Intelligent warehouse multi-task real-time scheduling method, system and device

Abstract

The application provides a kind of intelligent warehousing multitask real-time scheduling method, system and equipment, belongs to logistics automation technical field.The method includes: real-time acquisition of warehouse order task and the running state data of each AGV;According to the target storage location coordinates of task, the spatial distance is calculated, and the task correlation matrix is constructed;Based on the task correlation matrix and the total number of tasks to be allocated and the number of idle AGVs, the tasks are aggregated into multiple task packages;Each idle AGV uses an improved consistency binding algorithm to bid for each task package and conflict resolution, and outputs the allocation mapping relationship;According to the allocation result, task instruction is issued to the corresponding AGV for execution.The application reduces the scale of allocation problem by task aggregation, combines the bidding mechanism driven by marginal incremental cost and the localized consensus communication between adjacent AGVs, reduces the communication overhead and computing load, and improves the real-time performance and global optimization capability of scheduling.

Description

Technical Field

[0001] This invention belongs to the field of logistics automation technology, and more specifically relates to an intelligent warehousing multi-task real-time scheduling method, system and equipment. Background Technology

[0002] With the booming development of e-commerce and the rapid popularization of intelligent warehousing systems, outbound order tasks within warehouses are exhibiting significant characteristics of massive volume, concurrency, and high dynamism. Automated Guided Vehicles (AGVs), as the core execution unit of intelligent warehousing, directly impact the overall throughput and response speed of the warehousing system through their task scheduling efficiency. How to achieve efficient matching between massive tasks and multiple AGVs under limited computing resources and communication bandwidth has become a core technical problem urgently needing to be solved in the field of intelligent warehousing.

[0003] Currently, distributed task allocation methods achieve decentralized decision-making through communication and negotiation among multiple agents, exhibiting good scalability and are widely used in multi-robot task allocation scenarios. However, these methods face significant communication bottlenecks in large-scale warehousing scenarios: as the number of AGVs and the scale of tasks increase, the amount of data that needs to be exchanged between agents increases dramatically, and the number of iteration rounds also increases accordingly, resulting in a significant extension of algorithm solution time, making it difficult to meet the real-time requirements of warehousing operations.

[0004] To alleviate the aforementioned communication pressure, existing technologies typically aggregate tasks into larger task packages based on spatial proximity before allocation, and then allocate tasks in units of these packages. This reduces the number of allocation objects from individual tasks to task packages, effectively decreasing the data scale that subsequent allocation algorithms need to process. However, existing grouping schemes are mostly static and rigid; once tasks are aggregated into task packages, they cannot be further split or adjusted during the entire allocation process. When the task space is unevenly distributed, static grouping can easily lead to some task packages containing too many or too scattered tasks, forcing the AGVs receiving those task packages to travel long distances to execute individual tasks, thus increasing global path costs.

[0005] Furthermore, the choice of granularity for task aggregation is itself a complex trade-off. Overly large task packages may overload a single AGV, while overly small packages offer only limited reduction in communication overhead. Simultaneously, relying solely on spatial proximity for aggregation cannot guarantee load balancing among task packages. Therefore, how to effectively reduce the problem size while ensuring the rationality of task aggregation and load balancing is a prerequisite problem that must be addressed when adopting a grouping allocation method. Summary of the Invention

[0006] To address the above problems, the present invention aims to provide an intelligent warehousing multi-task real-time scheduling method, system, and device. By aggregating tasks into task packages based on spatial proximity to reduce the scale of the allocation problem, implementing dynamic bidding under the real-time status perception of AGVs through a bidding mechanism driven by marginal incremental costs, and cooperating with localized consensus communication limited to adjacent AGVs, the invention improves the global optimality of the allocation results and the system's adaptability to dynamic environments while reducing communication overhead and computational load.

[0007] To achieve the above objectives, the present invention employs the following technical solution: In a first aspect, embodiments of this application provide a real-time scheduling method for multi-task intelligent warehousing, comprising the following steps: S1: Real-time acquisition of outbound order task data and operating status data of each AGV within the warehousing system, and statistical analysis of the total number of tasks to be assigned and the number of currently idle AGVs; the outbound order task data includes the three-dimensional coordinates of the target outbound storage location for each outbound order task to be assigned, and the AGV operating status data includes the current position coordinates and remaining power of each AGV. S2: Based on the three-dimensional coordinates of the target outbound storage location of each outbound order task to be assigned, calculate the spatial distance between any two outbound order tasks, and use the spatial distance as a measure of the spatial correlation between outbound order tasks to construct a task correlation matrix. S3: Based on the task association matrix, the total number of tasks to be assigned, and the current number of idle AGVs, aggregate all outbound order tasks to be assigned into multiple task packages and output them to all currently idle AGVs; wherein, the total number of aggregated task packages does not exceed the current number of idle AGVs, and the number of outbound order tasks contained in each task package is not less than the preset threshold for the minimum number of tasks contained in a single task package. S4: Based on the task package and the current position coordinates and remaining power of the AGV, each currently idle AGV adopts an improved consistency binding algorithm to bid for and resolve conflicts of each task package with the goal of minimizing path cost, and outputs the allocation mapping relationship between each AGV and the task package. In the allocation mapping relationship, each task package is uniquely assigned to one AGV. S5: Based on the allocation mapping relationship and the current position coordinates of the AGV, the task package allocated to each AGV is broken down into a single task instruction sequence according to the execution order. Combined with the warehouse environment map, an actual navigation path is planned for each AGV, and the single task instruction sequence and the corresponding actual navigation path are sent to the corresponding AGV for execution.

[0008] In an optional implementation, step S1 includes: The warehouse management system obtains the list of outbound order tasks to be assigned in real time through its external interface. Each outbound order task in the list includes a task number, a target outbound location number, the three-dimensional coordinates of the target outbound location, the order priority, and the task generation timestamp. Extract the three-dimensional coordinates of the target outbound storage location for each outbound order task from the obtained outbound order task list; The status information of all AGVs in the warehouse is obtained in real time through the status monitoring interface of the AGV scheduling system. The status information includes AGV number, current position coordinates, current movement status, remaining power, and currently assigned task list; the current position coordinates and remaining power of each AGV are extracted from the obtained status information. The total number of tasks to be assigned in the obtained outbound order task list is counted, and the number of AGVs currently in an idle state in the status information is counted as the current number of idle AGVs.

[0009] In an optional implementation, step S2 includes: Extract the three-dimensional coordinates of the target outbound storage location corresponding to each outbound order task from the data of each outbound order task. Assume there are a total of The number of outbound order tasks to be assigned, the first The three-dimensional coordinates of the target outbound storage location for each outbound order task are as follows: ,in For any two outbound order tasks and The spatial distance between the target outbound storage locations is calculated using the following formula. :

[0010] The calculated spatial distance As an outbound order task With outbound order task Spatial correlation measurement; organize the spatial distances between all outbound order tasks in a matrix form to construct... Task association matrix :

[0011] matrix elements Indicates the first The outbound order task and the first The spatial distance between each outbound order task, when hour .

[0012] In an optional implementation, step S3 includes: Based on the total number of outbound order tasks to be assigned and the number of currently idle AGVs The total number of task packages formed by aggregation is determined according to the following formula. :

[0013] in The minimum number of tasks a single task package must contain is a preset threshold. Indicates rounding up; Set spatial distance threshold The task association matrix Medium spatial distance less than Task pairs are marked as aggregateable; a task association graph is constructed using the task pairs marked as aggregateable as edges, and the connected components in the task association graph are extracted as the initial task clusters; Count the number of initial task clusters ,like Then, in order of increasing number of tasks within the initial task cluster, smaller clusters are merged into their nearest spatially adjacent clusters, until the number of tasks in the initial task cluster equals the number of clusters in the previous cluster. ;like Then, the cluster with the largest number of tasks within the initial task cluster is selected for splitting. The splitting criterion is the maximum spatial distance between tasks within the cluster. The two tasks corresponding to the maximum spatial distance are used as seeds for two new clusters, and the remaining tasks are redistributed according to the principle of closest spatial distance, until the number of initial task clusters equals the number of tasks in the cluster. Each of the finalized task clusters will be treated as a task package.

[0014] In an optional implementation, step S4 includes: The following bidding process will be conducted: All task packages are treated as a set of task packages to be auctioned, which is the set of task packages that have not yet been assigned to any AGV and are currently available for auction; each currently idle AGV is initialized. Task package bidding list Empty list, task package allocation list empty sequence, timestamp vector The zero vector, the The task packages in the AGV are arranged according to the execution order of the AGV; For AGV and any task package in the set of task packages to be auctioned Let the AGV currently The existing task package sequence is Calculate the task package according to the following formula. Insert sequence at optimal position The resulting marginal increase in the total path :

[0015] in Indicates AGV According to sequence The shortest path total length of the shortest path to the unloading point is obtained by sequentially accessing all target outbound storage locations in each task package and returning the unloading point. The access order of the target outbound storage locations of each outbound order task in the task package is determined according to the shortest path principle. Indicates the task package Insert sequence at optimal position The resulting new sequence has the optimal position determined according to the shortest path principle. This represents the total length of the shortest path executed according to the new sequence; Determine AGV For any task package in the set of task packages to be auctioned bid If the task package Insertion sequence If the remaining battery power of the AGV exceeds its maximum travel distance, then... Let it be negative infinity; Select the highest-bid task package from the set of task packages to be auctioned and add it to each currently idle AGV. Whenever a new task pack is added Then, the task package is removed from the set of task packages to be auctioned, and the results are updated according to the shortest path principle. The execution order of each task package is determined, and the virtual current position of the AGV is updated to the position after the update is completed. After the endpoint of all task packages, update the virtual remaining power of the AGV to the remaining value after deducting the power consumed along that path; Repeat the above bidding calculation and selection operations until the virtual remaining power of the AGV is insufficient to support the insertion of any remaining task package into the set of task packages to be bid on, or until the set of task packages to be bid on is empty, to obtain the task package allocation list of each currently idle AGV. As a candidate allocation scheme, and compared with the task package bidding list and timestamp vector Packaged into consensus messages.

[0016] In an optional implementation, step S4 further includes: Perform the following conflict resolution process: Receive consensus messages from each currently idle AGV and obtain the set of task packages to be auctioned; Set conflict perception radius The topological distance is calculated by taking the shortest navigation path distance between any two AGVs in the warehouse environment map; only when the topological distance between the two AGVs is less than 1. At that time, a strong connection consensus link is established between the two AGVs and consensus messages are exchanged; for topological distances greater than or equal to The two AGVs do not establish a strong consensus link, but exchange consensus messages in an asynchronous weak consensus manner; Within a strongly connected consensus link, consensus messages are received from neighboring AGVs. For task packages that are simultaneously won by multiple AGVs, the bids of each AGV for the task package are compared, and the AGV with the lowest bid is assigned a task package from its own task package allocation list. Release the task package in the middle; if the bids are the same, compare the timestamps and the AGV with the later timestamp releases the task package. The released task package is re-added to the set of task packages to be auctioned; Repeat the bidding process and conflict resolution process until the convergence condition is met, and output the allocation mapping relationship between each AGV and the task package. In the allocation mapping relationship, each task package is uniquely assigned to one AGV.

[0017] In an optional implementation, step S5 includes: Based on the aforementioned allocation mapping relationship, each outbound order task in each task package assigned to each AGV is allocated according to the task package allocation list of that AGV. The execution sequence is broken down into individual task instructions, and each individual task instruction includes at least the three-dimensional coordinates of the target outbound storage location, the task execution sequence number, and the expected arrival time window. Starting from the current position coordinates of each AGV, and taking the three-dimensional coordinates of the target outbound location of each outbound order task as the waypoint, combined with the constraints of aisles and obstacles in the warehouse environment map, the actual navigation path is planned for each AGV. The planned actual navigation path and the corresponding sequence of individual task instructions are sent to the onboard control system of the corresponding AGV; The system monitors the task execution progress of each AGV in real time, and feeds back the status information to the scheduling system when an AGV completes all tasks or encounters an abnormality.

[0018] In an optional implementation, the convergence condition is: all task packages have been uniquely assigned to a single AGV and no AGV has made a higher bid for the assigned task package, or the number of iterations has reached a preset maximum number of iterations.

[0019] Secondly, embodiments of this application also provide an intelligent warehousing multi-task real-time scheduling system, including: The data sensing module is used to acquire outbound order task data and the operating status data of each AGV in the warehousing system in real time, and to count the total number of tasks to be assigned and the number of currently idle AGVs. The outbound order task data includes the three-dimensional coordinates of the target outbound location of each outbound order task to be assigned, and the AGV operating status data includes the current position coordinates and remaining power of each AGV. The task association module is used to calculate the spatial distance between any two outbound order tasks based on the three-dimensional coordinates of the target outbound location of each outbound order task to be assigned, and to construct a task association matrix using the spatial distance as a measure of the spatial correlation between outbound order tasks. The dynamic grouping module is used to aggregate all outbound order tasks to be assigned into multiple task packages and output them to all currently idle AGVs based on the task association matrix, the total number of tasks to be assigned, and the current number of idle AGVs. The total number of aggregated task packages does not exceed the current number of idle AGVs, and the number of outbound order tasks contained in each task package is not less than a preset threshold for the minimum number of tasks contained in a single task package. An improved CBBA solving module is used to bid on and resolve conflicts for each task package based on the task package, the current position coordinates and remaining power of the AGV, and each currently idle AGV adopts an improved consistency binding algorithm with the goal of minimizing path cost. The module outputs the allocation mapping relationship between each AGV and the task package, in which each task package is uniquely assigned to one AGV. The path execution module is used to break down the task package allocated to each AGV into a single task instruction sequence according to the execution order based on the allocation mapping relationship and the current position coordinates of the AGV, plan the actual navigation path for each AGV in combination with the warehouse environment map, and send the single task instruction sequence and the corresponding actual navigation path to the corresponding AGV for execution.

[0020] Thirdly, embodiments of this application also provide an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the intelligent warehousing multi-task real-time scheduling method as described in any of the above.

[0021] As can be seen from the above technical solutions, the present invention has the following advantages: The intelligent warehousing multi-task real-time scheduling method provided in this application combines a two-stage architecture that integrates task clustering with distributed bidding, along with a marginal incremental cost-driven bidding mechanism and a localized consensus communication strategy. This approach reduces the scale of large-scale task allocation problems while maintaining the flexibility of task allocation, effectively improving the real-time scheduling, allocation balance, and system robustness of the intelligent warehousing multi-AGV system.

[0022] This application aggregates massive outbound order tasks into task packages based on spatial proximity, enabling the distributed consensus binding algorithm to auction and resolve conflicts using task packages rather than individual tasks as processing units. Each AGV only needs to perform bidding calculations and consensus communication on an order-of-volume scale of task packages. Simultaneously, each AGV establishes strong connectivity consensus links only between neighboring AGVs with a topological distance less than a preset conflict detection radius, avoiding communication congestion caused by global broadcasting. The synergistic effect of reducing the number of processing objects and localizing the communication range significantly reduces the system's communication overhead and the computational load per iteration, greatly shortening the algorithm's solution time and meeting the real-time requirements of task allocation in warehousing operations.

[0023] This application determines the total number of task packages based on the total number of tasks to be assigned and the current number of idle AGVs, ensuring that the number of task packages matches the number of available AGVs. During the aggregation process, tasks with similar geographical locations are grouped into the same task package based on spatial proximity, shortening the travel path required for AGVs to execute the task package. Through dual constraints—an upper limit on the total number of task packages and a lower limit on the number of tasks per package—along with granular control of spatial distance thresholds, extreme situations are avoided, such as task packages being too large leading to excessive load on a single AGV or task packages being too small resulting in limited reduction of communication overhead. This achieves a balance between reasonable task aggregation and system load balancing.

[0024] This application employs a marginal incremental cost-driven bidding mechanism. Each AGV's bid for a task package is calculated based on the marginal increase in the total path cost caused by inserting the task package into the existing task package sequence. This ensures that the bid truly reflects the actual incremental cost of executing the task package, rather than the cost of an independent path. Simultaneously, through repeated iterations of the bidding and conflict resolution processes, released task packages re-enter the bidding set to participate in subsequent rounds of bidding. This allows task packages to be acquired by AGVs with lower path costs, avoiding the local optima problem easily caused by static grouping or single-round greedy allocation, and improving the global optimality of the final allocation result.

[0025] In this application, task package aggregation is deeply coupled with the real-time status of AGVs. The total number of task packages depends on the current number of idle AGVs, allowing the grouping granularity to adaptively adjust as the number of available AGVs changes. When each AGV executes the auction, it incorporates its current position, remaining power, and currently held task package sequence into the calculation of marginal incremental cost. The bidding result changes dynamically with the real-time status of the AGVs. Whenever a new task package is added to the allocation list, the virtual position and virtual remaining power are immediately updated, ensuring that subsequent bidding calculations are based on the latest status. Distributed auctions achieve dynamic matching between task packages and AGVs, avoiding the rigid allocation caused by static grouping and enabling the system to adapt to scenarios with real-time changes in AGV status and dynamic emergence of tasks.

[0026] This application employs a distributed, improved consensus binding algorithm architecture. Each AGV independently calculates its own candidate allocation scheme, resolving conflicts through local communication. This eliminates the need for a global centralized scheduler and removes the risk of single-point failures. For AGVs with a topological distance greater than or equal to the conflict perception radius, asynchronous weak consensus is used, reducing the size of the strong connectivity consensus links. This ensures that as the number of AGVs increases, the communication overhead grows locally rather than explodes globally, exhibiting good scalability. The iterative mechanism allows the system to gradually correct allocation results through subsequent rounds of bidding and conflict resolution even when some AGVs experience temporary anomalies in bidding calculation or communication, enhancing the system's operational stability in highly dynamic warehousing environments. Attached Figure Description

[0027] To more clearly illustrate the technical solution of the present invention, the accompanying drawings used in the description will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0028] Figure 1 A flowchart illustrating the intelligent warehousing multi-task real-time scheduling method provided in this application.

[0029] Figure 2 A schematic diagram of the structure of the intelligent warehousing multi-task real-time scheduling system provided in this application.

[0030] Figure 3 A schematic diagram of the structure of the electronic device provided in this application. Detailed Implementation

[0031] The various embodiments of this disclosure will be described more fully in the detailed steps of the intelligent warehousing multi-task real-time scheduling method described below. This disclosure may have various embodiments, and adjustments and changes may be made therein. However, it should be understood that there is no intention to limit the various embodiments of this disclosure to the specific embodiments disclosed herein, but rather this disclosure should be understood to cover all adjustments, equivalents, and / or alternatives falling within the spirit and scope of the various embodiments of this disclosure.

[0032] In the following, the terms “comprising” or “may include”, which may be used in various embodiments of this disclosure, indicate the presence of the disclosed functions, operations, or elements, and do not limit the addition of one or more functions, operations, or elements. Furthermore, as used in various embodiments of this disclosure, the terms “comprising,” “having,” and their cognates are intended only to indicate a particular feature, number, step, operation, element, component, or combination of the foregoing, and should not be construed as primarily excluding the presence of one or more other features, numbers, steps, operations, elements, components, or combinations of the foregoing, or the possibility of adding one or more combinations of the foregoing.

[0033] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0034] Please see Figure 1 The diagram shows a flowchart of a real-time multi-task scheduling method for intelligent warehousing in a specific embodiment. The method includes the following steps: S1: Real-time acquisition of outbound order task data and operating status data of each AGV within the warehousing system; statistical analysis of the total number of tasks to be assigned and the number of currently idle AGVs; the outbound order task data includes the three-dimensional coordinates of the target outbound storage location for each outbound order task to be assigned; the AGV operating status data includes the current position coordinates and remaining power of each AGV.

[0035] In a specific implementation, after the intelligent warehouse scheduling process is initiated, a data sensing process is executed first. Through the external data interface of the warehouse management system, a task data acquisition request is sent to the warehouse management system according to a preset sampling period to obtain a real-time list of all pending outbound order tasks within the warehouse system. The preset sampling period can be configured based on the warehouse task arrival rate, with a typical value of 500 milliseconds. Each outbound order task in the acquired list contains at least the following fields: task number, target outbound location number, three-dimensional coordinates of the target outbound location in the warehouse space, order priority, and task generation timestamp. The three-dimensional coordinates of the target outbound location for each outbound order task are extracted from the acquired list and used as input data for subsequent task association construction steps.

[0036] Meanwhile, the real-time operating status information of all AGVs within the warehousing system is obtained through the status monitoring interface of the AGV scheduling system. The acquired AGV status information includes at least the AGV's ID, precise location coordinates on the current warehousing environment map, current movement status, current remaining power, and a list of currently assigned but not yet completed tasks. The current movement status can be categorized as idle, executing a task, charging, and faulty. The current location coordinates and remaining power of each AGV are extracted from the acquired status information for subsequent scheduling decisions. Furthermore, the number of task entries in the acquired outbound order task list is counted to obtain the total number of tasks to be assigned; simultaneously, the number of AGVs marked as idle in the status information is counted to obtain the current number of idle AGVs. The acquired task data and AGV status data are fused and aligned using a unified timestamp to construct a complete system status snapshot for the current scheduling cycle, for use in subsequent steps. If the current number of idle AGVs is zero, the scheduling process can be paused and the process awaits the next sampling cycle.

[0037] S2: Based on the three-dimensional coordinates of the target outbound storage location for each outbound order task to be assigned, calculate the spatial distance between any two outbound order tasks, and use the spatial distance as a measure of the spatial correlation between outbound order tasks to construct a task correlation matrix.

[0038] In a specific implementation, after obtaining the three-dimensional coordinates of the target outbound storage location for each outbound order task to be assigned, the task association construction phase begins. Assume there are a total of [number missing] tasks within the current scheduling cycle. There are 10 outbound order tasks to be assigned, and for each outbound order task... Establish the three-dimensional coordinate vector of its target outbound storage location. ,in For any two outbound order tasks and The spatial distance between the target outbound storage locations is calculated using the Euclidean distance formula in three-dimensional space. ,Right now:

[0039] The smaller the spatial distance value, the closer the target outbound storage locations of the two tasks are in the physical location of the warehouse space, and the higher the spatial correlation between the two tasks.

[0040] Organize the calculated pairwise spatial distances between all tasks in matrix form to construct a... OK Column task association matrix :

[0041] In this matrix, the position located at the Line 1 Column elements Indicates the first The outbound order task and the first Spatial distance between outbound order tasks, diagonal elements of the matrix A uniform setting of 0 indicates that the distance between the task and itself is zero. This task association matrix comprehensively depicts the spatial distribution relationships between all outbound order tasks, serving as the core basis for subsequent dynamic grouping steps. As an optional implementation, if obstacles such as aisles and pillars in the warehousing environment affect the actual travel path of the AGV, the Euclidean distance can be corrected when calculating the distance, introducing an A... The algorithm pre-calculates the actual navigation path distance and Euclidean distance, and then performs a weighted fusion to make the task relevance more consistent with the actual driving cost.

[0042] S3: Based on the task association matrix, the total number of tasks to be assigned, and the current number of idle AGVs, aggregate all outbound order tasks to be assigned into multiple task packages and output them to all currently idle AGVs; wherein, the total number of aggregated task packages does not exceed the current number of idle AGVs, and the number of outbound order tasks contained in each task package is not less than the preset threshold for the minimum number of tasks contained in a single task package.

[0043] In a specific implementation, after obtaining the task association matrix, the process proceeds to the dynamic grouping strategy stage. The core objective of this stage is to... Individual outbound order tasks are aggregated based on spatial proximity. This involves creating task packages to reduce the size of objects that the subsequent distributed allocation algorithm needs to process. First, based on the total number of tasks to be allocated obtained in step S1... and the number of currently idle AGVs The total number of task packages formed by aggregation is determined according to the following formula. :

[0044] in, This is a preset threshold for the minimum number of tasks a single task package can contain. Its value can be configured by the system based on the warehouse size; typical values ​​are 3 or 5. This indicates rounding up. The formula represents the total number of task packages. No more than the current number of idle AGVs At the same time, all tasks should be completed in the minimum number of available tasks. The number of packages required for sub-packaging ensures that each idle AGV is assigned at most one task package, while also preventing the grouping from becoming meaningless due to an insufficient number of tasks within a single task package.

[0045] Determine the total number of task packages Then, perform the specific clustering operation. Set a spatial distance threshold. This threshold can be configured based on the actual size of the warehouse, with a typical value being five percent of the longest side length of the warehouse. The task association matrix... Medium spatial distance less than Task pairs are marked as aggregateable. A task association graph is constructed using the aggregateable task pairs as edges. Nodes in the graph represent outbound order tasks, and edges represent the spatial distance between two tasks that is less than a threshold. Extract all connected components from the task association graph, and use each connected component as an initial task cluster.

[0046] At this point, the initial number of task clusters This may not equal the total number of target task packages. Perform cluster size adjustment operation: If If the number of initial task clusters exceeds the target number, then smaller clusters within the initial task clusters are merged into their nearest spatially adjacent clusters, in ascending order of the number of tasks. This process is repeated until the number of clusters equals the target number. ;like If the initial number of task clusters is less than the target number, the cluster with the most tasks within it is selected for splitting. Specifically, the two tasks with the greatest spatial distance within that cluster are identified as the seeds for two new clusters. The remaining tasks are then redistributed to the two new clusters based on proximity. This process is repeated until the number of clusters equals the target number. After adjustments, each finalized task cluster is defined as a task package, and all generated task package information is broadcast to each currently idle AGV.

[0047] S4: Based on the task package and the current position coordinates and remaining power of the AGV, each currently idle AGV adopts an improved consistency binding algorithm to bid for and resolve conflicts of each task package with the goal of minimizing path cost, and outputs the allocation mapping relationship between each AGV and the task package. In the allocation mapping relationship, each task package is uniquely assigned to one AGV.

[0048] In a specific implementation, after each currently idle AGV receives the task package information, it enters the improved consistency binding algorithm solution phase. This phase consists of two sub-phases: an auction process and a conflict resolution process, which iterate cyclically until convergence. Each AGV executes the auction process independently, employing a marginal incremental cost-driven bidding mechanism. Specifically, all task packages are first organized into a set of task packages to be auctioned, where each task package has not yet been assigned to any AGV and is in an auctionable state. For each currently idle AGV... Initialize its local data structure: task package bidding list Empty list, task package allocation list empty sequence, timestamp vector Let be the zero vector, where The task packages in the AGV will be arranged according to the execution order of the AGV.

[0049] For AGV and any one of the task packages in the set of task packages to be auctioned. Let the AGV currently The execution sequence consisting of existing task packages is as follows The sequence is initially empty. The task package is calculated according to the following formula. Insert sequence at optimal position The resulting marginal increase in the total path :

[0050] in, Indicates AGV According to sequence Visit all target outbound locations in each task package in sequence and return the total shortest path length to the unloading point; Indicates the task package Insert sequence at optimal position The optimal position in the resulting new sequence is the position that minimizes the total path length after insertion. The smallest insertion position; This represents the total length of the shortest path executed according to the new sequence. The order of accessing the target outbound storage location for each outbound order task in the task package is determined according to the shortest path principle, which can be solved using dynamic programming or the nearest neighbor heuristic algorithm. .

[0051] Determine the AGV based on marginal increment. For task package The bid is:

[0052] The smaller the marginal increment of the path, the higher the bid. At the same time, the AGV also needs to be checked. Remaining power constraint: If the task package Insertion sequence If the total path length for all tasks exceeds the maximum travel distance supported by the AGV's remaining battery power, the bid for that task package is set to negative infinity, indicating that the AGV cannot accept the task package. Each AGV selects the highest-bid task package from the set of available task packages and adds it to its own list. Simultaneously, the task package is removed from the set of task packages to be auctioned, and the virtual current position and virtual remaining power are updated. Each AGV repeats the above bidding calculation and selection operation until its virtual remaining power is insufficient to support the continued execution of any remaining task packages or the set of task packages to be auctioned is empty.

[0053] After each AGV completes its local bidding process, it will include a list of task package bids. and timestamp vector The consensus message is sent to the conflict resolution process. During the conflict resolution phase, after receiving the consensus message from each AGV, conflict resolution is performed in a distributed environment. A conflict detection radius is set. This radius can be configured based on the warehouse size and AGV density, with a typical value of 10 meters. The topological distance between any two AGVs is defined as the shortest navigation path distance between them in the warehouse environment map. Only when the topological distance between two AGVs is less than... At that time, a strong consensus link is established between the two AGVs and frequent consensus message exchanges are performed; for topological distances greater than or equal to The two AGVs do not establish a strong connection link, but only use an asynchronous weak consensus method to exchange information.

[0054] Within a strongly connected consensus link, each AGV receives consensus messages from neighboring AGVs. For a task package that is simultaneously won by multiple AGVs, the bids of each AGV for that task package are compared, and the AGV with the lowest bid wins the task package from its own... The task package is released during the bidding process. If bids are the same, the timestamps are compared, and the task package is released by the AGV with the later timestamp. The released task package is then added back to the set of task packages to be auctioned and participates in the next round of bidding. The bidding process and conflict resolution process are repeated until all task packages are uniquely assigned to a single AGV and no AGV offers a higher bid for the assigned task packages, or the number of iterations reaches the preset maximum number of iterations, typically 50. After convergence, the allocation mapping relationship between each AGV and the task packages is output, in which each task package is uniquely assigned to one AGV.

[0055] In one embodiment, a task package is included in the current sequence of an AGV. and task pack The auction collection includes task packages. . Task package Try inserting the AGV at the beginning, middle, and end of the sequence, and select the position that minimizes the total path increment as the optimal insertion position. If the total path increment is small after insertion, it indicates that the AGV's travel path is relatively convenient, the bid will be higher, and the AGV is more likely to win the auction.

[0056] S5: Based on the allocation mapping relationship and the current position coordinates of the AGV, the task package allocated to each AGV is broken down into a single task instruction sequence according to the execution order. Combined with the warehouse environment map, an actual navigation path is planned for each AGV, and the single task instruction sequence and the corresponding actual navigation path are sent to the corresponding AGV for execution.

[0057] In a specific implementation, after completing the improved CBBA solution and obtaining the allocation mapping relationship, the path execution phase begins. Based on the allocation mapping relationship output in step S4, the outbound order tasks in the task packages assigned to each AGV are executed according to the AGV's... The execution order is broken down into individual task instructions. Each individual task instruction contains at least the three-dimensional coordinates of the target outbound storage location, the execution order number of the task in the sequence, and the expected arrival time window.

[0058] Starting from the current position coordinates of each AGV obtained in step S1, and using the three-dimensional coordinates of the target outbound storage location for each outbound order task as the waypoints, combined with the aisle layout and obstacle information in the warehouse environment map, an actual navigation path is planned for each AGV, starting from its current position and sequentially traversing all its assigned task target points. This path planning can use A... The algorithm is implemented using known path search algorithms. The planned actual navigation path and its corresponding sequence of individual task instructions are then transmitted to the onboard control system of the corresponding AGV via a wireless communication network.

[0059] After receiving task instructions and navigation paths, the AGVs execute each outbound task sequentially according to the instruction sequence. The status monitoring interface of the AGV scheduling system tracks the task execution progress of each AGV in real time. Once an AGV completes all assigned tasks, its status is automatically updated to idle, and it is reinstated into the idle AGV statistics for subsequent scheduling cycles. If an AGV encounters an anomaly during execution, including but not limited to insufficient power requiring charging or path obstruction, the AGV will report the anomaly status information to the scheduling system in real time, triggering a rescheduling or manual intervention process to ensure the continuity and reliability of warehousing operations.

[0060] In this embodiment, tasks are aggregated into task packages based on spatial proximity to reduce the scale of the allocation problem; a bidding mechanism driven by marginal incremental cost ensures that the bids of each AGV reflect the actual incremental cost of executing the task package; localized consensus communication limited to neighboring AGVs reduces communication overhead; and a constraint on the number of idle AGVs is introduced when the total number of task packages is determined to ensure that the number of task packages matches the available AGV resources. Simultaneously, during the bidding process, the current position of the AGV, remaining power, and the sequence of task packages already held are included in the marginal incremental cost calculation to achieve dynamic updates of bids based on real-time status. Through repeated iterations of bidding and conflict resolution, released task packages are re-entered into the bidding set to optimize the allocation results. This effectively reduces communication overhead and computational load, improves the global optimality of the allocation results, the system's adaptability to dynamic environments, and the scalability and decentralized robustness of the multi-AGV system.

[0061] like Figure 2 As shown, the following are embodiments of the intelligent warehousing multi-task real-time scheduling system provided in this disclosure. This system and the intelligent warehousing multi-task real-time scheduling method of the above embodiments belong to the same inventive concept. For details not described in detail in the embodiments of the intelligent warehousing multi-task real-time scheduling system, please refer to the embodiments of the intelligent warehousing multi-task real-time scheduling method described above.

[0062] A smart warehouse multi-task real-time scheduling system includes: The data sensing module is used to acquire outbound order task data and the operating status data of each AGV in the warehousing system in real time, and to count the total number of tasks to be assigned and the number of currently idle AGVs. The outbound order task data includes the three-dimensional coordinates of the target outbound storage location for each outbound order task to be assigned, and the AGV operating status data includes the current position coordinates and remaining power of each AGV.

[0063] The task association module is used to calculate the spatial distance between any two outbound order tasks based on the three-dimensional coordinates of the target outbound storage location of each outbound order task to be assigned, and to construct a task association matrix using the spatial distance as a measure of the spatial correlation between outbound order tasks.

[0064] The dynamic grouping module is used to aggregate all outbound order tasks to be assigned into multiple task packages and output them to all currently idle AGVs based on the task association matrix, the total number of tasks to be assigned, and the number of currently idle AGVs. The total number of aggregated task packages does not exceed the number of currently idle AGVs, and the number of outbound order tasks contained in each task package is not less than a preset threshold for the minimum number of tasks contained in a single task package.

[0065] An improved CBBA solving module is used to bid on and resolve conflicts for each task package based on the task package, the current position coordinates of the AGV, and the remaining power of the AGV. Each currently idle AGV is assigned to an AGV using an improved consistency binding algorithm with the goal of minimizing path cost. The module outputs the allocation mapping relationship between each AGV and the task package, in which each task package is uniquely assigned to one AGV.

[0066] The path execution module is used to break down the task package allocated to each AGV into a single task instruction sequence according to the execution order based on the allocation mapping relationship and the current position coordinates of the AGV, plan the actual navigation path for each AGV in combination with the warehouse environment map, and send the single task instruction sequence and the corresponding actual navigation path to the corresponding AGV for execution.

[0067] The intelligent warehousing multi-task real-time scheduling system provided in this embodiment reduces the scale of the allocation problem by aggregating tasks into task packages based on spatial proximity. It adopts a marginal incremental cost-driven bidding mechanism so that the bid of each AGV reflects the actual incremental cost of executing the task package. Combined with localized consensus communication limited to adjacent AGVs, it improves the global optimality of the allocation result while reducing communication overhead and computational load. At the same time, by adaptively adjusting the total number of task packages according to the number of idle AGVs and participating in the bidding calculation with the real-time status of AGVs, the system has good dynamic environment adaptability, scalability and decentralized robustness.

[0068] Figure 3 A schematic diagram of the hardware structure of an electronic device for implementing various embodiments of the present invention.

[0069] The intelligent warehousing multi-task real-time scheduling method provided in this application embodiment can be applied to electronic devices. Those skilled in the art will understand that the electronic device structure involved in the embodiments of this invention does not constitute a limitation on the electronic device. An electronic device may include more or fewer components than illustrated, or combine certain components, or have different component arrangements. In the embodiments of this invention, the electronic device includes, but is not limited to, laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the embodiments of this application described and / or claimed herein.

[0070] Electronic devices may include processors, external memory interfaces, internal memory, universal serial bus (USB) interfaces, charging management modules, power management modules, batteries, wireless communication modules, audio modules, speakers, microphones, sensor modules, buttons, cameras, displays, and SIM card interfaces, etc.

[0071] A processor may include one or more processing units, such as: a central processing unit (CPU), an application processor (AP), a modem processor, a graphics processing unit (GPU), an image signal processor (ISP), a controller, memory, a video codec, a digital signal processor (DSP), a baseband processor, and / or a neural network processing unit (NPU). Different processing units may be independent devices or integrated into one or more processors.

[0072] The processor can serve as the nerve center and command center of an electronic device. The controller can generate operation control signals based on the instruction opcode and timing signals to control the fetching and execution of instructions.

[0073] The processor may also include memory for storing instructions and data. In some embodiments, the memory in the processor is a cache memory. This memory can store instructions or data that the processor has just used or that are used repeatedly. If the processor needs to use the instruction or data again, it can retrieve it directly from this memory. This avoids repeated accesses, reduces processor latency, and thus improves system efficiency.

[0074] An external storage interface (ESI) can be used to connect external memory cards, such as microSD cards, to expand the storage capacity of electronic devices. The external memory card communicates with the processor through the ESI to perform data storage functions, such as saving music and video files on the external memory card.

[0075] Internal memory can be used to store computer executable program code, which includes instructions. The processor executes various functional applications and data processing of electronic devices by running the instructions stored in internal memory. Internal memory can include a program storage area and a data storage area. Internal memory can include high-speed random access memory, and can also include non-volatile memory, such as at least one disk storage device, flash memory device, universal flash storage (UFS), etc.

[0076] Wireless communication functionality in electronic devices can be achieved through antennas, wireless communication modules, modem processors, and baseband processors.

[0077] Wireless communication modules can provide solutions for wireless communication applications in electronic devices, including wireless local area networks (WLANs) (such as wireless fidelity (Wi-Fi) networks), Bluetooth (BT), global navigation satellite system (GNSS), frequency modulation (FM), near field communication (NFC), and infrared (IR) technologies.

[0078] Electronic devices can implement audio functions through audio modules, speakers, receivers, microphones, headphone jacks, and application processors.

[0079] Electronic devices can achieve shooting functions through ISPs, cameras, video codecs, GPUs, displays, and application processors.

[0080] Electronic devices can achieve display functions through GPUs, displays, and application processors.

[0081] A GPU is a microprocessor for image processing, connected to the display screen and application processor. GPUs are used to perform mathematical and geometric calculations for graphics rendering. A processor may include one or more GPUs, which execute program instructions to generate or modify display information.

[0082] A display screen is used to display images, videos, etc. A display screen includes a display panel.

[0083] The aforementioned electronic device realizes the intelligent warehousing multi-task real-time scheduling method of this application, which reduces the scale of the allocation problem through task aggregation, a bidding mechanism driven by marginal incremental cost, localized consensus communication between neighboring AGVs, and adaptive adjustment of task packages according to the number of idle AGVs. This achieves the beneficial effects of reducing communication overhead and computational load, improving global optimality of allocation, and dynamic adaptability of the system.

[0084] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A real-time scheduling method for multi-task intelligent warehousing, characterized in that, Includes the following steps: S1: Real-time acquisition of outbound order task data and operating status data of each AGV within the warehousing system, and statistical analysis of the total number of tasks to be assigned and the number of currently idle AGVs; the outbound order task data includes the three-dimensional coordinates of the target outbound storage location for each outbound order task to be assigned, and the AGV operating status data includes the current position coordinates and remaining power of each AGV. S2: Based on the three-dimensional coordinates of the target outbound storage location of each outbound order task to be assigned, calculate the spatial distance between any two outbound order tasks, and use the spatial distance as a measure of the spatial correlation between outbound order tasks to construct a task correlation matrix. S3: Based on the task association matrix, the total number of tasks to be assigned, and the current number of idle AGVs, aggregate all outbound order tasks to be assigned into multiple task packages and output them to all currently idle AGVs; wherein, the total number of aggregated task packages does not exceed the current number of idle AGVs, and the number of outbound order tasks contained in each task package is not less than the preset threshold for the minimum number of tasks contained in a single task package. S4: Based on the task package and the current position coordinates and remaining power of the AGV, each currently idle AGV adopts an improved consistency binding algorithm to bid for and resolve conflicts of each task package with the goal of minimizing path cost, and outputs the allocation mapping relationship between each AGV and the task package. In the allocation mapping relationship, each task package is uniquely assigned to one AGV. S5: Based on the allocation mapping relationship and the current position coordinates of the AGV, the task package allocated to each AGV is broken down into a single task instruction sequence according to the execution order. Combined with the warehouse environment map, an actual navigation path is planned for each AGV, and the single task instruction sequence and the corresponding actual navigation path are sent to the corresponding AGV for execution.

2. The intelligent warehousing multi-task real-time scheduling method according to claim 1, characterized in that, Step S1 includes: The warehouse management system obtains the list of outbound order tasks to be assigned in real time through its external interface. Each outbound order task in the list includes a task number, a target outbound location number, the three-dimensional coordinates of the target outbound location, the order priority, and the task generation timestamp. Extract the three-dimensional coordinates of the target outbound storage location for each outbound order task from the obtained outbound order task list; The status information of all AGVs in the warehouse is obtained in real time through the status monitoring interface of the AGV scheduling system. The status information includes AGV number, current position coordinates, current movement status, remaining power, and currently assigned task list; the current position coordinates and remaining power of each AGV are extracted from the obtained status information. The total number of tasks to be assigned in the obtained outbound order task list is counted, and the number of AGVs currently in an idle state in the status information is counted as the current number of idle AGVs.

3. The intelligent warehousing multi-task real-time scheduling method according to claim 1, characterized in that, Step S2 includes: Extract the three-dimensional coordinates of the target outbound storage location corresponding to each outbound order task to be assigned from the data of each outbound order task. Let there be a total of The number of outbound order tasks to be assigned, the first The three-dimensional coordinates of the target outbound storage location for each outbound order task are as follows: ,in For any two outbound order tasks and The spatial distance between the target outbound storage locations is calculated using the following formula. : The calculated spatial distance As an outbound order task With outbound order task Spatial correlation measurement; organize the spatial distances between all outbound order tasks in a matrix form to construct... Task association matrix : matrix elements Indicates the first The outbound order task and the first The spatial distance between each outbound order task, when hour .

4. The intelligent warehousing multi-task real-time scheduling method according to claim 3, characterized in that, Step S3 includes: Based on the total number of outbound order tasks to be assigned and the number of currently idle AGVs The total number of task packages formed by aggregation is determined according to the following formula. : in The minimum number of tasks a single task package must contain is a preset threshold. Indicates rounding up; Set spatial distance threshold The task association matrix Medium spatial distance less than Task pairs are marked as aggregateable; a task association graph is constructed using the task pairs marked as aggregateable as edges, and the connected components in the task association graph are extracted as the initial task clusters; Count the number of initial task clusters ,like Then, in order of increasing number of tasks within the initial task cluster, smaller clusters are merged into their nearest spatially adjacent clusters, until the number of tasks in the initial task cluster equals the number of clusters in the previous cluster. ;like Then, the cluster with the largest number of tasks within the initial task cluster is selected for splitting. The splitting criterion is the maximum spatial distance between tasks within the cluster. The two tasks corresponding to the maximum spatial distance are used as seeds for two new clusters, and the remaining tasks are redistributed according to the principle of closest spatial distance, until the number of initial task clusters equals the number of tasks in the cluster. Each of the finalized task clusters will be treated as a task package.

5. The intelligent warehousing multi-task real-time scheduling method according to claim 4, characterized in that, Step S4 includes: The following bidding process will be conducted: All task packages are treated as a set of task packages to be auctioned, which is the set of task packages that have not yet been assigned to any AGV and are currently available for auction; each currently idle AGV is initialized. Task package bidding list Empty list, task package allocation list empty sequence, timestamp vector The zero vector, the The task packages in the AGV are arranged according to the execution order of the AGV; For AGV and any task package in the set of task packages to be auctioned Let the AGV currently The existing task package sequence is Calculate the task package according to the following formula. Insert sequence at optimal position The resulting marginal increase in the total path : in Indicates AGV According to sequence The shortest path total length of the shortest path to the unloading point is obtained by sequentially accessing all target outbound storage locations in each task package and returning the unloading point. The access order of the target outbound storage locations of each outbound order task in the task package is determined according to the shortest path principle. Indicates the task package Insert sequence at optimal position The resulting new sequence has the optimal position determined according to the shortest path principle. This represents the total length of the shortest path executed according to the new sequence; Determine AGV For any task package in the set of task packages to be auctioned bid If the task package Insertion sequence If the remaining battery power of the AGV exceeds its maximum travel distance, then... Let it be negative infinity; Select the highest-bid task package from the set of task packages to be auctioned and add it to each currently idle AGV. Whenever a new task pack is added Then, the task package is removed from the set of task packages to be auctioned, and the results are updated according to the shortest path principle. The execution order of each task package is determined, and the virtual current position of the AGV is updated to the position after the update is completed. After the endpoint of all task packages, update the virtual remaining power of the AGV to the remaining value after deducting the power consumed along that path; Repeat the above bidding calculation and selection operations until the virtual remaining power of the AGV is insufficient to support the insertion of any remaining task package into the set of task packages to be bid on, or until the set of task packages to be bid on is empty, to obtain the task package allocation list of each currently idle AGV. As a candidate allocation scheme, and compared with the task package bidding list and timestamp vector Packaged into consensus messages.

6. The intelligent warehousing multi-task real-time scheduling method according to claim 5, characterized in that, Step S4 further includes: Perform the following conflict resolution process: Receive consensus messages from each currently idle AGV and obtain the set of task packages to be auctioned; Set conflict perception radius The topological distance is calculated by taking the shortest navigation path distance between any two AGVs in the warehouse environment map; only when the topological distance between the two AGVs is less than 1. At that time, a strong connection consensus link is established between the two AGVs and consensus messages are exchanged; for topological distances greater than or equal to The two AGVs do not establish a strong consensus link, but exchange consensus messages in an asynchronous weak consensus manner; Within a strongly connected consensus link, consensus messages are received from neighboring AGVs. For task packages that are simultaneously won by multiple AGVs, the bids of each AGV for the task package are compared, and the AGV with the lowest bid is assigned a task package from its own task package allocation list. Release the task package in the middle; if the bids are the same, compare the timestamps and the AGV with the later timestamp releases the task package. The released task package is re-added to the set of task packages to be auctioned; Repeat the bidding process and conflict resolution process until the convergence condition is met, and output the allocation mapping relationship between each AGV and the task package. In the allocation mapping relationship, each task package is uniquely assigned to one AGV.

7. The intelligent warehousing multi-task real-time scheduling method according to claim 6, characterized in that, Step S5 includes: Based on the aforementioned allocation mapping relationship, each outbound order task in each task package assigned to each AGV is allocated according to the task package allocation list of that AGV. The execution sequence is broken down into individual task instructions, and each individual task instruction includes at least the three-dimensional coordinates of the target outbound storage location, the task execution sequence number, and the expected arrival time window. Starting from the current position coordinates of each AGV, and taking the three-dimensional coordinates of the target outbound location of each outbound order task as the waypoint, combined with the constraints of aisles and obstacles in the warehouse environment map, the actual navigation path is planned for each AGV. The planned actual navigation path and the corresponding sequence of individual task instructions are sent to the onboard control system of the corresponding AGV; The system monitors the task execution progress of each AGV in real time, and feeds back the status information to the scheduling system when an AGV completes all tasks or encounters an abnormality.

8. The intelligent warehousing multi-task real-time scheduling method according to claim 6, characterized in that, The convergence condition is: all task packages have been uniquely assigned to a single AGV and no AGV has made a higher bid for the assigned task package, or the number of iterations has reached the preset maximum number of iterations.

9. A smart warehousing multi-task real-time scheduling system, characterized in that, The system employs the intelligent warehousing multi-task real-time scheduling method as described in any one of claims 1 to 8; The system includes: The data sensing module is used to acquire outbound order task data and the operating status data of each AGV in the warehousing system in real time, and to count the total number of tasks to be assigned and the number of currently idle AGVs. The outbound order task data includes the three-dimensional coordinates of the target outbound storage location of each outbound order task to be assigned, and the operating status data of the AGV includes the current position coordinates and remaining power of each AGV. The task association module is used to calculate the spatial distance between any two outbound order tasks based on the three-dimensional coordinates of the target outbound location of each outbound order task to be assigned, and to construct a task association matrix using the spatial distance as a measure of the spatial correlation between outbound order tasks. The dynamic grouping module is used to aggregate all outbound order tasks to be assigned into multiple task packages and output them to all currently idle AGVs based on the task association matrix, the total number of tasks to be assigned, and the current number of idle AGVs. The total number of aggregated task packages does not exceed the current number of idle AGVs, and the number of outbound order tasks contained in each task package is not less than a preset threshold for the minimum number of tasks contained in a single task package. An improved CBBA solving module is used to bid on and resolve conflicts for each task package based on the task package, the current position coordinates and remaining power of the AGV, and each currently idle AGV adopts an improved consistency binding algorithm with the goal of minimizing path cost. The module outputs the allocation mapping relationship between each AGV and the task package, in which each task package is uniquely assigned to one AGV. The path execution module is used to break down the task package allocated to each AGV into a single task instruction sequence according to the execution order based on the allocation mapping relationship and the current position coordinates of the AGV, plan the actual navigation path for each AGV in combination with the warehouse environment map, and send the single task instruction sequence and the corresponding actual navigation path to the corresponding AGV for execution.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the intelligent warehousing multi-task real-time scheduling method as described in any one of claims 1 to 8.

Images