Corrugated board finished product multi-specification automatic sorting and intelligent scheduling method and system

Abstract

The invention relates to the technical field of corrugated board production, in particular to a corrugated board finished product multi-specification automatic sorting and intelligent scheduling method which comprises the following steps: acquiring an original paperboard image on a corrugated board production line, extracting characteristic specification parameters of paperboards according to the original paperboard image, and generating a paperboard attribute data set; the method comprises the following steps: acquiring spare capacity data and downstream process demand information of each outlet of a sorting line, and calculating a sorting target position priority weight according to the spare capacity data and the downstream process demand information; through multi-link collaborative optimization, the problems existing in the prior art are solved in a targeted manner, resource idling and emergency order delay are effectively avoided, the problem that transportation delay and equipment performance are not matched due to a static path is solved, the continuity of cache region circulation and preferential execution of emergency transfer tasks are ensured, and the service life of the cache region is prolonged. Finally, high efficiency, accuracy and stability of the whole process of sorting and dispatching the finished corrugated boards are achieved.

Description

Technical Field

[0001] This invention relates to the field of corrugated cardboard production technology, and in particular to a method and system for automatic sorting and intelligent scheduling of multi-specification finished corrugated cardboard products. Background Technology

[0002] In the corrugated cardboard production sector, with the packaging industry's increasing demands for product specification diversity, production efficiency, and delivery timeliness, the sorting and scheduling of finished corrugated cardboard products has become a crucial link between production and downstream processes. Currently, in mainstream corrugated cardboard production processes, after cardboard is processed and formed on an assembly line, it must first be sorted by specifications using sorting equipment. Then, it is transported by AGVs (Automated Guided Vehicles) or RGVs (Automated Guided Vehicles) to downstream processes such as stacking, packaging, or loading. The entire process requires coordinated execution based on multi-dimensional information including production needs, equipment status, and the workshop environment. The sorting stage typically relies on robotic arms and visual recognition technology to identify and sort cardboard specifications. The scheduling stage allocates tasks and plans routes based on transfer requests and the status of transport vehicles to achieve efficient flow of cardboard from sorting to downstream processes.

[0003] However, existing sorting and scheduling technologies for corrugated cardboard products still have many unresolved issues in practical applications, making it difficult to fully adapt to the production demands of multiple specifications and high timeliness. On the one hand, the priority determination of sorting target locations does not fully integrate the real-time available capacity of the sorting line exit with the dynamic needs of downstream processes (such as order delivery time and production cycle time), which can easily lead to over-utilization of some exit spaces while some exit resources remain idle, or delivery delays due to failure to prioritize urgent orders. On the other hand, the path planning process mostly generates fixed paths based on static environmental maps of the workshop, without adjusting them in real time according to the workshop's traffic heat map. This can easily lead to transportation delays when AGVs / RGVs enter highly congested areas, and the path generation does not fully consider equipment motion parameters (such as maximum speed and turning radius), which may result in a mismatch between the path and equipment performance, increasing the risk of equipment failure. Summary of the Invention

[0004] The main objective of this invention is to provide an automatic sorting and intelligent scheduling method and system for multi-specification corrugated cardboard finished products, aiming to solve the technical problems mentioned in the background art.

[0005] This invention proposes an automatic sorting and intelligent scheduling method for multi-specification corrugated cardboard finished products, characterized by comprising: Obtain original images of corrugated cardboard from the production line, extract characteristic specification parameters of the cardboard based on the original images, and generate a cardboard attribute dataset. Obtain the available capacity data and downstream process demand information of each exit of the sorting line, and calculate the priority weight of the sorting target position based on the available capacity data and the downstream process demand information; Based on the cardboard attribute dataset and the sorting target location priority weight, the sorting robot arm's gripping instructions and joint movement instructions are generated. Obtain the status data of the buffer area at the output end of the sorting line, obtain the cardboard stacking height based on the buffer area status data, and generate a transfer request containing cardboard specifications and quantity information when the cardboard stacking height reaches a preset threshold. Obtain transfer requests and real-time status data of AGV / RGV, and generate a task queue based on the transfer requests and the real-time status data of AGV / RGV; Obtain a workshop environment map and a workshop traffic heat map, generate a planned path based on the task queue, the workshop environment map, and the workshop traffic heat map, and schedule the corrugated cardboard based on the planned path.

[0006] Preferably, the step of calculating the priority weight of the sorting target location based on the available capacity data and the downstream process demand information includes: Calculate the space utilization rate of each exit based on the available capacity data of each exit of the sorting line; Extract order delivery time and production cycle parameters based on downstream process demand information; Calculate the space adaptation coefficient based on the space utilization rate; Calculate the time urgency factor based on the order delivery time and the production cycle parameters; Obtain the spatial adaptation coefficient and the time urgency coefficient, and calculate the initial priority weight using a weighted algorithm; Historical sorting efficiency data is obtained, and the initial priority weight is dynamically adjusted based on the historical sorting efficiency data to obtain the priority weight of the sorting target location.

[0007] Preferably, the step of generating the gripping instructions and joint movement instructions of the sorting robot arm based on the cardboard attribute dataset and the sorting target position priority weights includes: Obtain the size, weight parameters, and corrugation parameters of the cardboard attribute dataset. Determine the gripping force and number of gripping points of the robotic arm based on the size and weight parameters, and determine the placement angle and gentle placement buffer parameters of the robotic arm based on the corrugation parameters. The gripping force, the number of gripping points, the placement angle, and the gentle placement buffer parameters are integrated to generate the gripping instructions for the sorting robot arm; Based on the priority weight of the sorting target location, select the multiple candidate landing points with the highest weights; The motion trajectory parameters of the robotic arm are calculated based on the spatial coordinates of multiple candidate landing points, and joint motion commands for the sorting robotic arm are generated based on the motion trajectory parameters.

[0008] Preferably, the step of obtaining the cardboard stack height based on the buffer zone status data, and generating a transfer request containing cardboard specifications, quantity, and priority when the cardboard stack height reaches a preset threshold includes: Obtain a preset threshold for the height of the cardboard stack, wherein the preset threshold includes a first preset threshold and a second preset threshold; Compare the cardboard stack height with the preset threshold: When the height of the cardboard stack reaches a first preset threshold, a pre-reminder signal is generated and transmitted to the AGV / RGV scheduling center; When the height of the cardboard stack reaches the second preset threshold, the number of cardboard pieces and their corresponding specifications in the buffer area are counted. Obtain statistical results by comparing the cardboard stack height with the preset threshold, and generate a transfer request containing cardboard specifications, quantity, and priority based on the statistical results.

[0009] Preferably, the step of generating a task queue based on the transfer request and the real-time status data of the AGV / RGV includes: Determine the required type and quantity of transport vehicles based on the transfer request; Acquire real-time status data of AGV / RGV, including current location, power level, load status, and fault information; Based on the type of transport vehicle and the real-time status data of AGV / RGV, available transport equipment is selected; Obtain the priority of each of the aforementioned transfer requests and sort them from high to low priority; The sorted requests are assigned to specific AGVs / RGVs based on the required number of transport vehicles; Generate a task queue containing task number, target location, and execution time limit.

[0010] Preferably, the step of generating a planned route based on the task queue, the workshop environment map, and the workshop traffic heat map includes: The coordinates of passageways, intersections, and obstacles are extracted from the workshop environment map. Based on the workshop traffic heat map, identify high-congestion areas and smooth-flowing areas; Generate an initial path based on the target location in the task queue; Obtain congestion data for the areas traversed by the initial path, and adjust the weight of the initial path accordingly; Obtain the device motion parameters of the AGV / RGV; Based on the adjusted initial path and the device motion parameters, a planned path containing speed commands is generated.

[0011] Preferably, the step of generating an initial path based on the target location in the task queue includes: Obtain the current position coordinates of the AGV / RGV and the target position in the task queue, and determine the path start point and path end point based on the current position coordinates of the AGV / RGV and the target position; A grid model is constructed based on the workshop environment map, and the locations of passable nodes and obstacles are marked to determine the path search range; Obtain the current node, and define a cost function to evaluate the path cost based on the actual distance from the path start point to the current node and the estimated Manhattan distance from the current node to the path end point; The open list and closed list are initialized, the starting point of the path is added to the open list, and the cost function value corresponding to the starting point is calculated according to the cost function. Iteratively select the node with the smallest cost function value in the open list, expand the adjacent passable nodes of that node and update their costs, until the end of the path is found; An initial path consisting of consecutive coordinate points is generated by tracing back the parent-child relationships between nodes from the end point of the path to the beginning point of the path.

[0012] This invention also provides an automated sorting and intelligent scheduling system for multi-specification corrugated cardboard finished products, comprising: The data acquisition module acquires original images of corrugated cardboard on the production line, extracts characteristic specification parameters of the cardboard based on the original images, and generates a cardboard attribute dataset. The weight calculation module obtains the available capacity data of each exit of the sorting line and the demand information of downstream processes, and calculates the priority weight of the sorting target position based on the available capacity data and the demand information of downstream processes. The collaborative control module generates gripping instructions and joint movement instructions for the sorting robotic arm based on the cardboard attribute dataset and the priority weight of the sorting target position. The request generation module obtains the buffer status data at the output end of the sorting line, obtains the cardboard stacking height based on the buffer status data, and generates a transfer request containing cardboard specifications and quantity information when the cardboard stacking height reaches a preset threshold. The task generation module acquires transfer requests and real-time status data of AGVs / RGVs, and generates a task queue based on the transfer requests and the real-time status data of AGVs / RGVs. The scheduling module acquires a workshop environment map and a workshop traffic heat map, generates a planned path based on the task queue, the workshop environment map, and the workshop traffic heat map, and schedules the corrugated cardboard based on the planned path.

[0013] The present invention also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, characterized in that the processor executes the computer program to implement the steps of an automatic sorting and intelligent scheduling method for multi-specification corrugated cardboard finished products.

[0014] The present invention also provides a computer-readable storage medium having a computer program stored thereon, characterized in that, when the computer program is executed by a processor, it implements the steps of an automatic sorting and intelligent scheduling method for multi-specification corrugated cardboard finished products.

[0015] The beneficial effects of this invention are as follows: This invention addresses the problems existing in the prior art through multi-stage collaborative optimization. In determining the priority of sorting target locations, it calculates the space adaptation coefficient by combining the available capacity data of each sorting line exit, and calculates the time urgency coefficient by combining the order delivery time and production cycle time of downstream processes. Then, it dynamically corrects the priority weight using a weighted algorithm and historical sorting efficiency data, effectively avoiding resource idleness and delays in urgent orders, and improving the utilization rate of sorting line exit space and the response speed of urgent orders. In path planning, it first extracts physical constraint information from the workshop environment map, then identifies congested areas by combining the workshop traffic heat map. After generating an initial path based on the task queue, it adjusts the path weight according to congestion data, and generates a planned path containing speed commands by combining the motion parameters of the AGV / RGV equipment. This solves the problem of transportation delays and equipment performance mismatch caused by static paths, ensuring the smooth operation of AGV / RGV equipment. Efficiency and safety of operation; When triggering transfer requests in the buffer area, a first preset threshold and a second preset threshold are set. When the first threshold is reached, a pre-reminder signal is generated, and when the second threshold is reached, a transfer request with priority is generated. This avoids buffer area overflow and delays in emergency tasks, ensures the continuity of buffer area flow and the priority execution of emergency transfer tasks, and ultimately achieves high efficiency, accuracy and stability in the entire process of corrugated cardboard finished product sorting and scheduling. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of a method flow according to an embodiment of this application.

[0017] Figure 2 This is a schematic diagram of the system structure according to an embodiment of this application.

[0018] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0020] like Figure 1 As shown, this application provides a method for automatic sorting and intelligent scheduling of corrugated cardboard finished products in multiple specifications, including: S1, Obtain the original image of the corrugated cardboard on the production line, extract the characteristic specification parameters of the cardboard based on the original image, and generate a cardboard attribute dataset; S2, obtain the available capacity data and downstream process demand information of each exit of the sorting line, and calculate the priority weight of the sorting target position based on the available capacity data and the downstream process demand information. S3, Generate the gripping instructions and joint movement instructions of the sorting robot arm based on the cardboard attribute dataset and the sorting target position priority weight; S4, obtain the status data of the buffer area at the output end of the sorting line, obtain the cardboard stacking height based on the buffer area status data, and generate a transfer request containing cardboard specifications and quantity information when the cardboard stacking height reaches a preset threshold. S5, acquire the transfer request and the real-time status data of the AGV / RGV, and generate a task queue based on the transfer request and the real-time status data of the AGV / RGV; S6, obtain the workshop environment map and workshop traffic heat map, generate a planned path based on the task queue, the workshop environment map and the workshop traffic heat map, and schedule the corrugated cardboard based on the planned path.

[0021] As described in steps S1-S6 above, the present invention can realize the fully automated collaborative scheduling of corrugated cardboard from production line output to AGV / RGV transfer, solving the problems of high dependence on manual labor, disordered resource allocation, poor equipment action adaptability, and low transfer efficiency in traditional sorting and scheduling, and improving the overall accuracy and efficiency of sorting and transferring multi-specification cardboard.

[0022] Specifically: The first step involves acquiring raw images of corrugated cardboard from the production line. Based on these raw images, characteristic specifications are extracted, and a cardboard attribute dataset is generated. This is achieved by using a high-resolution 3D camera array on the production line to capture the raw images. The images are then processed by a visual recognition system (noise reduction, edge detection, texture matching, etc.) to extract parameters such as size, thickness, flute type, and load-bearing capacity, generating the cardboard attribute dataset. Machine vision replaces manual labor, significantly improving recognition efficiency and accuracy, and avoiding errors in subsequent sorting decisions. Specifically, the raw images are captured by a 3D camera covering the entire width of the production line, with the sampling frequency matched to the production line speed to ensure complete imaging of each cardboard sheet. The load-bearing capacity in the cardboard attribute dataset is derived from thickness, material density (preset industry standard values), and mechanical formulas.

[0023] The second step involves acquiring the available capacity data for each exit of the sorting line and the demand information for downstream processes. Based on the available capacity data and the demand information for downstream processes, the priority weight of the sorting target location is calculated. This is done by first acquiring the available capacity data for each exit of the sorting line (from the infrared sensor in the buffer area) and the demand information for downstream processes (extracted from the MES system, including order delivery time and production cycle time). Then, the space adaptation coefficient (negatively correlated with the exit space utilization rate) and the time urgency coefficient (negatively correlated with the remaining delivery time) are calculated based on this. The initial priority weight is obtained through a weighted algorithm. Finally, it is dynamically corrected by combining historical sorting efficiency data to ensure that the sorting location allocation is accurately matched with the capacity and demand, thereby improving the efficiency of downstream process connection.

[0024] The third step involves generating gripping and joint movement commands for the sorting robotic arm based on the cardboard attribute dataset and the priority weights of the sorting target positions. This addresses the issues of cardboard damage and landing point deviation caused by the fixed movements of traditional robotic arms. This step determines the gripping force and number of gripping points based on the cardboard size and weight, and determines the placement angle and gentle placement buffer parameters based on the flute shape, integrating these to generate gripping commands. Simultaneously, candidate landing point positions are filtered according to priority weights, the robotic arm's motion trajectory parameters are calculated, and joint movement commands are generated. This ensures that the robotic arm's movements are precisely matched to the cardboard specifications and landing point positions, reducing cardboard damage and improving gripping success rate.

[0025] The fourth step involves acquiring the status data of the buffer zone at the sorting line output end. Based on this data, the cardboard stacking height is determined. When the cardboard stacking height reaches a preset threshold, a transfer request containing cardboard specifications and quantity information is generated, avoiding the accumulation risk caused by traditional single-threshold warnings. This step first acquires a stacking height standard containing first and second preset thresholds (set based on the total height of the buffer zone and the AGV / RGV response time). Then, the real-time stacking height is acquired using a laser ranging sensor and compared with the thresholds for tiered processing. When the first threshold is reached, a pre-warning signal is sent to schedule the AGV / RGV in advance. When the second threshold is reached, the cardboard quantity and specifications are counted, and a priority transfer request is generated to ensure no accumulation in the buffer zone and avoid sorting line downtime.

[0026] The fifth step involves acquiring transfer requests and real-time status data of AGVs / RGVs. Based on these data, a task queue is generated to address the issue of uneven equipment workload in traditional scheduling. This step first determines the type and quantity of carriers according to the load-bearing capacity of the cardboard in the transfer request (e.g., RGVs are used for load-bearing capacities ≥ B, and AGVs are used otherwise). Then, the real-time status of the equipment (location, battery level, load, and fault information from the UWB positioning and vehicle-mounted system) is acquired. After filtering available equipment, tasks are assigned according to request priority, generating a queue containing task numbers, target locations, and execution time limits. This achieves a balanced match between equipment and tasks, improving equipment utilization and task completion rates.

[0027] Step 6: Obtain the workshop environment map and workshop traffic heat map. Generate a planned route based on the task queue, the workshop environment map, and the workshop traffic heat map. Schedule the corrugated cardboard based on the planned route to solve the transportation delay problem caused by ignoring congestion in traditional fixed routes. This step first extracts the workshop environment map (a pre-built digital twin map, including passageways, intersections, and obstacles) and the traffic heat map (generated based on real-time equipment location, marking congested / smooth areas). After generating an initial route through an algorithm, the route weight is adjusted based on congestion data. Then, AGV / RGV motion parameters (maximum speed, turning radius) are incorporated to generate a planned route with speed commands, ensuring efficient and collision-free transportation and shortening transportation time.

[0028] In summary, this invention forms a complete process of data acquisition, decision calculation, instruction generation, status monitoring, task allocation, and path planning, solving the pain points of traditional sorting and scheduling, and improving the automation level and overall efficiency of sorting and transferring multi-specification corrugated cardboard.

[0029] In one embodiment of the present invention, the step of calculating the priority weight of the sorting target location based on the available capacity data and the downstream process demand information includes: S21, Calculate the space utilization rate of each exit based on the available capacity data of each exit of the sorting line; S22, extract order delivery time and production cycle parameters based on downstream process demand information; S23, Calculate the space adaptation coefficient based on the space utilization rate; S24, Calculate the time urgency coefficient based on the order delivery time and the production cycle parameters; S25, obtain the spatial adaptation coefficient and the time urgency coefficient, and calculate the initial priority weight using a weighted algorithm; S26, Obtain historical sorting efficiency data, and dynamically adjust the initial priority weight based on the historical sorting efficiency data to obtain the priority weight of the sorting target location.

[0030] As described in steps S21-S26 above, this invention combines the actual capacity of each exit of the sorting line with the urgency of downstream process requirements to generate scientific and dynamic priority weights, providing a precise basis for the allocation of cardboard sorting positions. This solves the technical problem in traditional sorting scheduling where "exits are allocated only in a fixed order, without linkage between capacity and demand, resulting in some exits accumulating and some exits being idle, thus affecting the efficiency of downstream process connections," ensuring efficient matching of sorting resources and production needs.

[0031] In traditional corrugated cardboard sorting scenarios, the allocation of sorting target locations often relies on manual experience or fixed rotation rules. For example, cardboard is allocated sequentially according to the sorting line exit number. This approach neither considers the available capacity of each exit—if an exit already has a large amount of cardboard piled up, continuing to allocate new cardboard will exacerbate the pile-up and may even cause the sorting line to stop—nor does it consider the needs of downstream processes—if a downstream stacking process urgently needs to process cardboard of a specific size to meet order delivery, but cannot obtain that cardboard first due to disordered exit allocation, it will delay order delivery. To address these issues, this invention achieves precise optimization step by step through a layered design of "data collection - parameter calculation - weight generation - dynamic correction." First, the space utilization rate is calculated based on the available capacity data of each exit of the sorting line, providing a basic indicator for subsequent assessment of the exit's accepting capacity. The available capacity data for each exit of the sorting line comes from infrared ranging sensors installed on the side walls of each exit buffer area. These sensors continuously monitor the height of stacked cardboard within the buffer area and the total height of the buffer area, calculating the available capacity using the formula: "Available Capacity = (Total Buffer Height - Stacked Height) × Buffer Area". The data is updated every 2 seconds and transmitted to the control system. For example, if the total height of the buffer area at the exit of a sorting line is 2m and the base area is 1.5㎡, and the current stacked height is 0.8m, then the available capacity at this exit is (2-0.8)×1.5=1.8m³. The calculation logic for space utilization rate is "space utilization rate = 1-(available capacity / total buffer area capacity)", where the total buffer area capacity = total height × base area. Using the above example, the total buffer area capacity is 2×1.5=3m³, therefore the space utilization rate = 1-(1.8 / 3)=0.4, or 40%. This step transforms the abstract "available capacity" into the concrete "space utilization rate", making it easier to intuitively judge the acceptance capacity of each exit. The lower the space utilization rate, the stronger the exit's ability to accept new cardboard, providing a direct basis for the subsequent calculation of the space adaptation coefficient.

[0032] Next, we focus on extracting order delivery time and production cycle parameters from the downstream process demand information to clarify the urgency and processing capacity of the downstream demand for cardboard. The downstream process demand information here comes from the factory's Production Management System (MES system), which stores order information and production plans for each downstream process (such as stacking, packaging, and loading). The order delivery time refers to the final delivery deadline for the orders corresponding to the cardboard that the process needs to process. For example, if a batch of cardboard to be processed by the downstream stacking process requires packaging and loading within 3 hours, then the order delivery time for this process is "3 hours remaining." The production cycle parameter refers to the number of cardboard sheets that the downstream process can process per unit of time. For example, if the stacking process can process 60 sheets of cardboard per hour, the production cycle parameter is "60 sheets / hour." By extracting these two parameters, we can clarify the "time urgency" (the closer the delivery time, the more urgent) and "processing capacity limit" (the production cycle determines the rate at which the process can receive cardboard) of the downstream processes, providing key input for subsequent calculations of the time urgency coefficient.

[0033] Then, based on the space utilization rate calculated in step S21, the space adaptation coefficient is calculated to quantify the degree of adaptation of the outlet to accept new cardboard. The core logic of the space adaptation coefficient is that "the lower the space utilization rate, the stronger the outlet's ability to accept new cardboard, and the higher the adaptation coefficient." Its calculation formula is preset as "space adaptation coefficient = 1 - space utilization rate," ensuring that the parameter value range is between 0 and 1, which facilitates subsequent weighted calculations. For example, if the space utilization rate of an outlet in step S21 is 40% (i.e., 0.4), then the space adaptation coefficient of that outlet = 1 - 0.4 = 0.6; if the space utilization rate of another outlet is 70%, then the space adaptation coefficient = 1 - 0.7 = 0.3. Through this calculation, the space utilization rate is transformed into a coefficient that directly reflects the "outlet adaptability." Outlets with higher adaptation coefficients are more likely to be prioritized in subsequent priority weight calculations, effectively avoiding the allocation of too much cardboard to high-utilization (low adaptability) outlets, which could lead to accumulation.

[0034] Next, the time urgency coefficient is calculated by combining the order delivery time and production cycle parameters extracted in step S22 to quantify the urgency of the downstream processes' demand for cardboard. The calculation of the time urgency coefficient needs to consider both "remaining delivery time" and "production cycle matching degree": First, the basic urgency coefficient is calculated based on the order delivery time, with the formula "basic urgency coefficient = 1 / remaining delivery hours". For example, if the remaining delivery time is 3 hours, the basic urgency coefficient = 1 / 3 ≈ 0.33; if the remaining delivery time is 1 hour, the basic urgency coefficient is 1, and the urgency is significantly increased. Then, it is corrected by combining the production cycle parameters. If the production cycle of the downstream process is higher than the average level (for example, the factory's average production cycle is 50 sheets / hour, and this process is 60 sheets / hour), it indicates that its processing capacity is stronger, and the urgency coefficient can be appropriately increased to match its processing efficiency. The corrected formula is "time urgency coefficient = basic urgency coefficient × (production cycle of this process / average production cycle)". Using the above example, the time urgency coefficient = 0.33 × (60 / 50) ≈ 0.4; if the production cycle is lower than the average level, the urgency coefficient is reduced according to the same logic. This step uses a dual-parameter correction method to ensure that orders with tight delivery times are prioritized while also taking into account the actual processing capacity of downstream processes. This avoids allocating excessively urgent cardboard to processes with insufficient processing capacity, which could lead to process overload.

[0035] Next, a weighted algorithm is used to combine the spatial adaptation coefficient from step S23 with the time urgency coefficient from step S24 to calculate the initial priority weight. This weighted algorithm presets fixed weight coefficients α and β, where α is the weight of the spatial adaptation coefficient and β is the weight of the time urgency coefficient, satisfying α + β = 1. Considering the actual requirement in the corrugated cardboard sorting scenario that "capacity assurance takes precedence over urgent demand," α is preset to 0.6 and β to 0.4. The formula for calculating the initial priority weight is "initial priority weight = α × spatial adaptation coefficient + β × time urgency coefficient." For example, if the spatial adaptation coefficient of a certain exit is 0.6 and the corresponding time urgency coefficient of the downstream process is 0.4, then the initial priority weight = 0.6 × 0.6 + 0.4 × 0.4 = 0.36 + 0.16 = 0.52; if the spatial adaptation coefficient of another exit is 0.3 and the time urgency coefficient is 0.8, then the initial priority weight = 0.6 × 0.3 + 0.4 × 0.8 = 0.18 + 0.32 = 0.5. By using weighted calculations, the parameters of "export capacity" and "downstream demand" are integrated into a single priority weight. Exports with higher weights should be allocated cardboard first, providing a clear quantitative basis for sorting location decisions.

[0036] Finally, the initial priority weights are dynamically adjusted using historical sorting efficiency data to improve their actual adaptability. This historical sorting efficiency data comes from the sorting-related data of each exit stored in the control system over the past 72 hours. This includes the "sorting completion rate" (actual number of cardboard pieces sorted / planned sorting quantity), "stacking rate" (number of cardboard stacking occurrences / total sorting times), and "downstream process satisfaction" (timeliness of downstream processes obtaining cardboard as needed) for each exit under different initial priority weights. The dynamic adjustment logic is as follows: if the stacking rate of an exit is higher than 5% (a preset threshold) after tasks are assigned according to the initial priority weight, it indicates that the actual acceptance capacity of that exit is lower than the initial weight expectation. Its spatial adaptability coefficient needs to be lowered by 0.1, and the initial priority weight is recalculated as the final adjusted weight. If the downstream process satisfaction is lower than 90% (a preset threshold), it indicates that the urgency of downstream demand corresponding to that exit is not fully met. The time urgency coefficient needs to be increased by 0.1, and the final weight is recalculated. For example, if an export's initial priority weight is 0.52, but its historical backlog rate is 7%, then the spatial adaptation coefficient is first lowered from 0.6 to 0.5, and the weight is recalculated as 0.6 × 0.5 + 0.4 × 0.4 = 0.3 + 0.16 = 0.46, which is then used as the corrected priority weight for the sorting target location. This step uses historical data feedback for correction, avoiding deviations caused by the initial weight not considering actual equipment performance, environmental interference, and other factors, making the final weight more closely reflect actual production conditions.

[0037] In summary, steps S21-S26 form a complete logic for calculating the priority weight of sorting target locations: Steps S21 and S23 start from the dimension of "sorting line capacity" and quantify the outlet acceptance capacity through space utilization and space adaptation coefficient; Steps S22 and S24 start from the dimension of "downstream demand" and quantify the urgency of demand through order delivery time, production cycle time and time urgency coefficient; Step S25 integrates the two-dimensional parameters through a weighted algorithm to generate initial weights; Step S26 dynamically corrects the weights based on historical data to ensure the accuracy of the weights.

[0038] In one embodiment of the present invention, the step of generating the gripping instructions and joint motion instructions of the sorting robot arm based on the cardboard attribute dataset and the sorting target position priority weight includes: S31, obtain the size, weight parameters and corrugation parameters of the cardboard attribute dataset, determine the gripping force and number of gripping points of the robotic arm based on the size and weight parameters, and determine the placement angle and gentle placement buffer parameters of the robotic arm based on the corrugation parameters of the cardboard. S32, integrate the gripping force, the number of gripping points, the placement angle and the gentle placement buffer parameters to generate a gripping command for the sorting robot arm; S33, based on the priority weight of the sorting target location, select the multiple candidate landing locations with the highest weight; S34, calculate the motion trajectory parameters of the robotic arm based on the spatial coordinates of the multiple candidate landing points, and generate joint motion commands for the sorting robotic arm based on the motion trajectory parameters.

[0039] As described in steps S31-S34 above, this invention accurately generates gripping instructions and joint movement instructions for the sorting robotic arm based on the cardboard's own specifications and the priority of the sorting target position. This ensures that the robotic arm's movements can adapt to the physical characteristics (size, weight, flute type) of different specifications of corrugated cardboard and accurately match the sorting landing position with the highest priority. This solves the technical problems in traditional robotic arm sorting, such as "fixed motion parameters, mismatch with cardboard specifications leading to cardboard damage, and single landing position, lack of priority combination leading to low sorting efficiency." This enables non-destructive and efficient sorting of multiple cardboard specifications.

[0040] First, based on the key parameters in the cardboard attribute dataset, the core motion parameters for the robotic arm's grasping and placement are determined, providing a foundation for subsequent instruction generation. The cardboard attribute dataset here originates from step S1 above, where original images of the cardboard are acquired using a 3D camera on the production line. Visual recognition is then used to extract dimensions (e.g., length L, width W), weight (calculated from dimension × thickness × preset material density; the preset density for ordinary corrugated paper is 0.6 g / cm³), and cardboard flute parameters (e.g., A-flute, B-flute, C-flute), which are then integrated into a structured dataset. In S31, the above parameters are first extracted from the dataset, and then the robotic arm motion parameters are determined by dimension: For the gripping force and the number of gripping points, the size and weight parameters need to be considered. For example, when the cardboard weight is ≤1kg and the size is ≤500mm×500mm, because the cardboard is light and small, the gripping force is set to 30-40N to avoid falling off. At the same time, two gripping points are used (symmetrically distributed at 1 / 3 of the diagonal of the cardboard) to reduce contact damage to the cardboard. When the cardboard weight is 1-3kg and the size is 500mm×500mm-1200mm×800mm, the gripping force needs to be increased to 60-80N to ensure stable gripping. At the same time, the number of gripping points is increased to 4 (evenly distributed near the midpoint of the four sides of the cardboard) to avoid excessive force on a single point, which would cause the cardboard to deform. When the cardboard weight is >3kg and the size is >1200mm×800mm, the gripping force is set to 100-120N and the number of gripping points is set to 6 to further improve gripping stability. The placement angle and gentle placement buffer parameters need to be considered in conjunction with the corrugation parameters of the cardboard. A-flute cardboard has a larger flute height (approximately 4.5-5mm) and wider flute pitch (approximately 10mm), making it less resistant to vertical impacts. Horizontal placement can easily cause deformation of the flute tops due to stress. Therefore, the placement angle is set to 3-5° (slightly tilted along the flute length), and the gentle placement buffer parameter (i.e., the end speed of the robotic arm when it descends to the landing point) is set to 0.05m / s to reduce the impact force at the moment of contact. B-flute cardboard has a smaller flute height (approximately 2.5-3mm) and closer flute pitch (approximately 5mm), making it more resistant to impacts. A placement angle of 0° (horizontal placement) is sufficient, and the gentle placement buffer parameter is set to 0.1m / s to improve placement efficiency while ensuring no damage. C-flute cardboard has performance between A-flute and B-flute, with a placement angle of 2-3° and a gentle placement buffer parameter of 0.08m / s. For example, if a piece of cardboard has the following properties: size 1200mm×800mm, weight 2.5kg, flute type B, then according to the above rules, the gripping force is determined to be 70N, the number of gripping points is 4, the placement angle is 0°, and the gentle placement buffer parameter is 0.1m / s. These parameters directly determine the adaptability of the robotic arm's gripping and placement, thus avoiding damage to the cardboard caused by parameter mismatch from the source.

[0041] Next, by integrating the gripping-related parameters determined in S31, a complete gripping instruction for the sorting robot arm is generated. In the traditional process, parameters such as gripping force and number of gripping points are often set separately and need to be manually adjusted one by one, which is prone to parameter omission or conflict. However, in S32, the gripping force, number of gripping points, placement angle, and gentle placement buffer parameters determined in S31 are structurally integrated to form an instruction data package containing specific execution values. For example, the gripping instruction for the 1200mm×800mm B-flute cardboard mentioned above will clearly indicate "Gripping force: 70N; Number of gripping points: 4; Gripping point coordinates (based on the lower left corner of the cardboard as the origin): (100mm, 100mm), (1100mm, 100mm), (100mm, 700mm), (1100mm, 700mm); Placement angle: 0°; Gentle placement buffer speed: 0.1m / s". The instruction also includes gripping trigger conditions (such as triggering gripping when the cardboard reaches the center of the robotic arm's working area) and placement confirmation conditions (such as triggering gentle placement buffer when the pressure sensor detects the cardboard contacting the landing platform). This instruction data packet is transmitted to the robotic arm controller via the industrial bus. The controller can directly parse and drive the robotic arm to execute the command without manual intervention, effectively avoiding execution errors caused by parameter dispersion, while ensuring the integrity and consistency of the gripping action, reducing the cardboard damage rate during the gripping process to below 0.5%.

[0042] Then, based on the priority weight of the sorting target location, the multiple candidate landing points with the highest weights are selected to solve the sorting interruption problem caused by the traditional single landing point. The priority weight of the sorting target location here comes from step S2 above. In S2, the priority weight of each exit is calculated by integrating the available capacity data of each exit of the sorting line (collected by infrared sensors) and the demand information of downstream processes (obtained by the MES system). For example, among the 8 exits of the sorting line, the priority weight of exit 3 is 0.85, exit 5 is 0.78, exit 7 is 0.72, and the weights of the remaining exits are all below 0.7. In S33, candidate landing points need to be selected based on this weight. The selection rule is preset to "select the top 3 exits with the highest weights as candidate landing points", which ensures the high priority of the landing points while retaining redundant options to deal with unexpected situations. For example, in the above example, the candidate landing points are exits 3, 5, and 7. The spatial coordinates of each candidate landing point are preset in the system database (e.g., the coordinates of exit 3 are (12m, 6m, 0.8m), exit 5 is (15m, 8m, 0.8m), and exit 7 is (18m, 6m, 0.8m)). These coordinates are calibrated by laser ranging in the early stage, and the accuracy can reach ±5mm, ensuring the accuracy of subsequent motion trajectory calculation. If only one candidate landing point is selected, when the exit cannot be received due to sensor failure or cardboard accumulation, the robotic arm needs to wait for manual handling, causing sorting to stop. However, setting three candidate landing points allows the robotic arm to automatically switch to the second highest weight landing point when the current landing point is unavailable, reducing the sorting interruption rate to below 1% and significantly improving sorting continuity.

[0043] Finally, the robot arm's motion trajectory parameters are calculated based on the spatial coordinates of the candidate landing points, and joint motion commands are generated to ensure the robot arm moves precisely to the target landing point. First, the spatial coordinates of multiple candidate landing points selected by S33 are retrieved from the system database, and the current position coordinates of the robot arm are obtained (collected in real time by the robot arm's built-in encoder, including the spatial position corresponding to each joint angle). Taking the current position of the robot arm as the starting point and the candidate landing point position as the ending point, the motion trajectory parameters are calculated through forward and inverse kinematics algorithms. The motion trajectory parameters include the rotation angle, motion speed, and acceleration of each joint of the robot arm (such as the base, upper arm, forearm, and wrist). During the calculation process, fixed obstacles in the workshop (such as sorting line supports and sensor mounting rods, whose coordinates are preset in the motion trajectory calculation model) must be avoided. For example, if the current position coordinates of the robotic arm are (8m, 6m, 0.8m) and the coordinates of the candidate landing point exit 3 are (12m, 6m, 0.8m), then it is necessary to calculate the base rotation of 0° (because the starting point and the ending point are on the same horizontal straight line), the upper arm rotation of 30°, and the lower arm rotation of 45°, so that the end of the robotic arm moves from the starting point to the ending point. At the same time, the movement speed is set to 0.3m / s (a higher speed can be used to improve efficiency in open areas), and the acceleration is set to 0.1m / s² to avoid the robotic arm vibrating due to sudden speed changes. If the candidate landing point is exit 5 (15m, 8m, 0.8m), then it is necessary to add a base rotation of 37° (calculated according to the side-angle relationship of a right triangle, with a horizontal distance of 3m, a vertical distance of 2m, and an included angle arctan(2 / 3)≈37°), and at the same time adjust the rotation angles of the upper arm and lower arm to ensure that the trajectory does not interfere with other equipment. After the motion trajectory parameters are calculated, they are converted into specific control commands for each joint of the robotic arm (e.g., base joint rotation of 37° at a speed of 0.3m / s, upper arm joint rotation of 35° at a speed of 0.25m / s, etc.). These joint motion commands are then transmitted to the robotic arm controller, which drives each joint to execute the commands, ensuring the end effector accurately reaches the landing position. Through this step, the positioning error of the robotic arm's landing position can be controlled within ±10mm, fully meeting the accuracy requirements for corrugated cardboard sorting. Furthermore, the optimized motion trajectory reduces motion time by 10%-15% compared to traditional fixed trajectories, further improving sorting efficiency.

[0044] In summary, this invention reduces the damage rate of corrugated cardboard sorting, improves sorting efficiency, and reduces the need for manual intervention, providing key technical support for the automated sorting of multi-specification cardboard.

[0045] In one embodiment of the present invention, the step of obtaining the cardboard stack height based on the buffer status data, and generating a transfer request containing cardboard specifications, quantity, and priority when the cardboard stack height reaches a preset threshold includes: S41, Obtain a preset threshold for the cardboard stacking height, wherein the preset threshold includes a first preset threshold and a second preset threshold; S42, compare the cardboard stack height with the preset threshold: S43, when the height of the cardboard stack reaches the first preset threshold, a pre-reminder signal is generated and transmitted to the AGV / RGV scheduling center; S44, when the cardboard stacking height reaches the second preset threshold, count the number of cardboards in the buffer area and their corresponding specifications; S45, obtain the statistical results of comparing the cardboard stack height with the preset threshold, and generate a transfer request containing cardboard specifications, quantity and priority based on the statistical results.

[0046] As described in steps S41-S45 above, this invention, through real-time monitoring and threshold judgment of the stacking height of cardboard in the buffer area at the output end of the sorting line, triggers pre-warnings and formal transfer requests in stages. At the same time, it accurately counts the cardboard specifications and quantities and assigns priorities, ensuring that there is no accumulation in the buffer area and that transfer requests are efficiently transmitted to the AGV / RGV scheduling center. This solves the technical problems in traditional buffer area management, such as "single threshold warnings leading to delayed response, incomplete transfer request information (missing specifications / quantities), and chaotic AGV / RGV scheduling due to lack of priority sorting," and achieves seamless connection between the buffer area and the transfer process.

[0047] First, a preset threshold for the cardboard stacking height (including the first and second preset thresholds) is obtained to provide a criterion for subsequent height comparison. This preset threshold is not a fixed value, but rather a comprehensive setting considering the sorting line output efficiency, the average response time of the AGV / RGV, and the maximum carrying capacity of the buffer zone. The data is stored in the system parameter configuration module and can be dynamically adjusted according to the production rhythm. Specifically, the setting logic is as follows: the first preset threshold is 60%-70% of the maximum carrying height of the buffer zone, used to trigger pre-warnings and reserve preparation time for AGV / RGV scheduling; the second preset threshold is 80%-90% of the maximum carrying height of the buffer zone, used to trigger formal transfer requests to avoid near-full load leading to accumulation. For example, the maximum carrying height of the buffer area at the output end of a sorting line is 2m (capable of stacking approximately 50 standard corrugated cardboard sheets). The average response time of the AGV / RGV from receiving a request to arriving at the buffer area is 3 minutes. The sorting line outputs 3 cardboard sheets per minute, and 9 new cardboard sheets will be added within 3 minutes. Therefore, the first preset threshold is set to 1.2m (corresponding to 30 cardboard sheets, 60% of the maximum height). When a pre-alert is triggered at this time, the AGV / RGV can arrive within 3 minutes, preventing the stacking height from quickly reaching full capacity. The second preset threshold is set to 1.6m (corresponding to 40 cardboard sheets, 80% of the maximum height). If this height is reached, the AGV / RGV must be immediately dispatched for transfer to prevent the stacking height from exceeding 2m within the following 3 minutes. By setting dual thresholds, compared to the traditional single threshold (such as only setting 1.8m), preparatory actions can be triggered in advance, avoiding the risk of accumulation caused by AGV / RGV response delays.

[0048] Next, the real-time cardboard stack height is compared with the preset threshold set in S41 to provide a basis for grading response. The cardboard stack height here comes from a laser rangefinder sensor installed at the top of the buffer area at the sorting line output. The sensor emits a laser signal to the bottom of the buffer area every second, and the height is calculated using the formula: "Stack height = Total height of buffer area - Laser rangefinder value." This data is transmitted to the control system in real time with an accuracy of ±2mm, ensuring accurate height monitoring. For example, if the total height of the buffer area is 2m and the laser rangefinder sensor currently detects a distance of 0.8m, then the cardboard stack height = 2 - 0.8 = 1.2m. At this point, the control system will call the preset thresholds in S41 (1.2m is the first preset threshold, and 1.6m is the second preset threshold) for comparison, determining that the current stack height has reached the first preset threshold; if the laser rangefinder value is 0.4m, the stack height = 2 - 0.4 = 1.6m, then it is determined that the second preset threshold has been reached. This step transforms abstract "height data" into explicit "threshold trigger signals" through real-time comparison, providing a direct basis for subsequent graded actions of S43 and S44, and avoiding the lag and errors of manual inspection.

[0049] Then, when the cardboard stack height reaches the first preset threshold, a pre-alert signal is generated and transmitted to the AGV / RGV scheduling center, enabling the early initiation of transfer preparation. The pre-alert signal is not a formal transfer instruction, but rather a warning message containing "buffer zone number, current stack height, and estimated time to reach the second preset threshold." The "estimated time to reach the second preset threshold" is calculated from the difference between the current sorting line output rate (obtained from the MES system, such as 3 sheets / minute) and the current height relative to the second preset threshold. For example, if the current stack height is 1.2m (30 sheets), the second preset threshold is 1.6m (40 sheets), the difference is 10 sheets, and the output rate is 3 sheets / minute, then the estimated time = 10 ÷ 3 ≈ 3.3 minutes. This information helps the AGV / RGV scheduling center plan idle equipment in advance, avoiding equipment shortages caused by temporary scheduling. Signal transmission utilizes industrial Ethernet, with latency controlled within 100ms to ensure rapid reception by the dispatch center. Upon receiving the signal, the system interface displays a yellow warning and automatically marks the pre-scheduling requirements for the corresponding buffer zone, such as "Buffer Zone 3, current height 1.2m (first threshold), estimated transfer in 3.3 minutes." Dispatchers can then pre-arrange for AGVs / RGVs that have completed their current tasks to move towards the vicinity of this buffer zone, shortening the equipment arrival time during subsequent formal transfers. Through this step, the average response time of AGVs / RGVs is reduced from the traditional 3 minutes to 1.5 minutes, effectively improving the timeliness of transfers.

[0050] Then, when the cardboard stack height reaches the second preset threshold, the number of cardboard sheets and their corresponding specifications in the buffer zone are counted to provide core data support for the formal transfer request. The quantity statistics here are cross-validated in two ways: First, based on the stack height, i.e., "Quantity = Current stack height ÷ Single cardboard sheet thickness (obtained from the cardboard attribute dataset of S1, such as 4mm / sheet)". For example, if the current stack height is 1.6m and the single sheet thickness is 4mm, then the quantity = 1600 ÷ 4 = 40 sheets. Second, through the photoelectric counter installed at the entrance of the buffer zone, the counter increments by 1 for each cardboard sheet detected entering the buffer zone. The error between the two methods must be ≤1 sheet to ensure accurate quantity. Specification parameter statistics are achieved by reading the RFID tags on the cardboard within the buffer area (tag information is bound to the cardboard attribute dataset of S1, including size, flute type, load-bearing capacity, etc.). The RFID reader installed on the side of the buffer area can read tag data in batches without scanning each card individually, and the statistical time is ≤2 seconds. If the RFID tag of a cardboard fails, the system will automatically trigger the 3D camera (from the same source as S1) on the top of the buffer area to take an image and supplement the specification parameters through visual recognition to avoid data loss. For example, the statistical result is "Quantity 40 sheets, Specifications: 1200mm×800mm, B flute, load-bearing capacity B (30 sheets); 1000mm×600mm, C flute, load-bearing capacity A (10 sheets)". This data will be used as key content for subsequent transfer requests to ensure that the AGV / RGV scheduling can match the corresponding vehicle type (e.g., RGV is required for load-bearing capacity A), avoiding transfer failures caused by mismatch between the vehicle and the cardboard specifications.

[0051] Finally, the comparison result between the cardboard stacking height and the preset threshold is obtained (i.e., whether the second preset threshold is reached, the quantity and specifications are statistically analyzed), and a transfer request containing cardboard specifications, quantity and priority is generated, completing the closed loop from "threshold triggering" to "instruction generation". The priority setting of the transfer request is based on two factors: one is the difference between the stacking height of the buffer area and the maximum load-bearing height. The smaller the difference (closer to full load), the higher the priority. For example, if the current height of a buffer area is 1.8m (0.2m difference from the maximum height of 2m), the priority is set to "high". The other factor is the urgency of the downstream process demand (obtained from the MES system, such as a batch of cardboard that needs to be loaded urgently). If the cardboard in the buffer area corresponds to the urgent downstream demand, the priority is automatically increased by one level. For example, among the 40 cardboard sheets statistically analyzed above, 10 Class A load-bearing cardboard sheets correspond to an urgent order in the downstream loading process (delivery time remaining 1 hour). Then the priority of this transfer request is set to "highest", the specifications and quantity follow the statistical results of S44, and "target location (workstation 5 in the downstream stacking area, obtained from the MES system)" is added. After a transfer request is generated, it is transmitted to the AGV / RGV dispatch center via an encrypted protocol. Upon receiving the request, the center automatically marks it as a red urgent task and prioritizes equipment allocation. For example, "Buffer Zone 3, Transfer Request: 40 sheets (30 sheets of B-flute 1200×800, 10 sheets of A-grade C-flute 1000×600), highest priority, target location: stacking area, workstation 5." The dispatch system immediately filters available and matching carriers (e.g., 1 RGV for A-grade cardboard, 1 AGV for B-grade cardboard) and generates a preliminary task allocation plan. Through this step, the information completeness of the transfer request is improved from the traditional 70% to 100%, and the task execution accuracy of AGV / RGV is improved to over 99.5%.

[0052] In summary, this invention can improve the scheduling efficiency of AGV / RGV, ensuring the continuous flow of corrugated cardboard from sorting to transfer, while reducing the cost of manual intervention and meeting the needs of automated production.

[0053] In one embodiment of the present invention, the step of generating a task queue based on the transfer request and the real-time status data of the AGV / RGV includes: S51, determine the required type and quantity of transport vehicles based on the transfer request; S52 acquires real-time status data of AGV / RGV, including current location, power level, load status, and fault information; S53, Based on the type of transport vehicle and the real-time status data of the AGV / RGV, select available transport equipment; S54, obtain the priority of each of the transfer requests and sort them from high to low priority; S55, based on the required number of transport vehicles, assigns the sorted requests to specific AGVs / RGVs; S56 generates a task queue containing task number, target location, and execution time limit.

[0054] As described in steps S51-S56 above, this invention completes the entire process of "vehicle matching - equipment screening - request sorting - task allocation - queue generation" based on the specific requirements of the transfer request and the real-time status of the AGV / RGV. This ensures that the transfer task and the transportation equipment are accurately matched and efficiently allocated, solving the technical problems in traditional AGV / RGV scheduling such as "transfer failure due to mismatch between vehicle type and cardboard specifications, waste of resources due to lack of real-time synchronization of equipment status, and delay of emergency tasks due to lack of priority sorting". This achieves optimal utilization of transportation resources and orderly execution of transfer tasks.

[0055] First, the required type and quantity of transport vehicles are determined based on the transfer request, providing clear criteria for subsequent equipment selection. This transfer request originates from step S45 above and includes key information such as cardboard specifications (size, flute type, load-bearing capacity), quantity, and priority. The load-bearing capacity is the core basis for determining the type of transport vehicle—the system presets matching rules between vehicle type and load-bearing capacity: When the load-bearing capacity is Class A (single sheet load ≥ 50kg) or the total weight of stacked cardboard is ≥ 300kg, a higher-capacity RGV (Automated Guided Vehicle) should be selected because its fixed track and high load-bearing stability prevent cardboard displacement caused by bumps during AGV (Automated Guided Vehicle) movement; when the load-bearing capacity is Class B (single sheet load 20-50kg) and the total weight is < 300kg, an AGV can be used, offering greater flexibility and suitability for multi-path transfers. The required number of transport vehicles is calculated based on the rated load capacity of a single vehicle and the total number of cardboard sheets in the transfer request. For example, if a transfer request contains 40 sheets of Grade B cardboard, and a single AGV has a rated load capacity of 20 sheets, then the required number of transport vehicles = 40 ÷ 20 = 2 vehicles. If a transfer request contains 30 sheets of Grade A cardboard, and a single RGV has a rated load capacity of 15 sheets, then the required number = 30 ÷ 15 = 2 vehicles. For example, if a transfer request is "Specifications: 1200mm × 800mm, Grade A load capacity (55kg per sheet), quantity 30 sheets, total weight 1650kg", according to the matching rules, the vehicle type is determined to be RGV, and a single RGV has a rated load capacity of 15 sheets, therefore the required number of transport vehicles = 30 ÷ 15 = 2 vehicles. This step, by clarifying the type and quantity of carriers, avoids type mismatch (such as overload failure caused by using AGV to transport heavy Grade A cardboard) or insufficient quantity (such as one carrier not being able to complete the transport of 30 cardboard sheets, requiring two round trips and delaying time) caused by the traditional scheduling method of "allocating carriers based on experience" and lays the foundation for accurate equipment selection in the future.

[0056] Next, real-time status data of the AGV / RGV is acquired to comprehensively understand the current availability of the equipment and provide data support for selecting usable equipment. Real-time status data is collected in two ways: first, equipment location data, acquired by UWB positioning modules installed on the AGV / RGV, with a positioning accuracy of ±10cm, providing real-time feedback of the equipment's current coordinates; second, equipment operating status data (battery level, load status, fault information), transmitted in real-time from the vehicle control system to the dispatch center via an industrial bus—battery level data is collected by the battery management system, and when it is below 20%, it is considered low battery and requires priority charging; load status is detected by the vehicle-mounted weight sensor, displaying "no load," "half load," and "full load," with only "no load" equipment able to participate in new tasks; fault information is generated by the equipment self-checking system, such as motor failures and sensor malfunctions, with faulty equipment automatically marked as "unavailable." For example, real-time status data obtained by the dispatch center shows: AGV1 (coordinates: 5m, 8m), 65% battery, no load, no fault; AGV2 (coordinates: 12m, 6m), 18% battery, no load, no fault; RGV1 (coordinates: 8m, 10m), 70% battery, fully loaded, no fault; RGV2 (coordinates: 15m, 9m), 60% battery, no load, no fault. This data is updated every second to ensure the dispatch center has the latest equipment status and avoids task interruption caused by using low-battery or faulty equipment.

[0057] Then, based on the transport vehicle type determined in S51 and the real-time status data obtained in S52, available transport equipment that meets the requirements is selected, and equipment that does not meet the conditions is removed. The selection rules are executed in two steps: First, the selection is based on the vehicle type, retaining only equipment of the type determined in S51. For example, if S51 determines the vehicle type to be RGV, then RGV1, RGV2, and other RGV type equipment are selected from all equipment. Second, the selection is based on the real-time status data, retaining equipment with "power ≥ 20% (ensuring the completion of a single transfer task, with a preset average power consumption of 15% per transfer), no load status, and no fault information". For example, continuing the above example, if S51 determines that the vehicle type is RGV and the quantity is 2 units, first filter out RGV1 and RGV2; then check the status data: RGV1 is fully loaded, which does not meet the "no-load" requirement, so it is eliminated; RGV2 has 60% battery, is no-load, and has no faults, which meets the requirements. At the same time, other RGVs need to be filtered—if there is also RGV3 in the system (coordinates: 10m, 7m), with 55% battery, no-load, and no faults, it also meets the requirements. Finally, two usable devices, RGV2 and RGV3, are selected. If S51 determines that the vehicle type is AGV and the quantity is 2 units, first filter out AGV1 and AGV2; AGV2's battery is 18% < 20%, so it is eliminated; AGV3 (coordinates: 9m, 5m) needs to be added to the filter, with 72% battery, no-load, and no faults. Finally, AGV1 and AGV3 are selected. This step, through double screening, ensures that available equipment fully meets the transfer requirements, avoiding the problems of "selecting low-power equipment leading to power outages" or "selecting load-bearing equipment that cannot receive new cardboard" in traditional scheduling. The accuracy rate of available equipment screening reaches 100%.

[0058] Next, the priority of each transfer request is obtained and sorted from high to low to ensure that urgent tasks are executed first. The priority of transfer requests comes from step S45, and the priority is divided into four levels: "highest", "high", "medium", and "low". The determination criteria are as follows: when the difference between the stack height of the buffer area and the maximum carrying height is <0.3m and the corresponding downstream urgent order (delivery time remaining <2 hours), the priority is "highest"; when the difference is 0.3-0.5m or the delivery time remaining is 2-4 hours, the priority is "high"; when the difference is 0.5-0.8m or the delivery time remaining is 4-8 hours, the priority is "medium"; when the difference is >0.8m and the delivery time remaining is >8 hours, the priority is "low". When the system receives multiple transfer requests simultaneously, they must be sorted by priority. For example, if there are three transfer requests: Request A (highest priority, buffer difference 0.2m, delivery time 1.5 hours), Request B (higher priority, buffer difference 0.4m, delivery time 3 hours), and Request C (medium priority, buffer difference 0.6m, delivery time 6 hours), the sorting result would be Request A > Request B > Request C. If there are requests with the same priority, they are then sorted by their creation time, with the earlier creation time taking precedence. This priority sorting avoids the delays in urgent tasks caused by the "out-of-order execution of all requests" in traditional scheduling (such as assigning low-priority requests to idle devices while high-priority requests wait, missing order delivery times), reducing the average waiting time for the highest priority request from the traditional 15 minutes to less than 5 minutes.

[0059] Subsequently, based on the required number of transport vehicles, the sorted transfer requests from S54 are allocated to specific AGVs / RGVs, achieving precise matching between tasks and equipment. The allocation logic follows the principle of "priority first + proximity": high-priority requests are processed first, and from the available devices filtered by S53, the device closest to the corresponding buffer of the request is selected first to shorten the time for the device to arrive at the buffer; if the number of available devices is greater than or equal to the required number, the corresponding number of devices are directly allocated; if the number of available devices is insufficient, existing devices are allocated to perform part of the task, while new device scheduling is triggered (such as notifying devices that are charging and about to complete preparation). For example, for request A (vehicle type RGV, quantity 2 units, buffer area coordinates 12m, 6m), the available RGVs selected by S53 are RGV2 (coordinates 15m, 9m) and RGV3 (coordinates 10m, 7m). The straight-line distance between the devices and the buffer area is calculated as follows: RGV2 distance = √[(15-12)² + (9-6)²] = √18 ≈ 4.24m, RGV3 distance = √[(10-12)² + (7-6)²] = √5 ≈ 2.24m. Therefore, RGV3 and RGV2, which are closer, are prioritized to ensure that both devices can quickly reach the buffer area. If request A requires 2 units, but S53 only selects 1 available RGV (RGV3), then RGV3 is first assigned to perform the transfer task of 15 cardboard sheets. Simultaneously, the system detects that RGV4 is charging (90% battery, expected to complete in 5 minutes), so the remaining 15 cardboard sheets are pre-assigned to RGV4 and executed immediately after charging. This step, through a dual consideration of "priority + distance," ensures that urgent tasks are executed first, while also shortening equipment movement time by allocating equipment to nearby locations. The average arrival time of equipment is reduced from the traditional 8 minutes to 3 minutes, improving transfer efficiency.

[0060] Finally, a task queue containing task number, target location, and execution time limit is generated, providing a complete basis for task execution for subsequent path planning. The task number adopts the format of "date + request number + equipment number", such as "20230220-A02-RGV3", which facilitates system traceability and management; the target location is derived from the downstream process location corresponding to the transfer request (obtained from the MES system, such as workstation 5 in the stacking area, coordinates 20m, 12m), clearly defining the destination of the equipment transfer; the execution time limit is calculated based on the priority of the transfer request and the estimated execution time of the equipment—the execution time limit for the highest priority request = current time + 30 minutes (the transfer must be completed within 30 minutes), high priority is current time + 60 minutes, medium priority is current time + 120 minutes, and low priority is current time + 240 minutes, ensuring that the task has clear time constraints. For example, a task assigned to RGV3, with task number "20230220-A02-RGV3", has a target location of workstation 5 in the stacking area (20m, 12m), the highest priority, and a current time of 10:00, therefore the execution time limit is 10:30. The task queue is stored in a structured list format, containing information such as task number, equipment number, starting point (buffer area coordinates), target location, execution time limit, cardboard specifications and quantity, etc. This information can be synchronized in real-time to the AGV / RGV's onboard control system and the dispatch center monitoring interface, allowing dispatchers to visually view the execution status of each task (pending execution, in progress, completed). The task queue generated in this step avoids execution deviations caused by "fragmented task information" in traditional dispatching (such as unclear target location or execution time for equipment, requiring manual confirmation), improving task execution accuracy to 99.5%, and providing clear targets and time constraints for subsequent path planning in step S6.

[0061] In summary, this invention improves the accuracy of transport vehicle matching, increases equipment utilization, and enhances the on-time completion rate of emergency tasks. It also reduces the workload of manual scheduling and better meets the intelligent needs of automated production.

[0062] In one embodiment of the present invention, the step of generating a planned route based on the task queue, the workshop environment map, and the workshop traffic heat map includes: S61, extract the coordinate information of passageways, intersections and obstacles from the workshop environment map; S62, Based on the workshop traffic heat map, identify high-congestion areas and smooth-flowing areas; S63, Generate an initial path based on the target position in the task queue; S64, Obtain congestion data of the areas traversed by the initial path, and adjust the weight of the initial path; S65, obtain the device motion parameters of AGV / RGV; S66, Based on the adjusted initial path and the device motion parameters, generate a planned path containing speed commands.

[0063] As described in steps S61-S66 above, this invention combines the execution requirements of the task queue, the physical constraints of the workshop environment, and real-time traffic conditions. Through a progressive process of "environmental analysis - congestion identification - initial path generation - path optimization - parameter adaptation - instruction generation," it generates planned paths for AGVs / RGVs that take into account safety, timeliness, and equipment adaptability. This solves the technical problems in traditional path planning, such as "failure to consider dynamic congestion in the workshop leading to transportation delays, failure to consider the motion characteristics of equipment leading to unexecutable paths, and failure to extract key environmental information leading to collision risks." This enables efficient and safe operation of AGVs / RGVs during corrugated cardboard transfer.

[0064] First, the coordinates of passageways, intersections, and obstacles are extracted from the workshop environment map to build a basic physical environment framework for subsequent path planning, avoiding equipment collisions caused by ignoring fixed obstacles during path planning. The workshop environment map here is a high-precision map generated in the early stage using laser SLAM (simultaneous localization and mapping) technology, stored in the system database, containing the coordinate information of all static facilities in the workshop, and can be updated according to the workshop layout. When extracting information, the system uses an image segmentation algorithm to structure the map: passageways are divided into main passageways (width ≥ 3m, allowing two-way traffic) and branch passageways (width 1.5-2.5m, allowing one-way traffic). The starting point, ending point coordinates, and direction of travel of each passageway must be marked during extraction. The center point coordinates and the number of passageways connected to intersections (such as T-junctions and crossroads) must be marked at intersections. Deceleration or avoidance logic must be set at these intersections in subsequent path planning. Obstacles include fixed equipment (such as sorting line supports and shelves), walls, and columns. The boundary coordinates of obstacles must be marked during extraction (such as the coordinate range of a shelf being (10m-12m, 8m-10m)) to ensure that the path planning avoids this area. For example, from the workshop environment map, the following can be extracted: Main aisle 1 (starting point (0m, 5m), ending point (20m, 5m), two-way traffic), branch aisle 2 (starting point (10m, 5m), ending point (10m, 15m), one-way traffic), crossroads (coordinates (10m, 5m)), and obstacle 1 (shelf, coordinates (8m-10m, 12m-14m)). This step transforms the abstract map into structured coordinate information, providing clear "feasible areas" and "prohibited areas" for subsequent path searching. This avoids information omissions caused by relying on manual obstacle labeling in traditional planning and reduces the risk of equipment collisions.

[0065] Next, the system identifies high-congestion and smooth-flowing areas based on the workshop traffic heatmap, providing dynamic traffic condition data for route optimization and preventing AGVs / RGVs from entering congested areas and causing delays. The workshop traffic heatmap data comes from the real-time location data of the AGVs / RGVs (synchronously acquired from the real-time status data in step S52, updated every second). The system uses a density clustering algorithm to count the number of devices in each area (divided into 5m×5m grids): when the number of devices in a grid is ≥3, it is identified as a high-congestion area, marked in red on the heatmap; when the number of devices is 1-2, it is a normal traffic area, marked in yellow; when the number of devices is 0, it is a smooth-flowing area, marked in green. Simultaneously, the system supplements and corrects the real-time heatmap by incorporating historical traffic data (e.g., the concentration of devices near intersections (10m, 5m) from 10:00-11:00 daily, indicating potential congestion). For example, a heat map at a certain moment might show that there are 4 AGVs / RGVs within a grid (10m-15m, 5m-10m), which is identified as a high-congestion area; while there are no devices within a grid (15m-20m, 10m-15m), which is a smooth-flowing area. This step allows for real-time monitoring of traffic dynamics within the workshop, avoiding the uncontrollable transportation time caused by traditional route planning that "only plans based on the shortest distance without considering congestion," and providing a basis for subsequent adjustments to route weights.

[0066] Then, an initial path is generated based on the target position in the task queue to complete the basic path search from the "starting point to the ending point". The task queue comes from step S56 and contains the task starting point of the AGV / RGV (corresponding to the buffer coordinates, obtained from the transfer request in step S45) and the target position (corresponding to the downstream process position, obtained from the MES system). The initial path generation adopts an improved A* algorithm (see steps S631-S636 for details). The core is to balance the path length and search efficiency through a cost function to ensure that the generated initial path is a feasible path with "shorter distance and no obstacles". For example, the starting point of an AGV's task is buffer zone 3 (coordinates (12m, 6m)), and the target location is workstation 5 in the stacking area (coordinates (20m, 12m)). Combining the passageway and obstacle information extracted by S61, the generated initial path is: from (12m, 6m) along the main passageway 1 (20m, 5m) to (15m, 6m), turn into branch passageway 3 (15m, 6m-12m), and finally reach (20m, 12m). This path avoids obstacle 1 (8m-10m, 12m-14m) and has a shorter total distance. The generation of the initial path does not consider real-time congestion, but only ensures "physical feasibility," laying the foundation for subsequent optimization based on congestion conditions and avoiding excessive computation caused by directly generating complex paths.

[0067] Next, acquire congestion data for the areas traversed by the initial path (extracted from the heatmap in step S62) and adjust the weights of the initial path to achieve congestion avoidance path optimization. The adjustment logic is as follows: assign a congestion weight to each grid area traversed by the initial path—a weight of 5 for highly congested areas, 2 for generally passable areas, and 1 for smoothly passable areas; calculate the total weight of the initial path (the sum of the weights of each grid); if the total weight is ≥10 (i.e., the path contains many highly congested areas), trigger a path re-search, prioritizing alternative paths in smoothly passable areas; if the total weight is <10, retain the initial path, only marking "requires slowing down and avoiding" sections passing through highly congested areas. For example, the initial path of the AGV mentioned above passes through grid (10m-15m, 5m-10m) (high congestion area, weight 5) and grid (15m-20m, 10m-15m) (smooth area, weight 1), with a total weight of 5+1=6<10. The initial path is retained, but the section passing through the high congestion area (12m-15m, 6m-10m) is marked "decelerate to 0.2m / s". If the initial path passes through two high congestion areas, the total weight is 5+5=10. Then the path is searched again. For example, it is adjusted to travel from (12m, 6m) along branch channel 4 (12m, 6m-12m) to (12m, 12m), and then turn into main channel 2 (12m-20m, 12m) to reach the target position. The new path passes through smooth areas, and the total weight is reduced to 3. This step integrates dynamic congestion information into route planning, avoiding delays caused by the initial route passing through highly congested areas and improving the timeliness of route travel.

[0068] Subsequently, the motion parameters of the AGV / RGV are acquired to provide a basis for generating speed commands, ensuring that the planned path matches the equipment performance. These motion parameters are stored in the system's equipment database, categorized by vehicle type: AGV motion parameters include maximum travel speed (typically 1.2 m / s), maximum acceleration (0.3 m / s²), and minimum turning radius (1.5 m); RGVs, traveling along tracks, have parameters including track speed (fixed at 1.5 m / s) and start / stop acceleration (0.5 m / s²). These parameters are provided by the equipment manufacturer and verified through on-site testing before being entered into the system to ensure consistency with the actual equipment performance. For example, the motion parameters for AGV1 performing the task are: maximum speed 1.2 m / s, maximum acceleration 0.3 m / s², and minimum turning radius 1.5 m; if the task is performed by RGV2, the parameters are track speed 1.5 m / s and acceleration 0.5 m / s². This step clarifies the "upper limit of the equipment's motion capability," preventing subsequent speed commands from exceeding the equipment's performance range, which could lead to equipment malfunction or unstable operation.

[0069] Finally, based on the adjusted initial path and equipment motion parameters, a planned path containing speed commands is generated, completing the transformation from "path coordinates" to "executable commands". The speed command generation logic is designed according to road segments: In smooth traffic areas (weight 1), if the road segment is straight and has no turns, the speed is set to the maximum travel speed of the equipment (e.g., 1.2 m / s for AGV); in general traffic areas (weight 2), the speed is reduced to 70% of the maximum speed (e.g., 0.84 m / s); in highly congested areas (weight 5) or intersections, the speed is reduced to 30% of the maximum speed (e.g., 0.36 m / s), while setting an acceleration limit (not exceeding 50% of the maximum acceleration) to avoid rapid acceleration causing equipment vibration; in turning sections (the turning angle is calculated based on the path coordinates, ≥90° is considered a sharp turn), the speed is set to 50% of the maximum speed (e.g., 0.6 m / s), and the turning radius is ensured to be ≥ the minimum turning radius of the equipment. For example, the adjusted path includes three segments: the first segment (12m, 6m) to (15m, 6m) (smooth area, straight line), with the AGV speed set to 1.2m / s; the second segment (15m, 6m) to (15m, 12m) (smooth area, 90° turn), with the speed set to 0.6m / s; and the third segment (15m, 12m) to (20m, 12m) (general area, straight line), with the speed set to 0.84m / s. Simultaneously, the planned path will mark the coordinates of the speed switching points for each segment (e.g., switching from 1.2m / s to 0.6m / s at (15m, 6m), forming a complete "coordinate + speed" instruction set. This step ensures that the planned path not only includes the travel route but also specifies the travel speed for each segment, avoiding the instability caused by traditional planning that only provides path coordinates and the equipment traveling at a fixed speed. This ensures that the AGV / RGV travels at the appropriate speed, improving transportation safety and stability.

[0070] In summary, this invention improves the stability of AGV / RGV path travel time, significantly enhances the success rate of congestion avoidance, reduces the incidence of equipment collision failures, and ensures that the path and equipment performance are compatible, providing key guarantees for the efficient and safe execution of corrugated cardboard transportation.

[0071] In one embodiment of the present invention, the step of generating an initial path based on the target location in the task queue includes: S631, obtain the current position coordinates of the AGV / RGV and the target position in the task queue, and determine the path start point and path end point based on the current position coordinates of the AGV / RGV and the target position; S632, construct a grid model based on the workshop environment map, mark the locations of passable nodes and obstacles, and determine the path search range; S633, Obtain the current node. Define a cost function to evaluate the path cost based on the actual distance from the path start point to the current node and the estimated Manhattan distance from the current node to the path end point. The cost function is: ; In the formula, This represents the cost function value. Indicates the path from the starting point to the current node. The actual distance Indicates the current node Estimated Manhattan distance to the end of the route; S634, Initialize the open list and the closed list, add the starting point of the path to the open list, and calculate the cost function value corresponding to the starting point according to the cost function; S635 Iteratively select the node with the smallest cost function value in the open list, expand the adjacent passable nodes of the node and update their costs until the end of the path is found. S636, by tracing back the parent-child relationship between each node from the end point of the path to the starting point of the path, an initial path composed of continuous coordinate points is generated.

[0072] As described in steps S631-S636 above, this invention, based on the real-time location of AGV / RGV and the target location in the task queue, combined with the physical constraints of the workshop environment map, generates an initial path with no obstacles and optimal path cost through a systematic path search logic of "coordinate positioning - mesh modeling - cost definition - list initialization - node iteration - path backtracking". This solves the technical problems in traditional path search such as "path deviation caused by ambiguous starting and ending point positioning, computational redundancy caused by unclear search range, and non-optimal path caused by single cost evaluation", laying a precise and efficient foundational framework for subsequent path optimization based on congestion data.

[0073] First, the current coordinates of the AGV / RGV and the target position in the task queue are obtained to clearly define the path start and end points, ensuring that the path search has accurate "starting points" and "ending points" to avoid the path deviating from the task requirements due to ambiguous coordinates. The current coordinates of the AGV / RGV are obtained from the real-time status data in step S52, collected in real-time by the UWB positioning module installed on the AGV / RGV, with a positioning accuracy of ±10cm, updated every second to ensure accurate reflection of the equipment's current location. The target position in the task queue is obtained from the task queue in step S56. This target position corresponds to the downstream transfer destination of the corrugated cardboard (such as stacking area workstations, loading areas, etc.), and its coordinate information is pre-stored in the MES system and synchronously associated with specific tasks when the task queue is generated, ensuring consistency with the coordinate system in the workshop environment map. For example, if an AGV's task queue shows that it needs to transfer cardboard from buffer area 3 to workstation 5 in the stacking area, the current coordinates of the AGV are obtained from step S52 as (12m, 6m), and the coordinates of the target location (workstation 5 in the stacking area) are obtained from the task queue as (20m, 12m). Based on these two coordinates, the path start point is determined to be (12m, 6m), and the path end point is determined to be (20m, 12m). This step transforms the abstract "equipment location" and "task destination" into specific coordinate points, providing clear directional guidance for subsequent path searching. This avoids coordinate errors caused by manually setting the start and end points in traditional methods, ensuring that the path search direction perfectly matches the task requirements.

[0074] Next, a grid model is constructed based on the workshop environment map, marking the locations of passable nodes and obstacles and determining the path search range. This transforms the complex workshop environment into structured grid cells, facilitating efficient node expansion and path search in the future. (Workshop environment map and land used in step S61) Figure 1The initial high-precision static map, generated using laser SLAM technology, contains the coordinates of all fixed facilities within the workshop. When constructing the mesh model, the system uses an equidistant mesh division method, dividing the workshop environment map into multiple square mesh units of a preset size (e.g., 0.5m × 0.5m), with each mesh unit considered a "node." Subsequently, combining the obstacle coordinates extracted in step S61, the mesh units containing obstacles are labeled as "impassable nodes," while other obstacle-free mesh units are labeled as "passable nodes." Mesh units corresponding to passageways and intersections are also labeled to ensure the mesh model accurately reflects walkable and prohibited areas within the workshop. The path search range is determined based on the coordinates of the path's start and end points, typically defined as a rectangular area with the start and end points as diagonal vertices. The search range boundary extends 2-3 mesh units beyond the line connecting the start and end points on both sides, avoiding excessive computational redundancy while ensuring no potential optimal paths are missed. For example, based on the starting point (12m, 6m) and ending point (20m, 12m) determined in step S631, the area (10m-22m, 4m-14m) in the workshop environment map is divided into 0.5m × 0.5m grid cells, forming a total of (22-10) / 0.5 × (14-4) / 0.5 = 24 × 20 = 480 grid nodes. Combining this with the obstacle 1 (shelf, coordinates 8m-10m, 12m-14m) extracted in step S61, it is found that this obstacle is partially located at the edge of the search range. The corresponding grid cells (10m, 12m), (10m, 12.5m), etc., are marked as impassable nodes, while the remaining grid cells are marked as passable nodes. Through this step, the complex workshop environment is discretized into standardized grid nodes, clearly distinguishing between passable and impassable areas, while limiting a reasonable search range. This reduces the computational complexity of subsequent path search and ensures that the search process does not exceed the environmental range required by the task, thus improving path search efficiency.

[0075] Then, based on the actual distance from the path start point to the current node and the estimated Manhattan distance from the current node to the path end point, a cost function is defined to evaluate the path cost, providing a quantitative evaluation standard for subsequent selection of the optimal node and generation of the optimal path. The actual distance from the path start point to the current node refers to the cumulative distance from the path start point along all searched traversable nodes to the current node. This is obtained by calculating and summing the straight-line distances between adjacent nodes (the grid cell is a square, and the distance between adjacent nodes is the grid side length, such as 0.5m), ensuring a true reflection of the traveled path length. The estimated Manhattan distance from the current node to the path end point is a simplified distance estimation method, calculated as "Manhattan distance = |current node X coordinate - end point X coordinate| + |current node Y coordinate - end point Y coordinate|". This estimation method has low computational cost and effectively reflects the approximate distance between the node and the end point, quickly guiding the path search towards the end point and preventing the path search from deviating from the target. The cost function combines both factors, comprehensively evaluating the "cost already traveled" and "remaining estimated cost" of the current node. This ensures that the selected node guarantees a short traveled path while quickly approaching the destination, thus generating a path with the optimal overall cost. For example, if the current node coordinates are (15m, 9m), the path start point is (12m, 6m), and the path end point is (20m, 12m), then the actual distance from the path start point to the current node is the cumulative distance of adjacent nodes traversed from (12m, 6m) to (15m, 9m). Assuming a straight line traversing 6 grid nodes, the actual distance = 6 × 0.5m = 3m. The estimated Manhattan distance from the current node to the end point = |15-20| + |9-12| = 5 + 3 = 8 (unit: grid side length, i.e., 8 × 0.5m = 4m). According to the cost function, the calculated cost function value = 3m + 4m = 7m. This step establishes a cost evaluation system that considers both "actual driving distance" and "destination proximity." Compared to a cost function that only considers actual distance, this system can more efficiently guide the path search towards the destination, reduce the expansion of invalid nodes, and improve path search speed.

[0076] Next, the open and closed lists are initialized. The starting point of the path is added to the open list, and its cost function value is calculated to establish the initial data structure for subsequent node iterative searches. The open list stores traversable nodes that have been discovered but not yet expanded, while the closed list stores nodes that have been expanded and do not require further processing. Dynamic updates to the two lists avoid redundant node expansion and improve search efficiency. During initialization, both lists are first cleared to ensure no interference from historical data. Then, the grid node corresponding to the starting point of the path determined in step S631 is added to the open list. At this point, this node is the only node to be expanded. Simultaneously, the cost function value of the starting point is calculated according to the cost function defined in step S633. Since the actual distance from the starting point to itself is 0, the estimated Manhattan distance from the starting point to the end point can be directly calculated from the coordinates. Therefore, the cost function value of the starting point = 0 + the estimated Manhattan distance. For example, given a path starting point (12m, 6m) and ending point (20m, 12m), the estimated Manhattan distance from the starting point to the ending point is |12-20| + |6-12| = 8 + 6 = 14 (grid side length), which is 14 × 0.5m = 7m. Therefore, the cost function value of the starting point is 0 + 7m = 7m, and this value is associated with the starting point node and stored in the open list. This step establishes the initial data structure for path search, clarifies the first node to be expanded and its real-time cost, provides a clear starting state for subsequent iterative searches, and avoids search anomalies caused by chaotic initial data.

[0077] The process iteratively selects the node with the smallest cost function value from the open list, expands its adjacent passable nodes, and updates the cost, until the destination of the path is found. Through continuous iterative optimization, the process gradually approaches the destination and finds a feasible path. The iterative process is executed in a fixed loop: First, the node with the smallest cost function value is selected from the open list and designated as the current node to be expanded; Second, this node is moved from the open list to the closed list and marked as "expanded" to avoid repeated processing; Third, the adjacent nodes of the current node are determined (usually grid nodes in the four directions of up, down, left, and right; diagonal nodes are not considered adjacent nodes to ensure that the path conforms to the straight-line travel characteristics of AGV / RGV), and nodes marked as "passable nodes" that are not in the closed list are selected; Fourth, the cost function value of each adjacent node is calculated—the actual distance between adjacent nodes = the actual distance of the current node + The distance between the current node and the adjacent node (e.g., 0.5m) and the estimated Manhattan distance of the adjacent node are calculated according to the formula in step S633. The two are added together to obtain the cost function value of the adjacent node. In the fifth step, if the adjacent node is not in the open list, it is added to the open list and associated with the calculated cost function value. If the adjacent node is already in the open list, the newly calculated cost function value is compared with the original value. If the new value is smaller, the cost function value of the node in the open list is updated. In the sixth step, it is determined whether the open list contains the node corresponding to the end point of the path. If it contains it, the iteration stops. If it does not contain it, the above steps are repeated. For example, the node with the lowest cost in the open list is (13m, 7m), with a cost function value of 8m. After moving it to the closed list, its neighboring nodes (12.5m, 7m), (13.5m, 7m), (13m, 6.5m), and (13m, 7.5m) are expanded. The nodes (13.5m, 7m) and (13m, 7.5m) that are passable and not in the closed list are selected. The cost function values ​​of these two nodes are calculated to be 8.5m and 8.5m respectively, and they are added to the open list. This process is iterated until, in a certain round of expansion, the node corresponding to the path endpoint (20m, 12m) is added to the open list. At this point, the iteration stops, indicating that a feasible path from the starting point to the endpoint has been found. Through this step, guided by the cost function value, nodes closer to the endpoint and with shorter travel distances are prioritized for expansion, ensuring that the overall cost of the generated path is optimal. Simultaneously, the dynamic management of the open and closed lists avoids redundant node processing, improving the efficiency of iterative search.

[0078] Finally, by tracing the parent-child relationships between nodes from the path's endpoint to its starting point, an initial path composed of continuous coordinate points is generated, transforming the node relationships obtained through iterative search into an intuitive and executable sequence of path coordinates. During the node expansion process in step S635, the system synchronously records the "parent node" of each node (i.e., which node it expanded from), forming a parent-child relationship between nodes—for example, if node (13.5m, 7m) expands from node (13m, 7m), then (13m, 7m) is the parent node of (13.5m, 7m). During backtracking, starting from the node corresponding to the path's endpoint, the system sequentially searches for the parent node of each node until it reaches the path's starting point. Then, the backtracked nodes are arranged in the order of "starting point → ... → ending point," and the center coordinates of each node are extracted to form a continuous sequence of coordinate points, which constitutes the initial path. For example, starting from the endpoint (20m, 12m), the parent node is (19.5m, 12m), and the parent node of (19.5m, 12m) is (19m, 12m), and so on, until tracing back to the starting point (12m, 6m). These node coordinates are then arranged sequentially as (12m, 6m) → (12.5m, 6.5m) → ... → (19.5m, 12m) → (20m, 12m), forming an initial path containing multiple consecutive coordinate points. This step transforms discrete node relationships into continuous path coordinates, clarifying the specific travel route of the AGV / RGV from the starting point to the endpoint. This provides a clear basic path structure for subsequent adjustments to path weights based on congestion data, preventing inaccurate optimization due to unclear path representation.

[0079] In summary, this invention improves the accuracy of initial path generation, significantly shortens search time, and ensures that the path strictly avoids fixed obstacles, further guaranteeing the efficiency and safety of AGV / RGV transport of corrugated cardboard.

[0080] This invention also provides an automated sorting and intelligent scheduling system for multi-specification corrugated cardboard finished products, comprising: The data acquisition module acquires original images of corrugated cardboard on the production line, extracts characteristic specification parameters of the cardboard based on the original images, and generates a cardboard attribute dataset. The weight calculation module obtains the available capacity data of each exit of the sorting line and the demand information of downstream processes, and calculates the priority weight of the sorting target position based on the available capacity data and the demand information of downstream processes. The collaborative control module generates gripping instructions and joint movement instructions for the sorting robotic arm based on the cardboard attribute dataset and the priority weight of the sorting target position. The request generation module obtains the buffer status data at the output end of the sorting line, obtains the cardboard stacking height based on the buffer status data, and generates a transfer request containing cardboard specifications and quantity information when the cardboard stacking height reaches a preset threshold. The task generation module acquires transfer requests and real-time status data of AGVs / RGVs, and generates a task queue based on the transfer requests and the real-time status data of AGVs / RGVs. The scheduling module acquires a workshop environment map and a workshop traffic heat map, generates a planned path based on the task queue, the workshop environment map, and the workshop traffic heat map, and schedules the corrugated cardboard based on the planned path.

[0081] The present invention also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, characterized in that the processor executes the computer program to implement the steps of an automatic sorting and intelligent scheduling method for multi-specification corrugated cardboard finished products.

[0082] The present invention also provides a computer-readable storage medium having a computer program stored thereon, characterized in that, when the computer program is executed by a processor, it implements the steps of an automatic sorting and intelligent scheduling method for multi-specification corrugated cardboard finished products.

[0083] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, apparatus, article, or method. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, apparatus, article, or method that includes that element.

[0084] The above description is merely a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for automatic sorting and intelligent scheduling of corrugated cardboard finished products in multiple specifications, characterized in that, include: Obtain original images of corrugated cardboard from the production line, extract characteristic specification parameters of the cardboard based on the original images, and generate a cardboard attribute dataset. Obtain the available capacity data and downstream process demand information of each exit of the sorting line, and calculate the priority weight of the sorting target position based on the available capacity data and the downstream process demand information; Based on the cardboard attribute dataset and the sorting target location priority weight, the sorting robot arm's gripping instructions and joint movement instructions are generated. Obtain the status data of the buffer area at the output end of the sorting line, obtain the cardboard stacking height based on the buffer area status data, and generate a transfer request containing cardboard specifications and quantity information when the cardboard stacking height reaches a preset threshold. Obtain transfer requests and real-time status data of AGV / RGV, and generate a task queue based on the transfer requests and the real-time status data of AGV / RGV; Obtain a workshop environment map and a workshop traffic heat map, generate a planned path based on the task queue, the workshop environment map, and the workshop traffic heat map, and schedule the corrugated cardboard based on the planned path.

2. The automatic sorting and intelligent scheduling method for multi-specification corrugated cardboard finished products according to claim 1, characterized in that, The step of calculating the priority weight of the sorting target location based on the available capacity data and the downstream process demand information includes: Calculate the space utilization rate of each exit based on the available capacity data of each exit of the sorting line; Extract order delivery time and production cycle parameters based on downstream process demand information; Calculate the space adaptation coefficient based on the space utilization rate; Calculate the time urgency factor based on the order delivery time and the production cycle parameters; Obtain the spatial adaptation coefficient and the time urgency coefficient, and calculate the initial priority weight using a weighted algorithm; Historical sorting efficiency data is obtained, and the initial priority weight is dynamically adjusted based on the historical sorting efficiency data to obtain the priority weight of the sorting target location.

3. The automatic sorting and intelligent scheduling method for multi-specification corrugated cardboard finished products according to claim 2, characterized in that, The step of generating the gripping instructions and joint movement instructions of the sorting robot arm based on the cardboard attribute dataset and the sorting target position priority weight includes: Obtain the size, weight parameters, and corrugation parameters of the cardboard attribute dataset. Determine the gripping force and number of gripping points of the robotic arm based on the size and weight parameters, and determine the placement angle and gentle placement buffer parameters of the robotic arm based on the corrugation parameters. The gripping force, the number of gripping points, the placement angle, and the gentle placement buffer parameters are integrated to generate the gripping instructions for the sorting robot arm; Based on the priority weight of the sorting target location, select the multiple candidate landing points with the highest weights; The motion trajectory parameters of the robotic arm are calculated based on the spatial coordinates of multiple candidate landing points, and joint motion commands for the sorting robotic arm are generated based on the motion trajectory parameters.

4. The automatic sorting and intelligent scheduling method for multi-specification corrugated cardboard finished products according to claim 3, characterized in that, The step of obtaining the cardboard stack height based on the buffer status data, and generating a transfer request containing cardboard specifications, quantity, and priority when the cardboard stack height reaches a preset threshold, includes: Obtain a preset threshold for the height of the cardboard stack, wherein the preset threshold includes a first preset threshold and a second preset threshold; Compare the cardboard stack height with the preset threshold: When the height of the cardboard stack reaches a first preset threshold, a pre-reminder signal is generated and transmitted to the AGV / RGV scheduling center; When the height of the cardboard stack reaches the second preset threshold, the number of cardboard pieces and their corresponding specifications in the buffer area are counted. Obtain statistical results by comparing the cardboard stack height with the preset threshold, and generate a transfer request containing cardboard specifications, quantity, and priority based on the statistical results.

5. The automatic sorting and intelligent scheduling method for multi-specification corrugated cardboard finished products according to claim 4, characterized in that, The step of generating a task queue based on the transfer request and the real-time status data of the AGV / RGV includes: Determine the required type and quantity of transport vehicles based on the transfer request; Acquire real-time status data of AGV / RGV, including current location, power level, load status, and fault information; Based on the type of transport vehicle and the real-time status data of AGV / RGV, available transport equipment is selected; Obtain the priority of each of the aforementioned transfer requests and sort them from high to low priority; The sorted requests are assigned to specific AGVs / RGVs based on the required number of transport vehicles; Generate a task queue containing task number, target location, and execution time limit.

6. The automatic sorting and intelligent scheduling method for multi-specification corrugated cardboard finished products according to claim 5, characterized in that, The step of generating a planned route based on the task queue, the workshop environment map, and the workshop traffic heat map includes: The coordinates of passageways, intersections, and obstacles are extracted from the workshop environment map. Based on the workshop traffic heat map, identify high-congestion areas and smooth-flowing areas; Generate an initial path based on the target location in the task queue; Obtain congestion data for the areas traversed by the initial path, and adjust the weight of the initial path accordingly; Obtain the device motion parameters of the AGV / RGV; Based on the adjusted initial path and the device motion parameters, a planned path containing speed commands is generated.

7. The automatic sorting and intelligent scheduling method for multi-specification corrugated cardboard finished products according to claim 6, characterized in that, The step of generating an initial path based on the target location in the task queue includes: Obtain the current position coordinates of the AGV / RGV and the target position in the task queue, and determine the path start point and path end point based on the current position coordinates of the AGV / RGV and the target position; A grid model is constructed based on the workshop environment map, and the locations of passable nodes and obstacles are marked to determine the path search range; Obtain the current node, and define a cost function to evaluate the path cost based on the actual distance from the path start point to the current node and the estimated Manhattan distance from the current node to the path end point; The open list and closed list are initialized, the starting point of the path is added to the open list, and the cost function value corresponding to the starting point is calculated according to the cost function. Iteratively select the node with the smallest cost function value in the open list, expand the adjacent passable nodes of that node and update their costs, until the end of the path is found; An initial path consisting of consecutive coordinate points is generated by tracing back the parent-child relationships between nodes from the end point of the path to the beginning point of the path.

8. A multi-specification automatic sorting and intelligent scheduling system for corrugated cardboard finished products, characterized in that, include: The data acquisition module acquires original images of corrugated cardboard on the production line, extracts characteristic specification parameters of the cardboard based on the original images, and generates a cardboard attribute dataset. The weight calculation module obtains the available capacity data of each exit of the sorting line and the demand information of downstream processes, and calculates the priority weight of the sorting target position based on the available capacity data and the demand information of downstream processes. The collaborative control module generates gripping instructions and joint movement instructions for the sorting robotic arm based on the cardboard attribute dataset and the priority weight of the sorting target position. The request generation module obtains the buffer status data at the output end of the sorting line, obtains the cardboard stacking height based on the buffer status data, and generates a transfer request containing cardboard specifications and quantity information when the cardboard stacking height reaches a preset threshold. The task generation module acquires transfer requests and real-time status data of AGVs / RGVs, and generates a task queue based on the transfer requests and the real-time status data of AGVs / RGVs. The scheduling module acquires a workshop environment map and a workshop traffic heat map, generates a planned path based on the task queue, the workshop environment map, and the workshop traffic heat map, and schedules the corrugated cardboard based on the planned path.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images