determined. Based on the multi-objective optimization algorithm, the stacking relationship, and the sorting situation, a priority sequence for package separation is generated to determine the optimal separation order; The starting point and target pose of each package are determined according to the optimal separation sequence. Generate a motion path corresponding to the starting point of the motion and the target pose.

[0011] Optionally, adjusting the operating speed of each of the tilting transport devices according to the optimal separation sequence, so that each package is sequentially separated into individual packages along the movement path, includes: The optimal separation sequence is used to determine the separation time of each package and the speed parameters of the corresponding tilting transport device, and the target operating speed of each tilting transport device is determined. Obtain the current operating speed of each of the tilting transport devices; The difference between the current operating speed and the target operating speed is determined, and the drive signal of the tilting transport device is adjusted so that each package is separated into individual packages sequentially along the movement path.

[0012] Furthermore, to achieve the above objectives, the present invention also provides a control device, the control device comprising: a memory, a processor, and a control program for a stacking separation device stored in the memory and executable on the processor, the control program for the stacking separation device being configured to implement the control method for the stacking separation device as described above.

[0013] In addition, to achieve the above objectives, the present invention also provides a stacking separation device, including the control device described above.

[0014] The stacked package separation device of this invention includes multiple inclined transport devices arranged sequentially and multiple camera components positioned above the inclined transport devices. The control method of the stacked package separation device continuously acquires image data of multiple packages on the inclined transport devices using the camera components. This determines the three-dimensional position, size, and contour features of each package in each image data, generates package features corresponding to each package, calculates the spatial overlap and contact area between the package features, determines the degree of adhesion and stacking relationship between adjacent packages, and determines the optimal separation order and corresponding movement path for each package based on the degree of adhesion and stacking relationship. Finally, the operating speed of each inclined transport device is adjusted according to the optimal separation order so that each package is sequentially separated into individual packages along the movement path. Thus, through three-dimensional positioning and dynamic path planning technology, combined with multi-dimensional package feature analysis, precise separation control of stacked packages is achieved, offering advantages such as improved package separation accuracy and efficiency, and reduced package damage rate. Attached Figure Description

[0015] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic flowchart of a control method for a stacked component separation device according to an embodiment of the present invention; Figure 2 This is a schematic flowchart of a control method for a stacked component separation device according to another embodiment of the present invention; Figure 3 This is a schematic flowchart of a control method for a stacked component separation device according to another embodiment of the present invention; Figure 4 This is a schematic flowchart of a control method for a stacking separation device according to another embodiment of the present invention; Figure 5 This is a schematic flowchart of a control method for a stacked component separation device according to another embodiment of the present invention; Figure 6 This is a schematic flowchart of a control method for a stacked component separation device according to another embodiment of the present invention; Figure 7 This is a schematic flowchart of a control method for a stacked component separation device according to another embodiment of the present invention; Figure 8 This is a schematic flowchart of a control method for a stacking separation device according to another embodiment of the present invention; Figure 9 This is a schematic diagram of the stack separation device of the present invention; Figure 10 for Figure 9 A schematic diagram of the cross-sectional structure.

[0018] Explanation of reference numerals in the attached figures: 10 - Inclined transport device, 20 - Camera assembly.

[0019] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Well-known modules, units, and their connections, links, communications, or operations are not shown or described in detail. Furthermore, the described features, architectures, or functions can be combined in any way in one or more embodiments. Those skilled in the art should understand that the various embodiments described below are only for illustrative purposes and not for limiting the scope of protection of the present invention. It is also readily understood that the modules, units, or processing methods in the various embodiments described herein and shown in the accompanying drawings can be combined and designed in various different configurations. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] The definitions of various terms or methods used in the following embodiments are, except where logically impossible, generally defined as broad concepts that can be implemented under the premise of the content disclosed in the embodiments. Under this understanding, all specific subordinate limitations of the terms or methods should be considered as part of the invention and should not be narrowly interpreted or biased simply because the specification does not disclose such a specific limitation. Similarly, provided that it is logically feasible, the order of the steps in the method is flexible and varied, and all specific subordinate limitations in the broad concepts of various terms or methods fall within the scope of protection of this invention.

[0022] Stacked package separation equipment is a key front-end device in modern automated sorting systems, primarily used to address the low sorting efficiency caused by the stacking and adhesion of packages or components during transportation. This type of equipment typically consists of multiple sequentially arranged tilting conveyor devices and a supporting vision acquisition system. It acquires images of the packages through camera components and combines image processing and motion control algorithms to automatically identify, locate, and separate stacked packages. Its core objective is to separate disorderly stacked packages one by one, forming a single-item flow to facilitate efficient subsequent sorting or processing steps. Traditional stacked package separation equipment has improved the level of sorting automation to some extent, but due to the diversity of package shapes, materials, and stacking methods, many technical challenges still exist.

[0023] In actual operation, when packages stick together due to surface materials (such as plastic bags or foam boxes) or external environmental factors (such as humidity or static electricity), traditional equipment relies solely on two-dimensional image analysis or fixed threshold separation strategies, making it difficult to accurately determine the actual adhesion status between packages. For example, in cases of slight adhesion or partial bonding, the system may fail to identify it as a unit requiring priority separation and instead continue operating according to a preset sequence, leading to pulling, package damage, or even equipment jamming during the separation process. Furthermore, existing methods are mostly based on static image analysis, lacking the ability to dynamically evaluate and adaptively adjust the adhesion relationship under the movement of packages, resulting in decreased separation efficiency, increased error rate, and severely impacting the overall system throughput and reliability.

[0024] The main solution in this application's embodiments is: as follows Figure 9 and Figure 10As shown, by controlling the camera component 20 to continuously acquire multiple package image data on the inclined transport device 10, the three-dimensional position, size and contour features of each package in each package image data are determined, package features corresponding to each package are generated, the spatial overlap and contact area between each package feature are calculated, the degree of adhesion and stacking relationship between adjacent packages are determined, and the optimal separation order and corresponding movement path of each package are determined according to the degree of adhesion and stacking relationship. Finally, the running speed of each inclined transport device 10 is adjusted according to the optimal separation order so that each package is separated into individual packages along the movement path.

[0025] In this embodiment, for ease of description, the following description will focus on the control device as the executing entity.

[0026] This application provides a solution that uses three-dimensional positioning and dynamic path planning technology, combined with multi-dimensional package feature analysis, to achieve precise separation control of stacked packages. This solution has the advantages of improving package separation accuracy and efficiency and reducing package damage rate.

[0027] Therefore, the present invention proposes a control method for a stacking separation device; it is understood that the stacking separation device is provided with a control device for storing and executing the following method. The control device can be implemented by a main controller, such as an MCU (Micro Controller Unit), a DSP (Digital Signal Processor), an FPGA (Field Programmable Gate Array), or a SOC (System On Chip).

[0028] In existing technologies, stacked package separation equipment in automated sorting systems often faces the problem of low separation efficiency due to package adhesion. Traditional methods rely on two-dimensional image analysis and fixed threshold strategies, which cannot accurately identify slight adhesion or localized sticking, easily leading to package damage or equipment stalling. For example, in high-humidity environments, plastic bags may form surface contact due to electrostatic adsorption. Existing stacked package separation equipment struggles to dynamically assess such adhesion relationships, resulting in unreasonable separation sequence planning.

[0029] To address the aforementioned issues, a dynamic analysis model of the three-dimensional spatial relationships of packages is needed. Traditional two-dimensional images cannot reflect the true contact state of the packages; therefore, three-dimensional vision technology should be introduced to acquire three-dimensional information about the packages. By analyzing the spatial overlap and contact area between packages, the degree of adhesion can be quantified. Furthermore, a topological model of the stacked relationships needs to be established to optimize the separation sequence and transportation path, enabling dynamic speed adjustment.

[0030] Based on the above, referring to Figure 1In one embodiment of the present invention, the stacked component separation device includes a plurality of inclined transport devices 10 arranged sequentially and a plurality of camera components 20 disposed above the inclined transport devices 10; the control method of the stacked component separation device includes steps S100-S500, wherein: S100: Control the camera component to continuously acquire image data of multiple packages on the tilting transport device; S200: Determine the three-dimensional position, size, and contour features of each package in each package image data, and generate package features corresponding to each package; S300. Calculate the spatial overlap and contact area between the features of each package to determine the degree of adhesion and stacking relationship between adjacent packages. S400: Determine the optimal separation sequence and corresponding movement path for each package according to the degree of adhesion and the stacking relationship; S500. Adjust the operating speed of each of the inclined transport devices according to the optimal separation sequence so that each package is separated into individual packages sequentially along the movement path.

[0031] The inclined transport device 10 refers to a conveyor mechanism with an inclined design, which can be implemented using a segmented roller conveyor. The inclined design utilizes gravity to assist in package separation. The camera component 20 refers to a multi-view image acquisition system, which can be implemented using a binocular stereo camera array. The camera component 20 is used to acquire three-dimensional spatial data of the package. The three-dimensional position refers to the spatial coordinates of the package's centroid, which can be calculated using a point cloud model centroid. The three-dimensional position is used to determine the spatial distribution of the package. Spatial overlap refers to the ratio of the intersection to the union of the package's projected areas, which can be calculated using a bounding box collision detection algorithm, reflecting the package's stacking density. The contact area refers to the actual contact area on the package surface, which can be obtained through point cloud density analysis. The contact area is used to assess the physical adhesion strength. The motion path refers to the trajectory of the package from its current position to the separation position. The motion path can be generated using an inverse kinematics algorithm to ensure collision-free separation.

[0032] Multiple inclined conveyor devices 10 form a stepped conveying channel, with camera components 20 continuously acquiring images of stacked packages. After preprocessing, the image data is used to reconstruct a 3D model of the packages using a stereo vision algorithm, extracting dimensions and contour features. The spatial overlap between packages is calculated using bounding box intersection, and the contact area is obtained through point cloud contact area density analysis. An adhesion assessment module, combined with a material property database, calculates the mechanical parameters required for actual separation. A layered topology model analyzes the package support state and determines the priority separation order of the bottom layer. A path planning module generates the movement trajectory of each package according to the separation order, and a speed control module dynamically adjusts the speed difference between each section of the conveyor, ensuring that the packages are separated layer by layer according to the planned path.

[0033] Compared to existing technologies, traditional methods rely on two-dimensional images, leading to a high rate of false positives for adhesion. This embodiment accurately quantifies the contact state through three-dimensional spatial analysis. Existing systems using a fixed separation sequence are prone to tearing and breakage. This embodiment optimizes the separation priority based on a dynamic topology model. Traditional speed control strategies cannot adapt to complex stacking conditions. This embodiment achieves coordinated speed adjustment through path planning, avoiding secondary stacking of packages.

[0034] Through the above technical solutions, the accuracy of package adhesion identification is significantly improved, and the damage rate during separation is effectively reduced. The dynamic speed adjustment mechanism adapts to different stacking configurations, enhancing equipment operational stability. Three-dimensional spatial relationship modeling makes the separation of complex stacked packages possible, substantially improving the sorting system's processing capacity.

[0035] The stacked package separation device of this embodiment includes multiple inclined transport devices 10 arranged sequentially and multiple camera components 20 disposed above the inclined transport devices 10. The control method of the stacked package separation device involves controlling the camera components 20 to continuously acquire image data of multiple packages on the inclined transport devices 10, determining the three-dimensional position, size, and contour features of each package in each image data, generating package features corresponding to each package, calculating the spatial overlap and contact area between the package features, determining the degree of adhesion and stacking relationship between adjacent packages, and determining the optimal separation order and corresponding movement path for each package according to the degree of adhesion and stacking relationship. Finally, the operating speed of each inclined transport device 10 is adjusted according to the optimal separation order so that each package is sequentially separated into individual packages along the movement path. Thus, through three-dimensional positioning and dynamic path planning technology, combined with multi-dimensional package feature analysis, precise separation control of stacked packages is achieved, which has the advantages of improving package separation accuracy and efficiency and reducing package damage rate.

[0036] Optionally, refer to Figure 2 Another embodiment of the present invention provides a control method for a stacking separation device, based on the above. Figure 1 The illustrated embodiment determines the three-dimensional position, size, and contour features of each package in each package image data, and generates package features corresponding to each package, including steps S210-S260, wherein: S210. Preprocess each package image data in chronological order. The preprocessing includes noise filtering, image correction and brightness equalization. S220. Identify the package edges of each package image data based on an image segmentation algorithm, and determine the package position and background position in the package image data; S230. Generate a point cloud model for each package corresponding to the package location and the background location, and calculate the three-dimensional coordinates of each package. S240. Determine the size parameters of each package based on the three-dimensional coordinates, wherein the size parameters include length, width, and height values; S250. Generate a contour feature corresponding to the edge of the package, wherein the contour feature is generated by curvature calculation and shape descriptor; S260, the wrapping feature generated from the three-dimensional coordinates, size parameters, and contour features.

[0037] The preprocessing of package image data involves noise filtering, image correction, and brightness equalization of the original image. Preprocessing can utilize median filtering to eliminate salt-and-pepper noise, affine transformation to correct image distortion, and histogram equalization to adjust brightness distribution, thereby improving the accuracy of subsequent image segmentation. Image segmentation algorithms are techniques for identifying package edges and distinguishing the package from the background. These can employ a U-Net-based deep learning model for pixel-level semantic segmentation, combined with the Canny edge detection algorithm to extract the package contour, and morphological closing operations to repair broken edges, thus fully locating the package area. The point cloud model is the geometric representation of the package in three-dimensional space. Depth images can be acquired using binocular vision or structured light sensors. The point cloud model uses a 3D mapping algorithm to convert two-dimensional pixel coordinates into three-dimensional spatial coordinates, generating point cloud data containing location information. Three-dimensional coordinates refer to the positional parameters of the package's centroid in three-dimensional space, which can be determined by calculating the geometric center or weighted centroid of the point cloud model. These three-dimensional coordinates are used to quantify the spatial distribution of the package. Size parameters refer to the measured length, width, and height of the package. These can be calculated using the minimum bounding cube algorithm or principal component analysis algorithm of the point cloud model. Size parameters are used to evaluate the package volume and stacking status. Contour features refer to the geometric shape description of the package edges. These can be calculated by calculating the curvature of contour points, extracting Fourier descriptors for frequency domain dimensionality reduction, and combining key feature points to construct a low-dimensional feature vector. Contour features are used to distinguish the morphological differences between different packages.

[0038] The package image data is first preprocessed in time series. Noise filtering eliminates sensor noise interference, image correction compensates for lens distortion, and brightness equalization unifies lighting conditions. Then, a deep learning segmentation model is used to identify the package region in the preprocessed image. Edge detection algorithms are combined to extract the package contour, and morphological repair eliminates contour breaks. Based on the segmentation results, depth data of the package region is extracted, and a point cloud model is generated through 3D mapping, with centroid coordinates calculated as the 3D position. The minimum bounding cube of the point cloud model is used to determine the package size parameters. Simultaneously, the contour point sequence is extracted for curvature calculation and Fourier transform, generating a low-dimensional contour descriptor that integrates geometric features. Finally, the 3D coordinates, size parameters, and contour features are integrated into package feature data, providing multi-dimensional input for subsequent adhesion analysis.

[0039] Compared to existing technologies, traditional methods rely on two-dimensional image analysis, which makes it difficult to accurately determine the three-dimensional stacking relationship of packages. This embodiment, however, achieves precise quantification of the three-dimensional spatial position and morphological features of packages through point cloud model construction and contour feature extraction. Existing technologies using fixed threshold segmentation are susceptible to changes in lighting. This embodiment, through the combined application of deep learning models and morphological repair, significantly improves the robustness of package edge recognition in complex scenes. Existing methods lack the ability to describe the geometric features of package surfaces. This embodiment enhances the feature discrimination of packages of different shapes through the fusion of curvature calculation and Fourier descriptors. This embodiment effectively solves the problem of misjudging package adhesion caused by the limitations of two-dimensional image analysis in traditional devices. Through precise extraction of three-dimensional coordinates and contour features, the spatial overlap and contact area between packages can be accurately identified, providing reliable data support for subsequent separation order decisions. Optimization of the preprocessing workflow further improves image segmentation accuracy, avoids package positioning deviations caused by noise interference or uneven lighting, and ensures the stability of subsequent feature extraction.

[0040] Optionally, refer to Figure 3 Another embodiment of the present invention provides a control method for a stacking separation device, based on the above. Figure 2 The illustrated embodiment identifies the package edges of each package image data based on an image segmentation algorithm, and determines the package location and background location in the package image data, including steps S221-S224, wherein: S221. Analyze the preprocessed package image data based on a deep learning segmentation model, and output pixel-level labels for the package region and background region. S222. Based on the pixel-level label, use an edge detection algorithm to extract the outline of the package edge; S223. Determine the location of the broken edge in the outline of the package, and use morphological repair and curve interpolation algorithms to reconstruct the continuity of the broken edge to generate complete edge outline data. S224. Determine the package position and background position according to the complete edge contour data.

[0041] Generate a point cloud model for each package corresponding to the package location and the background location, and calculate the three-dimensional coordinates of each package, including steps S231-S233, wherein: S231. Based on the package location and background location corresponding to each package, extract the depth image data of the corresponding package area for three-dimensional mapping to generate point cloud data containing three-dimensional spatial coordinates. S232. Generate a point cloud model corresponding to the point cloud data; S233. Calculate the centroid coordinates of the point cloud model as the three-dimensional position of the package, and determine the three-dimensional coordinates of each package.

[0042] The deep learning segmentation model refers to an algorithm that performs pixel-level classification of package regions in an image by training a multi-layer neural network. It can be implemented using a convolutional neural network combined with an encoder-decoder structure. This model addresses the sensitivity of traditional threshold segmentation to complex backgrounds and variations in package surface material. The edge detection algorithm is a computational method for extracting contour boundaries from the segmented package region. It can be implemented using the Canny or Sobel operator combined with non-maximum suppression techniques. This algorithm accurately identifies transition areas between the package and the background. The morphological repair and curve interpolation algorithm involves performing dilation and erosion operations on broken edges and then restoring continuity through polynomial fitting. This can be implemented using morphological closing operations combined with cubic spline interpolation. This algorithm eliminates incomplete contours caused by uneven lighting or occlusion. The point cloud model is a set of three-dimensional points generated by transforming depth image data. It can be generated by acquiring depth information using binocular vision or structured light imaging techniques and then transforming the coordinates. The point cloud model is used to establish the spatial mapping relationship of the package on the transport device. Centroid coordinates refer to the geometric center position of all three-dimensional coordinate points in the point cloud model. They can be calculated by weighted averaging the mean values ​​of each coordinate axis. Centroid coordinates are used to characterize the actual three-dimensional positioning of the device wrapped on the inclined transport device 10.

[0043] The package image data, after noise filtering and brightness equalization preprocessing, is input into a pre-trained deep learning segmentation model, which outputs a probability distribution map of each pixel belonging to the package or the background. Binarization is performed based on the probability distribution map to obtain a mask image of the package region. Edge detection is performed on the mask image to extract the initial contour line and detect broken areas within the contour line. Small holes are filled using morphological closing operations, and cubic spline interpolation is performed on the broken edges to reconstruct a continuous closed contour curve. The package location region and the background region are divided based on the complete contour curve. For each package location region, depth information is extracted from the disparity map obtained from the binocular camera. The two-dimensional pixel coordinates and depth values ​​are converted into three-dimensional spatial coordinates using a coordinate transformation formula, forming a point cloud dataset. After outlier filtering and downsampling processing on the point cloud data, a three-dimensional point cloud model of the package is constructed. Finally, the average coordinates of all points in the point cloud model are calculated to obtain the three-dimensional coordinates of the package's centroid.

[0044] Existing methods rely on fixed threshold segmentation, resulting in low recognition rates for reflective or semi-transparent packages, and two-dimensional image analysis cannot accurately determine the spatial position of the packages. This embodiment achieves adaptive segmentation through a deep learning model, combined with 3D point cloud modeling technology, which can accurately identify the three-dimensional contours of packages of different materials and accurately determine the actual position of the package on the inclined transport device 10 by calculating the centroid coordinates, providing reliable three-dimensional spatial data support for subsequent separation sequence decisions. This embodiment effectively solves the problem of package adhesion misjudgment caused by the limitations of two-dimensional image analysis in traditional equipment, significantly improves the completeness of package edge detection and 3D positioning accuracy in complex stacking scenarios, and provides an accurate spatial data foundation for subsequent calculation of the actual contact area and stacking relationship between packages, thereby reducing the risk of package tearing and damage during separation.

[0045] Optionally, refer to Figure 4 In another embodiment of the present invention, a control method for a stacking separation device is provided, based on the above. Figure 3 The illustrated embodiment generates contour features corresponding to the edge of the package, including steps S251-S255, wherein: S251. Extract the contour point sequence corresponding to the edge of the package; S252. Calculate the curvature value corresponding to the contour point sequence and identify significant geometric feature points in the contour; S253. Perform frequency domain transformation on the contour point sequence using Fourier descriptors to generate the frequency domain feature coefficient sequence of the contour. S254. Perform dimensionality reduction processing on the frequency domain feature coefficient sequence to construct a low-dimensional contour feature vector corresponding to the dimensionality-reduced frequency domain feature coefficient sequence. S255. The low-dimensional contour feature vector is fused with curvature value and geometric feature point information to generate the contour feature.

[0046] The contour point sequence refers to the set of coordinate points arranged sequentially along the contour line of the package edge. It can be extracted using edge tracking algorithms and is used to describe the geometric distribution of the package shape. The curvature value refers to the curvature value of each point in the contour point sequence, which can be calculated using the finite difference method or polynomial fitting. The curvature value is used to quantify the degree of local curvature of the contour. Significant geometric feature points are points with curvature values ​​exceeding a set threshold or extreme curvature points, such as inflection points or corners. They can be identified using curvature thresholding or extreme value detection algorithms and are used to capture key morphological changes in the contour. The Fourier descriptor refers to the frequency domain coefficients obtained after performing a Fourier transform on the contour point sequence. It can be implemented using discrete Fourier transform and is used to convert the contour shape into frequency domain features to eliminate noise interference. Dimensionality reduction refers to truncating or performing principal component analysis on the high-frequency domain coefficients. This can be achieved by retaining the first N low-frequency coefficients. Dimensionality reduction reduces the data dimensionality while preserving the main shape information. The low-dimensional contour feature vector is a vector composed of the frequency domain feature coefficients after dimensionality reduction and is used to characterize the core frequency domain features of the package contour. Geometric feature point information fusion refers to the feature concatenation or weighted combination of frequency domain feature vectors, curvature values, and geometric feature points. It can be achieved by feature concatenation or neural network feature fusion methods. Geometric feature point information fusion is used to comprehensively describe the local details and overall shape features of a contour.

[0047] After the contour point sequence of the package edge is extracted, the curvature value of each point is first calculated to identify key geometric feature points, such as package corners or concave areas. Then, a Fourier transform is performed on the contour point sequence to convert it into a frequency domain coefficient sequence, for example, by generating coefficients containing harmonic components from order 0 to K through discrete Fourier transform. Next, the high-frequency coefficients are truncated, for example, retaining only the first 20 low-frequency coefficients to construct a low-dimensional feature vector. Finally, the low-dimensional frequency domain features are fused with the curvature extremum locations and curvature distribution statistics to form a composite contour descriptor that simultaneously contains global frequency domain features and local geometric features. This descriptor can effectively distinguish contour deformation caused by adhesion from differences in the shape of the package itself; for example, it can accurately distinguish between the contours of two partially overlapping packages and the complex contours of a single package.

[0048] Existing methods typically rely solely on geometric feature points or curvature analysis for contour matching, which can easily lead to misjudgments when package edges undergo local deformation due to adhesion. For example, when the contours of two packages in contact area experience abrupt curvature changes due to compression, traditional methods may incorrectly identify them as the inherent shape features of a single package. This embodiment, however, introduces frequency domain features to capture the overall shape characteristics of the contour, reducing local noise interference. Simultaneously, it enhances sensitivity to key deformation areas by combining geometric feature points, thereby improving the contour recognition accuracy of adhered packages. This embodiment effectively distinguishes contour distortion caused by package adhesion from the difference in the true shape of the package, avoiding misjudging adhered packages as single packages and thus improving the accuracy of separation order decisions. For example, in a package stacking scenario, when two square packages experience local contour indentation due to lateral contact, this embodiment can identify the overall rectangular features through frequency domain features and, combined with curvature abrupt changes in the contact area, determine them as two independent packages, thereby correctly triggering the separation operation.

[0049] Optionally, refer to Figure 5 The present invention also provides a control method for a stacking separation device, based on the above. Figure 4 The illustrated embodiment calculates the spatial overlap and contact area between each package feature to determine the adhesion and stacking relationship between adjacent packages, including steps S310-S360, wherein: S310. Based on the three-dimensional position of each package, calculate the intersection area and union area of ​​the package bounding boxes, and generate the corresponding spatial overlap values. S320. Analyze the contact area of ​​the package surface using the point cloud model, measure the point density distribution, and determine the actual contact area. S330. Obtain the friction coefficient and surface characteristic parameters from the package material database, and establish an adhesion force calculation model; S340. Based on the spatial overlap value and the actual contact area, evaluate the adhesion strength level between adjacent packages and determine the degree of adhesion. S350. Construct a topology diagram of the package stacking and analyze the support relationship and stacking order of each package in the vertical direction; S360. Generate the stacking relationship according to the support relationship and the stacking order.

[0050] Spatial overlap refers to the ratio of the intersection area to the union area of ​​the 3D bounding boxes of packages on the projection plane. It can be achieved using bounding box collision detection algorithms combined with geometric operations. Spatial overlap is used to quantify the degree of positional overlap between packages. Contact area refers to the area of ​​the actual contact region on the package surface. It can be achieved by measuring the density distribution of the overlapping area of ​​adjacent package point clouds in a point cloud model. Combining this with surface curvature analysis improves the accuracy of contact area identification. Adhesion force calculation model refers to the mechanical relationship established based on the friction coefficient and material elastic parameters. It can be achieved using Coulomb's law of friction combined with the material deformation energy equation. The adhesion force calculation model is used to simulate the adhesive force generated between packages due to physical properties. Topology diagram refers to the spatial hierarchical relationship network of the package stacking structure. It can be achieved by constructing a directed graph through 3D coordinate analysis of the support contact points and vertical projection relationships between packages. The topology diagram is used to describe the hierarchical dependency order of package stacking.

[0051] The calculation of the intersection and union areas of the 3D bounding boxes of the packages reflects the spatial overlap of the packages. Combined with the statistical analysis of the point density in the contact area using a point cloud model, the actual contact area of ​​the package surface can be accurately identified. The adhesion force calculation model dynamically corrects the adhesion force estimate based on the contact area by introducing the friction coefficient and elastic parameters from the material database. The adhesion strength level assessment weights the spatial overlap with the corrected adhesion force to form a quantitative classification of the adhesion state between packages. The topology diagram establishes the hierarchical dependency of the stacked structure by analyzing the distribution of support points of the packages in the vertical direction, providing a spatial logical basis for the separation sequence.

[0052] Existing methods rely on two-dimensional image analysis, which fails to identify the three-dimensional spatial overlap of packages. This embodiment, however, combines a three-dimensional bounding box with a point cloud model to dynamically assess the contact area and spatial relationship of packages. Existing technologies lack dynamic correction for material properties; this embodiment integrates friction coefficients and elastic parameters through an adhesion force calculation model, adapting to the adhesion analysis of packages made of different materials. Existing methods do not construct stacking topology relationships; this embodiment generates a hierarchical dependency graph through vertical support point analysis, optimizing the separation sequence and avoiding pulling problems caused by lower-layer packages not being separated. This embodiment accurately identifies the three-dimensional spatial overlap and surface contact areas between packages, dynamically assesses adhesion strength based on material properties, and effectively distinguishes between slightly adhered and tightly adhered packages. By constructing a stacking topology graph, separation path conflicts caused by misjudgment of stacking order can be avoided, thereby reducing the risk of package damage and improving the success rate of separation operations.

[0053] Optionally, refer to Figure 6 Another embodiment of the present invention provides a control method for a stacking separation device, based on the above. Figure 5The embodiment shown evaluates the adhesion strength level between adjacent packages based on the spatial overlap value and the actual contact area, and determines the degree of adhesion, including steps S341-S344, wherein: S341. Obtain the friction coefficient, surface roughness, and elastic modulus parameters corresponding to each package; S342. Based on the actual contact area and friction coefficient, calculate the estimated value of static adhesion force; S343. Based on the surface roughness and elastic modulus parameters, correct the estimated static adhesion force and generate the corrected adhesion force value. S344. The corrected adhesion force value and the spatial overlap value are weighted and calculated to evaluate the adhesion strength level between adjacent packages and determine the degree of adhesion.

[0054] The coefficient of friction refers to the sliding resistance parameter generated when the surface of a package comes into contact with an adjacent package. This can be achieved by measuring the static friction coefficients of different material samples using a material testing instrument and establishing a database for matching. The coefficient of friction is used to quantify the sliding resistance between packages. Surface roughness refers to the degree of microscopic unevenness of the package contact surface. This can be achieved by scanning the surface morphology with a laser profilometer and calculating the average roughness value. Surface roughness is used to correct the actual effective area of ​​the contact area. The elastic modulus parameter refers to the deformation capacity of the package material under stress. This can be achieved by performing stress-strain curve tests on standard samples using a compression testing machine. The elastic modulus parameter is used to evaluate the deformation recovery characteristics of the contact area. The estimated static adhesive force is the estimated value of the adhesive force generated by contact when the package is at rest. This can be calculated using the Coulomb friction formula, which uses the product of the coefficient of friction and the normal contact force as the basis. The estimated static adhesive force is used to preliminarily determine the external force required for separation. The corrected adhesion force value refers to the dynamic adhesion force considering the influence of surface morphology and material elasticity. It can be achieved by iteratively correcting the roughness parameters and elastic modulus using a contact mechanics model and incorporating them into Hertzian contact theory. The corrected adhesion force value is used to more accurately reflect the actual contact state. The weighted calculation refers to a comprehensive evaluation method that integrates spatial overlap and the corrected adhesion force according to a preset ratio. This can be achieved by mapping the two types of parameters to a unified dimension and then superimposing them using a linear weighting function. The weighted calculation is used to balance the coupling effect of geometric overlap and physical adhesion.

[0055] The friction coefficient, surface roughness, and elastic modulus parameters of the package are obtained from a pre-set material database. Combined with the actual contact area measured by the vision system, the initial adhesion force is calculated using the Coulomb friction model. Due to the microscopic irregularities and elastic deformation of the actual contact surface, a roughness compensation factor and an elastic deformation correction coefficient are introduced based on Hertzian contact theory to dynamically adjust the initial adhesion force. The adjusted adhesion force and the three-dimensional spatial overlap data are input into a weighted model. Through normalization and coefficient allocation, a comprehensive adhesion strength level is generated, ultimately outputting a quantified adhesion degree index to guide separation priority decisions.

[0056] Existing methods rely solely on two-dimensional image analysis of spatially overlapping areas, neglecting the influence of material properties on actual contact forces. This can easily lead to slight adhesion being misjudged as a lack of adhesion. This embodiment, by fusing physical properties of the package surface with three-dimensional geometric data, establishes a dynamically corrected adhesion force calculation model. This model accurately distinguishes between physical adsorption and geometric overlap, avoiding misjudgments caused by differences in material hardness or environmental factors. This embodiment solves the problem of misjudging adhesion strength caused by neglecting material properties in traditional equipment, effectively reducing the risk of package tearing due to misjudgment of adhesion, improving the accuracy of separation sequence decisions in complex stacking scenarios, and ensuring the stability and sorting efficiency of continuous equipment operation.

[0057] Optionally, refer to Figure 7 Another embodiment of the present invention provides a control method for a stacking separation device, based on the above. Figure 1 The illustrated embodiment determines the optimal separation order and corresponding movement path for each package according to the degree of adhesion and the stacking relationship, including steps S410-S440, wherein: S410. Determine the degree of adhesion and separate and sort each package to determine the sorting status of each package; S420. Based on the multi-objective optimization algorithm, the stacking relationship, and the sorting situation, a priority sequence for package separation is generated to determine the optimal separation order; S430. Determine the starting point and target pose of each package according to the optimal separation sequence; S440. Generate a motion path corresponding to the motion starting point and the target pose.

[0058] The degree of adhesion refers to the strength of the adhesive force between packages due to contact area, friction coefficient, and surface characteristics. It can be calculated by analyzing the contact area using a point cloud model and combining it with material database parameters. The degree of adhesion is used to determine the threshold of mechanical force required for separation. The layering relationship refers to the topological diagram of the support structure of the packages in the vertical direction. It can be constructed through the intersection analysis of three-dimensional coordinates and bounding boxes. The layering relationship is used to identify the support state of the lower packages for the upper packages. The multi-objective optimization algorithm is a decision model that simultaneously considers separation efficiency, path conflict avoidance, and energy consumption control. It can be implemented using a genetic algorithm or particle swarm optimization algorithm. The multi-objective optimization algorithm is used to generate a priority sequence that takes into account multiple constraints. The motion path refers to the trajectory of the package from its current position to the separation target pose. It can be generated by a path planning algorithm combined with the operating parameters of the inclined transport device 10. The motion path is used to control the direction and speed of the package's movement during the separation process.

[0059] The data on the degree of adhesion of the packages and the topology of their stacking relationships are input into a multi-objective optimization algorithm. The algorithm iteratively calculates a comprehensive score for different separation orders based on preset separation efficiency weights, path conflict penalty factors, and energy consumption thresholds. For example, when two packages are highly adhered and located on the same support layer, the algorithm prioritizes their separation to avoid cascading motion; when packages are located on different layers and have no direct contact, the algorithm can postpone their separation to reduce frequent speed adjustments by the transport device. The optimized priority sequence is converted into the motion starting point and target pose coordinates of each package. The path planning module generates the package movement trajectory based on the tilt angle and operating speed constraints of the transport device, ensuring that the separation paths of adjacent packages do not conflict in time and space.

[0060] Existing separation order decision-making methods are typically based on fixed rules or single parameter thresholds, such as arranging packages solely by size or arrival order, which cannot dynamically adapt to complex stacking scenarios. This embodiment integrates multi-dimensional physical characteristic data with real-time optimization algorithms, enabling separation order decision-making to synchronously respond to the dynamic mechanical relationships between packages and the equipment's operating status. This effectively solves the path conflicts and secondary adhesion problems caused by static decision-making in traditional methods. This embodiment achieves dynamic optimization decision-making for the separation order of adhered packages, avoiding mechanical pulling or package congestion caused by fixed separation strategies, and improving the separation success rate in high-density stacking scenarios. Simultaneously, through coordinated control of path planning and transportation device speed adjustment, mechanical wear caused by sudden stops and starts of the equipment is reduced, ensuring the continuity and stability of the separation process.

[0061] Optionally, refer to Figure 8 In another embodiment of the present invention, a control method for a stacking separation device is provided, based on the above. Figure 1The illustrated embodiment adjusts the operating speed of each of the tilting transport devices 10 according to the optimal separation sequence so that each package is sequentially separated into individual packages along the movement path, including steps S510-S530, wherein: S510. Determine the separation time of each package and the speed parameters of the corresponding tilting transport device based on the optimal separation sequence, and determine the target operating speed of each tilting transport device. S520, Obtain the current operating speed of each of the inclined transport devices; S530. Determine the difference between the current operating speed and the target operating speed, and adjust the drive signal of the tilting transport device so that each package is separated into individual packages along the movement path.

[0062] The separation time refers to the point in time when the package begins the separation operation. This can be achieved using a timestamp or a trigger signal. The separation time ensures that the package initiates the separation action at the correct time according to its priority sequence. The speed parameter refers to the operating speed value that the tilting transport device 10 needs to achieve during the separation process. This can be set using a preset speed curve or a real-time calculation model. The speed parameter is used to match the movement path of the package and the separation requirements. The drive signal refers to the electrical signal that controls the drive motor of the tilting transport device 10. This signal can be generated using pulse width modulation or voltage regulation. The drive signal is used to precisely adjust the operating speed of the transport device.

[0063] After determining the optimal separation order, the control module assigns a corresponding separation time window and a target speed for the inclined transport device 10 to each package based on its priority sequence and movement path planning. For example, when package A needs to be separated first, the system calculates the acceleration, deceleration, and steady-state speed of the inclined transport device 10 at the moment of separation, generating a target speed curve. Subsequently, the system monitors the current speed of the transport device in real time using sensors, calculates the difference between the current speed and the target speed, and generates a speed error signal. Based on this error signal, the control module dynamically adjusts the input voltage or current of the drive motor, for example, by using a proportional-integral-differential algorithm to correct the drive signal, so that the actual operating speed of the transport device gradually approaches the target speed. During this process, package A moves along a preset path at the adjusted transport speed, disengages from adjacent packages, and completes the separation. Subsequent packages are processed sequentially according to the same logic, forming a continuous single-item flow.

[0064] Existing methods typically employ fixed speeds or speed adjustment strategies based on simple thresholds, which cannot adapt to the complex conditions during dynamic package separation. For example, existing technologies may simply distribute transport speed evenly based on the number of packages, ignoring the impact of the degree of adhesion and stacking relationship between packages on the separation sequence, leading to separation conflict or package delays. This embodiment, however, dynamically correlates the separation sequence with speed parameters, combining real-time speed feedback and closed-loop control of drive signals to precisely match the package's motion characteristics with separation requirements, avoiding package pulling or equipment jamming due to speed mismatch.

[0065] This embodiment solves the problems of low separation efficiency and package damage caused by inaccurate speed control in the prior art. By dynamically adjusting the operating speed of the tilting transport device 10, it ensures that each package completes the separation action at the optimal time and at an appropriate speed, effectively reducing mutual interference between packages and improving the reliability of the separation process and system throughput.

[0066] The present invention also proposes a control device, the control device comprising: a memory, a processor, and a control program for a stacked component separation device stored in the memory and executable on the processor, the control program for the stacked component separation device being configured to implement the control method for the stacked component separation device as described above.

[0067] The memory refers to the storage medium used to store the control program and data generated during operation. It can be implemented using a solid-state drive or flash memory chip. Its function is to store image processing algorithms, motion path planning parameters, and equipment operation logs. The processor is the arithmetic unit that executes the control program instructions. It can be implemented using a multi-core central processing unit or an embedded microcontroller. Its function is to process package image data in real time and generate speed adjustment instructions for the transport device. The control program of the stacking and separating equipment refers to the software module containing image analysis, spatial modeling, and motion control logic. It can be written in Python or C++ and compiled into an executable file. Its function is to dynamically evaluate the adhesion status of the packages and optimize the separation sequence.

[0068] The control device continuously acquires images of packages on the inclined transport device 10 via the camera component 20. The processor calls the image segmentation algorithm in the control program to identify the edges of the packages and generate a 3D point cloud model. Based on the point cloud model, the spatial overlap and contact area of ​​each package are calculated, and the adhesion strength level is dynamically evaluated by combining the friction coefficient parameters in the material database. The control program uses a multi-objective optimization algorithm to generate the optimal separation sequence and adjusts the drive signal according to the difference between the current transport speed and the target speed, so that the inclined transport device 10 separates the packages sequentially along a predetermined path. For example, when it is detected that the contact area of ​​two packages has increased abnormally due to electrostatic adsorption, the control program will prioritize adjusting the acceleration parameters of the corresponding transport segment to reduce the risk of adhesion.

[0069] In some specific implementations, the memory can be configured as a dual-channel redundant storage structure to improve data reliability, the processor can integrate a graphics processing unit to accelerate three-dimensional coordinate calculations, and the control program can support remote update functionality to adapt to parameter adjustment requirements for different packaging materials.

[0070] It is worth noting that since the control device of the present invention is based on the control method of the stacked parts separation device described above, the embodiments of the control device of the present invention include all the technical solutions of all embodiments of the control method of the stacked parts separation device described above, and the technical effects achieved are exactly the same, so they will not be repeated here.

[0071] like Figure 9 and Figure 10 As shown, the present invention also proposes a stacking separation device, which includes a control device as described in the above embodiments.

[0072] The control device refers to the hardware module used to execute the package separation control logic. It can be implemented using an embedded industrial computer or a programmable logic controller (PLC). It processes image data and motion parameters in real time by running a preset algorithm program. The memory refers to the physical medium that stores the control program and running data. It can be implemented using a solid-state drive (SSD) or flash memory chip. The memory is used to store image processing algorithms, motion path planning models, and package feature databases. The processor is the core component that executes the control program calculations. It can be implemented using a multi-core CPU or GPU chip. It accelerates package feature extraction and adhesion analysis through parallel computing. The control program of the stacked package separation device refers to the software module that realizes package recognition, separation decision-making, and motion control. It can be developed by combining deep learning frameworks with kinematic simulation algorithms. The control program of the stacked package separation device is used to dynamically adjust the operating parameters of the tilting transport device 10.

[0073] The stacking separation device acquires real-time package image data from the camera component 20 via a processor in the control unit, and calls the control program in the memory to perform noise filtering and 3D coordinate reconstruction on the images. After generating package contour features, the program calculates the contact area and spatial overlap of adjacent packages based on a point cloud model, and evaluates the adhesion strength level by combining the friction coefficient in the material parameter library. Subsequently, the program generates a separation priority sequence according to the stacking topology and determines the movement path of each package through a multi-objective optimization algorithm. Finally, the processor sends a speed adjustment command to the tilting transport device 10, causing it to separate the packages one by one according to the optimal separation order.

[0074] It is worth noting that since the stacking separation device of the present invention is based on the above-mentioned control device, the embodiments of the stacking separation device of the present invention include all the technical solutions of all the embodiments of the above-mentioned control device, and the technical effects achieved are exactly the same, so they will not be repeated here.

[0075] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0076] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0077] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0078] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A control method for a stacking separation device, characterized in that, The stacked parts separation device includes multiple inclined transport devices arranged in sequence and multiple camera components disposed above the inclined transport devices; the control method of the stacked parts separation device includes: The camera assembly is controlled to continuously acquire image data of multiple packages on the tilting transport device; The three-dimensional position, size, and contour features of each package in each package image data are determined, and package features corresponding to each package are generated; Calculate the spatial overlap and contact area between the features of each package to determine the degree of adhesion and stacking relationship between adjacent packages; The optimal separation order and corresponding movement path for each package are determined according to the degree of adhesion and the layering relationship. The operating speed of each of the tilting transport devices is adjusted according to the optimal separation sequence so that each package is sequentially separated into individual packages along the movement path. The step of determining the three-dimensional position, size, and contour features of each package in each package image data, and generating package features corresponding to each package, includes: Each package image data is preprocessed sequentially according to time. Based on the image segmentation algorithm, the package edges of each package image data are identified, and the package position and background position in the package image data are determined; Generate a point cloud model for each package corresponding to the package location and the background location, and calculate the three-dimensional coordinates of each package; The size parameters of each package are determined based on the three-dimensional coordinates, and the size parameters include length, width, and height values; Generate contour features corresponding to the edge of the package, the contour features being generated through curvature calculation and shape descriptor; The wrapping feature is generated from the three-dimensional coordinates, size parameters, and contour features; The calculation of the spatial overlap and contact area between the features of each package, and the determination of the adhesion and stacking relationship between adjacent packages, includes: Based on the three-dimensional position of each package, calculate the intersection area and union area of ​​the package bounding boxes, and generate the corresponding spatial overlap value. The point cloud model is used to analyze the contact area on the surface of the package, measure the point density distribution, and determine the actual contact area. Obtain the friction coefficient and surface characteristic parameters from the packaging material database, and establish an adhesive force calculation model; Based on the spatial overlap value and the actual contact area, the adhesion strength level between adjacent packages is evaluated, and the degree of adhesion is determined. Construct a topology diagram of the stacked packages and analyze the vertical support relationships and stacking order of each package; The stacking relationship is generated based on the support relationship and the stacking order; The step of evaluating the adhesion strength level between adjacent packages based on the spatial overlap value and the actual contact area, and determining the degree of adhesion, includes: Obtain the friction coefficient, surface roughness, and elastic modulus parameters for each package; Based on the actual contact area and friction coefficient, calculate the estimated static adhesion force. Based on the surface roughness and elastic modulus parameters, the estimated static adhesion force is corrected to generate the corrected adhesion force value. The corrected adhesion force value is weighted and calculated with the spatial overlap value to evaluate the adhesion strength level between adjacent packages and determine the degree of adhesion.

2. The control method for the stacked component separation device as described in claim 1, characterized in that, The step of identifying the package edges of each package image data based on the image segmentation algorithm, and determining the package position and background position in the package image data, includes: The preprocessed package image data is analyzed based on a deep learning segmentation model, and pixel-level labels for the package region and background region are output. Based on the pixel-level labels, an edge detection algorithm is used to extract the outline of the package edge; The location of the broken edge in the outline of the package is determined, and the broken edge is reconstructed continuously using morphological repair and curve interpolation algorithms to generate complete edge outline data; Determine the package location and background location based on complete edge contour data; The process of generating a point cloud model for each package corresponding to the package location and the background location, and calculating the three-dimensional coordinates of each package, includes: Based on the package location and background location corresponding to each package, the depth image data of the corresponding package area is extracted and three-dimensionally mapped to generate point cloud data containing three-dimensional spatial coordinates. Generate a point cloud model corresponding to the point cloud data; The centroid coordinates of the point cloud model are calculated as the three-dimensional position of the package, and the three-dimensional coordinates of each package are determined.

3. The control method for the stacking separation device as described in claim 2, characterized in that, The generation of contour features corresponding to the wrapping edge includes: Extract the sequence of contour points corresponding to the edge of the package; Calculate the curvature value corresponding to the contour point sequence and identify significant geometric feature points in the contour; The frequency domain characteristic coefficient sequence of the contour is generated by performing frequency domain transformation on the contour point sequence using Fourier descriptors. The frequency domain feature coefficient sequence is subjected to dimensionality reduction processing to construct a low-dimensional contour feature vector corresponding to the dimensionality-reduced frequency domain feature coefficient sequence; The low-dimensional contour feature vector is fused with curvature values ​​and geometric feature point information to generate the contour feature.

4. The control method for the stacked component separation device as described in claim 1, characterized in that, Determining the optimal separation order and corresponding movement path for each package based on the degree of adhesion and the layering relationship includes: The degree of adhesion is determined to separate and sort each package, and the sorting status of each package is determined. Based on the multi-objective optimization algorithm, the stacking relationship, and the sorting situation, a priority sequence for package separation is generated to determine the optimal separation order; The starting point and target pose of each package are determined according to the optimal separation sequence. Generate a motion path corresponding to the starting point of the motion and the target pose.

5. The control method for the stacking separation device as described in claim 1, characterized in that, The step of adjusting the operating speed of each of the tilting transport devices according to the optimal separation sequence, so that each package is sequentially separated into individual packages along the movement path, includes: The optimal separation sequence is used to determine the separation time of each package and the speed parameters of the corresponding tilting transport device, and the target operating speed of each tilting transport device is determined. Obtain the current operating speed of each of the tilting transport devices; The difference between the current operating speed and the target operating speed is determined, and the drive signal of the tilting transport device is adjusted so that each package is separated into individual packages sequentially along the movement path.

6. A control device, characterized in that, The control device includes: a memory, a processor, and a control program for the stacking device stored in the memory and executable on the processor, the control program for the stacking device being configured to implement the control method for the stacking device as described in any one of claims 1 to 5.

7. A stacking separation device, characterized in that, Includes the control device as described in claim 6.

Images