Multi-agent cooperative crowdsourcing task processing method and system

Abstract

The invention discloses a crowdsourcing task processing method and system based on multi-agent cooperation, and relates to the technical field of artificial intelligence, and the method comprises the steps: receiving a user order which comprises a service type, a starting point address, an end point address and the arrival time; decomposing all user orders according to a preset rule to generate a task package; determining the predicted arrival time of each task package according to the pickup address and the delivery address, and generating an optimal allocation scheme according to the real-time state of the worker and the predicted arrival time of each task package; and performing task pushing according to the optimal allocation scheme. Intelligent decomposition and dynamic allocation of tasks are achieved through cooperation of large language model agents, the completion efficiency and quality of crowdsourcing tasks are remarkably improved, predicted arrival time is dynamically calculated through path planning agents, execution of time-sensitive tasks is optimized, and delay is reduced.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and particularly to a method and system for processing crowdsourcing tasks with multi-agent collaboration. Background Art

[0002] The statements in this part merely provide background technical information related to the present invention and do not necessarily constitute prior art.

[0003] Crowdsourcing is a way of outsourcing tasks, problems or projects to a large group of solution providers through the Internet, rather than using a dedicated company to complete them. It utilizes the power of a large number of individuals to jointly solve problems and execute tasks, and usually can obtain a large amount of feedback or results in a short time. For example, the ride-hailing platform matches the travel needs of passengers with the transport capacity of drivers through the crowdsourcing model, greatly improving the travel efficiency; the food delivery platform assigns delivery tasks to a large number of riders through the crowdsourcing platform to meet the immediate delivery needs of users. The advantages of crowdsourcing include low cost, fast execution, etc., which can promote social development and improve people's quality of life.

[0004] In recent years, with the development of artificial intelligence technology, large language models have shown great potential in natural language processing and task automation. A large language model can be regarded as an agent, and a multi-agent system is a system in which multiple agents interact and collaborate to achieve a common goal. These agents can create subtasks, search for information, and seek help from each other, thus showing higher performance when dealing with complex tasks. Some crowdsourcing systems have begun to use large language models to optimize task allocation.

[0005] However, these systems are often limited to the application of a single model and cannot fully utilize the collaborative ability of multiple large language model agents to handle more complex tasks. In scenarios that require multi-step decomposition and dynamic coordination, existing systems are difficult to achieve efficient task decomposition and real-time status update, resulting in limitations in task completion efficiency and quality. Summary of the Invention

[0006] To solve the above problems, the present invention proposes a method and system for processing crowdsourcing tasks with multi-agent collaboration, which realizes intelligent decomposition and dynamic allocation of tasks through the collaboration of large language model agents, significantly improves the completion efficiency and quality of crowdsourcing tasks, uses path planning agents to dynamically calculate the estimated time of arrival, optimizes the execution of time-sensitive tasks, and reduces delays.

[0007] To achieve the above object, the present invention adopts the following technical solutions:

[0008] In the first aspect, the present invention provides a method for processing crowdsourcing tasks with multi-agent collaboration, including:

[0009] Receive a user order, where the user order includes service type, starting address, ending address, and delivery time;

[0010] Decompose all user orders according to preset rules to generate task packages;

[0011] Screen candidate workers based on the obtained worker status; among them, workers in the idle state are candidate workers; for workers in the non-idle state, with the current task volume and historical task completion rate as constraints, workers who can complete the current task volume before the delivery time of the task package are candidate workers; where the minimum delivery time within the task package is defined as the delivery time of the task package;

[0012] Based on the current location of the candidate worker, the currently executing task, the starting address and ending address of the task package, determine the estimated arrival time of the task package, and combine the set scoring metrics of distance, worker rating, task load, and path coherence to obtain the score of the candidate worker-task package combination, thereby determining the optimal allocation plan and pushing tasks according to the optimal allocation plan.[[ID=,10]]

[0013] As an alternative implementation, the worker status includes current location, idle state, current task volume, task queue, and historical task completion rate;

[0014] For workers in the non-idle state, when the current task volume is less than the set maximum threshold and the historical task completion rate is higher than the set minimum threshold, calculate the estimated completion time of the current task volume of the non-idle state worker, and thus screen out workers who can complete the current task volume before the deadline of the task package delivery time and use them as candidate workers.

[0015] As an alternative implementation, the process of determining the estimated arrival time of the task package includes: calculating the distance d1 and estimated time t1 from the current location of the candidate worker to the starting address of the new task, and the distance d2 and estimated time t2 from the starting address to the ending address; and for candidate workers with a current task, determine the time to start executing the new task according to the current task completion time; at the same time, introduce additional buffer time according to the set peak hours and non-peak hours; then the estimated arrival time is the sum of the time to start executing the new task, the estimated time t1, the estimated time t2, and the buffer time.

[0016] As an alternative implementation, the timeliness score is where ETA is the estimated arrival time, and the units of the task package delivery time and ETA are minutes;

[0017] The distance score is where the units of d1 and d2 are kilometers, and the score range is from 0 to 10;

[0018] The worker score is the historical task completion rate × 10, where the historical task completion rate is expressed as a percentage and the score ranges from 0 to 10;

[0019] The task load score = 10 - min(10, current task volume × 2), where the current task volume is an integer and the score ranges from 0 to 10;

[0020] The path coherence score is as follows: if the distance between the end address of the candidate worker's current task and the start address of the new task is less than or equal to the set distance threshold, a score of 2 is obtained; otherwise, a score of 0 is obtained.

[0021] According to the set circles of timeliness score, distance score, worker score, task load score, and path coherence score, the weighted total score is obtained through weighted calculation;

[0022] Sort all worker - task package combinations in descending order of the weighted total score, and sequentially traverse the sorted combination list to establish the matching between candidate workers and task packages, ensuring that each candidate worker and each task package are only assigned once until all task packages are assigned or there are no available workers.

[0023] As an alternative implementation, the preset rules include geographical clustering or load prediction;

[0024] Geographical clustering specifically means: using a clustering algorithm to merge user orders within a set distance radius into one task package;

[0025] Load prediction specifically means: based on historical order data, using a time - series analysis model to predict the regional order volume within a set future time period. If the predicted order volume exceeds the set threshold, the order task is split into multiple task packages.

[0026] As an alternative implementation, after receiving a user order, the user order is verified. The verification process includes: checking whether the required fields are empty, checking whether the service type is a preset type, checking whether the delivery time is later than the current time, and the time interval between the delivery time and the current time is not less than the set interval threshold.

[0027] In a second aspect, the present invention provides a crowdsourcing task processing system for multi - agent collaboration, including:

[0028] A management module configured to receive a user order, where the user order includes a service type, a start address, an end address, and a delivery time;

[0029] A decomposition module configured to decompose all user orders according to preset rules to generate task packages;

[0030] A screening module, configured to screen candidate workers according to the obtained worker status; among them, workers in the idle state are candidate workers; for workers in the non-idle state, with the current task volume and historical task completion rate as constraint conditions, workers who can complete the current task volume before the task package delivery time are candidate workers; among them, the minimum delivery time in the task package is defined as the task package delivery time.

[0031] An allocation module, configured to determine the estimated arrival time of the task package according to the current location of the candidate worker, the current task being executed, the starting address and the ending address of the task package, combine the set scoring metrics of distance, worker score, task load and path coherence, obtain the score of the candidate worker-task package combination, thereby determine the optimal allocation plan, and push tasks according to the optimal allocation plan.

[0032] In a third aspect, the present invention provides an electronic device, including a memory and a processor, as well as computer instructions stored on the memory and running on the processor. When the computer instructions are run by the processor, the method described in the first aspect is completed.

[0033] In a fourth aspect, the present invention provides a computer-readable storage medium for storing computer instructions. When the computer instructions are executed by a processor, the method described in the first aspect is completed.

[0034] In a fifth aspect, the present invention provides a computer program product, including a computer program. When the computer program is executed by a processor, the method described in the first aspect is implemented. [[ID=—13]]

[0035] Compared with the prior art, the beneficial effects of the present invention are:

[0036] The present invention discloses a crowdsourcing system based on multi-agent collaboration of large language models, including a management agent, a task decomposition agent, an allocation agent, a path planning agent and a worker scheduling agent. The user submits task information through the mobile APP. After verification by the management agent, it is handed over to the task decomposition agent to split or merge orders to generate task packages. The allocation agent combines the worker status and the estimated arrival time of the path planning agent, and outputs the optimal allocation plan through a multi-objective optimization model. The worker scheduling agent binds the tasks and pushes them to the workers for execution. Through the collaboration of large language model agents, intelligent decomposition and multi-objective dynamic allocation of tasks are realized. Compared with traditional single-model systems, the completion efficiency and quality of mobile crowdsourcing tasks are significantly improved, and the efficiency and flexibility of crowdsourcing task processing are improved; the path planning agent is used to dynamically calculate the estimated arrival time, optimize the execution of time-sensitive tasks, and reduce delays.

[0037] The advantages of the additional aspects of the present invention will be partially given in the following description, partially become obvious from the following description, or be understood through the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the accompanying drawings required for the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only the embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can also be obtained based on the provided drawings.

[0039] Figure 1 It is a flowchart of a crowdsourcing task processing method for multi-agent collaboration provided in Embodiment 1 of the present invention. Detailed implementation manners

[0040] The following will further illustrate the present invention in conjunction with the accompanying drawings and embodiments.

[0041] It should be noted that the following detailed descriptions are all exemplary and are intended to provide further illustrations of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present invention belongs.

[0042] It should be noted that the terms used herein are only for describing specific implementation manners and are not intended to limit the exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that the terms "comprising" and "including" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or device that includes a series of steps or units does not necessarily limit to those clearly listed steps or units, but may include other steps or units not clearly listed or inherent to these processes, methods, products, or devices.

[0043] In the case of no conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other.

[0044] Embodiment 1

[0045] This embodiment provides a crowdsourcing task processing method for multi-agent collaboration. Among them, the multi-agents include a management agent, a decomposition agent, an allocation agent, a path planning agent, and a worker scheduling agent; combined with Figure 1 , the method includes:

[0046] S1: Receive a user order, where the user order includes a service type, a starting address, an ending address, and a delivery time.

[0047] Among them, the service type includes takeaway, express delivery, errand running, taxi, or housekeeping, etc.

[0048] Exemplarily, the user fills in structured information such as service type, starting address, ending address, and delivery time through the mobile user APP. After receiving the structured information, the user APP stores it in the form of a form and transmits the obtained task form to the management agent.

[0049] Exemplarily, the user selects a service type (such as takeout or express delivery), fills in the starting address (such as "No. 2000, Shunhua Road, Licheng District, Jinan City, Shandong Province"), the ending address (such as "No. 1500, Shunhua Road, Licheng District, Jinan City, Shandong Province"), and the delivery time (such as "12:00 on March 10, 2025");

[0050] The task form is stored in JSON format. For example:

[0051] {

[0052] "service_type": "takeout",

[0053] "pickup_address": "No. 2000, Shunhua Road, Licheng District, Jinan City, Shandong Province",

[0054] "delivery_address": "No. 1500, Shunhua Road, Licheng District, Jinan City, Shandong Province",

[0055] "delivery_time": "2025-03-10 12:00:00"

[0056] }。[[ID=2,5]]

[0057] In this embodiment, after receiving the task form, the management agent first verifies the task form. After the verification is completed, the task form is transmitted to the task decomposition agent, and a decomposition request is sent to the task decomposition agent.

[0058] As an alternative implementation, the verification process of the task form includes:

[0059] Check whether the required fields are empty, such as service type, starting address, ending address, or delivery time;

[0060] Check whether the service type is correct, that is, whether it is a preset type, and only allow preset types such as takeout, express delivery, errand running, taxi, or housekeeping;

[0061] Check the time validity, that is, whether the delivery time is later than the current time, and the time interval between the delivery time and the current time is not less than the set interval threshold, such as the delivery time is at least 30 minutes later than the current time.

[0062] As an alternative implementation, if the task form verification fails, the management agent returns an error message to the user APP, such as "The delivery time is invalid. Please select again"; if it passes, the task form is passed to the task decomposition agent along with a decomposition request.

[0063] S2: Decompose all user orders according to preset rules to generate task packages.

[0064] Among them, the task decomposition agent splits or merges orders according to preset rules and passes the obtained task packages to the management agent.

[0065] The preset rules include geographical clustering or load prediction, specifically:

[0066] (1) Geographical clustering: Use the DBSCAN algorithm (Density-Based Spatial Clustering of Applications with Noise, a density-based unsupervised clustering algorithm) to merge user orders within a set distance radius range (such as 500 meters) into one task package (such as multiple takeaway orders in the same office building); assume there are 3 takeaway orders with pickup addresses "No. 10 Quancheng Road", "No. 12 Quancheng Road", and "No. 2000 Shunhua Road" respectively, then the first two are merged into one task package, and the latter becomes a separate task package.

[0067] (2) Load prediction: Based on historical data, use a time series analysis model to predict the regional order volume within a set future time period and split tasks in high-load areas in advance; if the prediction result shows high load, that is, the order volume exceeds the set threshold, the order tasks are split into multiple small task packages in advance to avoid overloading workers.

[0068] For example, based on the order volume in the same period in the past 7 days, use the ARIMA model to predict the order volume in the Shunhua Road area within the next 30 minutes; if the prediction result shows high load (such as exceeding 20 orders), the task is split into multiple small task packages in advance.

[0069] Exemplarily, after decomposition, the task packages are returned to the management agent in the form of a list, for example:

[0071] {"task_id":"T001","orders":["order1","order2"],"area":"Quancheng Road"},

[0072] {"task_id":"T002","orders":["order3"],"area":"Shunhua Road"}

[0073] 。 ​

[0074] S3: Screen candidate workers based on the obtained worker status; among them, workers with an idle status are candidate workers; for workers with a non-idle status, with the current task volume and historical task completion rate as constraints, workers who can complete the current task volume before the task package delivery time are candidate workers.

[0075] Specifically:

[0076] The management agent passes the task package to the allocation agent and makes an allocation request. The allocation agent performs the following operations:

[0077] Obtain the real-time status of workers from the worker scheduling agent, including: current location (latitude and longitude coordinates, such as "longitude: 117.129, latitude: 36.651"), idle status (represented by a boolean value, "idle" = 1, "busy" = 0), current task volume (an integer value, such as "3" indicating that there are 3 tasks to be completed currently), task queue, and historical task completion rate (the percentage of successfully completed tasks in the total number of received tasks, such as "95%" indicating the historical task completion rate).

[0078] Subsequently, the allocation agent screens candidate workers based on the above information. The screening rules are as follows:

[0079] Prioritize workers with an idle status (idle marked as 1) as candidate workers;

[0080] For workers with a busy status (busy marked as 0, i.e., non-idle status), perform the following screening process:

[0081] First, limit the current task volume of candidate busy workers to be less than the set maximum threshold (such as 3), and ensure that the historical task completion rate of candidate workers is higher than the set minimum threshold (such as 80%). Then calculate the estimated completion time of the current task volume of busy workers, and screen out workers who can complete the current task volume before the task package delivery time deadline, and use them as candidate workers; among them, define the minimum delivery time in the task package as the task package delivery time.

[0082] It can be understood that the estimated completion time of the current task volume of busy workers is generated based on the current location, end address, etc., based on map data.

[0083] S4: Determine the estimated arrival time of the task package based on the current location of the candidate worker, the current task being executed, the starting address and ending address of the task package, and combine the set scoring metrics of distance, worker score, task load, and path coherence to obtain the score of the candidate worker-task package combination, thereby determining the optimal allocation plan and pushing the task according to the optimal allocation plan.

[0084] Specifically:

[0085] The allocation agent calls the path planning agent and receives the data returned by it: When calling the path planning agent, the current location of each candidate worker (or the end location of its last task), the starting address of the task package, and the ending address of the task package are passed in.

[0086] Based on the map data, the path planning agent generates accurate distance and time estimates, calculates the distance d1 and the estimated time t1 from the current location of the candidate worker (or the end location of its last task) to the starting address of the new task, and the distance d2 and the estimated time t2 from the starting address to the ending address. The specific implementation can obtain the real-time path planning result by calling the map application API.

[0087] To adapt to changes in traffic conditions, a time buffer mechanism is introduced: During the set peak hours (such as 11:00 - 12:00 and 17:00 - 19:00 every day), an additional 25% buffer time is added; during non-peak hours, an additional 10% buffer time is added.

[0088] For candidate workers with current tasks, the estimated arrival time needs to additionally consider the time required to complete the current task; then the final estimated time of arrival (ETA) is calculated and returned by the path planning agent, and its formula is: ETA = the time when the candidate worker starts to execute the new task + t1 + t2 + buffer time.

[0089] The allocation agent calculates a comprehensive score for each candidate worker-task combination. The scoring dimensions include timeliness, distance, worker score, task load, and path coherence.

[0090] Specifically:

[0091] The timeliness score is calculated based on the task delivery time of the task package, and the calculation formula is: Among them, the units of the task package delivery time and ETA are minutes.

[0092] The distance score is calculated based on the total travel distance, and the calculation formula is: Among them, the units of d1 and d2 are kilometers, and the score range is from 0 to 10.

[0093] The worker score is calculated based on the worker's historical task completion rate, and the calculation formula is: worker score = historical task completion rate × 10, where the historical task completion rate is expressed as a percentage, and the score range is from 0 to 10.

[0094] The task load score is calculated based on the worker's current task volume, and the calculation formula is: task load score = 10 - min(10, current task volume × 2), where the current task volume is an integer, and the score range is from 0 to 10.

[0095] The calculation of the path coherence score follows the following rules: If the distance between the end address of the current task of the candidate worker and the start address of the new task is less than or equal to the set distance threshold (such as 2 kilometers), then 2 points are obtained; otherwise, 0 points are obtained.

[0096] Then calculate the weighted total score of the above scores: Weighted total score = timeliness score × timeliness weight + distance score × distance weight + worker rating × worker rating weight + task load score × load weight + path coherence score × path coherence weight, where the range of each weight is between 0 and 1, and the sum of the weights is 1, which can be custom set according to experience.

[0097] Finally, the allocation intelligent agent uses the greedy allocation algorithm for allocation. The specific steps are as follows: Sort all worker-task package combinations in descending order of the weighted total score; traverse the sorted combination list in turn to establish the matching between the candidate worker and the task; during the matching process, ensure that each candidate worker and each task are only allocated once; repeat the steps until all tasks are allocated or there are no available workers.

[0098] The final allocation plan is returned to the management intelligent agent in JSON format, including the worker ID and the estimated time of arrival (ETA) corresponding to each task package. The output format is as follows:

[0099] {

[0100] "T001":{"worker_id":"W001","ETA":"2025-03-10 11:50:00"},

[0101] "T002":{"worker_id":"W002","ETA":"2025-03-10 11:55:00"}

[0102] }

[0103] S5: Push tasks according to the optimal allocation plan.

[0104] Specifically, after receiving the optimal allocation plan, the management intelligent agent sends it to the worker scheduling intelligent agent. The worker scheduling intelligent agent binds the task with the worker and pushes the task notification through the worker APP; the worker APP displays the task details (such as the start address, end address, delivery time, and remuneration, etc.). After the worker confirms receiving the order, the status is updated to "in execution", and the task execution starts.

[0105] In summary, through the method of this embodiment, the efficiency and flexibility of crowdsourcing task processing can be improved.

[0106] It should be noted that the acquisition of all data is based on compliance with laws and regulations and user consent, and the data is legally applied.

[0107] Embodiment 2

[0108] This embodiment provides a crowdsourcing task processing system for multi-agent collaboration, including:

[0109] A management module, configured to receive user orders, where the user orders include service type, starting address, ending address, and delivery time;

[0110] A decomposition module, configured to decompose all user orders according to preset rules to generate task packages;

[0111] A screening module, configured to screen candidate workers according to the obtained worker status; among them, workers in the idle state are candidate workers; for workers in the non-idle state, with the current task volume and historical task completion rate as constraint conditions, workers who can complete the current task volume before the delivery time of the task package are candidate workers; among them, the minimum delivery time in the task package is defined as the delivery time of the task package;

[0112] An allocation module, configured to determine the estimated arrival time of the task package according to the current location of the candidate worker, the current task being executed, the starting address and ending address of the task package, combine the set scoring metrics of distance, worker rating, task load, and path coherence, obtain the score of the candidate worker-task package combination, thereby determine the optimal allocation plan, and push the task according to the optimal allocation plan.

[0113] In this embodiment, the process of determining the estimated arrival time of the task package includes: calculating the distance d1 and estimated time t1 from the current location of the candidate worker to the starting address of the new task, and the distance d2 and estimated time t2 from the starting address to the ending address; and for candidate workers with current tasks, determine the time to start executing the new task according to the completion time of the current task; at the same time, according to the set peak hours and non-peak hours, introduce additional buffer time; then the estimated arrival time is the sum of the time to start executing the new task, the estimated time t1, the estimated time t2, and the buffer time.

[0114] In this embodiment, the timeliness score is where ETA is the estimated arrival time, and the units of the task package delivery time and ETA are minutes;

[0115] The distance score is where the units of d1 and d2 are kilometers, and the score range is 0 to 10;

[0116] The worker rating is the historical task completion rate × 10, where the historical task completion rate is expressed as a percentage, and the score range is 0 to 10;

[0117] The task load score = 10 - min(10, current task volume × 2), where the current task volume is an integer and the score range is from 0 to 10;

[0118] The path coherence score is as follows: If the distance between the end address of the current task of the candidate worker and the start address of the new task is less than or equal to the set distance threshold, then the score is 2; otherwise, the score is 0.

[0119] According to the set circles of timeliness score, distance score, worker rating, task load score, and path coherence score, the weighted total score is obtained through weighted calculation;

[0120] Sort all worker - task package combinations in descending order of the weighted total score, traverse the sorted combination list in sequence, establish the matching of candidate workers and task packages, and ensure that each candidate worker and each task package are only assigned once until all task packages are assigned or there are no available workers.

[0121] In this embodiment, the preset rules include geographical clustering or load prediction;

[0122] Geographical clustering is specifically: Using a clustering algorithm, merge user orders within a set distance radius into one task package;

[0123] Load prediction is specifically: Based on historical order data, use a time - series analysis model to predict the regional order volume within a set future time period. If the predicted order volume exceeds the set threshold, then split the order task into multiple task packages.

[0124] In this embodiment, after receiving a user order, verify the user order. The verification process includes: checking whether the required fields are empty, checking whether the service type is the preset type, checking whether the delivery time is later than the current time, and the time interval between the delivery time and the current time is not less than the set interval threshold.

[0125] It should be noted here that the above - mentioned modules correspond to the steps described in Embodiment 1. The examples and application scenarios implemented by the above - mentioned modules and the corresponding steps are the same, but are not limited to the content disclosed in the above - mentioned Embodiment 1. It should be noted that the above - mentioned modules, as part of the system, can be executed in a computer system such as a set of computer - executable instructions.

[0126] In more embodiments, there is also provided:

[0127] An electronic device, including a memory, a processor, and computer instructions stored on the memory and running on the processor. When the computer instructions are run by the processor, the method described in Embodiment 1 is completed. For the sake of brevity, it will not be elaborated here.

[0128] It should be understood that in this embodiment, the processor may be a central processing unit (CPU), or the processor may also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc.

[0129] The memory may include a read-only memory and a random access memory, and provide instructions and data to the processor. A part of the memory may also include a non-volatile random access memory. For example, the memory may also store information about the device type.

[0130] A computer-readable storage medium for storing computer instructions, which when executed by the processor, implement the method described in Embodiment 1.

[0131] The method in Embodiment 1 can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules in the processor. The software module may be located in a mature storage medium in the art, such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. This storage medium is located in the memory, and the processor reads the information in the memory and combines its hardware to complete the steps of the above method. To avoid repetition, it will not be described in detail here.

[0132] A computer program product including a computer program, which when executed by the processor, implements the method described in Embodiment 1.

[0133] The present invention also provides at least one computer program product tangibly stored on a non-transitory computer-readable storage medium. The computer program product includes computer-executable instructions, such as instructions included in program modules, which are executed in a device on a target real or virtual processor to perform the process / method as described above. Generally, program modules include routines, programs, libraries, objects, classes, components, data structures, etc. that perform specific tasks or implement specific abstract data types. In various embodiments, the functions of program modules may be combined or divided as needed. The machine-executable instructions for program modules may be executed locally or within a distributed device. In a distributed device, program modules may be located in local and remote storage media.

[0134] The computer program code for implementing the method of the present invention can be written in one or more programming languages. This computer program code can be provided to the processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when the program code is executed by the computer or other programmable data processing device, the functions / operations specified in the flowchart and / or block diagram are implemented. The program code can be executed entirely on the computer, partially on the computer, as a stand-alone software package, partially on the computer and partially on a remote computer, or entirely on a remote computer or server.

[0135] In the context of the present invention, the computer program code or related data can be carried by any suitable carrier so that the device, apparatus, or processor can perform the various processes and operations described above. Examples of carriers include signals, computer-readable media, and the like. Examples of signals can include electrical, optical, radio, acoustic, or other forms of propagated signals, such as carrier waves, infrared signals, etc.

[0136] Those of ordinary skill in the art can realize that the units and algorithm steps of the examples described in conjunction with this embodiment can be implemented by electronic hardware or a combination of computer software and electronic hardware. Whether these functions are executed in hardware or software depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0137] Although the specific implementation manners of the present invention have been described above in conjunction with the accompanying drawings, this is not a limitation on the protection scope of the present invention. Those skilled in the art should understand that, based on the technical solution of the present invention, various modifications or deformations that can be made by those skilled in the art without creative efforts are still within the protection scope of the present invention.

Claims

1. A method for processing crowdsourcing tasks with multi-agent collaboration, characterized in that, Including: Receiving a user order, where the user order includes service type, starting address, ending address, and delivery time; Decomposing all user orders according to preset rules to generate task packages; Screening candidate workers based on the obtained worker status; among them, workers in the idle state are candidate workers; for workers in the non-idle state, with the current task volume and historical task completion rate as constraints, workers who can complete the current task volume before the delivery time of the task package are candidate workers; where the minimum delivery time within the task package is defined as the delivery time of the task package; Based on the current location of the candidate worker, the current task being executed, the starting address and ending address of the task package, determining the estimated arrival time of the task package, and combining the set scoring metrics of distance, worker score, task load, and path coherence to obtain the score of the candidate worker-task package combination, thereby determining the optimal allocation plan and pushing tasks according to the optimal allocation plan.

2. The multi-agent collaborative crowdsourcing task processing method according to claim 1, characterized in that, Worker status includes current location, idle state, current task volume, task queue, and historical task completion rate; For workers in the non-idle state, when the current task volume is less than the set maximum threshold and the historical task completion rate is higher than the set minimum threshold, calculate the estimated completion time of the current task volume of the non-idle state worker, and thereby screen out workers who can complete the current task volume before the deadline of the task package delivery time and use them as candidate workers.

3. A crowdsourcing task processing method for multi-agent collaboration according to claim 1, characterized in that, The process of determining the estimated arrival time of the task package includes: calculating the distance d1 and estimated time t1 from the current location of the candidate worker to the starting address of the new task, as well as the distance d2 and estimated time t2 from the starting address to the ending address; and for candidate workers with a current task, determining the time to start executing the new task according to the current task completion time; at the same time, introducing additional buffer time according to the set peak period and non-peak period; then the estimated arrival time is the sum of the time to start executing the new task, the estimated time t1, the estimated time t2, and the buffer time.

4. A crowdsourcing task processing method for multi-agent collaboration according to claim 1, characterized in that The aging score is where ETA is the estimated time of arrival, and the units of the task package delivery time and ETA are minutes; The distance score is where d1 and d2 are in kilometers, and the score ranges from 0 to 10; The worker score is the historical task completion rate × 10, where the historical task completion rate is expressed as a percentage, and the score range is from 0 to 10; The task load score = 10 - min(10, current task volume × 2), where the current task volume is an integer, and the score range is from 0 to 10; The path coherence score is: if the distance between the ending address of the candidate worker's current task and the starting address of the new task is less than or equal to the set distance threshold, then the score is 2 points, otherwise the score is 0 points; According to the set circle of timeliness score, distance score, worker score, task load score, and path coherence score, the weighted total score is obtained through weighted calculation; Sort all worker-task package combinations from high to low according to the weighted total score, traverse the sorted combination list in sequence, establish the matching between candidate workers and task packages, and ensure that each candidate worker and each task package are only allocated once until all task packages are allocated or there are no available workers.

5. A crowdsourcing task processing method for multi-agent collaboration according to claim 1, characterized in that, The preset rules include geographical clustering or load prediction; Geographical clustering specifically is: using a clustering algorithm to merge user orders within a set distance radius into one task package; The load prediction is specifically as follows: Based on historical order data, a time series analysis model is used to predict the regional order volume within a set future time period. If the predicted order volume exceeds the set threshold, the order task is split into multiple task packages.

6. The crowdsourcing task processing method for multi-agent collaboration according to claim 1, wherein, After receiving a user order, the user order is verified. The verification process includes: checking whether the required fields are empty, checking whether the service type is a preset type, checking whether the delivery time is later than the current time, and the time interval between the delivery time and the current time is not less than the set interval threshold.

7. A crowdsourcing task processing system for multi-agent collaboration, characterized in that, It includes: A management module, configured to receive a user order, where the user order includes a service type, a starting address, an ending address, and a delivery time; A decomposition module, configured to decompose all user orders according to preset rules to generate task packages; A screening module, configured to screen candidate workers according to the obtained worker status; among them, workers in the idle state are candidate workers; for workers in the non-idle state, with the current task volume and the historical task completion rate as constraint conditions, workers who can complete the current task volume before the delivery time of the task package are candidate workers; among them, the minimum delivery time within the task package is defined as the task package delivery time; An allocation module, configured to determine the estimated arrival time of the task package according to the current location, the current executing task of the candidate worker, the starting address and the ending address of the task package, and combine the set scoring metrics of distance, worker score, task load, and path coherence to obtain the score of the candidate worker-task package combination, thereby determining the optimal allocation plan and pushing the task according to the optimal allocation plan.

8. An electronic device, characterized in that, It includes a memory, a processor, and computer instructions stored on the memory and running on the processor. When the computer instructions are run by the processor, the method described in any one of claims 1-6 is completed.

9. A computer-readable storage medium, characterized in that, For storing computer instructions, when the computer instructions are executed by the processor, the method described in any one of claims 1-6 is completed.

10. A computer program product, characterized in that It includes a computer program, and when the computer program is executed by the processor, the method described in any one of claims 1-6 is implemented.

Images