A heuristic-based robot hybrid palletizing method and device and related medium

Abstract

The application discloses a heuristic-based mechanical hand mixed stacking method and device and related medium, the method comprises the following steps: obtaining material original information and performing structured modeling to obtain a structured material data set; performing heuristic stacking type planning processing on the structured material data set to obtain a stacking pose planning result; performing task allocation processing according to the stacking pose planning result to obtain a task allocation result; performing stacking sequence processing on at least two mechanical hands by using the task allocation result to obtain a stacking sequence result; generating an outbound sorting queue according to the stacking sequence result based on an interleaving and merging rule, and mapping the outbound sorting queue to a scheduling object; and controlling the mechanical hands to perform stacking operations by using the scheduling object. The application generates an outbound sorting queue according to a calculated stacking sequence result and maps the outbound sorting queue to a scheduling object, and then controls multiple mechanical hands to perform stacking operations simultaneously by using the scheduling object, so that the working efficiency of the mechanical hand stacking mode is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of logistics automation, and in particular to a heuristic-based manipulator mixed palletizing method and device and related medium. BACKGROUND

[0002] Palletizing, as a key operation at the end of logistics automation, is directly related to the overall transfer efficiency of warehouse in-out and conveying links. With the rapid growth of demand for mixed palletizing of irregular materials in businesses such as supermarkets, fresh food, and cold chain, the traditional palletizing mode relying on a single manipulator needs to frequently switch paths and poses in a large operation space, with long paths and limited beats, making it difficult to meet the throughput requirements in high-flow scenarios while ensuring the stability of the pile type and space utilization. Currently, the traditional single manipulator palletizing mode has low work efficiency and cannot effectively support efficient out-of-warehouse and palletizing processing during peak periods. SUMMARY

[0003] Embodiments of the present application provide a heuristic-based manipulator mixed palletizing method, device and related medium, aiming to solve the technical problem of low work efficiency of the single manipulator palletizing mode in the prior art.

[0004] In a first aspect, embodiments of the present application provide a heuristic-based manipulator mixed palletizing method, comprising:

[0005] Obtaining raw material information in an order and structuring modeling to obtain a structured material data set;

[0006] Performing heuristic pile type planning processing on the structured material data set to generate candidate placement poses and hierarchical identifiers, and integrating to obtain a palletizing pose planning result;

[0007] Performing task allocation processing according to the palletizing pose planning result to obtain a task allocation result;

[0008] Performing palletizing sequencing processing on at least two manipulators using the task allocation result to obtain a palletizing sequence result;

[0009] Generating an out-of-warehouse sequencing queue based on an interleaving and merging rule according to the palletizing sequence result, and mapping the out-of-warehouse sequencing queue as a scheduling object;

[0010] Controlling a manipulator to perform a palletizing operation using the scheduling object.

[0011] In a second aspect, embodiments of the present application provide a heuristic-based manipulator mixed palletizing device, comprising:

[0012] A data modeling unit configured to obtain raw material information in an order and perform structured modeling to obtain a structured material data set;

[0013] a data integration unit configured to perform heuristic stacking type planning processing on the structured material data set to generate a candidate placement pose and a hierarchical identifier, and to integrate to obtain a palletizing pose planning result;

[0014] a task allocation unit configured to perform task allocation processing according to the palletizing pose planning result, and to obtain a task allocation result;

[0015] a palletizing sequencing unit configured to perform palletizing sequencing processing on at least two robots using the task allocation result, and to obtain a palletizing sequence result;

[0016] a parameter scheduling unit configured to generate an outbound sequencing queue according to the palletizing sequence result based on an interleaving and merging rule, and to map the outbound sequencing queue as a scheduling object;

[0017] an operation output unit configured to control the robots to perform palletizing operations using the scheduling object.

[0018] In a third aspect, an embodiment of the present application provides a computer device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, and the processor implements the heuristic-based robot hybrid palletizing method of the first aspect when executing the computer program.

[0019] In a fourth aspect, an embodiment of the present application provides a computer readable storage medium, wherein the computer readable storage medium stores a computer program, and the computer program is executable on a processor to implement the heuristic-based robot hybrid palletizing method of the first aspect.

[0020] An embodiment of the present application provides a heuristic-based robot hybrid palletizing method, including obtaining material original information in an order and performing structured modeling to obtain a structured material data set; performing heuristic stacking type planning processing on the structured material data set to generate a candidate placement pose and a hierarchical identifier, and integrating to obtain a palletizing pose planning result; performing task allocation processing according to the palletizing pose planning result, and obtaining a task allocation result; performing palletizing sequencing processing on at least two robots using the task allocation result, and obtaining a palletizing sequence result; generating an outbound sequencing queue according to the palletizing sequence result based on an interleaving and merging rule, and mapping the outbound sequencing queue as a scheduling object; and controlling the robots to perform palletizing operations using the scheduling object. The palletizing sequence result obtained by calculation is used to generate an outbound sequencing queue and map the outbound sequencing queue as a scheduling object, and then the scheduling object is used to control multiple robots to perform palletizing operations, thereby improving the working efficiency of the robot palletizing mode.

[0021] An embodiment of the present application also provides a heuristic-based robot hybrid palletizing device, a computer device, and a storage medium, which also have the above beneficial effects. Attached Figure Description

[0022] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 A flowchart illustrating a heuristic-based hybrid palletizing method using a robotic arm, provided for an embodiment of the present invention;

[0024] Figure 2 This is a schematic block diagram of a heuristic-based robotic hybrid palletizing device provided for an embodiment of the present invention. Detailed Implementation

[0025] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0026] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0027] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0028] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0029] Please see below. Figure 1 , Figure 1 The flowchart of a heuristic robotic hybrid palletizing method provided in this embodiment of the invention specifically includes steps S101 to S106.

[0030] S101, acquire material original information in an order and perform structured modeling to obtain a structured material data set;

[0031] S102, perform heuristic stacking type planning processing on the structured material data set to generate a candidate placement pose and a hierarchical identifier, and integrate to obtain a palletizing pose planning result;

[0032] S103, perform task allocation processing according to the palletizing pose planning result to obtain a task allocation result;

[0033] S104, perform palletizing sequencing processing on at least two manipulators using the task allocation result to obtain a palletizing sequence result;

[0034] S105, generate a warehouse-out sequencing queue based on the interleaving and merging rule according to the palletizing sequence result, and map the warehouse-out sequencing queue as a scheduling object;

[0035] S106, control the manipulator to perform palletizing operation using the scheduling object.

[0036] In step S101, the material original information from the order is received, normalized and structured modeling is performed to obtain a structured material data set including physical properties (such as length, width, height, weight), stacking constraints (such as pallet length, width, maximum stacking height) and positioning attributes (such as visual recognition coordinates), which provides basic data support for subsequent stacking type calculation and grasping positioning.

[0037] In an embodiment, the step S101 comprises:

[0038] The material original information is received and normalized parsed according to order line identifier, material identifier and timestamp to obtain an original information object;

[0039] The original information object is field extracted to obtain preliminary structured data; wherein the preliminary structured data at least includes physical properties, stacking constraints and positioning attributes;

[0040] The preliminary structured data is stored in a structured manner to obtain structured storage data;

[0041] Based on a preset order data connection mechanism, the dimensional fields of the structured storage data are associated with the corresponding order line to generate a connection association object;

[0042] The connection association object is calibrated for cargo fixture parameters to generate a calibration result object;

[0043] The calibration result object is written back to the field set corresponding to the connection association object to obtain a structured material data set.

[0044] In the embodiment, the material original information from the business system is received, and normalized analysis is performed taking "order line identifier, material identifier, timestamp" as the ternary primary key to construct an original information object containing source system number, unit of measurement and precision constraint, thereby providing a stable entry for field extraction. Then, field extraction is performed on the original information object to obtain preliminary structured data; the data covers at least three types of information dimensions: one is physical property, including length L, width W, height H and weight M; the other is stacking constraint, including pallet length, width and maximum stacking height, which are used to limit the planning boundary and the upper limit of the number of layers; the third is positioning attribute, including visual recognition coordinates X / Y / Z and attitude angle θ, which are used to express the spatial pose of grabbing and placing. To ensure consistency of data caliber from different sources, the unit of measurement and coordinate system are unified during the extraction process, and the field source and integrity marker are recorded. Then, the preliminary structured data is stored in the database in the form of "primary key-field set" to obtain structured storage data, and the mapping relationship between the material dimension (physical property), the constraint dimension (pallet and height constraint) and the pose dimension (X / Y / Z / θ) is established in the storage layer, thereby realizing consistent modeling of the information dimension of the material through order data conversion.

[0045] Further, based on the preset order data connection mechanism, the dimensions and fields of the structured storage data are associated with the corresponding order line one by one to generate a connection association object; the object aggregates multiple arrival, replenishment or split records of the same order line, fills in the missing fields and eliminates duplicates, thereby ensuring that the subsequent planning stage faces consistent material granularity. Subsequently, the connection association object is subjected to a pallet parameter calibration (matching material attributes): the pallet / clamp capacity parameters are read, wherein the maximum support length W_max is used to determine the maximum width range of the supportable material, and the maximum load M_max is used to determine whether the material can be stacked by the robot; W and M of the material are matched and verified with W_max and M_max respectively, and a calibration result object is output, which includes a stackable identifier, a recommended grabbing direction / support attitude marker and warning information (such as overwidth and overweight). Finally, the calibration result object is written back to the field set corresponding to the connection association object in a field-level manner to obtain a structured material data set, which serves as an input for subsequent heuristic stacking planning and task allocation.

[0046] In step S102, the stacking order can be determined according to experience rules such as "heavy below and light above, large below and small above", combined with position planning strategies such as hierarchical filling, minimum gap priority, edge alignment and center aggregation, and 0° to 90° attitude adjustment is performed on the rotatable material to adapt to the gap, thereby generating a candidate placement pose and labeling the hierarchical identifier, and the stacking pose planning result is obtained by summarizing.

[0047] In an embodiment, the step S102 comprises:

[0048] The structured material data set is established according to the weight, size, and stability to obtain a heuristic parameter weight table; wherein the priority of the weight, size, and stability decreases in turn;

[0049] The hierarchical target generation of the material is performed according to the heuristic parameter weight table to construct a layered filling rule;

[0050] The gap matching calculation of the material is performed according to the heuristic parameter weight table to construct a minimum gap priority rule;

[0051] The alignment and aggregation constraints of the material are superimposed according to the heuristic parameter weight table to construct an edge alignment rule;

[0052] The area division processing of the material is performed according to the heuristic parameter weight table to construct an area division rule;

[0053] The posture adjustment is set according to the heuristic parameter weight table, and the stacking coordinate parameter and hierarchical number are established to integrate a material posture adjustment rule;

[0054] The layered filling rule, the minimum gap priority rule, the edge alignment rule, the area division rule, and the material posture adjustment rule are integrated to obtain a stacking pose planning result.

[0055] In the embodiment, based on the physical attributes (length L, width W, height H, weight M) of each material in the structured material data set, the stacking constraints (tray length and width, maximum stacking height), and the positioning attributes (X / Y / Z / θ), a heuristic parameter weight table is established: taking weight, size (based on the bottom area), and stability (whether easy to damage) as three core parameters, and configuring the weights in turn according to the priority of weight>size>stability. The corresponding decision meaning is: the material with large weight is preferentially placed on the bottom layer, and the material with small weight is placed on the upper layer; the material with large bottom area is preferentially placed on the bottom layer to provide support for the upper layer, and the material with small bottom area is used to fill the gap later; the material with stable structure is preferentially placed on the bottom layer, and the material with poor stability or easy to damage is placed on the upper layer to reduce the risk of being pressed. Thus, the heuristic parameter weight table for sorting and scoring is obtained.

[0056] Furthermore, hierarchical targets are generated according to the heuristic parameter weight table, and layered filling rules are constructed. Specifically, the effective usable area and volume of the current working layer are calculated based on the pallet, and a layer filling threshold (e.g., 80%) is set. Material delivery to the current layer continues until the threshold is reached; once the threshold is reached, the system automatically switches to the next layer until the maximum palletizing height limit is met. The generation and updating of hierarchical targets proceed in a bottom-up order, and the layer number for each material delivery is recorded as F (F=1 represents the bottom layer, increasing sequentially). Then, the idle areas on the pallet are enumerated according to the heuristic parameter weight table, and the minimum gap priority rule is executed. This involves scanning all candidate idle areas, calculating the dimensional differences between each area and the material to be delivered in both length and width dimensions, and summing these differences (length difference + width difference). The area with the smallest difference is prioritized to reduce space waste. When multiple equivalent areas exist, parallel constraints and scoring are applied using the rules described later.

[0057] Furthermore, edge alignment rules and center aggregation rules are superimposed to refine and filter candidate placement poses. For edge alignment, the pallet coordinate system is discretized into a fixed grid (e.g., 5cm × 5cm), forcing candidate poses (X, Y) to be rounded to integer multiples of the grid, and aligning the material edges with the pallet edges or the edges of already placed materials in the horizontal / vertical directions to reduce interference and accumulated errors caused by skew. For center aggregation, the distance from the candidate position to the geometric center of the pallet is calculated, prioritizing positions with closer distances to improve overall center of gravity aggregation and stack balance without violating the aforementioned constraints.

[0058] Based on this, a region division rule is executed according to the heuristic parameter weight table to form layout constraints. That is, the pallet is divided into several sub-regions (such as left / right, front / back). Materials of the same type or materials in the same order line are arranged in a concentrated manner in the region to reduce the complexity of grasping, conveying and path planning caused by cross-region mixing. Region division in this step is only used for spatial allocation in the pose planning stage. Whether cross-region collaboration is required in the subsequent step is handled by the task allocation step. For materials that can be rotated, a posture adjustment rule is set and palletizing coordinate parameters and layer numbers are established: the matching degree between the 0° and 90° postures and the candidate free areas is evaluated respectively, and the posture with the largest contact area and gap is preferred. After the final posture is determined, three-dimensional pose parameters (X, Y, Z) and posture angle θ (which can be a discrete value of "parallel to the X-axis / perpendicular to the X-axis") are generated and bound to the corresponding layer number F.

[0059] According to the above rules, for each material, based on the six rules of hierarchical filling, minimum gap priority, edge alignment, center aggregation, region division and posture adjustment, a plurality of candidate placement poses are generated, and then weighted scoring and constraint checking (including tray size, layer height and boundary conditions of placed materials) are performed according to the heuristic parameter weight table, and finally the pose with the highest score and satisfying all constraint conditions is selected as the candidate placement pose of the material. The candidate poses of all materials and the hierarchical identifiers are integrated and sorted, and the palletizing pose planning result is output. In the data expression, for consistent calling in subsequent steps, the planning result is saved in the record structure P (L, W, H, M, X, Y, Z, θ, F, R), and the execution identifier R is recorded as 0 when the manipulator is not assigned in this step. The above process ensures that the heuristic strategy driven by weight forms a synergy in hierarchical utilization, gap matching and space alignment, thereby providing a directly consumable pose planning result for subsequent task allocation and palletizing sorting.

[0060] In step S103, the materials falling into each region are generated into corresponding tasks by dividing the left and right regions with the tray center line, and the number of tasks in the same layer is counted. If necessary, cross-region cooperation is implemented at the center line adjacent position to balance the load of the double manipulators, and the task allocation result facing different regions is obtained.

[0061] In an embodiment, the step S103 comprises:

[0062] The candidate extraction and identifier aggregation processing are performed on the palletizing pose planning result to obtain a set of tasks to be allocated;

[0063] The region division processing is performed on the set of tasks to be allocated to obtain an initial region allocation object;

[0064] The hierarchical statistical processing is performed on the initial region allocation object to obtain a hierarchical statistical object;

[0065] The cross-region cooperation balancing processing is performed on the hierarchical statistical object to obtain a balanced allocation object;

[0066] According to the balanced allocation object, the materials tasks after region division and cross-region cooperation are temporarily stored in the corresponding task queue to obtain a task queue object;

[0067] The task queue object is processed by label writing to obtain a task allocation result.

[0068] In the embodiment, the pose planning results of the stacking are extracted and identified, the pose parameters and the level identification of each material are aggregated into the task items to be distributed, and a task set to be distributed is formed (each record structure P corresponds, and contains fields of L / W / H / M, X / Y / Z / θ, layer number F, and execution identification R, etc., wherein R is empty or 0 before distribution). Based on the geometric center line of the tray, the region division processing is performed, that is, the material center point is used as the basis to divide the tasks into left and right regions; if the center point is exactly on the center line, the default rule of left region attribution is used to enter the left side, so as to obtain the initial region distribution object. The division also stipulates the execution responsibilities: the left region is responsible for the manipulator 1, and the right region is responsible for the manipulator 2.

[0069] After the initial division is completed, the layer task quantity is statistically processed, and the layer statistical object is obtained. The statistical caliber takes the layer number F as the primary key, and the task quantities of the left and right regions in the same layer are counted respectively, which is used for subsequent load balancing decision. Then, the cross-region cooperation balancing processing is performed according to the layer statistical object, when the task quantity of the left region-layer task quantity of the right region≥2, a plurality of tasks closest to the center line are selected from the left region, and the number is “(left-right) / 2”, which is attributed to the right region; symmetrically, when the task quantity of the right region-layer task quantity of the left region≥2, “(right-left) / 2” tasks closest to the center line are selected from the right region and attributed to the left region. Through the above nearest center line task migration strategy, the dynamic balance of the dual manipulator load in the same layer is realized, and the balanced distribution object is obtained.

[0070] Based on the balanced distribution result, the tasks are temporarily stored in the corresponding task queue, and the task queue object is obtained. The left region task enters the “manipulator 1 task” queue, and the right region task enters the “manipulator 2 task” queue. The execution identification R is updated to 1 or 2 (corresponding to the manipulator 1 and the manipulator 2 respectively) by synchronously performing the annotation rewriting processing on each record structure P, so as to complete the traceable distribution mark and the consistency transmission between subsequent systems, and finally obtain the task distribution result.

[0071] In step S104, the stacking order result of each manipulator is obtained in the order from bottom to top, and the space relationship with the manipulator is sorted in the same layer to reduce the interference risk.

[0072] In an embodiment, the step S104 includes:

[0073] According to the task distribution result, the at least two manipulators are respectively extracted and layer-merged, and the to-be-sequenced object is obtained;

[0074] According to the to-be-sequenced object, a layer sequence set is generated in the order from the lower layer to the upper layer, and the layer sequence sorting result is obtained;

[0075] The layer sequence sorting result is processed from near to far to obtain a distance sorting set;

[0076] The first rule sorting and the second rule sorting are set based on the distance sorting set, and the first rule sorting and the second rule sorting are integrated to obtain a stacking sequence result.

[0077] In this embodiment, at least two mechanical hands are extracted based on the task allocation result, and the task items marked as mechanical hand 1 are aggregated into a first team, and the task items marked as mechanical hand 2 are aggregated into a second team; the tasks in each team are merged according to the layer number F to obtain the to-be-sequenced objects in units of layers, wherein each task item carries fields such as pose coordinates (X, Y, Z), attitude angle θ and layer number F, which are used for subsequent spatial sorting and conflict avoidance in the same layer.

[0078] After completing the teaming and hierarchical merging, a layer sequence set is generated in each team, and the layers are traversed in the order from the lower layer to the upper layer (F=1 is the bottom layer, and the layer number increases in turn), each task item in the same layer is first assigned an initial sequence number, and then the sequence numbers are arranged from small to large according to the unified arrangement principle, thereby obtaining the layer sequence sorting result. The setting of the layer sequence set ensures that the stacking process follows the operation rhythm of “first bottom layer, then upper layer”, avoiding the blocking or interference of the upper layer task on the lower layer task which has not been completed.

[0079] Then, the layer sequence sorting result is optimized in the distance dimension. In order to reduce the potential interference between the grabbing and placing paths and the placed materials, each team uses the distance sorting strategy of “from far to near” in the same layer: taking the reference position (or work position reference point) of each mechanical hand as the distance measurement reference, calculating the planar distance between the candidate task grabbing / placing pose and the reference position, and arranging the distance sorting set from large to small. The set serves as the first level constraint for spatial avoidance, and provides a basic sequence for the subsequent directional refinement rules.

[0080] Based on the distance-based sorting set, a first-rule sorting and a second-rule sorting are set respectively, and integrated to generate the final palletizing order result. For robot 1, a directional strategy of "right-to-left, top-to-bottom" is adopted: extract the planar coordinates (X, Y) of tasks in the same layer, and sort them first by X from largest to smallest; when the difference in X coordinates of several tasks falls within a preset threshold (e.g., ±100mm), a secondary sort is performed by Y from largest to smallest to ensure that tasks far from the robot and located on the right and top are processed first, and then gradually transition to tasks on the left and bottom. Correspondingly, for robot 2, a directional strategy of "left-to-right, top-to-bottom" is adopted: within the same layer, sort by X from smallest to largest; when the difference in X coordinates of several tasks falls within a threshold (e.g., ±100 mm), a secondary sort is performed by Y from largest to smallest, so that tasks far from the robot and located on the left and top are processed first, and then gradually transition to tasks on the right and bottom. The two sets of directional rules mentioned above are consistent with the distance sequence of "from far to near": far side priority and near side placement later, which can reduce the occlusion of subsequent paths and attitudes by placing materials first while maintaining the correctness of the layer sequence.

[0081] Finally, the directionally refined sequence is solidified into two palletizing sequence queues within each team: palletizing sequence K1 = (P1, P2, P3…Pi) for robot 1 and palletizing sequence K2 = (P1, P2, P3…Pj) for robot 2, where Pi and Pj represent the quantity of material tasks within each team. Each Pi and Pj corresponds to a task record containing the key fields (X, Y, Z, θ, F, R), ensuring field-level consistency with upstream allocation and downstream scheduling. The two palletizing sequence queues, as the output of this step, are provided to the upper-level control system for subsequent interleaving, merging, and outbound scheduling processing.

[0082] In step S105, the palletizing sequence of the two robotic arms is integrated into an outbound sorting queue according to the staggered merging rule, and the queue is transmitted to the upper-level warehouse management system for instruction processing, completing the mapping from palletizing sequence to scheduling object, which is used to drive the sequential issuance and transportation of materials.

[0083] In one embodiment, step S105 includes:

[0084] The palletizing order result is parsed to obtain a sequence parsing object;

[0085] The sequential parsing objects are processed using interleaving and merging rules to obtain the outbound sorting queue;

[0086] Generate a global sequence number for each queue item in the outbound sorting queue and inherit the corresponding hierarchical identifier and pose parameter to obtain an outbound queue object;

[0087] The warehouse-out queue object is converted into a warehouse-out instruction set and sent to the warehouse management system to obtain a warehouse-out instruction confirmation object;

[0088] Based on the warehouse-out instruction confirmation object, the warehouse management system is driven to sequentially issue materials to the conveying line in the order of the warehouse-out sorting queue, and the warehouse-out state and conveying position information of each material are returned to obtain a conveying state set;

[0089] The conveying state set is subjected to target robot determination and station mapping processing respectively to generate a scheduling object for subsequent control.

[0090] In this embodiment, the stacking sequence result is subjected to queue analysis, and the record structure P is read from the stacking sequence K1=(P1, P2…Pi) of robot 1 and the stacking sequence K2=(P1, P2…Pj) of robot 2, and the pose parameters (X, Y, Z), the attitude angle θ, the layer number F and the execution identifier R are extracted to obtain a sequence analysis object that is grouped by source and indexable by level for interleaving and instruction-based arrangement. The sequence analysis objects are merged based on the interleaving rule, and the same sequence positions of the two queues are taken as a group for interleaving and splicing. The nth item of K1 is placed in the 2n-1th position of the warehouse-out sorting queue K, and the nth item of K2 is placed in the 2nth position of the queue K to obtain K=(K1.P1, K2.P1, K1.P2, K2.P2…K1.Pi, K2.Pj). When a queue is exhausted in advance, the remaining queue is appended to the end of the queue K in its original order to obtain a complete warehouse-out sorting queue.

[0091] After obtaining the warehouse-out sorting queue, a global sequence number (incrementing from 1) is generated for each queue item in the queue and the level identifier F and the pose parameters (X, Y, Z, θ) are inherited to obtain a warehouse-out queue object that is traceable and recoverable for consistent control of cross-system transmission and state backfilling. Subsequently, the warehouse-out queue object is converted into a warehouse-out instruction set and issued to the warehouse management system, which includes the global sequence number coded according to K sequence, the target station, the conveying path and the material identifier. After the warehouse management system receives and returns an acknowledgement message, a warehouse-out instruction confirmation object is obtained.

[0092] Based on the outbound instruction confirmation object, the warehouse management system drives the material to be sequentially issued to the conveying line in the order of queue K, and the real-time outbound state (issued / in transit / arrived / abnormal) and conveying position information (current line segment, arrival time stamp, etc.) of each material are returned, and the conveying state set is obtained to support the subsequent station mapping and execution closed loop. Finally, combined with the execution identifier R in the record structure P, the target manipulator is determined and the station is mapped: when R=1, the conveying destination station of the corresponding material is mapped to the grabbing station of manipulator 1; when R=2, it is mapped to the grabbing station of manipulator 2. The system completes the shunt control and arrival verification on the WCS system (Warehouse Control System) side accordingly, and outputs the scheduling object including the four-tuple of "global sequence number-pose parameter-hierarchical identifier-target station" as the direct input for the manipulator to execute the palletizing operation in step S106.

[0093] In step S106, the palletizing operation is controlled by using the calculated scheduling object to control the multiple manipulators to execute the palletizing operation, and the outbound and palletizing of the material can be quickly completed.

[0094] In an embodiment, after step S106, the following steps are included:

[0095] The region task mapping is obtained by regionally interpreting the scheduling object;

[0096] The spatial path planning result is obtained by performing spatial separation path planning processing on the region task mapping;

[0097] The center line adjacent interference detection is performed using the spatial path planning result, and the interference identifier is marked and summarized into a candidate interference task set;

[0098] The time separation scheduling processing is performed on the candidate interference task set to obtain the peak-shifting time sequence table;

[0099] The control instruction object is obtained by performing parameter arrangement processing on the peak-shifting time sequence table and the spatial path planning result;

[0100] The manipulator is driven to execute the palletizing operation based on the control instruction object, and the task completion mark and real-time pose data are written back to the scheduling object to obtain the updated scheduling object.

[0101] In this embodiment, the scheduling object is interpreted as a region, parsing its global sequence number, layer number F, pose parameters (X / Y / Z / θ), and execution identifier R. The task is mapped into two regions, left and right, according to the tray's geometric centerline, resulting in a region task mapping. The left region is assigned to robot 1 by default, and the right region to robot 2 by default. When the material's center point falls on the centerline, it is processed according to the default rule for the left region, ensuring consistent region division. Spatial separation path planning is performed for robot 1 and robot 2 respectively for the region task mapping, limiting their respective work boundaries to only cover the half-area of ​​the tray they are responsible for. An obstacle set is established (including the tray boundary and the outline of the stacked materials), and a path segment sequence consisting of "grabbing segment - transport segment - placement segment" is generated for each task. During path search, the end-effector pose is constrained by a fixed grid or sampled waypoints, maintaining a minimum safe clearance from obstacles, and the spatial path planning result, including key waypoints, velocity / acceleration constraints, and end-effector pose, is output.

[0102] After obtaining the spatial path planning results, the system performs interference detection on tasks located in the centerline proximity zone. A width threshold is set for this zone, and spatiotemporal overlap analysis is performed on the path segments of the two manipulators within this zone. When potential intersections or overlaps in the gripper's working domain exist, the relevant tasks are immediately marked with interference indicators and aggregated into a candidate interference task set to prevent head-on or parallel conflicts in narrow areas. For the candidate interference task set, the system implements time-separated scheduling to obtain a staggered timing table. Following the "first-come, first-served" queuing principle, adjacent conflicting tasks are time-slot misaligned based on the global sequence number. When a task is executed in the centerline proximity zone, the corresponding task of the other manipulator is placed in an interlocked waiting state until the former is completed, thus transforming the originally synchronous spatial conflict into sequential time-slice execution.

[0103] Furthermore, the staggered timing table and spatial path planning results are parameterized to generate a control command object for the equipment side. This object, granular at the task level, includes fields such as target robot number, global sequence number, layer number F, grasping / placement pose (X / Y / Z / θ), path segment sequence and its velocity / acceleration parameters, interlock / waiting conditions, and release trigger events, ensuring a unified expression of the time separation strategy and spatial path at the command level. Finally, the robot is driven to perform palletizing operations according to the control command object. The control system issues path and motion parameters item by item according to the programmed timing. After the robot completes the corresponding grasping and placement, it returns a task completion mark and real-time pose data. The system writes the above feedback back to the scheduling object in a field-level manner, updating the execution status, completion timestamp, and end pose record of the corresponding task, obtaining an updated scheduling object, providing a consistent data foundation for the continuous release and anomaly handling of subsequent tasks.

[0104] In addition, in a specific implementation scenario, the hardware configuration is specifically as follows:

[0105] Tray: Length * Width: 1200mm * 1000mm, maximum load 1000kg;

[0106] Manipulator: 2 ABB IRB6700-200 / 2.6 manipulators, load 200kg;

[0107] Vision: 3D industrial camera;

[0108] Clamp: maximum load 15kg;

[0109] The implementation example specifically refers to the following table and steps:

[0110]

[0111] Step 1: Obtain the above material data through an order, and perform preprocessing and modeling;

[0112] Step 2: Heuristic rule stacking type sorting: C-A-B-D-E, stacking type material center point position task information is: C (600, 500, 350, 250, 900, 0, 1, 1, 0), A (500, 400, 300, 750, 1000, 0, 0, 1, 0), B (320, 280, 220, 160, 460, 0, 0, 1, 0), D (220, 185, 160, 610, 707, 0, 0, 1, 0), E (400, 310, 260, 200, 1045, 20, 0, 2, 0);

[0113] Step 3: Task allocation

[0114] Manipulator 1 task: C (600, 500, 350, 250, 900, 0, 1, 1, 1), B (320, 280, 220, 160, 460, 0, 0, 1, 1), E (400, 310, 260, 200, 1045, 20, 0, 2, 1);

[0115] Manipulator 2 task: A (500, 400, 300, 750, 1000, 0, 0, 1, 2), D (220, 185, 160, 610, 707, 0, 0, 1, 1);

[0116] Step 4: Manipulator task sorting

[0117] Manipulator 1 task order: C-B-E, Manipulator 2 task order: A-D;

[0118] Step 5: material out-of-warehouse sorting: C-A-B-D-E, in order to out-of-warehouse, C, B, E are transported to the mechanical hand 1 stacking station, and A, D are transported to the mechanical hand 2 stacking station;

[0119] Step 6: execute stacking.

[0120] Compared with the traditional single mechanical hand stacking mode, the double mechanical hand collaborative stacking is adopted in the present application, and under the same working condition, the overall stacking efficiency is increased by about 60%-80% compared with the single machine mode, effectively relieving the single machine beat bottleneck. With the stacking type driving operation, the target stacking type meeting high stability and high space utilization is calculated first, and then the stacking position and posture are determined according to the stacking type to execute the out-of-warehouse sorting, so as to ensure the stability of the stacking type while improving the tray space utilization.

[0121] In combination with Figure 2 As shown in the figure, Figure 2 A schematic block diagram of a heuristic-based mechanical hand hybrid stacking device provided by an embodiment of the present application, the heuristic-based mechanical hand hybrid stacking device 200, comprising:

[0122] A data modeling unit 201 is configured to obtain original information of materials in an order and perform structured modeling to obtain a structured material data set;

[0123] A data integration unit 202 is configured to perform heuristic stacking type planning processing on the structured material data set to generate candidate placement poses and hierarchical identifiers, and integrate to obtain a stacking pose planning result;

[0124] A task allocation unit 203 is configured to perform task allocation processing according to the stacking pose planning result to obtain a task allocation result;

[0125] A stacking sorting unit 204 is configured to perform stacking sorting processing on at least two mechanical hands using the task allocation result to obtain a stacking sequence result;

[0126] A parameter scheduling unit 205 is configured to generate an out-of-warehouse sorting queue according to the stacking sequence result based on an interleaving and merging rule, and map the out-of-warehouse sorting queue as a scheduling object;

[0127] An operation output unit 206 is configured to control the mechanical hand to perform stacking operation using the scheduling object.

[0128] In the embodiment, the data modeling unit 201 obtains material original information in an order and performs structured modeling to obtain a structured material data set; the data integration unit 202 performs heuristic stacking type planning processing on the structured material data set to generate a candidate placement pose and a hierarchical identifier, and integrates to obtain a stacking pose planning result; the task allocation unit 203 performs task allocation processing according to the stacking pose planning result to obtain a task allocation result; the stacking sequencing unit 204 performs stacking sequencing processing on at least two robots using the task allocation result to obtain a stacking sequence result; the parameter scheduling unit 205 generates an outbound sequencing queue based on the interleaving and merging rule according to the stacking sequence result, and maps the outbound sequencing queue as a scheduling object; and the operation output unit 206 controls the robot to perform stacking operation using the scheduling object.

[0129] In an embodiment, the data modeling unit 201 is specifically configured to:

[0130] receive the material original information and perform normalized analysis according to order line identifiers, material identifiers, and timestamps to obtain an original information object;

[0131] perform field extraction on the original information object to obtain preliminary structured data; wherein the preliminary structured data at least includes physical attributes, stacking constraints, and positioning attributes;

[0132] perform structured storage on the preliminary structured data to obtain structured storage data;

[0133] associate dimensions of the structured storage data with corresponding order lines based on a preset order data connection mechanism to generate a connection association object;

[0134] perform pallet parameter calibration on the connection association object to generate a calibration result object;

[0135] write the calibration result object back to a field set corresponding to the connection association object to obtain a structured material data set.

[0136] In an embodiment, the data integration unit 202 is specifically configured to:

[0137] establish a heuristic parameter weight table according to weight, size, and stability of the structured material data set to obtain a heuristic parameter weight table; wherein the priority of weight, size, and stability decreases in turn;

[0138] generate a hierarchical target of the material according to the heuristic parameter weight table to construct a hierarchical filling rule;

[0139] perform gap matching calculation of the material according to the heuristic parameter weight table to construct a minimum gap priority rule;

[0140] According to the heuristic parameter weight table, alignment and aggregation constraints of the material are superimposed to construct edge alignment rules;

[0141] According to the heuristic parameter weight table, a region division process of the material is performed to construct region division rules;

[0142] According to the heuristic parameter weight table, a pose adjustment is set, and a stacking coordinate parameter and a hierarchical number are established, and a material pose adjustment rule is integrated;

[0143] The hierarchical filling rule, the minimum gap priority rule, the edge alignment rule, the region division rule, and the material pose adjustment rule are integrated to obtain a stacking pose planning result.

[0144] In an embodiment, the task allocation unit 203 is specifically configured to:

[0145] The stacking pose planning result is subjected to candidate extraction and label aggregation processing to obtain a task set to be allocated;

[0146] The task set to be allocated is subjected to region division processing to obtain an initial region allocation object;

[0147] The initial region allocation object is subjected to hierarchical statistical processing to obtain a hierarchical statistical object;

[0148] The hierarchical statistical object is subjected to cross-region collaboration balancing processing to obtain a balanced allocation object;

[0149] According to the balanced allocation object, the material tasks subjected to region division and cross-region collaboration are respectively stored in corresponding task queues to obtain a task queue object;

[0150] The task queue object is subjected to label rewriting processing to obtain a task allocation result.

[0151] In an embodiment, the stacking sorting unit 204 is specifically configured to:

[0152] According to the task allocation result, at least two robots are subjected to team extraction and hierarchical merging processing to obtain a to-be-sorted object;

[0153] According to the to-be-sorted object, a layer sequence set is generated in the order from lower layer to upper layer and in the principle of increasing sequence number in each layer to obtain a layer sequence sorting result;

[0154] The layer sequence sorting result is subjected to near-to-far sorting processing to obtain a distance sorting set;

[0155] Based on the distance sorting set, a first rule sorting and a second rule sorting are respectively set, and the first rule sorting and the second rule sorting are integrated to obtain a stacking sequence result.

[0156] In an embodiment, the parameter scheduling unit 205 is specifically configured to:

[0157] perform queue analysis on the stacking sequence result to obtain a sequence analysis object;

[0158] perform interleaving and merging rule processing on the sequence analysis object to obtain a warehouse-out sorting queue;

[0159] generate a global sequence number for each queue item in the warehouse-out sorting queue and inherit the corresponding hierarchical identifier and pose parameter to obtain a warehouse-out queue object;

[0160] convert the warehouse-out queue object into a warehouse-out instruction set and send it to a warehouse management system to obtain a warehouse-out instruction confirmation object;

[0161] drive the warehouse management system to sequentially issue materials to a conveying line in the order of the warehouse-out sorting queue based on the warehouse-out instruction confirmation object, and return the warehouse-out state and conveying position information of each item of material to obtain a conveying state set;

[0162] respectively perform target robot judgment and station mapping processing on the conveying state set to generate a scheduling object for subsequent control.

[0163] In an embodiment, the heuristic-based robot hybrid stacking device 200 is further configured to:

[0164] perform region interpretation on the scheduling object to obtain a region task mapping;

[0165] perform spatial separation path planning processing on the region task mapping to obtain a spatial path planning result;

[0166] perform center line proximity interference detection using the spatial path planning result, label interference identifiers, and aggregate them into a candidate interference task set;

[0167] perform time separation scheduling processing on the candidate interference task set to obtain a peak-shaving time sequence table;

[0168] perform parameter arrangement processing on the peak-shaving time sequence table and the spatial path planning result to obtain a control instruction object;

[0169] drive the robot to perform stacking operations based on the control instruction object, and write the task completion marker and real-time pose data back to the scheduling object to obtain an updated scheduling object.

[0170] Since the embodiments of the device part correspond to the embodiments of the method part, the embodiments of the device part are described in the description of the embodiments of the method part, which will not be described here.

[0171] The embodiment of the present application further provides a computer readable storage medium, which has a computer program stored thereon, and the computer program can realize the steps provided by the above embodiment when being executed. The storage medium can include a U disk, a mobile hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk and various storage medium capable of storing program codes.

[0172] The embodiment of the present application further provides a computer device, which can include a memory and a processor, the memory has a computer program stored therein, and the processor can realize the steps provided by the above embodiment when calling the computer program in the memory. Of course, the computer device can further include various network interfaces, power supplies, graphic card devices and the like, and the model can be operated by means of the performance of the graphic card, such as inference and training.

[0173] The embodiments in the specification are described in a progressive manner, and each embodiment focuses on the difference from other embodiments. The same or similar parts of each embodiment can be referred to each other. For the system disclosed by the embodiments, since it corresponds to the method disclosed by the embodiments, the description is relatively simple, and the related parts can be referred to the method part. It should be pointed out that, for those skilled in the art, without departing from the principles of the present application, some improvements and modifications can be made to the present application, and these improvements and modifications also fall within the protection scope of the claims of the present application.

[0174] It should be further noted that in the specification, the relationship terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between the entities or operations. Moreover, the terms "include", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or device. Without more limitations, the element defined by the statement "including a" does not exclude the presence of another identical element in the process, method, article or device including the element.

Claims

1. A heuristic-based robot mixed palletizing method, characterized by, The method comprises the following steps: obtaining raw information of materials in an order and structurally modeling to obtain a structured material data set; performing heuristic stacking type planning processing on the structured material data set to generate candidate placement poses and hierarchical identifiers, and integrating to obtain a stacking pose planning result; performing task allocation processing according to the stacking pose planning result to obtain a task allocation result; performing stacking sequence processing on at least two robots using the task allocation result to obtain a stacking sequence result; generating an outbound sorting queue according to the stacking sequence result based on an interleaving and merging rule, and mapping the outbound sorting queue as a scheduling object; controlling the robot to perform stacking operation using the scheduling object; The heuristic stacking type planning processing on the structured material data set to generate candidate placement poses and hierarchical identifiers, and integrating to obtain a stacking pose planning result comprises: establishing a heuristic parameter weight table according to the weight, size and stability of the structured material data set to obtain a heuristic parameter weight table; wherein the priority of weight, size and stability decreases in turn; generating hierarchical targets of the materials according to the heuristic parameter weight table to construct a hierarchical filling rule; calculating the gap matching of the materials according to the heuristic parameter weight table to construct a minimum gap priority rule; superimposing the alignment and aggregation constraints of the materials according to the heuristic parameter weight table to construct an edge alignment rule; performing regional division processing of the materials according to the heuristic parameter weight table to construct a regional division rule; setting the posture adjustment according to the heuristic parameter weight table, and establishing stacking coordinate parameters and hierarchical numbers to integrate to obtain a material posture adjustment rule; and integrating the hierarchical filling rule, the minimum gap priority rule, the edge alignment rule, the regional division rule and the material posture adjustment rule to obtain a stacking pose planning result; The task allocation processing according to the stacking pose planning result to obtain a task allocation result comprises: performing candidate extraction and identifier aggregation processing on the stacking pose planning result to obtain a set of tasks to be allocated; performing regional division processing on the set of tasks to be allocated to obtain an initial regional allocation object; performing hierarchical statistical processing on the initial regional allocation object to obtain a hierarchical statistical object; performing cross-region cooperation balancing processing on the hierarchical statistical object to obtain a balanced allocation object; temporarily storing the material tasks after regional division and cross-region cooperation to corresponding task queues according to the balanced allocation object to obtain a task queue object; and performing label rewriting processing on the task queue object to obtain a task allocation result. The palletizing sequence result is obtained by using the task allocation result to sort at least two mechanical hands, including: according to the task allocation result, the at least two mechanical hands are respectively subjected to team extraction and hierarchical merging processing to obtain a to-be-sequenced object; according to the to-be-sequenced object, a layer sequence set is generated in the order from a lower layer to an upper layer and in the order of a serial number from small to large in each layer to obtain a layer sequence sorting result; the layer sequence sorting result is subjected to a near-to-far sorting processing to obtain a distance sorting set; based on the distance sorting set, a first rule sorting and a second rule sorting are respectively set, and the first rule sorting and the second rule sorting are integrated to obtain the palletizing sequence result.

2. The heuristic based robot mixed case palletizing method of claim 1, wherein, The material original information in the order is obtained and structured modeling is performed to obtain a structured material data set, including: Receiving the material original information and performing normalized analysis according to order line identification, material identification and timestamp to obtain an original information object; Field extraction is performed on the original information object to obtain preliminary structured data; wherein the preliminary structured data at least includes physical properties, palletizing constraints and positioning attributes; The preliminary structured data is stored in a structured manner to obtain structured storage data; Based on a preset order data connection mechanism, the dimensional fields of the structured storage data are associated with the corresponding order lines to generate a connection association object; The connection association object is subjected to a cargo equipment parameter calibration to generate a calibration result object; The calibration result object is written back to the field set corresponding to the connection association object to obtain the structured material data set.

3. The heuristic based robot mixed case palletizing method of claim 1, wherein, The out-of-warehouse sorting queue is generated based on the interleaving merging rule according to the palletizing sequence result, and the out-of-warehouse sorting queue is mapped to a scheduling object, including: The palletizing sequence result is subjected to queue analysis to obtain a sequence analysis object; The sequence analysis object is subjected to interleaving merging rule processing to obtain an out-of-warehouse sorting queue; A global sequence number is generated for each queue item in the out-of-warehouse sorting queue and the corresponding hierarchical identifier and pose parameter are inherited to obtain an out-of-warehouse queue object; The out-of-warehouse queue object is converted into an out-of-warehouse instruction set and sent to a warehouse management system to obtain an out-of-warehouse instruction confirmation object; Based on the out-of-warehouse instruction confirmation object, the warehouse management system is driven to sequentially issue materials to a conveying line according to the order of the out-of-warehouse sorting queue, and the out-of-warehouse state and conveying position information of each material are returned to obtain a conveying state set; The conveying state set is subjected to target mechanical hand determination and station mapping processing to generate a scheduling object for subsequent control.

4. The heuristic-based robot mixed-case palletizing method of claim 1, wherein, After the scheduling object is used to control the mechanical hand to perform the palletizing operation, including: The scheduling object is subjected to area interpretation to obtain an area task mapping; The area task mapping is subjected to spatial separation path planning processing to obtain a spatial path planning result; The spatial path planning result is used for center line adjacent interference detection, interference identifiers are marked and summarized into a candidate interference task set; The candidate interference task set is subjected to time separation scheduling processing to obtain a peak-shifting time sequence table; The staggered timing table is parameterized with the space path planning result to obtain a control instruction object; Based on the control instruction object, a robot is driven to perform a stacking operation, and a task completion mark and real-time pose data are written back to the scheduling object to obtain an updated scheduling object.

5. A heuristic based robot hybrid palletizing device, characterized by, Comprise: A data modeling unit is configured to obtain raw material information in an order and perform structured modeling to obtain a structured material data set; A data integration unit is configured to perform heuristic stacking type planning processing on the structured material data set to generate candidate placement poses and hierarchical identifiers, and integrate to obtain a stacking pose planning result; A task allocation unit is configured to perform task allocation processing according to the stacking pose planning result to obtain a task allocation result; A stacking sequencing unit is configured to perform stacking sequencing processing on at least two robots using the task allocation result to obtain a stacking sequence result; A parameter scheduling unit is configured to generate an outbound sequencing queue based on an interleaving and merging rule according to the stacking sequence result, and map the outbound sequencing queue to a scheduling object; An operation output unit is configured to control a robot to perform a stacking operation using the scheduling object; The data integration unit is specifically configured to establish a heuristic parameter weight table according to weight, size, and stability of the structured material data set to obtain a heuristic parameter weight table; Wherein the priority of weight, size, and stability decreases in turn; hierarchical target generation of the material is performed according to the heuristic parameter weight table to construct a hierarchical filling rule; gap matching calculation of the material is performed according to the heuristic parameter weight table to construct a minimum gap priority rule; alignment and aggregation constraints of the material are superimposed according to the heuristic parameter weight table to construct an edge alignment rule; region division processing of the material is performed according to the heuristic parameter weight table to construct a region division rule; pose adjustment is set according to the heuristic parameter weight table, and stacking coordinate parameters and hierarchical numbers are established to integrate to obtain a material pose adjustment rule; the hierarchical filling rule, the minimum gap priority rule, the edge alignment rule, the region division rule, and the material pose adjustment rule are integrated to obtain a stacking pose planning result; The task allocation unit is specifically configured to perform candidate extraction and identifier aggregation processing on the stacking pose planning result to obtain a set of tasks to be allocated; region division processing is performed on the set of tasks to be allocated to obtain an initial region allocation object; hierarchical statistical processing is performed on the initial region allocation object to obtain a hierarchical statistical object; cross-region collaboration balancing processing is performed on the hierarchical statistical object to obtain a balanced allocation object; the material tasks after region division and cross-region collaboration are respectively stored in corresponding task queues according to the balanced allocation object to obtain a task queue object; and annotation write-back processing is performed on the task queue object to obtain a task allocation result; The palletizing sequencing unit is specifically configured to perform team extraction and hierarchical merging processing on the at least two mechanical arms respectively according to the task allocation result, to obtain a to-be-sequenced object; generate a layer sequence set according to the to-be-sequenced object in the order from a lower layer to an upper layer and according to the principle that the sequence number in each layer is from small to large, to obtain a layer sequence sequencing result; perform near-to-far sequencing processing on the layer sequence sequencing result, to obtain a distance sequencing set; set a first rule sequencing and a second rule sequencing respectively based on the distance sequencing set, and integrate the first rule sequencing and the second rule sequencing, to obtain a palletizing sequence result.

6. A computer device, comprising: A computer program product comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the heuristic-based mechanical arm hybrid palletizing method according to any one of claims 1 to 4 when executing the computer program.

7. A computer readable storage medium characterized in that, The computer program is stored on the computer readable storage medium, and the computer program is executed by the processor to implement the heuristic-based mechanical arm hybrid palletizing method according to any one of claims 1 to 4.

Images