Intelligent boxing method and system based on object compression characteristic response

Abstract

The invention discloses an intelligent boxing method and system based on object compression characteristic response, and relates to the technical field of logistics management. The method comprises the steps that object parameters and box type parameters are obtained; a compression response and three-dimensional physical stability coupled unified boxing optimization model is constructed under a mixed integer programming framework, box type selection, rotation selection, position, spatial separation, a support relation, a compression ratio and an actual size variable are set, and spatial, physical and logic constraints with the minimum box type cost as a target are constructed and solved. A boxing scheme containing the target box type, the posture, the position, the supporting relation and the compression rate is obtained and output; the system comprises a data acquisition module, an intelligent optimization module, an execution guidance module and a feedback verification module. By means of the technical scheme, cooperative constraint and global optimization of compression size change and stacking stability of the compressible articles are achieved, the box type cost is reduced, and boxing feasibility and transportation stability are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of logistics management, and particularly relates to an intelligent boxing method based on compression characteristic response of articles and an intelligent boxing system based on compression characteristic response of articles. BACKGROUND

[0002] With the development of e-commerce logistics and intelligent warehousing, three-dimensional boxing technology is widely used to improve the space utilization rate of the box and reduce the packaging cost. The existing boxing scheme usually optimizes the selection of the box type, the posture and placement position of the articles, the non-overlapping constraint, and the like. Some schemes further introduce support relationship or gravity center constraint to improve the stability of the stacking. However, in the actual fulfillment scene, the articles to be loaded usually include rigid articles and compressible articles (such as soft packaging, foam, and fabric), and the size change of the compressible articles after being compressed will directly affect the available space boundary and the stacking contact state, so that the boxing result under the traditional “rigid body assumption” deviates from the actual boxing process.

[0003] The existing technology still has the following problems in processing the compressible articles: firstly, the compression modeling lacks dynamicity, and is usually approximated by using a fixed compression ratio or a pre-compressed manner, which is difficult to depict the passive compression response of the compression rate changing with the upper load, so that the article size used in the boxing calculation is inconsistent with the actual size after being compressed; secondly, the stability evaluation system is incomplete or disconnected from the compression process, and usually only considers simple support or gravity center constraint, and does not uniformly incorporate the size update, support area change, gravity center projection and overturning risk caused by compression into the same constraint framework; thirdly, the compression processing and stability constraint are usually modeled separately, and the optimization process cannot form a consistent coupling decision among the box type cost, space loading, compression response and three-dimensional stability, so that problems such as insufficient space, large gap, compression damage or unstable stacking may occur. SUMMARY

[0004] In view of the above problems, the present application provides an intelligent boxing method and system based on the compression characteristic response of articles, which constructs a unified boxing optimization model coupling the compression characteristic response and the three-dimensional physical stability under the mixed integer programming solving framework, and solves the model with the minimum box type cost as the target, so as to realize the collaborative optimization of the box type selection, the article placement posture, the placement position, the support relationship and the compression rate, reduce the packaging box type cost and improve the scheme executability under the premise of meeting the loading and stability constraints. By establishing the passive compression response constraint between the load and the compression rate determined by the article support relationship and the weight, the actual size change of the compressible articles after being compressed is quantitatively depicted, the space modeling accuracy is improved, the three-dimensional physical stability constraint established based on the actual three-dimensional size is introduced and uniformly acts with the support constraint, the stability of the stacking is systematically controlled, the risk of gravity center deviation, insufficient support and overturning is reduced, and the safety and reliability of the transportation process are improved.

[0005] To achieve the above object, the application provides an intelligent packing method based on compression characteristic response of articles, comprising: obtaining article parameters of each article in a set of articles to be loaded and box type parameters of each box type in a set of optional box types, wherein the article parameters include initial three-dimensional size, weight, compression coefficient and maximum compression rate, and the box type parameters include internal size of the box type and cost of the box type; under a solution framework of mixed integer programming, a unified packing optimization model coupling compression characteristic response and three-dimensional physical stability is constructed, in which box type selection variables, article rotation selection variables, article space position variables, article space separation variables, article support relationship variables, article compression rate variables and article actual three-dimensional size variables are set; wherein the article actual three-dimensional size variables are associated with the article compression rate variables and the article rotation selection variables through size association constraints; a target function and constraint conditions are constructed in the unified packing optimization model, the target function is used to minimize the cost of the box type under the premise of meeting the requirements of space loading and stability; the constraint conditions include space constraints, physical constraints and logical constraints; wherein the space constraints include box boundary constraints based on the article actual three-dimensional size variables and non-overlapping constraints based on the article space separation variables; the physical constraints include support constraints based on the article support relationship variables, passive compression response constraints based on the load determined by the article support relationship variables and the weight of the article, and three-dimensional physical stability constraints established based on the article actual three-dimensional size variables; the logical constraints include box type unique selection constraints, rotation unique selection constraints, size association constraints and symmetry breaking constraints; the unified packing optimization model is solved to obtain a packing scheme, the packing scheme includes a target box type, placement posture, placement position, stacking support relationship and corresponding compression rate of each article; the packing scheme is output for packing execution and packing result display.

[0006] In the above technical solution, preferably, the unified packing optimization model is a mixed integer programming model, and a branch and bound algorithm is used to solve the mixed integer programming model to obtain a packing scheme with minimum box type cost.

[0007] In the above technical solution, preferably, the symmetry breaking constraints include: dividing articles with the same initial three-dimensional size, weight and material category into the same article group, and imposing a sequence constraint on the article space position variables of the articles in the same article group to eliminate equivalent packing solutions.

[0008] In the technical solution, preferably, the size-related constraint comprises: for each allowed rotation direction of each item, constructing a size-related constraint based on the item rotation selection variable, so that when the item rotation selection variable selects the allowed rotation direction, the item actual three-dimensional size variable is equal to the product of the initial three-dimensional size in the allowed rotation direction and the compression factor, and when the allowed rotation direction is not selected, the size-related constraint does not impose a constraint on the item actual three-dimensional size variable, wherein the compression factor is (1-the item compression rate variable).

[0009] In the technical solution, preferably, the passive compression response constraint comprises: determining a set of directly stacked items above each item based on the item support relationship variable, and summing the weights of the items in the set of directly stacked items to obtain the load bearing of the item. The item compression rate variable is determined based on the load bearing and the compression coefficient, and the item compression rate variable is upper limited by the maximum compression rate. The item compression rate variable of the item with a compression coefficient of zero is constrained to be zero.

[0010] In the technical solution, preferably, the non-overlapping constraint comprises: setting six-directional item-to-item space separation variables for any two items, and establishing a position separation inequality between the two items in the corresponding direction when the corresponding directional item-to-item space separation variable is activated, and establishing a separation direction uniqueness constraint to ensure that the two items are separated in at least one of the six directions.

[0011] In the technical solution, preferably, the support constraint comprises: setting a box bottom support identification variable and an item-to-item support relationship variable for each item, so that each item satisfies a uniqueness constraint between its box bottom support identification variable and its item-to-item support relationship variable, to determine whether the item is supported by the box bottom or another item.

[0012] In the technical solution, preferably, the three-dimensional physical stability constraint comprises an overall center of gravity balance constraint, a lateral support area constraint, and a roll moment prevention constraint, the overall center of gravity balance constraint is used to limit the overall center of gravity projection of all items after loading to be located within the box bottom support area, the lateral support area constraint is used to limit the projection overlap area between the bottom surface of the supported item and the support surface to satisfy a preset proportion, and the roll moment prevention constraint is used to limit the center of gravity projection of the supported item to be located within the inwardly recessed safety area of the top surface of the support item.

[0013] The application further provides an intelligent packing system based on the compression characteristic response of an article, which applies the intelligent packing method based on the compression characteristic response of an article as disclosed in any of the technical solutions, and comprises a data acquisition module, an intelligent optimization module, an execution guidance module and a feedback checking module. The data acquisition module is configured to acquire article parameters and box type parameters. The intelligent optimization module is configured to construct a unified packing optimization model under a mixed integer programming solving framework and solve the model to output a packing scheme. The execution guidance module is configured to output the packing scheme for packing execution and packing result display. The feedback checking module is configured to acquire monitoring data related to bearing load, compression deformation and inclination angle during packing execution, and provide the monitoring data to the intelligent optimization module for consistency checking and adjustment.

[0014] In the above technical solution, preferably, the data acquisition module comprises a visual sensor and a weighing sensor, which are configured to acquire the initial three-dimensional size and weight of each article, and comprises a compression parameter determination unit, which is configured to determine the material category of the article and determine the compression coefficient and maximum compression rate of the compressible article. The feedback checking module comprises a force sensor, a displacement sensor and an inclination sensor. The execution guidance module generates motion control parameters for controlling a mechanical arm according to the corresponding compression rate in the packing scheme.

[0015] Compared with the prior art, the application has the following beneficial effects: (1) By using the unified packing optimization model based on the coupling of the compression characteristic response of an article and three-dimensional physical stability, and globally solving the model under a mixed integer programming solving framework with the minimum box type cost as the target, integrated optimization is realized among the decisions of box type selection, article posture and position, and stacking support relationship, so that the box type cost is reduced and the executability of the packing scheme is improved under the premise of meeting the space loading and stability requirements.

[0016] (2) By introducing the bearing load determined by the support relationship and the weight of the article into the model, and establishing the passive compression response constraint between the bearing load and the compression rate, the actual three-dimensional size of the compressible article after being compressed is quantitatively described, the problems of not being able to pack, too large void or distortion caused by simply regarding the article as a rigid body are avoided, and the consistency between the packing result and the actual packing state is improved.

[0017] (3) The actual three-dimensional size of the article is associated with the compression rate and the rotation selection by a size association constraint, and the box boundary constraint and the non-overlapping constraint are constrained based on the actual three-dimensional size variable, so that the real-time reflection of the space occupation of the compression deformation is realized, and the accuracy of the space constraint modeling and the loading space utilization efficiency are improved.

[0018] (4) The three-dimensional physical stability constraint based on the actual three-dimensional size is introduced into the physical constraint, and the stacking support relationship is uniformly constrained in combination with the support constraint, so that the overall stability control of the stacking process is realized, the risk of gravity center deviation, insufficient support and overturning is reduced, and the safety and reliability in the transportation and handling process are improved. BRIEF DESCRIPTION OF DRAWINGS

[0019] Figure 1 The flowchart of the intelligent packing method based on the compression characteristic response of the article according to an embodiment of the present application is disclosed; Figure 2 The closed-loop feedback flowchart of the compression-stability coupling mechanism according to an embodiment of the present application is disclosed; Figure 3 The module schematic diagram of the intelligent packing system based on the compression characteristic response of the article according to an embodiment of the present application is disclosed. DETAILED DESCRIPTION

[0020] To make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme of the embodiments of the present application will be described clearly and completely below with reference to the drawings of the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0021] The present application will be described in further detail below with reference to the drawings: As Figure 1 shown, according to the intelligent packing method based on the compression characteristic response of the article provided by the present application, when the intelligent packing method is executed, the article parameters of each article in the set of articles to be loaded and the box type parameters of each box type in the set of selectable box types are first obtained. The article parameters at least include initial three-dimensional size, weight, compression coefficient and maximum compression rate; the box type parameters at least include box type internal size and box type cost, and form a data structure that can be called by an optimization model.

[0022] Subsequently, a unified bin packing optimization model coupling compression response and three-dimensional physical stability is constructed within the solution framework of mixed integer programming. The model sets variables for bin type selection, item rotation selection, item spatial position, spatial separation between items, item support relationship, item compression ratio, and actual three-dimensional dimensions of the items. Specifically, a correspondence is established between the actual three-dimensional dimensions of the items and the item compression ratio and rotation selection variables through size-related constraints. This ensures that the actual occupied dimensions of the items under different rotational attitudes and compression levels can be uniformly represented and invoked by subsequent spatial and stability constraints.

[0023] In this unified packing optimization model, an objective function and constraints are constructed. The objective function is used to minimize the cost of the packing box while meeting the requirements of space loading and stability.

[0024] Constraints are categorized into spatial constraints, physical constraints, and logical constraints. Spatial constraints include box boundary constraints based on the actual three-dimensional dimensions of the items and non-overlapping constraints based on the spatial separation variables between items. Physical constraints include support constraints based on the support relationship variables of the items, passive compression response constraints based on the load determined by the support relationship variables and the weight of the items, and three-dimensional physical stability constraints established based on the actual three-dimensional dimensions of the items. Logical constraints include box-type unique selection constraints, rotational unique selection constraints, size association constraints, and symmetry breaking constraints, which are used to ensure consistency in discrete decisions and reduce equivalent solutions.

[0025] After solving the unified packing optimization model, a packing scheme is obtained; the packing scheme includes at least the target box type, the placement posture of each item, the placement position, the stacking support relationship and the corresponding compression ratio.

[0026] Finally, the packing plan is output for packing execution and display of packing results.

[0027] During implementation, the parameters are set as follows: I: Collection of items to be loaded Indicates the item index; B: Optional set of box types, representing the box type index; K i :thing i The set of allowed rotation directions (considering symmetry optimization, cube items| K i | =1, items with equal two dimensions | K i | =3, General Items| K i | =6; D : Set of spatial separation directions, D={0,1,2,3,4,5}, representing the six separation directions: left of the X-axis, right of the X-axis, front of the Y-axis, back of the Y-axis, down of the Z-axis, and up of the Z-axis, respectively.

[0028] Item parameters include: :thing i The initial three-dimensional dimensions (length, width, height); m i :thing i The weight; :thing i The compressibility coefficient represents the compressibility caused by a unit load. :thing i The maximum compression ratio.

[0029] Box-type parameters include: : Box type b Internal dimensions (length, width, height) C b : Box type b The cost.

[0030] Decision variables include: Box selection variables, y b =1 indicates that the box type is selected. b ; Rotate the selected variable. Represents items i Select the k Rotation direction; Ground support variables Represents items i Place it directly at the bottom of the box; Supporting variables between items Represents items i Placed on items j Above; Spatially separated variables r ijd =1 indicates an item i With items j In direction d Upper separation; :thing i The actual compression ratio; :thing i The actual three-dimensional dimensions after compression; :thing i The three-dimensional coordinates (position) of the lower left front corner point; :thing i The total weight of the items stacked on top; :thing i The stacking layer number it is located in.

[0031] The solution model built into the intelligent optimization module aims to minimize the box-type cost.

[0032] The objective function considers selecting the lowest-cost container type while meeting all item loading and stability requirements, thereby reducing packaging and transportation costs. The optimization preference can be flexibly adjusted by setting different container types based on their cost.

[0033] In this implementation, by uniformly solving the box type selection, spatial layout, support stacking, compression response and stability constraints within the same model, the synergistic optimization of packing cost, spatial feasibility and physical stability is achieved, thereby improving the feasibility of the solution and reducing the box type cost.

[0034] In the above implementation, preferably, the unified bin packing optimization model is set as a mixed integer programming model and solved using a branch and bound algorithm. During the solution process, the bin type selection variable, rotation selection variable, and discrete separation / support related variables are used as branching objects. At each node, the continuous variables (position, compression ratio, actual size, etc.) are relaxed and the bounds are calculated. Pruning is performed when the lower bound of a node is not better than the current optimal upper bound, thereby obtaining the bin packing scheme that minimizes the bin type cost in a provably convergent search process.

[0035] In this implementation, under the premise of satisfying space and stability constraints, branch and bound are used to realize global search and system pruning, reduce invalid traversal in the solution process, improve solution efficiency and ensure optimization quality.

[0036] Specifically, a mixed-integer programming framework is adopted to transform the complex bin packing problem into a standard mathematical optimization problem, which can be solved using mature commercial solvers. The solver systematically searches the solution space through a branch-and-bound algorithm, and can find a provable global optimal solution or provide an optimality gap guarantee in finite time. Through symmetry breaking constraints and rotational symmetry optimization, the number of equivalent solutions is significantly reduced, and the solution efficiency is greatly improved. The global optimal solution framework provides provable optimization quality.

[0037] Regarding alternative solutions to the solution algorithm, besides using a mixed-integer programming framework, a hierarchical solution strategy can be adopted for ultra-large-scale problems. The first stage uses a column generation algorithm or Lagrange relaxation method to quickly generate several candidate box types and coarse layout schemes. The second stage uses mixed-integer programming for fine optimization and stability verification of each candidate scheme. The third stage selects the optimal scheme with the lowest cost from all feasible schemes. Furthermore, for scenarios with extremely high real-time requirements (such as online box packing in automated warehouses), reinforcement learning algorithms can be used to train the strategy network offline. In the online stage, loading action sequences are directly generated based on the current items and box type configurations. The control system of the device needs to integrate the trained neural network model accordingly.

[0038] In the above implementation, preferably, symmetry breaking constraints are used to eliminate equivalent bin packing solutions. Specifically, items with the same initial three-dimensional dimensions, weight, and material type are grouped into the same item group, and order constraints are applied to the spatial position variables of items within the same item group, so that the items within the group are monotonically ordered in a preset coordinate dimension, thereby preventing the solution process from repeatedly searching among equivalent permutation solutions.

[0039] In this implementation, by reducing the number of equivalent solutions, the combinatorial explosion caused by the symmetry of the solution space is reduced, thereby improving the convergence speed and stability of mixed integer programming solutions.

[0040] In the above implementation, preferably, the size-related constraints are used to link "allowed rotation direction - rotation selection - compression ratio - actual size". For each allowed rotation direction of each item, a size-related constraint is constructed based on the item's rotation selection variable: when the rotation selection variable selects the allowed rotation direction, the item's actual three-dimensional size variable is equal to the product of the initial three-dimensional size under the allowed rotation direction and the compression factor, where the compression factor is (1 - item compression ratio variable); when the allowed rotation direction is not selected, this set of size-related constraints does not impose constraints on the item's actual three-dimensional size variable, so that only the selected rotation direction has a constraint effect on the actual size.

[0041] In this implementation, a unified mapping of the object's rotational posture and compression deformation to its actual occupied size is achieved, ensuring that subsequent boundary, non-overlap, and stability constraints are based on a consistent "actual size," thereby improving modeling consistency and solution feasibility.

[0042] In the above embodiments, preferably, passive compression response constraints are used to establish the response relationship between load-bearing capacity and compression rate. Based on the item support relationship variables, the set of directly stacked items above each item is determined, and the load-bearing capacity of that item is obtained by summing the weights of each item in that set.

[0043] The compression ratio variable of an item is determined based on the load and the compression coefficient, and an upper limit constraint is imposed on the compression ratio variable of the item by the maximum compression ratio; for items with a compression coefficient of zero, the compression ratio variable of the item is constrained to zero so that the item maintains a rigid dimension in the model.

[0044] Specifically, for rigid objects, the present invention achieves rigid constraints by setting the compression coefficient to zero in the algorithm model; for compressible objects, the compression rate is dynamically adjusted according to the weight of the objects stacked on top of them, so that the present invention can handle rigid and compressible mixed material objects in the same optimization model at the same time, without the need for manual intervention to classify objects or design different algorithms for optimization.

[0045] In this implementation, a passive response is achieved where the greater the load, the greater the compression, making the compression ratio self-consistent with the stacking structure, reducing the risk of distortion caused by a uniform fixed compression ratio, and avoiding the risk of unexecutability or damage due to excessive compression by limiting the maximum compression ratio.

[0046] In the above implementation, preferably, the non-overlapping constraint is used to ensure that any two items do not intersect in space. Six spatial separation variables are set for any two items, and when the spatial separation variable in the corresponding direction is enabled, a positional separation inequality is established for the two items in that direction, ensuring that the two items satisfy the inequality relationship of "relative position + actual size interval" along that direction. Simultaneously, a unique separation direction constraint is established to ensure that the two items are separated in at least one of the six directions, thereby guaranteeing no overlap in three-dimensional space.

[0047] In this implementation, a mechanism of six-way separation and at least one-way separation is used to express the three-dimensional non-overlapping relationship in the form of linear constraints, and the space occupancy change after compression is automatically reflected based on the actual three-dimensional size variables, thereby improving the loading feasible domain and space utilization efficiency.

[0048] In the above implementation, preferably, the support constraint determines the support source of each item through the bottom support identifier variable and the support relationship variable between items. A bottom support identifier variable and a support relationship variable between items are set for each item, and a unique constraint is applied, ensuring that each item satisfies either being supported by the bottom of the box or by another item. This determines the value of the stacking support relationship variable for the items and is used for subsequent load-bearing and stability constraint calculations.

[0049] In this implementation, each item is ensured to have a clear source of support, forming a support topology that can be invoked by compressible response constraints and stability constraints, thereby improving the interpretability and executability of the stacking relationship.

[0050] In the above embodiments, preferably, the three-dimensional physical stability constraints include overall center of gravity balance constraints, lateral support area constraints, and overturning moment prevention constraints. The overall center of gravity balance constraint limits the overall center of gravity projection of all loaded items to be within the support area at the bottom of the container, and controls the allowable offset range of the center of gravity through safety margin parameters to prevent overall tipping due to center of gravity shift during transportation. The lateral support area constraint limits the projection overlap between the bottom surface of the supported item and the support surface to meet a preset ratio requirement, ensuring sufficient contact area for each stack and avoiding collapse caused by insufficient local support. The overturning moment prevention constraint limits the center of gravity projection of the supported item to be within the inward-recessed safety area of ​​the top surface of the supported item, ensuring that the center of gravity does not approach the support boundary, thereby preventing individual items from tipping over due to unsupported center of gravity and suppressing the risk of tipping over.

[0051] In this implementation, stability control is achieved at three levels: overall, local support, and individual anti-overturning. This ensures that the packing scheme not only meets spatial feasibility requirements but also possesses physical feasibility and long-term stability, thus systematically guaranteeing the physical stability of the loading scheme.

[0052] Because all stability constraints in the algorithm model use actual size variables instead of the original dimensions, and these actual dimensions are dynamically bound to the compression ratio through size-related constraints, when the optimizer adjusts the stacking position of an item (changing the load above it), the compression ratio of that item will automatically respond to the change through the constraint system, thereby updating its actual size and ultimately triggering a re-evaluation of all relevant stability constraints. This closed-loop feedback mechanism ensures real-time synchronization between compression decisions and stability assessments, avoiding the iterative and suboptimal solutions caused by compressing first and then checking or laying out first and then adjusting in existing technologies. The dynamic coupling of compression and stability achieves the accuracy of algorithm optimization.

[0053] In the above implementation, the unified bin packing optimization model sets 11 types of constraints, which represent the spatial constraints, physical constraints and logical constraints of the bin packing problem.

[0054] Constraint 1: Unique selection constraint for box-type

[0055] This constraint ensures that exactly one box type is selected from the set of available box types.

[0056] Constraint 2: Rotational Unique Selection Constraint

[0057] This constraint ensures that each item has exactly one rotation direction selected. For symmetrical items, this is reduced during the preprocessing stage. K iThe size (e.g., a cube-shaped object retains only one rotation, while an object with two equal dimensions retains three rotations).

[0058] Constraint 3: Dimensional Association Constraint This constraint establishes a precise relationship between the actual dimensions of the item and its compression ratio and rotation direction. For each item... i and each rotation direction Let the original size under this rotation be... ,but:

[0059]

[0060]

[0061]

[0062]

[0063]

[0064] in M It is a maximum constant. The logic of this constraint is: when o ik =1 (select rotation) k The above inequalities are transformed into equality constraints. That is, the actual size is equal to the original size multiplied by the compression factor (1- k i );when o ik When = 0, the inequality is automatically satisfied and has no constraint effect. The significance of this set of constraints is that it models compression as isotropic (i.e., compression of length, width, and height in the same proportion), which conforms to the physical properties of most compressible materials; in addition, it uses the compressibility variable... k i As an intermediary, it dynamically correlates the original geometric dimensions of the item with the actual loading dimensions.

[0065] Constraint 4: Passive Compression Constraint This constraint establishes the dynamic response relationship between the item's compression ratio and its load-bearing capacity. For each item... i :

[0066] This constraint calculates items. i The total weight of all items directly stacked on top. (Using support variables) Only when the item j Placed on items i When above, the item j Only the weight will be included. .

[0067]

[0068] This constraint establishes a linear relationship between the compressibility and the load-bearing capacity. Physically, this means that the degree of compression of an object is determined by the total weight of the objects stacked above it; the heavier the load, the greater the compression. For rigid objects, by setting... Automatically degenerate into k i =0.

[0069]

[0070] This constraint prevents the item from being overcompressed beyond its material limits. For compressible items, Typically set between 0.3 and 0.6, this indicates that the item can be compressed to 40%-70% of its original size; for rigid items, The size of the item must remain unchanged.

[0071] Constraint 5: Boundary Constraints

[0072]

[0073]

[0074] This constraint ensures that each item remains entirely within the interior space of the selected box type and does not extend beyond the box's boundaries. (Due to actual dimensions...) The compression effect has already been included, so the constraint has taken into account the space occupied after compression.

[0075] Constraint 6: Non-overlapping constraint This constraint ensures that no two items overlap in space. For any two items i and j ( i ≠ j (), separation is required in at least one of six directions:

[0076]

[0077]

[0078]

[0079]

[0080]

[0081] Uniqueness constraint of separation direction:

[0082] when When the inequality in the corresponding direction is activated, it requires that the two items be strictly separated in that direction; when When, add to the right side of the inequality M This ensures that constraints are automatically satisfied. The separation direction uniqueness constraint ensures that each pair of items separates in at least one direction, thus guaranteeing no overlap.

[0083] The key to this constraint is that it uses actual size variables. Rather than the original size. For example, when an item... i Compression occurs h i When it decreases, it is in Z The space occupied along the axis is automatically reduced, providing more feasible placement space for other items.

[0084] Constraint 7: Support Constraint This constraint ensures that each item has physical support and prevents unreasonable situations such as cyclic support.

[0085]

[0086] This constraint requires that each item must either be placed on the ground ( ), or place it on top of at least one other item (if it exists). j make ).

[0087]

[0088] When items i When placed on the ground ( This constraint is mandatory. z i =0, meaning the bottom surface of the item must be in contact with the ground; when it is not on the ground... z i Positive values ​​are allowed.

[0089]

[0090]

[0091] when hour, level i Forced to 0; when hour, level i ≥1 provides the basis for subsequent layer sorting constraints.

[0092]

[0093] This constraint prohibits two items from supporting each other simultaneously.

[0094]

[0095]

[0096] when When these two inequalities are transformed into equations , i.e., items i The height of the base is equal to that of the item j The height of the top surface ensures vertical contact. Here h j Use actual height (including compression), so when the item j When compression occurs, the item i The height will automatically decrease, thus achieving the transmission of the compression effect in the vertical direction.

[0097]

[0098]

[0099] when At that time, forced This means that the number of layers of the items above is exactly 1 greater than the number of layers of the items below.

[0100] Constraint 8: Overall center of gravity balance constraint This constraint defines the position of the overall centroid as follows:

[0101]

[0102] in Represents items i center of gravity X Coordinates. This constraint requires that the overall center of gravity projection must fall within the safe area of ​​the box-shaped bottom surface:

[0103]

[0104]

[0105]

[0106] in α This is the center-of-gravity safety margin factor, indicating the safe area within which the center of gravity can shift near the center of the box-shaped structure. For example... α When = 1 / 3, the center of gravity X Coordinates are allowed in Within the designated area, the risk of tipping over due to the center of gravity being too far to one side is avoided.

[0107] Constraint 9: Lateral support area constraint This constraint assesses whether the support area between each pair of stacked items is sufficient from a local support perspective. For any two items... i and j ,when i Place j When above ( ),Require i bottom surface and j The top surfaces must have sufficient overlap: X-direction support constraint:

[0108]

[0109] Y-direction support constraint:

[0110]

[0111] in β x and β y The minimum support ratio coefficient indicates the proportion of the upper-layer items. X and Y At least 75% of the bottom side length must be within the projection range of the top surface of the lower item.

[0112] First inequality Required items i The left boundary cannot exceed the item. j The left boundary is too far (at most, it's suspended in mid-air). (distance), the second inequality requires the items i The right boundary cannot exceed the item. j The right boundary is too far. The two constraints work together to ensure the item... i exist X At least in the direction β x The proportions are supported.

[0113] Constraint 10: Overturning Moment Prevention Constraint This constraint, from the perspective of individual object stability, prevents the object from tipping over due to its center of gravity being suspended. For any two objects... i and j ,when i Place j When above ( ),Requirei The projection of the center of gravity must be located at j Within the safe area of ​​the top surface retraction: X-direction centroid constraint:

[0114]

[0115] Y-direction centroid constraint:

[0116]

[0117] in gamma This is the overturning safety margin factor. Represents items i center of gravity X Coordinates. This constraint requires the item... i The center of gravity projection should not be too close to the supporting surface (object). j The edge of the top surface must have at least gamma · l j Safety margin.

[0118] Constraint 11: Symmetry breaking constraint For multiple items with the same attributes (size, weight, material), their positions in the optimization space are completely symmetrical, leading to the existence of a large number of equivalent solutions and increasing the difficulty of solving the problem. This constraint eliminates symmetry by forcing the placement order of symmetrical items: Let the set of items... I Grouped by type { I 1, I 2,……, I T}, where the items in each group are exactly the same. For each group I T (| I T | >1), Sort the items in the group by ID as { i 1, i 2,……, I n},but: This constraint forces identical items to be... X Arranging the coordinates in ascending order improves the efficiency of the solution without affecting the quality of the optimal solution.

[0119] like Figure 2As shown, closed-loop feedback is achieved through three layers during implementation. The first layer is the compression sensing layer, which calculates the load above each item in real time based on the support relationship in the algorithm model, thereby determining the compression ratio. This layer establishes a direct correlation between the stacking topology of items and the compression response.

[0120] The second layer is the size update layer, which transforms the compression ratio into actual size changes through size-related constraints. This layer ensures that all constraints involving item size in the algorithm model use the compressed actual size.

[0121] The third layer is the stability verification layer. All stability constraints directly use actual size variables, ensuring that size changes caused by compression automatically trigger a re-evaluation of stability constraints. The mechanism operates as follows: when the optimization solver adjusts an item during the branch search process... j Its position makes it from the item i When the top is removed, the constraint system responds automatically. The reduction in the load on the top of the object leads to a decrease in the compression ratio, which in turn increases the actual height, length, width, and height. These changes are passed to the non-overlapping constraints, support constraints, and center of gravity constraints through the size-related constraints. The optimization solver systematically explores these chain effects through the branch and bound algorithm to ensure that the final solution simultaneously satisfies the spatial constraints and stability constraints.

[0122] Regarding enhancements to stability constraints, in addition to the basic three-dimensional stability constraint system, dynamic stability constraints can be added to the algorithm model for scenarios with more stringent transportation conditions (such as long-distance sea freight and mountain road transportation). These include acceleration response constraints, simulating acceleration, deceleration, and turning during transportation, introducing an inertial force term into the model and requiring that the inertial torque under maximum acceleration does not exceed the safety margin of the stabilizing torque; vibration frequency constraints, limiting the ratio of the overall center of gravity height to the width of the box-shaped bottom surface in the model to address resonance risks; and interlayer friction constraints, introducing the friction coefficient between items into the model and establishing constraints to ensure that the friction force is sufficient to prevent items from sliding at tilt angles, further improving the robustness of the algorithm-generated scheme under harsh transportation conditions.

[0123] By implementing the above-mentioned intelligent packing method based on the compression characteristic response of items, a unified optimization model that deeply couples compression characteristic response and physical stability is constructed, and a two-way feedback mechanism between the algorithm module and the device execution system is established, thus realizing intelligent packing of items made of rigid and compressible mixed materials.

[0124] First, a passive compression response modeling mechanism based on load feedback. This invention establishes a dynamic correlation model between the compression ratio of an item and its load, and obtains compression characteristic parameters through the device's sensing and detection module. Unlike existing technologies that use active pre-compression or fixed compression ratio methods, this invention introduces a compression ratio variable into the algorithm model. k iA passive response relationship is established with the load variables above. This mechanism can perceive the actual pressure borne by each item during stacking in real time at the algorithm level and automatically calculate its degree of compression deformation based on the item's material properties. The technical principle of this mechanism lies in achieving accurate modeling of the compression response through multi-layer linkage of the constraint system. In the constraint system of the algorithm model, the load above is accurately calculated through support relationship variables, ensuring the consistency between load calculation and stack topology. In the size mapping relationship, the actual size of the item is expressed as a function of the compression ratio, which conforms to the isotropic compression characteristics of most compressible materials. At the device level, the physical parameters and maximum compression ratio of the item are obtained through sensing and detection modules (including a vision measurement device to obtain initial size, a weighing sensor to measure weight, and a material testing device to apply standard load to determine the compression response), and these are used as inputs to the algorithm model. By establishing an equality constraint relationship between the compression ratio and the load, the degree of compression in the algorithm model is ensured to be dynamically adjusted according to the stacking state. This allows the bottom items with higher pressure to fully utilize their compression potential to save space during model optimization, while the top items with lower pressure maintain a lower compression ratio to avoid damage. The device execution system guides the placement force of the robotic arm based on the compression rate output by the algorithm, forming a synergy between algorithm modeling and device execution.

[0125] Second, the systematic construction of a three-dimensional physical stability assessment system. This invention establishes a three-dimensional stability constraint system within the algorithm model, encompassing overall center of gravity balance, lateral support area, and overturning moment prevention. This system comprehensively ensures the physical feasibility of the loading scheme from three levels: global balance, local support, and individual unit stability, and provides evaluation standards for the device's stability monitoring module. At the overall center of gravity balance level, the algorithm model calculates the weighted center of gravity position of all items, establishing constraints to ensure that the overall center of gravity projection falls within the safe area of ​​the box-shaped bottom surface, thereby preventing the risk of overall tipping due to center of gravity shift during transportation. At the lateral support area level, the algorithm model requires that the bottom surface of the upper item and the top surface of the lower item must meet a minimum overlap ratio. For items placed on top of another item, constraints are established in the X, Y, and Z directions respectively to ensure that at least a considerable proportion of the bottom side length of the item is within the projection range of the top surface of item j, preventing local instability and collapse due to insufficient support area. In terms of preventing overturning moment, the algorithm model prevents the item from overturning due to the center of gravity being suspended from the ground from the perspective of individual stability. When an item is placed on top of another item, a constraint is established that the projection of the item's center of gravity must be located within the safe area of ​​the top surface of the other item. The same applies in the Y direction, ensuring that the item's center of gravity is not too close to the edge of the support surface, thereby preventing the risk of overturning caused by the center of gravity being suspended from the ground.

[0126] Third, a deep coupling mechanism between compression deformation and stability constraints. This invention achieves automatic coupling between the actual size variable of the object and the stability assessment by using the actual size variable of the object as an intermediate bridge in the algorithm model. When a compressible object increases its compression ratio due to bearing more overhead load during the algorithm optimization process, its actual height decreases, which changes the position of the vertical contact surface. The changes in the actual length and width affect the overlapping area of ​​the support surface and the projection range of the center of gravity. These changes are transmitted in real time to the non-overlapping constraints, support constraints, and stability constraints through size-related constraints, forming a closed-loop feedback mechanism of compression response-size update-stability verification.

[0127] Fourth, the invention utilizes a globally optimal solution framework and device based on mixed-integer programming. This invention integrates discrete decisions such as box type selection, item rotation, and support relationships, as well as continuous variables such as compression ratio, actual size, and three-dimensional position, into a single mixed-integer programming model. With the goal of minimizing box type cost, it automatically generates the optimal loading scheme while satisfying all spatial, compression, and stability constraints. This model can be solved using a standard optimization solver, employing a branch-and-bound algorithm to systematically search the solution space.

[0128] like Figure 3 As shown, the present invention also proposes an intelligent packing system based on the response of item compression characteristics, which applies the intelligent packing method based on the response of item compression characteristics disclosed in any of the embodiments, including: a data acquisition module, an intelligent optimization module, an execution guidance module, and a feedback verification module.

[0129] The data acquisition module obtains item parameters and box type parameters and provides them to the intelligent optimization module; The intelligent optimization module constructs and solves a unified bin packing optimization model within the mixed integer programming framework, and outputs a bin packing solution. The execution guidance module outputs packing plans for packing execution and display of packing results; The feedback verification module collects monitoring data related to load, compression deformation and tilt angle during the packing process, and provides the monitoring data to the intelligent optimization module for consistency verification and adjustment.

[0130] In this implementation, a closed-loop link of "solution output - execution monitoring - consistency verification" is established to link the packing optimization with the execution status, thereby reducing the deviation between the theoretical solution and the actual execution and improving the system's robustness and feasibility.

[0131] In the above embodiments, preferably, the data acquisition module includes a vision sensor and a weighing sensor for acquiring the initial three-dimensional dimensions and weight of each item, and includes a compression parameter measurement unit for determining the material type of the item and measuring the compression coefficient and maximum compression ratio of the compressible item.

[0132] The feedback detection module includes a force sensor, a displacement sensor, and a tilt sensor, which are used to collect load-related signals, compression deformation-related signals, and tilt angle-related signals respectively during the packing process, and provide the collection results to the intelligent optimization module for consistency verification and adjustment.

[0133] The execution guidance module generates motion control parameters for controlling the robotic arm based on the corresponding compression ratio in the packing scheme, so that the robotic arm meets the placement force and posture control requirements corresponding to the compression ratio during placement and stacking.

[0134] In this implementation, the closed-loop consistency between compression response modeling and device execution control is achieved through the coordination of "size / weight / compression parameter measurement - force / deformation / attitude monitoring during execution - generation of robotic arm control parameters according to compression rate", thereby improving the accuracy of packing execution and reducing the risk of overpressure damage and stability deviation.

[0135] Specifically, the data acquisition module is responsible for acquiring the physical parameters of the object, wherein the size measurement device (which can be a laser rangefinder, structured light scanner, or vision recognition system) acquires the initial three-dimensional dimensions of the object. Weighing sensors measure the weight of items. m The material recognition unit automatically identifies whether an object is rigid or compressible by applying a standard load and measuring its deformation response, and determines the compressibility coefficient of compressible objects. k coef (Unit: kg) -1 ) and maximum compression ratio k max These parameters provide accurate physical basis data for subsequent algorithm optimization.

[0136] The intelligent optimization module is the core of this invention. It incorporates a passive compression modeling algorithm and a multi-level stability evaluation algorithm, employing a mixed-integer programming framework for global optimal solution. The module's innovation lies in establishing a dynamic correlation mechanism between compression ratio and load capacity within the algorithm model, achieving accurate modeling of passive compression; constructing a three-dimensional stability constraint system encompassing overall center of gravity balance, lateral support area, and overturning moment prevention; and achieving deep coupling between compression deformation and stability constraints through actual size variables as an intermediate bridge. This module runs on an industrial control computer or embedded computing platform, receiving input from the data acquisition module and outputting an optimized loading scheme.

[0137] The execution guidance module transforms the optimization results into executable operation instructions. This module can be implemented in two ways: first, by displaying the placement position, rotation direction, and expected compression of each item to the operator via a screen, allowing manual loading; second, by converting the packing plan into motion instructions for a robotic arm (including target position, rotation angle, placement force, etc.), which are then executed by an automated device. If the device is equipped with real-time monitoring capabilities (such as force sensors, displacement sensors, and tilt sensors), the actual status can be fed back to the intelligent optimization module for verification or adjustment during execution.

[0138] The intelligent packing system based on the response of the compression characteristics of the items disclosed in the above embodiments has the same function as each step of the intelligent packing method based on the response of the compression characteristics of the items disclosed in the above embodiments. In the implementation process, the operation is carried out with reference to the above embodiments, and will not be repeated here.

[0139] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A smart packing method based on the compression characteristics response of an item, characterized in that, include: Obtain the item parameters of each item in the set of items to be loaded and the box parameters of each box in the set of optional box types. The item parameters include initial three-dimensional dimensions, weight, compression coefficient, and maximum compression ratio. The box parameters include the internal dimensions of the box and the box cost. Within the framework of mixed integer programming, a unified bin packing optimization model coupling compression characteristic response and three-dimensional physical stability is constructed. This model includes bin type selection variables, item rotation selection variables, item spatial position variables, spatial separation variables between items, item support relationship variables, item compression ratio variables, and actual three-dimensional size variables of the items. The actual three-dimensional size variables are then correlated with the item compression ratio variables and the item rotation selection variables through size-related constraints. In the unified packing optimization model, an objective function and constraints are constructed. The objective function minimizes the box cost while satisfying spatial loading and stability requirements. The constraints include spatial constraints, physical constraints, and logical constraints. The spatial constraints include box boundary constraints based on the actual three-dimensional dimensions of the items and non-overlapping constraints based on the spatial separation variables between items. The physical constraints include support constraints based on the support relationship variables of the items, passive compression response constraints based on the load-bearing capacity determined by the support relationship variables and the weight of the items, and three-dimensional physical stability constraints established based on the actual three-dimensional dimensions of the items. The logical constraints include unique box selection constraints, rotational unique selection constraints, size association constraints, and symmetry breaking constraints. Solve the unified packing optimization model to obtain a packing scheme, which includes the target box type, the placement posture of each item, the placement position, the stacking support relationship and the corresponding compression ratio; The packing scheme is output for packing execution and packing result display.

2. The intelligent packing method based on the compression characteristics response of an item according to claim 1, characterized in that, The unified packing optimization model is a mixed integer programming model. The branch and bound algorithm is used to solve the mixed integer programming model to obtain the packing scheme that minimizes the cost of the box type.

3. The intelligent packing method based on the compression characteristics response of an item according to claim 1, characterized in that, The symmetry breaking constraint includes: grouping items with the same initial three-dimensional dimensions, weight, and material category into the same item group, and applying order constraints to the spatial position variables of items within the same item group to eliminate equivalent bin packing solutions.

4. The intelligent packing method based on the compression characteristics response of an item according to claim 1, characterized in that, The size-related constraint includes: for each allowed rotation direction of each item, a size-related constraint is constructed based on the item rotation selection variable, such that when the allowed rotation direction is selected in the item rotation selection variable, the actual three-dimensional size variable of the item is equal to the product of the initial three-dimensional size under the allowed rotation direction and the compression factor, and when the allowed rotation direction is not selected, the size-related constraint does not impose a constraint on the actual three-dimensional size variable of the item, wherein the compression factor is (1 - the item compression rate variable).

5. The intelligent packing method based on the compression characteristics response of an item according to claim 1, characterized in that, The passive compression response constraint includes: determining the set of directly stacked items above each item based on the item support relationship variable, and summing the weights of each item in the set of directly stacked items to obtain the load-bearing capacity of the item; The compression ratio variable of the item is determined based on the load-bearing capacity and the compression coefficient, and the upper limit constraint of the compression ratio variable of the item is imposed by the maximum compression ratio. For items with a compression factor of zero, constrain the item compression ratio variable to zero.

6. The intelligent packing method based on the compression characteristics response of an item according to claim 1, characterized in that, The non-overlapping constraint includes: setting six spatial separation variables between items for any two items, and establishing a positional separation inequality of the two items in the corresponding direction when the spatial separation variable between items in the corresponding direction is activated, while establishing a uniqueness constraint of separation direction to ensure that the two items are separated in at least one of the six directions.

7. The intelligent packing method based on the compression characteristics response of an item according to claim 1, characterized in that, The support constraints include: setting a bottom support identifier variable and an inter-item support relationship variable for each item, so that each item satisfies the unique constraint between its bottom support identifier variable and the inter-item support relationship variable supported by another item, in order to determine whether the item is supported by the bottom of the box or by another item.

8. The intelligent packing method based on the compression characteristics response of an item according to claim 1, characterized in that, The three-dimensional physical stability constraints include overall center of gravity balance constraints, lateral support area constraints, and overturning moment prevention constraints. The overall center of gravity balance constraints are used to limit the overall center of gravity projection of all items after loading to be located within the bottom support area of ​​the box. The lateral support area constraints are used to limit the overlapping area of ​​the projection between the bottom surface of the supported item and the support surface to meet a preset ratio. The overturning moment prevention constraints are used to limit the center of gravity projection of the supported item to be located within the inward safe area of ​​the top surface of the support item.

9. A smart packing system based on the compression characteristics response of an item, characterized in that, The intelligent packing method based on the compression characteristics of articles as described in any one of claims 1 to 8 includes: a data acquisition module, an intelligent optimization module, an execution guidance module, and a feedback verification module; The data acquisition module is used to acquire item parameters and box type parameters; The intelligent optimization module is used to construct and solve a unified bin packing optimization model within the framework of mixed integer programming, so as to output a bin packing scheme. The execution guidance module is used to output the packing plan for packing execution and packing result display; The feedback detection module is used to collect monitoring data related to load, compression deformation and tilt angle during the packing process, and provide the monitoring data to the intelligent optimization module for consistency verification and adjustment.

10. The intelligent packing system based on the compression characteristics response of an item according to claim 9, characterized in that, The data acquisition module includes a vision sensor and a weighing sensor to acquire the initial three-dimensional dimensions and weight of each item, and includes a compression parameter measurement unit to determine the material type of the item and measure the compression coefficient and maximum compression ratio of the compressible item. The feedback detection module includes a force sensor, a displacement sensor, and a tilt sensor; The execution guidance module generates motion control parameters for controlling the robotic arm based on the corresponding compression ratio in the packing scheme.

Images