Object stacking method, system and equipment based on visual perception and product

Abstract

The invention discloses an object stacking method, system, equipment and product based on visual perception, and relates to the field of logistics automation, and the method comprises the following steps: generating a second three-dimensional point cloud of a target object under world coordinates; projecting the second three-dimensional point cloud along three coordinate axes to obtain three two-dimensional density thermodynamic diagrams; calculating the global density of each two-dimensional density thermodynamic diagram; setting a circle with a preset radius by taking each two-dimensional point as a circle center, and calculating local density in each circle; if the local density in the circle is smaller than the global density, corresponding two-dimensional points are removed from the two-dimensional density thermodynamic diagram; respectively calculating a maximum two-dimensional connected region in each two-dimensional density thermodynamic diagram, and reserving three-dimensional points of which the projections simultaneously fall in the three maximum two-dimensional connected regions in the second three-dimensional point cloud to form a target three-dimensional point cloud; according to the target placement area of the target object in the stacking area of the target three-dimensional point cloud, the size of the object can be accurately calculated, so that accurate execution of stacking operation is facilitated.

Description

Technical Field

[0001] This application relates to the field of logistics automation, and in particular to a visual perception-based method, system, device, and product for palletizing objects. Background Technology

[0002] Against the backdrop of rapid automation in the logistics industry, intelligent palletizing systems, as a core component of modern warehousing and logistics, face new opportunities and challenges in technological innovation. A complete palletizing process requires the coordinated operation of three key stages: acquiring the three-dimensional pose information of the object at the end of the conveyor belt, accurately measuring the object's dimensions, and finally dynamically calculating the optimal placement position based on the object's dimensions. However, existing palletizing methods typically only implement one or two of these stages. While combining different palletizing methods can form a complete palletizing process, poor coordination or incompatibility leads to low palletizing efficiency.

[0003] More importantly, in the process of measuring object dimensions, it is necessary to construct an effective 3D point cloud of the object based on its RGB-D image. However, in addition to the target object, the image also includes background objects such as conveyor belts. Therefore, the preliminary 3D point cloud extracted from the RGB-D image contains a large number of noise points that need to be removed. The traditional approach is to set the height of the conveyor belt as a threshold, and then directly remove 3D points with a height less than the threshold from the preliminary 3D point cloud. The remaining noise points are then removed using the DBSCAN algorithm. This 3D point cloud construction method has at least two drawbacks: 1. Due to vibration and wear, conveyor belts exhibit non-uniform height deviations, meaning their surfaces are not perfectly horizontal. Furthermore, objects are tightly placed on the conveyor belt surface. Therefore, simply using a height threshold to remove noise points belonging to the conveyor belt surface not only results in low accuracy but may also remove valid 3D points belonging to the object, leading to lower accuracy in the 3D point cloud construction and affecting subsequent object size calculations.

[0004] II. Although the DBSCAN algorithm can identify noise, it is not suitable for... ε The DBSCAN algorithm is sensitive to the MinPts parameter; improper parameters can lead to missing noise points or misclassifying valid edge points as noise points. Furthermore, it requires setting different filtering thresholds for objects of different materials, resulting in low scene adaptability. Additionally, the DBSCAN algorithm has a time complexity of O(n^2). 2 The complexity cannot meet the real-time requirements of object detection in the palletizing process.

[0005] In summary, existing palletizing methods suffer from low accuracy in constructing effective 3D point clouds of objects and cannot meet real-time requirements. Summary of the Invention

[0006] The purpose of this application is to provide a visual perception-based object palletizing method, system, device, and product, which is a full-process object palletizing method that can accurately calculate the size of objects to facilitate the precise execution of palletizing operations.

[0007] To achieve the above objectives, this application provides the following solution: In a first aspect, this application provides a visual perception-based object palletizing method, comprising: Acquire an RGB-D image of the target object at the end of the conveyor belt, wherein the RGB-D image is obtained by taking a top-down view of the target object using a depth camera; Generate a first three-dimensional point cloud of the target object in the coordinate system of the depth camera based on the RGB-D image and transform it into a second three-dimensional point cloud in world coordinates; Three two-dimensional density heat maps are obtained by projecting the second three-dimensional point cloud along the three coordinate axes respectively. For any given two-dimensional density heatmap: calculate the global density of the two-dimensional density heatmap; set circles with a preset radius centered on each two-dimensional point in the two-dimensional density heatmap, the preset radius being 2-3 times the average point spacing of the two-dimensional density heatmap; calculate the local density within each circle; for any given two-dimensional point, if the local density within its corresponding circle is less than the global density, then remove the two-dimensional point from the two-dimensional density heatmap. Calculate the largest two-dimensional connected region in each of the two-dimensional density heatmaps, and retain the three-dimensional points in the second three-dimensional point cloud whose projections simultaneously fall into the three largest two-dimensional connected regions to form the target three-dimensional point cloud; The size of the target object is determined based on the target 3D point cloud, and the target placement area of ​​the target object in the palletizing area is determined based on the size of the target object.

[0008] Secondly, this application provides a visual perception-based object palletizing system, comprising: The image acquisition module is used to: acquire an RGB-D image of the target object at the end of the conveyor belt, wherein the RGB-D image is obtained by taking a top-down shot of the target object using a depth camera; The point cloud generation module is used to: generate a first three-dimensional point cloud of the target object in the coordinate system of the depth camera based on the RGB-D image and convert it into a second three-dimensional point cloud in world coordinates; project the second three-dimensional point cloud along the three coordinate axes to obtain three two-dimensional density heatmaps; for any two-dimensional density heatmap: calculate the global density of the two-dimensional density heatmap; set circles with a preset radius centered on each two-dimensional point in the two-dimensional density heatmap, the preset radius being 2-3 times the average point spacing of the two-dimensional density heatmap; calculate the local density within each circle; for any two-dimensional point, if the local density within its corresponding circle is less than the global density, then remove the two-dimensional point from the two-dimensional density heatmap; calculate the maximum two-dimensional connected region in each two-dimensional density heatmap, and retain the three-dimensional points in the second three-dimensional point cloud whose projections simultaneously fall within the three maximum two-dimensional connected regions to form the target three-dimensional point cloud; The feature recognition module is used to: determine the size of the target object based on the target 3D point cloud; The area determination module is used to: determine the target placement area of ​​the target object in the palletizing area based on the size of the target object.

[0009] Thirdly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the visual perception-based object palletizing method described in any one of the above.

[0010] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the visual perception-based object palletizing method described above.

[0011] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the visual perception-based object palletizing method described above.

[0012] According to the specific embodiments provided in this application, the following technical effects are disclosed: This application provides a visual perception-based object palletizing method, system, device, and product. The object palletizing method provides a complete processing step, including acquiring the 3D pose information of the object at the end of the conveyor belt, accurately measuring the object size, and finally dynamically calculating the optimal placement position based on the object size. Specifically, steps S130 and S140 effectively extract the 3D point cloud of the target object (express delivery) while removing noise points from the conveyor belt surface. The mechanism lies in utilizing the fundamental difference between express delivery as a 3D entity and the conveyor belt as a near-2D plane. The express delivery point cloud can form complete connected regions in all three projection planes, while noise points on the conveyor belt surface only form connected regions in the xy plane. In the yz and zx planes, they are compressed into thin, ineffectively connected linear regions. Through the filtering rule of the intersection of the maximum connected regions in the three planes, even if the conveyor belt has height deviations or is not perfectly horizontal, background points adhering to the package can be thoroughly filtered out, solving the problem of height deviation backgrounds that height threshold filtering cannot handle. Furthermore, this method employs an adaptive filtering rule combining global density and local density comparison. Without requiring manual parameter setting, it retains valid edge points of the target object, avoiding the traditional method's mistaken identification of valid edge points as noise. This ensures the complete extraction of point clouds from various express delivery packages such as cardboard boxes, plastics, and fabrics. Simultaneously, steps S130 and S140 both have a linear time complexity of O(n): first, density filtering quickly filters out most background noise, then connectivity analysis is performed on the clean point set. Compared to the traditional DBSCAN's O(n) time complexity... 2 The complexity and computational efficiency are improved by 1-2 orders of magnitude, which can meet the needs of real-time industrial measurement. Attached Figure Description

[0013] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 This is a flowchart illustrating a visual perception-based object palletizing method in one embodiment of this application. Figure 2 This is an example diagram of the feature recognition process of a target object in one embodiment of this application; Figure 3 This is an example diagram illustrating the determination of the center of gravity stability of a target object in one embodiment of this application; Figure 4 This is an example diagram showing the selection of the target placement area in one embodiment of this application; Figure 5 This is an example diagram of an object palletizing process in one embodiment of this application. Detailed Implementation

[0015] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0016] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0017] In one embodiment of this application, a visual perception-based object palletizing method is provided, which can be executed by a computer device and specifically includes steps S110 to S150.

[0018] Step S110: Obtain the RGB-D image of the target object at the end of the conveyor belt. The RGB-D image is obtained by taking a top-down shot of the target object using a depth camera.

[0019] Step S120: Generate a first three-dimensional point cloud of the target object in the coordinate system of the depth camera based on the RGB-D image and convert it into a second three-dimensional point cloud in world coordinates.

[0020] In this step, in order to improve the construction efficiency of the target 3D point cloud and reduce the interference of other objects in the RGB-D image, the target object can be identified and located first using a target detection algorithm. Then, the detection box output by the target detection algorithm is cropped to obtain the region of interest containing the target object. Finally, the target 3D point cloud is generated based on the region of interest.

[0021] For example, an Intel RealSense D435i binocular camera can be used to acquire RGB-D images of the target object, and then target detection algorithms such as R-CNN series, YOLO series, and DETR can be used to identify and locate the target object.

[0022] In this embodiment, the first three-dimensional point cloud of the target object in the coordinate system of the depth camera is generated based on the RGB-D image and then transformed into a second three-dimensional point cloud in world coordinates, specifically including steps S121, S122 and S123.

[0023] Step S121 involves using an object detection algorithm to detect the RGB-D image and generate detection boxes for the target object. Based on these detection boxes, the RGB-D image is cropped to obtain a region of interest containing the target object. This step reduces the amount of point cloud data processed during size recognition and shortens the point cloud processing time.

[0024] Step S122: Generate the three-dimensional coordinates of each pixel in the region of interest based on the depth information of the RGB-D image and the intrinsic parameter matrix of the depth camera, and generate the color of each pixel in the region of interest based on the color information of the RGB-D image. Combine the three-dimensional coordinates and colors of each pixel to form the first three-dimensional point cloud in the coordinate system of the depth camera.

[0025] Specifically, after cropping the RGB-D image, each pixel in the region of interest is traversed. For each pixel, its depth value is read, and using the intrinsic parameter matrix of the depth camera, the pixel's two-dimensional coordinates are calculated. u , v Back projection into three-dimensional coordinates ( X , Y , Z This process involves reading the color of the corresponding pixel from the RGB-D image and assigning it to each 3D point. All 3D points and color information are then combined to form the first 3D point cloud in the coordinate system of the depth camera. The formula for calculating the coordinates of a 3D point is:

[0026] in, c x , c y , f x and f y These are all intrinsic parameters of the depth camera. d This represents the depth value of a pixel.

[0027] Step S123: Convert the first 3D point cloud from the coordinate system of the depth camera to the second 3D point cloud in the world coordinate system.

[0028] The world coordinate system refers to the coordinate system used to determine the pose of the target object in the actual scene. It is also the working coordinate system of the robot arm, facilitating subsequent robot operation and global planning. Therefore, the first 3D point cloud needs to be transformed from the depth camera's coordinate system to the world coordinate system. The transformation principle is homogeneous coordinate transformation. Specifically, a 4×4 transformation matrix is ​​used to transform the first 3D point cloud from the depth camera coordinate system to the world coordinate system. The points in the depth camera coordinate system are defined as follows: The corresponding point in the world coordinate system is: The two satisfy the following relationship:

[0029] in, T Let be the transformation matrix. R It is a 3×3 rotation matrix. t It is a 3×1 translation vector.

[0030] Considering that there are still some background points in the second 3D point cloud, i.e., 3D points that do not belong to the target object, these background points will be filtered out in step S130 in this embodiment to improve the accuracy of subsequent processing.

[0031] Step S130: Project the second three-dimensional point cloud along the three coordinate axes to obtain three two-dimensional density heatmaps; for any two-dimensional density heatmap: calculate the global density of the two-dimensional density heatmap; set a circle with a preset radius centered on each two-dimensional point in the two-dimensional density heatmap, the preset radius being 2-3 times the average point spacing of the two-dimensional density heatmap; calculate the local density within each circle; for any two-dimensional point, if the local density within its corresponding circle is less than the global density, then remove the two-dimensional point from the two-dimensional density heatmap.

[0032] Specifically, the maximum two-dimensional connected region is calculated for each two-dimensional density heatmap. This includes: for any two-dimensional density heatmap, determining that two two-dimensional points whose Euclidean distance is less than the connectivity determination distance are connected; determining each connected region in the two-dimensional density heatmap based on each pair of connected two-dimensional points; and taking the connected region containing the most two-dimensional points as the maximum two-dimensional connected region. The connectivity determination distance is the average point spacing of the two-dimensional density heatmap.

[0033] Specifically, projecting the second 3D point cloud along the three coordinate axes involves projecting the second 3D point cloud onto three orthogonal planes: xy, xz, and zy. It should be noted that these three orthogonal planes are coordinate planes in the world coordinate system.

[0034] For example, the process of eliminating two-dimensional points is explained in detail below using projection onto the xy plane as an example: Projection yields a two-dimensional point set (the set of two-dimensional points in a two-dimensional density heatmap). For each two-dimensional point, calculate the local density inside the circle centered at that point:

[0035] in, Let be the local density inside the i-th circle. k i Let be the total number of two-dimensional points inside the i-th circle. d For the preset radius, S d To use a preset radius d The area of ​​the circle, where the i-th circle is the circle centered at the i-th two-dimensional point.

[0036] The formula for calculating the global density in a two-dimensional density heatmap is:

[0037]

[0038] in, The global density of the two-dimensional density heatmap. n This represents the total number of two-dimensional points in the two-dimensional density heatmap, and also the total number of three-dimensional points in the second three-dimensional point cloud. S The area of ​​the two-dimensional density heatmap. x max and x min These represent the maximum and minimum projection values ​​of the second 3D point cloud onto the x-axis, respectively. y max and y min These represent the maximum and minimum projection values ​​of the second 3D point cloud onto the y-axis, respectively.

[0039] When the local density of a circle is less than the global density, the center point of that circle is removed. Essentially, the above calculates the neighborhood density of each two-dimensional point.

[0040] The above describes the process of removing two-dimensional points from a two-dimensional density heatmap on the xy plane. The process for removing two-dimensional points on the other two planes is the same.

[0041] By following the steps above, three two-dimensional density heatmaps after removing noise points can be obtained. Then, the maximum two-dimensional connected region in each two-dimensional density heatmap is calculated.

[0042] For each two-dimensional density heatmap, the first step is to calculate and determine the connectivity between any two two-dimensional points, thereby identifying each connected region. Specifically, a Breadth-First Search (BFS) algorithm can be used to traverse all points in the two-dimensional density heatmap to find all connected two-dimensional points, which then constitute a connected region. For example, if two-dimensional point A is connected to two-dimensional point B, and two-dimensional point B is connected to two-dimensional points C and D, and two-dimensional point D is connected to two-dimensional point E, then two-dimensional points A / B / C / D / E constitute a connected region.

[0043] It should be noted that step S130 above involves the calculation of the average point spacing of the two-dimensional density heat map. The calculation method is as follows: first, calculate the distance between each two-dimensional point and its nearest two-dimensional point, and then average the distances corresponding to each two-dimensional point to obtain the average point spacing.

[0044] Step S140: Calculate the largest two-dimensional connected region in each two-dimensional density heatmap, and retain the three-dimensional points in the second three-dimensional point cloud whose projections fall on the three largest two-dimensional connected regions to form the target three-dimensional point cloud.

[0045] Through steps S130 and S140 (multi-view density connectivity analysis), the 3D point cloud of the target object (express delivery) can be effectively extracted, while noise points on the conveyor belt surface are removed. The mechanism lies in the fundamental difference between the express delivery being a 3D entity and the conveyor belt being a quasi-2D plane. The express delivery point cloud can form complete connected regions in all three projection planes, while the noise points on the conveyor belt surface only form connected regions in the xy plane. In the yz and zx planes, they are compressed into thin, ineffectively connected linear regions. Specifically, in overhead shots, depth cameras rarely capture the bottom of the target object, meaning the 3D points at the bottom of the target object generally do not exist. Therefore, although the noise points on the conveyor belt surface and the projection of the target object in the xy plane can form connected regions, their projections in the yz and zx planes are essentially discontinuous; that is, the projection of the conveyor belt surface is a relatively isolated straight line. By employing a filtering rule based on the intersection of the maximum connected regions in three planes, background points adhered to packages (adhered in the vertical projection) can be thoroughly filtered out even if the conveyor belt has height deviations or is not perfectly horizontal, thus solving the problem that height threshold filtering cannot handle backgrounds with height deviations. Addressing the uneven point cloud density caused by the diverse materials used in express delivery, this method uses an adaptive filtering rule combining global density and local density comparison. This retains valid edge points of the target object without requiring manual parameter settings. Edge points are categorized as valid and invalid. The density of valid edge points is slightly lower than, but close to, the point density of the main body of the target object. Invalid edge points are distorted and spread outwards; introducing invalid edge points increases the point cloud size, leading to an increase in the calculated object size. The density of invalid edge points and noise points is significantly lower than that of valid edge points and the main body of the target object. When calculating the global density, the presence of invalid and noise points results in a global density between that of invalid and valid edge points, effectively eliminating them. Since global density is used as the filtering threshold, the above filtering criteria are applicable to target objects of any material. In this algorithm, the global density of target objects made of different materials varies, and all fall between the effective and ineffective edge point densities of the target object. Traditional DBSCAN algorithms require manual setting of a filtering threshold, making the filtering accuracy heavily dependent on this threshold setting. If the threshold is set appropriately (e.g., precisely between the effective and ineffective edge point densities), the same technical effect as in this embodiment can be achieved. However, an inappropriate threshold setting severely impacts the point cloud filtering accuracy. For example, if the filtering threshold is set too low, noise points and ineffective edge points will not be sufficiently removed; if the threshold is set too high, effective edge points will be removed, and even some 3D points of the main body of the target object will be removed, resulting in the calculated object size being smaller than the actual value. Therefore, the accuracy of traditional DBSCAN algorithms is highly dependent on the setting of the filtering threshold.More importantly, the point density of the main body and the effective point density of the edges of target objects differ depending on the material. For example, the point density of foam wrapping is significantly lower than that of a wooden box. Therefore, a fixed filtering threshold cannot be applied to target objects of different materials. For instance, a higher filtering threshold can effectively remove noise points and invalid edge points from wooden boxes, but it may also remove valid edge points and even 3D points of the main body of foam wrapping. Conversely, a lower filtering threshold can effectively remove invalid edge points and noise points from foam wrapping, but it may not be able to remove invalid edge points from wooden boxes (a higher overall point density means a higher density of invalid edge points). In palletizing scenarios, the target objects on the conveyor belt are not of uniform material; one may be foam wrapping, and the next may be a wooden box. As a real-time recognition scenario, it is difficult to manually adjust the filtering threshold continuously based on the current target object, and this would also increase labor costs and the error rate.

[0046] Meanwhile, steps S130 and S140 both have a linear time complexity of O(n): first, density filtering quickly filters out most of the background noise, and then connectivity analysis is performed on the clean point set, which is much faster than the traditional DBSCAN's O(n) time complexity. 2 The complexity and computational efficiency are improved by 1-2 orders of magnitude, which can meet the needs of real-time industrial measurement.

[0047] By following the steps above, the 3D point cloud of the target object can be obtained relatively accurately (target 3D point cloud).

[0048] Step S150: Determine the size of the target object based on the target 3D point cloud, and determine the target placement area of ​​the target object in the palletizing area based on the size of the target object.

[0049] Specifically, determining the size of the target object based on the target 3D point cloud includes: The optimal bounding box of the target 3D point cloud is determined. The length and width of the optimal bounding box are used as the length and width of the target object, respectively. The difference between the maximum depth value of the target 3D point cloud and the height value of the conveyor belt is used as the height of the target object. The gripping point of the target object is determined based on the center point of the upper surface of the optimal bounding box. The optimal bounding box is the bounding box with the smallest volume of the target 3D point cloud. The target object is gripped by the robot arm and placed in the palletizing area through the gripping point.

[0050] In this step, a grid search algorithm is used to determine the various bounding boxes of the target 3D point cloud. The bounding box with the smallest volume is selected as the optimal bounding box. The principle is that the smallest bounding box should be the one that best fits the target 3D point cloud, and its size and shape are closest to the target 3D point cloud. Since the RGB-D image is a top-down image, the length and width of the target 3D point cloud are the same as the length and width of the target object. Therefore, the length and width of the optimal bounding box can be used as the length and width of the target object, respectively. However, in the height direction, the depth camera cannot capture the bottom and interior of the target object. The target 3D point cloud only represents the point cloud of the visible part of the target object from a top-down perspective (basically the top point cloud of the target object). Therefore, there is a difference between the height of the target 3D point cloud and the height of the target object. Considering that the target object is placed on a conveyor belt, the difference between the minimum depth value of the target 3D point cloud (corresponding to the highest point in the target object) and the depth value of the conveyor belt is used as the height of the target object.

[0051] In this embodiment, determining the optimal bounding box of the target 3D point cloud specifically includes steps S151, S152, and S153.

[0052] Step S151: Randomly select three 3D points multiple times from the target 3D point cloud to form multiple candidate planes; for each candidate plane, calculate the comprehensive score of all 3D points in the target 3D point cloud relative to the candidate plane:

[0053] in, S i Let i be the overall score of the i-th 3D point relative to the candidate plane. , and There are three weights, and the sum of the three weights is 1. d i and Let be the Euclidean distance from the i-th 3D point to the candidate plane and its standard deviation, respectively. and Let be the normal vectors of the candidate plane and the i-th 3D point, respectively. Let the density of the neighborhood points of the i-th 3D point be . is the average point density of the candidate plane.

[0054] Preferred, , and The values ​​are 0.5, 0.4, and 0.1, respectively. For example, for a three-dimensional point... to plane distance distance for:

[0055] Step S152: Determine the three-dimensional points whose comprehensive score relative to the candidate plane is greater than the score threshold as interior points of the candidate plane, and determine the candidate plane with the most interior points as the target plane.

[0056] Steps S151 and S152 are to determine the optimal plane representing the main surface of the target object, and then calculate the covariance matrix of the target 3D point cloud for all points on that plane.

[0057] It should be noted that the traditional RANSAC (RANdom Sampling Consensus) plane fitting method relies solely on the distance from a point to the plane as the criterion for determining interior points. This results in low discrimination between noise points and edge points, and random sampling is prone to sampling noise points, leading to insufficient fitting accuracy. This embodiment proposes a robust RANSAC plane fitting method with multi-feature constraints (corresponding to steps S151 and S152) for object point cloud volume measurement scenarios, effectively solving the aforementioned problems.

[0058] In express delivery scenarios, low-density noise points left on conveyor belts and effective sparse points on package edges may be highly similar in geometric features. For example, when a residual point happens to fall near the package edge, the angle between its normal vector and the normal vector of the package plane may meet a threshold, and its distance to the plane may also meet the standard. Existing normal vector + distance schemes rely solely on geometric features and cannot distinguish between these two types of points. This leads to misclassifying such noise points as effective interior points, causing the fitted plane to be skewed by background points, ultimately affecting the accuracy of package size measurement. However, by adding density features, this problem is fundamentally solved: the noise points left on the conveyor belt are essentially background interference, and their local density is much lower than the stable density of the main package points and effective edge points. In the comprehensive score calculation, the density weight of such points will approach 0. Even if the geometric features meet the standard, the comprehensive score will be lower than the threshold, thus being accurately eliminated. This design is not a simple feature superposition, but rather provides a feasible distinguishing criterion based on density to address the core pain point of highly similar geometric features between the adhered background and effective edge points in express delivery scenarios. This completely solves the technical problem that existing solutions cannot distinguish between geometrically similar but different 3D points.

[0059] Step S153: Calculate the covariance matrix of the target 3D point cloud based on all 3D points on the target plane, and determine the optimal bounding box of the target 3D point cloud based on the covariance matrix.

[0060] covariance matrix C The calculation formula is:

[0061] in, p iLet i be the i-th 3D point in the target plane. N The total number of three-dimensional points in the target plane. The center point is located in the target plane.

[0062] Then, the covariance matrix is ​​decomposed into eigenvalues. .in, v j For the first j The directions of the three largest eigenvalues ​​are used as the three axes to construct the local coordinate system of the target 3D point cloud. Essentially, the three axes of the local coordinate system correspond to the length, width, and height directions of the bounding box of the target 3D point cloud.

[0063] Furthermore, the optimal bounding box of the target 3D point cloud is determined based on the covariance matrix, specifically including: performing eigenvalue decomposition on the covariance matrix to obtain a local coordinate system containing three coordinate axes for the target 3D point cloud; rotating the local coordinate system around the Z-axis of the local coordinate system by a preset step size of 360°; and at each rotation angle, determining the bounding box and calculating its volume based on the difference between the maximum and minimum values ​​of the projections of the target 3D point cloud onto the three coordinate axes, where the three differences are the length, width, and height of the bounding box, respectively.

[0064] For example, the preset step size is 1°. In this step, a grid search strategy is used to perform a 360-degree rotation search around the Z-axis with a step size of 1° to find the optimal rotation angle that minimizes the bounding box volume. The search process is implemented as follows: Construct the rotation matrix. The rotation matrix around the Z-axis is defined as follows:

[0065] in, It is a rotation angle around the Z-axis of . The rotation matrix, Let be the rotation angle about the Z-axis. , , and These are the cosine and sine values ​​of the rotation angle, respectively.

[0066] Initialize the local coordinate system ,make , as well as . v 1. v 2 and v 3 represents three coordinate axes, corresponding to the length, width, and height directions of the bounding box, respectively, which are the length axis, width axis, and height axis.

[0067] Grid search optimization: for each rotation angle Perform the calculation of the rotated axis vectors:

[0068] in, This is the local coordinate system after rotation.

[0069] Calculation of the projection of the target 3D point cloud onto the three coordinate axes:

[0070] in, It is the projection of the target's 3D point cloud onto the coordinate system. The coordinate matrix after that, P Let the coordinate matrix be the target 3D point cloud itself. The center point of the target 3D point cloud.

[0071] Through the projection calculations described above, the projected coordinates of the target 3D point cloud on the three coordinate axes can be obtained, thus determining the maximum and minimum projection values ​​on each axis. For each coordinate axis, the difference between the maximum and minimum projection values ​​is the size of the bounding box along that axis. Ultimately, the dimensions of the bounding box in the three directions can be determined, and the product of these three dimensions is the volume of the bounding box.

[0072] By iterating through each rotation angle, the bounding box volume at each rotation angle can be determined, thus identifying the optimal rotation angle that minimizes the bounding box volume. :

[0073] in, Let be the volume of the bounding box under the rotation angle. The above formula indicates that the rotation angle that minimizes the volume of the bounding box is taken as the optimal rotation angle.

[0074] The bounding box at the optimal rotation angle is the optimal bounding box, and the optimal local coordinate system is determined based on the optimal rotation angle. The three axes of the optimal local coordinate system correspond to the length, width, and height of the optimal bounding box. The optimal local coordinate system can be used to determine the orientation of the optimal bounding box and the target object.

[0075] The bounding box with the smallest volume is then determined as the optimal bounding box. The dimensions of the optimal bounding box are calculated as follows:

[0076] in, Let the projection of the target 3D point cloud onto the three coordinate axes of the optimal local coordinate system be defined. Then, the maximum and minimum values ​​of the projection on the three coordinate axes are calculated using the following functions:

[0077]

[0078] in, and Represented on the coordinate axes Minimum and maximum projection values ​​on the surface. . and They represent in Taken on the coordinate axis The minimum and maximum values ​​of the projection on the surface.

[0079] Finally, the length and width of the optimal bounding box are used as the length and width of the target object, respectively:

[0080]

[0081] Furthermore, the difference between the minimum depth value of the target 3D point cloud and the depth value of the conveyor belt is taken as the height of the target object. It should be noted that the height of the target object can be calculated directly from the corresponding depth value from the depth camera, or it can be calculated by subtracting the height value of the conveyor belt from the maximum value projected onto the height axis of the target 3D point cloud.

[0082] in, , and Let be the length, width, and height of the optimal bounding box, respectively. a This represents the height of the conveyor belt. In an example scenario, it is set to 0.2. and These represent the maximum and minimum projection values ​​of the target 3D point cloud onto the length axis, respectively. and These represent the maximum and minimum projection values ​​of the target 3D point cloud onto the width axis, respectively. This represents the maximum projection of the target 3D point cloud onto the height axis.

[0083] Once the optimal rotation angle is determined, the pose of the optimal bounding box is the pose of the target object. The longest axis within this bounding box is then calculated. v Angle 1 represents the orientation angle of the target object. :

[0084] in, and They are respectively The element in row 1, column 0 and the element in row 0, column 0. This represents the arctangent function in the four quadrants. It should be noted that... The row and column indices in the array start from 0.

[0085] Typically, the gripping position of a robotic arm is the center point of the upper surface of the target object. Since the center point of the upper surface of the optimal bounding box corresponds to the center point of the upper surface of the target object, the gripping point of the target object is determined based on the center point of the upper surface of the optimal bounding box. Specifically: the position of the center point of the upper surface of the optimal bounding box in the world coordinate system is determined as the gripping point of the target object.

[0086] First, we need to calculate the coordinates of the center point of the upper surface of the optimal bounding box in the optimal local coordinate system. c top :

[0087] Then, the coordinates of the center point of the upper surface of the optimal bounding box are transformed from the optimal local coordinate system to the world coordinate system.

[0088]

[0089] in, p grasp This indicates the coordinates of the center point in the world coordinate system.

[0090] Through the above steps, based on the optimal bounding box of the target object, information such as the target object's size, gripping point, and orientation can be determined. During the subsequent gripping process, the robotic arm grasps the target object through the gripping point and can adjust the target object's orientation according to its placement position.

[0091] Steps S120 and S150 above constitute the feature recognition process for the target object. (Refer to...) Figure 2 , Figure 2 The document illustrates an example of the feature recognition process for a target object. First, the RGB image and depth image (RGB-D image) of the package (target object) are cropped based on the detection bounding box, and a point cloud is extracted. Then, multi-view density connectivity analysis is performed on the point cloud to remove noise from the point cloud (a set of background points in the point cloud). Next, a grid search algorithm based on directed bounding boxes is used to find the bounding box with the smallest volume. Finally, the bounding box features are analyzed to obtain the size, pose, and grasping point of the package.

[0092] In the step of determining the target placement area of ​​the target object in the palletizing area based on the size of the target object: firstly, based on the stability judgment criteria, candidate placement areas that meet the stability placement requirements are identified, and then the target placement area is determined from the candidate placement areas. Specifically, this includes steps S154 and S155.

[0093] Step S154: In the first pose state and the second pose state of the target object, based on the height map of the stacking area, determine the contact surface between the current bottom surface and the stacking area when the target object is placed in different sub-regions of the stacking area. The sub-regions that make the center of gravity of the target object fall on the contact surface and the contact surface is greater than half of the current bottom surface are selected as candidate placement areas. The sub-regions have the same size as the current bottom surface. The first pose state and the second pose state are two pose states formed by the interchange of the length and width of the bottom surface of the target object.

[0094] For the height map of the palletizing area, the RGB-D image of the palletizing area can be cropped to obtain the region of interest (ROI), and each pixel within the ROI can be traversed. For each pixel coordinate (x, y), its depth value is read, and the 2D pixel coordinates are back-projected into 3D space according to the depth camera intrinsic parameter matrix to obtain the 3D point coordinates (X, Y, Z). Simultaneously, the color information of the corresponding pixel is obtained from the RGB-D image and assigned to the 3D point. All 3D points and their color information are integrated to generate a complete point cloud of the palletizing area. The point cloud is voxelized and discretized into a length × width × height 3D grid, and the height value of the point cloud within each grid cell is calculated. Then, the 3D grid is projected along the height direction to generate a 2D height table, i.e., the height map of the palletizing area. Each position in the 2D height table stores the height value of its corresponding grid cell, and the height value of each grid cell is the height value of the corresponding position in the palletizing area.

[0095] After determining the dimensions of the target object and the height of the stacking area, the placement position of the target object is calculated. The selection of the placement position must ensure both the feasibility of placing the object and the space utilization rate. Therefore, this embodiment uses a joint determination method based on the object's center of gravity and the contact area of ​​its bottom surface to ensure the stability of the object.

[0096] Object stability assessment is primarily used in automated stacking and handling scenarios to ensure that placed objects remain balanced at their target location and do not tip over due to insufficient support or a shift in the center of gravity. The core principle is that the object's base must have sufficient supporting area, and its center of gravity must fall within this supporting area. When selecting an object's placement location, the actual contact area between the object's base and its supporting surface is first calculated.

[0097] Determine the contact surface between the current bottom surface and the stacking area when the target object is placed in different sub-areas of the stacking area. Specifically, this includes: first, determining a sub-area in the stacking area that is the same as the bottom surface of the target object; then, determining the height of each position point in the sub-area based on the height map of the stacking area; determining the highest position point in the sub-area; and all the highest position points constitute the support area, which is the contact surface between the stacking area and the current bottom surface.

[0098] After obtaining the contact surface, determine whether the geometric center of the target object's bottom surface falls within the convex hull or boundary of the supporting area. If so, the target object's center of gravity is considered to fall on the contact surface. If not, the target object's center of gravity is considered not to fall on the contact surface. In this case, the object may tip over and cannot be placed. Figure 3 Examples are provided regarding the contact surface and whether the center of gravity of the target object falls within the contact surface.

[0099] For example, assume the coordinates of the geometric center of the bottom surface of the target object are... The coordinates of each point on the contact surface are The determination condition is:

[0100]

[0101] in, x min and x max respectively x i For the minimum and maximum values ​​in, y min and y max respectively y i Let be the minimum and maximum values ​​in .

[0102] If the area of ​​the contact surface is less than half of the current bottom surface, the support is considered insufficient and placement is not allowed. Only sub-regions that simultaneously meet all the above conditions are considered stable and feasible placement areas.

[0103] Through the above steps, we can obtain all the feasible placement areas corresponding to the target object in different pose states (namely, the two pose states formed by the interchange of the length and width of the bottom surface, which is achieved by rotating the target object around the height axis by 90 degrees). These feasible placement areas constitute the set of candidate placement areas.

[0104] Step S155: Select the candidate placement area with the lowest height as the first candidate placement area, and determine the target placement area within the first candidate placement area.

[0105] In this step, to ensure placement stability, the target object is preferentially placed at a lower position. Therefore, the candidate placement area with the lowest height is selected as the first candidate placement area. If there is only one first candidate placement area, that area is selected as the target placement area. If multiple first candidate placement areas exist, one of them is selected as the target placement area. Furthermore, in the horizontal direction, the priority of the X-axis and Y-axis in the world coordinate system can be adjusted according to the actual scene. For example, when the X-axis priority is higher than the Y-axis, the area with the smallest X-coordinate among the first candidate placement areas is selected as the second candidate placement area. If there is only one second candidate placement area, that area is selected as the target placement area. If multiple second candidate placement areas exist, the second candidate placement area with the smallest Y-coordinate is selected as the target placement area.

[0106] Reference Figure 4 It provides a selection of candidate placement areas ( Figure 4 Select the target placement region from the candidate position (which represents a vertex of the candidate placement region). Figure 4 An example of a target location (where the target is placed at a vertex).

[0107] It should be noted that after the target placement area is determined, the robotic arm needs to adjust the posture of the target object so that the corresponding surface is used as the bottom surface to place the target object on the target placement area. The bottom surface of the target object needs to be perfectly matched with the target placement area.

[0108] The above fully illustrates a specific embodiment of a vision-based object palletizing method. It achieves closed-loop control throughout the entire process, from object recognition, size estimation, angle estimation, and grasp point estimation to stability determination and position planning, solving the problem of fragmented processes in traditional palletizing systems. Specifically, the steps for object size recognition are given: projecting the second 3D point cloud along the three coordinate axes to obtain three 2D density heatmaps; for any 2D density heatmap: calculating the global density of the 2D density heatmap; setting circles with a preset radius centered on each 2D point in the 2D density heatmap, the preset radius being 2-3 times the average point spacing of the 2D density heatmap; calculating the local density within each circle; for any 2D point, if the local density within its corresponding circle is less than the global density, then removing the 2D point from the 2D density heatmap; calculating the maximum 2D connected region in each 2D density heatmap, retaining 3D points in the second 3D point cloud whose projections simultaneously fall within the three maximum 2D connected regions to form the target 3D point cloud, which can efficiently and accurately determine the object size. The algorithm provides specific steps for determining stability using a dual-criteria approach (the contact area must be greater than half the bottom surface of the object and the object's center of gravity must be within the contact surface), significantly improving palletizing stability. The heightmap-based heuristic placement strategy (prioritizing the placement areas with the smallest height, smallest X-coordinate, and smallest Y-coordinate in that order) exhibits strong generalization ability, handling objects of different sizes, weights, and packaging materials, ensuring a space utilization rate of over 73%, which is superior to conventional heuristic algorithms.

[0109] Based on this object palletizing method, this application provides an exemplary object palletizing process. (Refer to...) Figure 5 First, the system detects a sensor at the end of the conveyor belt, using the signal from this sensor to determine if there is an object at the end of the conveyor belt. If so, a detection model (object detection algorithm) identifies the object category and draws a detection bounding box. Different types of objects are palletized using different types of pallets. Then, a bounding box optimization algorithm based on grid search is used to find the minimum bounding box (optimal bounding box). The bounding box features are analyzed to determine the package's posture, size, and gripping position (grip point). Second, the system determines the height map of the palletizing area based on the point cloud of the palletizing area. Combining the height map of the palletizing area and the package size, the system traverses the height map of the palletizing area, determining the center of gravity and contact area for each position (sub-region) to determine if the package is stable at that position. If stable, the position is added to the candidate position set. A heuristic placement strategy based on the height map is used to obtain the target position and place the package. Finally, it checks if there are any more objects to be placed. If so, the above process is repeated until the end.

[0110] In one exemplary embodiment, a vision-based object palletizing system is also provided, including: an image acquisition module, a point cloud generation module, a feature recognition module, and a region determination module.

[0111] The image acquisition module is used to: acquire the RGB-D image of the target object at the end of the conveyor belt. The RGB-D image is obtained by taking a top-down shot of the target object using a depth camera. The point cloud generation module is used to: generate a first 3D point cloud of the target object in the coordinate system of the depth camera based on the RGB-D image and convert it into a second 3D point cloud in world coordinates; project the second 3D point cloud along the three coordinate axes to obtain three 2D density heatmaps; for any 2D density heatmap: calculate the global density of the 2D density heatmap; set circles with a preset radius centered on each 2D point in the 2D density heatmap, the preset radius being 2-3 times the average point spacing of the 2D density heatmap; calculate the local density within each circle; for any 2D point, if the local density within its corresponding circle is less than the global density, then remove the 2D point from the 2D density heatmap; calculate the maximum 2D connected region in each 2D density heatmap, and retain the 3D points in the second 3D point cloud whose projections simultaneously fall within the three maximum 2D connected regions to form the target 3D point cloud. The feature recognition module is used to: determine the size of the target object based on the target 3D point cloud; The area determination module is used to determine the target placement area of ​​the target object in the palletizing area based on the size of the target object.

[0112] It should be noted that each module in the vision-based object palletizing system is used to implement each step of the vision-based object palletizing method in the above embodiments. Therefore, for a detailed description of the object palletizing system, please refer to the above object palletizing method.

[0113] In one exemplary embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0114] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0115] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0116] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0117] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0118] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0119] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0120] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A visual perception-based object palletizing method, characterized in that, include: Acquire an RGB-D image of the target object at the end of the conveyor belt, wherein the RGB-D image is obtained by taking a top-down view of the target object using a depth camera; Generate a first three-dimensional point cloud of the target object in the coordinate system of the depth camera based on the RGB-D image and transform it into a second three-dimensional point cloud in world coordinates; Three two-dimensional density heatmaps are obtained by projecting the second three-dimensional point cloud along the three coordinate axes respectively. For any two-dimensional density heatmap: the global density of the two-dimensional density heatmap is calculated; circles with a preset radius centered on each two-dimensional point in the two-dimensional density heatmap are set, and the preset radius is 2-3 times the average point spacing of the two-dimensional density heatmap; the local density within each circle is calculated; for any two-dimensional point, if the local density within its corresponding circle is less than the global density, the two-dimensional point is removed from the two-dimensional density heatmap. Calculate the largest two-dimensional connected region in each of the two-dimensional density heatmaps, and retain the three-dimensional points in the second three-dimensional point cloud whose projections simultaneously fall into the three largest two-dimensional connected regions to form the target three-dimensional point cloud; The size of the target object is determined based on the target 3D point cloud, and the target placement area of ​​the target object in the palletizing area is determined based on the size of the target object.

2. The object palletizing method based on visual perception according to claim 1, characterized in that, Calculate the maximum two-dimensional connected region for each of the two-dimensional density heatmaps, specifically including: For any given two-dimensional density heatmap: two two-dimensional points whose Euclidean distance is less than the connectivity determination distance are considered connected; based on each pair of connected two-dimensional points, each connected region in the two-dimensional density heatmap is determined; and the connected region containing the most two-dimensional points is taken as the maximum two-dimensional connected region; wherein, the connectivity determination distance is the average point spacing of the two-dimensional density heatmap.

3. The object palletizing method based on visual perception according to claim 1, characterized in that, Determining the size of the target object based on the target 3D point cloud specifically includes: The optimal bounding box of the target 3D point cloud is determined, and the length and width of the optimal bounding box are used as the length and width of the target object, respectively. The difference between the minimum depth value of the target 3D point cloud and the depth value of the conveyor belt is used as the height of the target object. The gripping point of the target object is determined based on the center point of the upper surface of the optimal bounding box. The optimal bounding box is the bounding box with the smallest volume of the target 3D point cloud. The target object is gripped by the robot arm and placed in the palletizing area through the gripping point.

4. The object palletizing method based on visual perception according to claim 3, characterized in that, Determining the optimal bounding box of the target 3D point cloud specifically includes: Multiple candidate planes are formed by randomly selecting three 3D points from the target 3D point cloud; for each candidate plane, a comprehensive score of all 3D points in the target 3D point cloud relative to the candidate plane is calculated: ； in, S i The comprehensive score of the i-th 3D point relative to the candidate plane is given. , and There are three weights, and the sum of the three weights is 1. d i and Let be the Euclidean distance from the i-th 3D point to the candidate plane and its standard deviation, respectively. and Let be the normal vectors of the candidate plane and the i-th 3D point, respectively. Let the density of the neighborhood points of the i-th 3D point be . The average point density of the candidate plane; Three-dimensional points whose comprehensive score relative to the candidate plane is greater than the score threshold are determined as interior points of the candidate plane, and the candidate plane with the most interior points is determined as the target plane; The covariance matrix of the target 3D point cloud is calculated based on all 3D points on the target plane, and the optimal bounding box of the target 3D point cloud is determined based on the covariance matrix.

5. The object palletizing method based on visual perception according to claim 4, characterized in that, Determining the optimal bounding box of the target 3D point cloud based on the covariance matrix specifically includes: The covariance matrix is ​​decomposed into eigenvalues ​​to obtain a local coordinate system containing three coordinate axes for the target 3D point cloud. The local coordinate system is rotated 360° around the Z-axis of the local coordinate system with a preset step size. At each rotation angle, the bounding box is determined and its volume is calculated based on the difference between the maximum and minimum values ​​of the projection of the target 3D point cloud onto the three coordinate axes, respectively. The three differences are the length, width and height of the bounding box, respectively. The bounding box with the smallest volume is determined as the optimal bounding box.

6. The object palletizing method based on visual perception according to claim 5, characterized in that, The preset step size is 1°.

7. The object palletizing method based on visual perception according to claim 1, characterized in that, Determining the target placement area of ​​the target object in the palletizing area based on the size of the target object specifically includes: Based on the height map of the palletizing area, in the first and second pose states of the target object, the contact surface between the bottom surface of the target object and the palletizing area is determined when the target object is placed in different sub-regions of the palletizing area. The sub-regions in which the center of gravity of the target object falls on the contact surface and the contact surface is greater than half of the bottom surface of the target object are selected as candidate placement areas. The sub-regions have the same bottom surface dimensions as the target object, and the first and second pose states are two pose states formed by interchanging the length and width of the bottom surface of the target object. The candidate placement area with the lowest height is selected as the first candidate placement area, and the target placement area is determined within the first candidate placement area.

8. A visual perception-based object palletizing system, characterized in that, include: The image acquisition module is used to: acquire an RGB-D image of the target object at the end of the conveyor belt, wherein the RGB-D image is obtained by taking a top-down shot of the target object using a depth camera; The point cloud generation module is used to: generate a first three-dimensional point cloud of the target object in the coordinate system of the depth camera based on the RGB-D image and convert it into a second three-dimensional point cloud in world coordinates; project the second three-dimensional point cloud along the three coordinate axes to obtain three two-dimensional density heatmaps; for any two-dimensional density heatmap: calculate the global density of the two-dimensional density heatmap; set circles with a preset radius centered on each two-dimensional point in the two-dimensional density heatmap, the preset radius being 2-3 times the average point spacing of the two-dimensional density heatmap; calculate the local density within each circle; for any two-dimensional point, if the local density within its corresponding circle is less than the global density, then remove the two-dimensional point from the two-dimensional density heatmap; calculate the maximum two-dimensional connected region in each two-dimensional density heatmap, and retain the three-dimensional points in the second three-dimensional point cloud whose projections simultaneously fall within the three maximum two-dimensional connected regions to form the target three-dimensional point cloud; The feature recognition module is used to: determine the size of the target object based on the target 3D point cloud; The area determination module is used to: determine the target placement area of ​​the target object in the palletizing area based on the size of the target object.

9. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the visual perception-based object palletizing method according to any one of claims 1-7.

10. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the visual perception-based object palletizing method according to any one of claims 1-7.

Images