Multi-agent task progress management method and device, medium, equipment and product

Abstract

The invention discloses a multi-agent task progress management method and device, a medium, equipment and a product, and relates to the technical field of computers. According to the method, a first agent sends sub-schedules corresponding to sub-tasks to a corresponding second agent according to a main schedule, then the second agent executes the sub-tasks according to the sub-schedules, the sub-schedules are updated based on task execution results of the sub-tasks, and when the sub-tasks are completed, the main schedule is updated according to the sub-schedules. Therefore, the first intelligent agent can accurately memorize the global task state and the historical information and can timely and effectively perform task planning adjustment or resource redistribution again; and the completion state of the task target can be accurately transmitted among the agents in the multi-agent collaboration system, and the task progress management and collaboration among the agents can be accurately carried out.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technology, and in particular to a multi-agent task progress management method, apparatus, medium, equipment, and product. Background Art

[0002] In multi-agent systems (MAS), effective context management and accurate task progress tracking are key to ensuring efficient and reliable system operation. As task complexity increases, agents need to process and transmit large amounts of dynamic information. In multi-agent collaboration or long task chains, critical contextual information is easily lost, distorted, or inconsistent during transmission, leading to subsequent decisions based on incomplete or erroneous information. Furthermore, multi-agent systems lack real-time and accurate perception of task progress, making it difficult for planners in these systems to understand the true status of the task, hindering effective dynamic adjustments and interventions. Summary of the Invention

[0003] This summary is provided to briefly introduce concepts that will be described in detail in the detailed description below. This summary is not intended to identify key features or essential features of the claimed technical solution, nor is it intended to limit the scope of the claimed technical solution.

[0004] In a first aspect, the present disclosure provides a multi-agent task progress management method, comprising: Decomposing the target task into a plurality of subtasks by the first agent, and constructing a master schedule corresponding to the target task, wherein the master schedule includes the task objectives corresponding to each of the subtasks and the completion status corresponding to the task objectives; Sending, by the first agent, a sub-schedule corresponding to each sub-task to a corresponding second agent according to the main schedule; the sub-schedule including a task goal corresponding to the sub-task and a completion status of the task goal; executing the subtask according to the subschedule by the second agent, and updating the completion status of the task target in the subschedule based on the task execution result of the subtask; In response to the subtask being completed, the completion status of the task target corresponding to the subtask in the main schedule is updated according to the sub-schedule corresponding to the subtask.

[0005] In a second aspect, the present disclosure provides a multi-agent task progress management device, comprising: The task decomposition module is configured to decompose the target task into a plurality of subtasks through the first agent, and construct a master schedule corresponding to the target task, wherein the master schedule includes the task objectives corresponding to each of the subtasks and the completion status corresponding to the task objectives; A task sending module is configured to send, through the first agent, a sub-schedule corresponding to each sub-task to the corresponding second agent according to the main schedule; the sub-schedule includes a task goal corresponding to the sub-task and a completion status of the task goal; a first updating module configured to execute the subtask according to the subschedule through the second agent, and update the completion status of the task target in the subschedule based on the task execution result of the subtask; The second updating module is configured to update the completion status of the task target corresponding to the subtask in the main schedule according to the sub-schedule corresponding to the subtask in response to the completion of the subtask.

[0006] In a third aspect, the present disclosure provides a computer-readable medium having a computer program stored thereon, which, when executed by a processing device, implements the steps of the method described in the first aspect.

[0007] In a fourth aspect, the present disclosure provides an electronic device, comprising: a storage device having a computer program stored thereon; A processing device is used to execute the computer program in the storage device to implement the steps of the method described in the first aspect.

[0008] In a fifth aspect, the present disclosure provides a computer program product, comprising a computer program, which implements the steps of the method described in the first aspect when executed by a processor.

[0009] According to the technical solution, the target task is decomposed into multiple subtasks by the first agent, a main schedule table including the task target and the completion state of the subtasks is constructed, the first agent sends the sub-schedule table corresponding to each subtask to the corresponding second agent according to the main schedule table, then the second agent executes the subtask according to the sub-schedule table, and the completion state of the task target in the sub-schedule table maintained by the second agent is updated based on the task execution result of the subtask. In response to the execution of the subtask, the second agent updates the completion state of the task target corresponding to the main schedule table according to the sub-schedule table corresponding to the subtask. Not only can the first agent accurately remember the global task state and historical information, and the first agent can timely and effectively re-plan the task or re-allocate resources, but also the completion state of the task target can be accurately transmitted between the agents in the multi-agent collaboration system, and the accuracy of the task progress management and collaboration between the agents is ensured.

[0010] Other features and advantages of the present disclosure will be described in detail in the following detailed description section. BRIEF DESCRIPTION OF DRAWINGS

[0011] The above and other features, advantages, and aspects of embodiments of the present disclosure will become more apparent by describing in detail some embodiments with reference to the attached drawings. The same or similar components have the same or similar reference numbers throughout the drawings. It should be understood that the drawings are schematic and elements are not necessarily drawn to scale. In the drawings: Figure 1 is a flowchart of a task progress management method of a multi-agent according to some embodiments.

[0012] Figure 2 is an architecture diagram of a multi-agent collaboration system according to some embodiments.

[0013] Figure 3 is a schematic diagram of a main schedule table according to some embodiments.

[0014] Figure 4 is a schematic diagram of a sub-schedule table according to some embodiments.

[0015] Figure 5 is Figure 1 is a detailed flowchart of the step 130.

[0016] Figure 6 is a working logic schematic diagram of a second agent according to some embodiments.

[0017] Figure 7 is a structural schematic diagram of a task progress management device of a multi-agent according to some embodiments.

[0018] Figure 8 is a structural diagram of an electronic device according to some embodiments. DETAILED DESCRIPTION

[0019] Embodiments of the present disclosure will be described in more detail with reference to the drawings. While certain embodiments of the present disclosure will be illustrated in the drawings, it is understood that the present disclosure can be embodied in various forms and should not be interpreted in a limited sense by the drawings set forth herein. Rather, the embodiments are provided so that the present disclosure can be more thoroughly understood and complete.

[0020] It should be understood that each step described in the method embodiments of the present disclosure can be performed in different orders and / or in parallel. In addition, the method embodiments can include additional steps and / or omit the steps shown. The scope of the present disclosure is not limited in this respect.

[0021] The term "comprising" and variations thereof as used herein are used inclusively, i.e., "comprising, but not limited to". The term "based on" is "based, at least in part, on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". Related definitions will be given in the description below.

[0022] It should be noted that the terms "first", "second", and the like in the present disclosure are merely used to distinguish different devices, modules or units, and do not imply the order or interdependence of the functions performed by these devices, modules or units.

[0023] It should be noted that the terms "one", "multiple" in the present disclosure are illustrative and not restrictive, and those skilled in the art should understand that "one or more" should be understood unless otherwise explicitly indicated in the context.

[0024] The names of the messages or information exchanged between the devices in the embodiments of the present disclosure are merely for illustrative purposes, and are not intended to limit the scope of the messages or information.

[0025] Figure 1 is a flowchart of a multi-agent task progress management method according to some embodiments. As shown in Figure 1 The embodiments of the present disclosure provide a multi-agent task progress management method, which can be specifically performed by a multi-agent task progress management device, which can be implemented by software and / or hardware. As shown in Figure 1 The method can include the following steps.

[0026] In step 110, by the first agent, the target task is decomposed into a plurality of subtasks, and a main schedule corresponding to the target task is constructed, the main schedule including task objectives corresponding to the subtasks and completion states corresponding to the task objectives.

[0027] Here, Figure 2 is an architecture diagram of a multi-agent collaboration system according to some embodiments. As Figure 2 shown, in the multi-agent collaboration system, a first agent as a planner and a second agent as an executor are included. The first agent is responsible for receiving a target task and guiding the second agent to collaborate to complete the target task according to the task objectives and the current state of the target task, and the second agent is responsible for actually performing specific actions or operations to promote the completion of the target task. It should be noted that the first agent and the second agent can be respectively deployed in the same or different servers.

[0028] The first agent can determine the target task that the user wants to solve by processing the natural language input by the user. Then, the first agent decomposes the target task into a plurality of subtasks, and each subtask decomposed can be assigned to a different second agent for execution.

[0029] Moreover, the first agent can construct a structured main schedule belonging to the target task, which includes task objectives corresponding to all subtasks and completion states corresponding to the task objectives. It should be noted that the task objectives corresponding to the subtasks can include necessary task objectives and optional task objectives, the necessary task objectives are task objectives that must be completed, the optional task objectives are task objectives that are added on the basis of the necessary task objectives, and the completion or non-completion of the optional task objectives does not affect the implementation of the necessary task objectives, but can add additional value to the task. The completion state corresponding to the task objective includes a completed state and an uncompleted state.

[0030] Figure 3 is a schematic diagram of a main schedule according to some embodiments. As Figure 3 shown, “Objective 1” and “Objective 2” respectively represent a main task objective, “Sub-objective 1.1” and “Sub-objective 1.2” represent two subtask objectives under the main task objective “Objective 1”, “Sub-objective 2.1” and “(Optional) Sub-objective 2.2” represent two subtask objectives under the main task objective “Objective 2”, and “(Optional) Sub-objective 2.2” is an optional task objective.

[0031] In addition, the completion status of each task goal can be visually displayed in the master schedule through specific marks. For example, the completion status of the task goal can be marked by a check box. Figure 3 In the master schedule shown, "[x]" indicates that the corresponding objective has been completed. For example, "[x]Sub-objective 1.2" indicates that the objective "Sub-objective 1.2" has been completed. "[ ]" indicates that the corresponding objective has not been completed. For example, "[ ]Sub-objective 2.1" indicates that the objective "Sub-objective 2.1" has not been completed.

[0032] That is to say, the master schedule not only includes the task objectives of each subtask obtained by hierarchically decomposing the target task, but also includes the real-time completion status of each task objective.

[0033] It's worth noting that the master schedule can be maintained within the first agent, serving as the single source of truth for the multi-agent collaborative system. This ensures that both the first and second agents in the multi-agent collaborative system have a unified and up-to-date understanding of the target task's current progress, as well as its completed, pending, and optional parts. Furthermore, the master schedule itself can serve as a contextual carrier, carrying critical contextual information. For example, because subtasks are derived through a hierarchical decomposition of the target task, the master schedule implicitly contains contextual information about the dependencies between subtasks.

[0034] In step 120, the sub-schedule corresponding to each sub-task is sent to the corresponding second agent according to the main schedule through the first agent; the sub-schedule includes the task goal corresponding to the sub-task and the completion status of the task goal.

[0035] Here, a multi-agent collaborative system can include multiple second agents for handling different types of tasks. It should be understood that a second agent can be considered a sub-agent of the first agent (serving as the main agent). For example, second agents can be pre-created based on different task types to specifically handle a single type of task. Different second agents are used to handle different atomic tasks, which are the smallest units of tasks that can be broken down.

[0036] The first agent can send the sub-schedule corresponding to each sub-task to the corresponding second agent based on the dependency relationship between the task objectives included in the main schedule.

[0037] For example, the dependencies contained in the master schedule can be used to determine the corresponding predecessor subtasks for each subtask. The predecessor subtasks are then assigned to the corresponding second agent for processing. Only after the predecessor subtasks are processed are the subtasks that depend on the predecessor subtasks assigned to the second agent. In other words, if subtask A depends on subtask B, then subtask B is a predecessor subtask of subtask A and must be completed before subtask A. Therefore, subtask B can be assigned to the corresponding second agent first, and then subtask A can be assigned to the second agent after subtask B is completed.

[0038] It should be noted that the first agent can assign subtasks to the second agent one by one in sequence. Of course, the first agent can also assign subtasks that can be executed concurrently to the corresponding second agent in parallel to increase the task execution speed.

[0039] The sub-schedule includes the task objectives corresponding to the sub-tasks and the completion status of the task objectives. In other words, the sub-schedule includes the task objectives within the scope of the sub-tasks and the completion status of the task objectives.

[0040] A sub-schedule is a partial schedule extracted from the main schedule that includes the task objectives within the scope of the sub-task and the completion status of the task objectives. Figure 4 FIG is a schematic diagram of a sub-schedule according to some embodiments. Figure 4 As shown in the figure, the sub-schedule corresponding to the sub-task of "Objective 2" is as follows Figure 4 shown.

[0041] In step 130, the second agent executes the subtask according to the subschedule, and based on the task execution result of the subtask, updates the completion status of the task target in the subschedule.

[0042] Here, the second agent receives the subschedule issued by the first agent and, based on the received subschedule, executes the subtask corresponding to the subschedule. It is important to note that the subschedule is maintained within the second agent to effectively maintain the contextual information when the second agent executes the subtask.

[0043] When the second agent executes the subtask according to the subschedule, the second agent may update the completion status of the task target in the subschedule maintained internally by the second agent according to the task execution result of the subtask.

[0044] Exemplarily, in a case that the task execution result of the subtask indicates that the task target corresponding to the subtask has been completed, the completion state of the task target in the sub progress table can be marked as a completed state. In a case that the task execution result of the subtask indicates that the task target corresponding to the subtask cannot be completed, a note for the task target or a new task target can be added in the sub progress table to update the sub progress table maintained by the second agent internally.

[0045] In step 140, in response to the completion of the subtask, the completion state of the task target corresponding to the subtask in the main progress table is updated according to the sub progress table corresponding to the subtask.

[0046] Here, the completion of the subtask refers to the execution of the task by the second agent. It should be noted that the completion of the subtask does not mean that the task target is achieved. For example, when the second agent executes the subtask overtime, the subtask can also be considered as completed. After the second agent executes the subtask responsible for it, the second agent can update the completion state of the task target in the main progress table according to the sub progress table maintained by the second agent, so as to update the task target and the completion state of the task target in the sub progress table maintained by the second agent to the main progress table.

[0047] It should be noted that the note or the task target existing in the sub progress table but not existing in the main progress table can be updated to the main progress table.

[0048] The completion state of the task target in the main progress table is updated by the sub progress table maintained by the second agent, which can synchronize the real-time progress of the subtask of the second agent to the main progress table, so that the first agent can have unified and latest cognition of the task progress of the target task, thereby enabling the first agent to timely and effectively make planning adjustment or resource reallocation.

[0049] For example, if a second agent cannot complete a task target, the completion state of the task target in the main progress table will be marked as an uncompleted state, and accordingly, the first agent can reallocate the subtask corresponding to the task target or make planning adjustment again.

[0050] It should be understood that since the first agent sends the sub progress table corresponding to each subtask to the corresponding second agent according to the main progress table, after the main progress table is updated, the first agent can allocate the subtask to the second agent according to the updated main progress table. Moreover, the sub progress table sent by the first agent to the second agent can be sent based on the updated main progress table, so as to transfer the completion state of the task target among the multiple second agents.

[0051] It is worth mentioning that the main schedule maintained in the first agent can ensure that each agent in the multi-agent collaboration system accurately remembers the global task state and historical information after multiple rounds of interaction or task switching, and can ensure that the decision quality of the agent does not decrease and that the agent does not deviate from the task. Moreover, through the main schedule, the first agent as a planner can accurately determine the real progress of the target task and the encountered bottlenecks, thereby timely and effectively re-planning the task or reallocating resources. The sub-schedule maintained in the second agent can be used to maintain the context and progress of the second agent, to guide the second agent to execute the sub-tasks.

[0052] Therefore, by the first agent, the target task is decomposed into multiple sub-tasks, and a main schedule including the task target and completion state of the sub-tasks is constructed, by the first agent, the sub-schedule corresponding to each sub-task is sent to the corresponding second agent according to the main schedule, then by the second agent, the sub-tasks are executed according to the sub-schedule, and based on the task execution result of the sub-tasks, the completion state of the task target in the sub-schedule maintained by the second agent is updated, and in response to the execution of the sub-tasks, the completion state of the task target corresponding to the main schedule is updated by the second agent according to the sub-schedule corresponding to the sub-tasks. Not only can the first agent accurately remember the global task state and historical information and the first agent can timely and effectively re-plan the task or reallocate resources, but also can ensure that the completion state of the task target can be accurately transmitted between each agent in the multi-agent collaboration system, and the accurate progress management and collaboration between agents can be ensured.

[0053] Figure 5 is Figure 1 As shown in FIG. 13, in some implementable embodiments, step 130 can include the following steps. Figure 5

[0054] In step 501, the decision logic corresponding to the task target is determined according to the task target and the current state of the second agent.

[0055] Here, the second agent can select a task target from the sub-schedule according to the task execution order, and then think of the decision logic of the second agent for the task target according to the task target and the current state of the second agent.

[0056] In the thinking stage of the second agent, the second agent can reason and decide based on the task target and the current state of the second agent. The second agent analyzes the current state, evaluates the pros and cons of different action paths, and determines the next decision logic.

[0057] ​The decision-making logic includes updating the sub-schedules and executing actions to achieve the mission objectives. During the second agent's thinking phase, the second agent analyzes whether the current state can achieve the mission objectives. If the mission objectives cannot be achieved or have already been achieved, the second agent updates the sub-schedules. If the mission objectives can be achieved, the second agent can execute the actions to achieve the mission objectives.

[0058] In other words, the second agent can evaluate whether it can complete the task objective based on the task objective and current status. If the second agent determines that it cannot complete the task objective or that the target task has been completed, the corresponding decision logic is to update the sub-schedule and select the next task objective for processing. If the second agent determines that it can complete the task objective, the corresponding decision logic is to execute the action required to complete the task objective.

[0059] In some embodiments, the thinking process of the second agent in obtaining the decision logic can be recorded inside the second agent, and the thinking process is used to show the process of the second agent thinking and obtaining the decision logic.

[0060] By recording the thinking process of the second agent, the transparency and debuggability of the second agent's behavior can be enhanced. Moreover, by recording the thinking process of the second agent, the basis for the second agent's corresponding decision-making logic can be explained through the thinking process.

[0061] It should be understood that recording the thinking process can be recording the key steps in the thinking process, or it can be completely recording each step of the second intelligent agent's thinking to obtain the decision logic.

[0062] In step 502, it is determined whether the decision logic indicates that the sub-schedule should be updated.

[0063] Here, the decision logic includes updating the completion status of the task objectives in the sub-schedule and executing the corresponding actions. The corresponding actions can be actions that the second agent deduces to complete the task objectives. The second agent can determine different actions based on the decision logic derived from its deductions.

[0064] In step 503 , in response to the decision logic being to update the sub-schedule, the completion status of the task objectives in the sub-schedule is updated based on the decision logic.

[0065] Here, when the decision logic is to update the sub-progress table, the second agent updates the completion status of the task target in the sub-progress table based on the decision logic. Wherein, in the case that the decision logic represents that the task target has been completed, the completion status of the corresponding task target in the sub-progress table is updated to the completed state. In the case that the decision logic represents that the second agent cannot complete the task target, a note for the task target is added in the sub-progress table or a new task target is added.

[0066] That is, when the decision logic is to update the sub-progress table, different actions of updating the sub-progress table are performed for different decision logics. For example, if a sub-task is completed, the task target corresponding to the sub-task is marked as completed in the sub-progress table maintained in the second agent. If the second agent encounters an obstacle when performing the task target, a note for the task target is added in the sub-progress table or a new task target is added in the sub-progress table. Illustratively, if the task target cannot be completed, a note describing that the task target cannot be completed can be added. Illustratively, if the task target can be implemented by implementing other task targets first, a new task target can be added. For example, assuming that a task target needs to be completed, a task target of querying data needs to be performed, and then a task target of calling a tool by using the queried data needs to be performed. If it is found in the process of calling the tool that the tool has not been registered, the tool needs to be registered first, a task target of registering the tool---querying data---calling the tool by using the queried data is established, and the task target is updated to the sub-progress table. Then, by using the task targets included in the updated sub-progress table, a new execution action is thought.

[0067] In step 504, in response to the decision logic being to perform an action action corresponding to the task target, the action action is performed to update the current state of the second agent, and the execution step 501 is returned.

[0068] Here, in the thinking stage of the second agent, if the decision logic thought by the second agent is to perform an action action for completing a task target, it means that the current state of the second agent can complete the task target by performing the thought action action, and accordingly the second agent can perform the thought action action.

[0069] For example, in the thinking stage, the second agent thinks "which tool should the second agent call to complete the task target assigned to the second agent? What are the parameters?", thereby obtaining a corresponding action action. In the action stage, the second agent performs the thought action action. For example, the action action performed by the second agent can be to call one or more tools to complete the task target responsible for the second agent.

[0070] When the second agent performs an action, its current state changes. At this point, the second agent returns to step 501 and reconsiders its next decision-making process based on its latest state and mission objectives. This means that the second agent can dynamically advance its mission objectives based on the actual progress and feedback from its previous actions.

[0071] It is worth noting that for each task target included in the sub-schedule, the above steps 501 to 504 can be executed. When a task target is executed to step 503, a new task target can be selected from the sub-schedule until every task target in the sub-schedule has been executed.

[0072] Figure 6 FIG. 1 is a schematic diagram of the working logic of the second agent according to some embodiments. Figure 6 As shown, in the thinking phase of the second intelligent agent, the decision logic obtained by the second intelligent agent is either to update the sub-schedule or to execute an action in the action phase, and after the action is executed, it will rethink based on the latest current state after the action is executed.

[0073] In other words, within the second agent, a continuous cycle of reflection and action is conducted for each mission objective, dynamically adjusting the mission objective based on the actual progress and feedback from executing actions. The second agent can determine whether to update its internally maintained sub-schedule based on the results of the previous phase's actions. It then deliberates and decides on its next action based on the latest sub-schedule, thus achieving self-management and decision-making optimization.

[0074] It is worth noting that the internal working logic of the first intelligent agent may be consistent with the internal working logic of the second intelligent agent.

[0075] Thus, through the above-described embodiment, by dynamically updating the sub-schedule maintained within the second agent, the second agent can effectively maintain the accuracy of contextual information. Furthermore, by updating the sub-schedule, a clear view of the current status of the task can be provided. Within the second agent, through a cycle of thinking, updating the sub-schedule, and acting, the sub-schedule can be dynamically and timely updated, allowing the second agent to make decisions based on the latest internal state, thereby improving the second agent's ability to adapt to dynamic changes during task execution. Furthermore, when problems arise during task execution, the second agent can also reflect its internal adjustments by updating the sub-schedule.

[0076] In some feasible implementations, in step 140, in response to the completion of the subtask, a sub-schedule corresponding to the sub-task can be sent to the first agent, and then the first agent can update the completion status of the task target corresponding to the sub-schedule in the main schedule based on the received sub-schedule.

[0077] Here, when the second agent completes a subtask, it can review and update the subschedules within the subtask it was responsible for. For example, it can clearly indicate which original task objectives in the subschedule have been completed, and which have not been completed or partially completed. In other words, upon completing the subtask, the second agent can obtain a final subschedule, which is used to update the global master schedule maintained by the first agent.

[0078] For example, the second agent can send a sub-schedule maintained by the second agent to the first agent, so that the first agent can update the completion status of the task objectives corresponding to the sub-schedule in the main schedule based on the received sub-schedule. It should be understood that how to update the main schedule based on the sub-schedule has been detailed in the above embodiment and will not be repeated here.

[0079] In some feasible implementations, in response to completing the subtask, the second agent can generate task conclusion information and / or data reference information corresponding to the subtask, and then send the task conclusion information and / or data reference information to the first agent. The task conclusion information and / or data reference information is used to be passed to other agents as context information of the target task.

[0080] Here, the task conclusion information is used to summarize the execution of the subtask. For example, the task conclusion information can clearly describe the execution results and specific achievements of the subtask, describe any problems, obstacles, or unexpected situations encountered during the execution of the subtask, and summarize the important observations, discoveries, or lessons learned by the second agent.

[0081] The second agent can use the task conclusion information to self-evaluate the quality of task execution. Furthermore, the second agent can provide the first agent and / or other second agents with richer contextual information beyond a simple "success / failure" status, enabling them to accurately assess task completion and decide on next steps.

[0082] Data reference information is used to describe the intermediate data generated by the second agent during the execution of the subtask. For example, data reference information can store and index the intermediate data generated during the task execution. For example, data reference information can include the path or unique identifier of the generated data file, a summary or link to the analysis report, used code snippets or configuration parameters, etc.

[0083] Through the data reference information, it can be determined that the output of the execution of the subtask can be systematically managed and traced, and can be conveniently used by other agents or subsequent processes.

[0084] After the second agent completes the subtask assigned to it, the second agent can generate task conclusion information and / or data reference information corresponding to the subtask based on the task execution. Then, the second agent sends the task conclusion information and / or data reference information to the first agent, so as to use the task conclusion information and / or data reference information as the context information of the target task to pass to other agents (such as the first agent and / or other second agents).

[0085] In some embodiments, the first agent associates the task conclusion information and / or data reference information with the corresponding task target in the master schedule.

[0086] The first agent can associate the received task conclusion information and / or data reference information with the corresponding task target in the master schedule. By associating the task conclusion information and / or data reference information with the corresponding task target, a rich and specific description of the task completion can be provided, thereby providing an objective and accurate basis for the evaluation of the work results of the agent.

[0087] It is worth noting that associating the task conclusion information and / or data reference information with the corresponding task target in the master schedule can be embedding the task conclusion information and / or data reference information into the corresponding task target in the master schedule. Of course, it can also be to establish an index between the task target and the task conclusion information and / or data reference information.

[0088] It is worth noting that after the second agent completes the subtask assigned to it, the second agent can send the second agent internally maintained sub-schedule, generated task conclusion information and data reference information to the first agent. The sub-schedule, task conclusion information and data reference information can be in a structured data format. Through the sub-schedule, task conclusion information and data reference information, it can be determined that the task execution result and related context information of the second agent for the subtask can be passed between different agents, so that the agents can better manage the progress and cooperate with each other. Even in a complex, multi-step scenario, context loss can be effectively reduced, so that the agent can perform subsequent operations based on the accurate current state.

[0089] Thus, the first agent is sent the sub-schedule, task conclusion information and data reference information, which clearly reflects the problems in task execution, so that the first agent can perform targeted troubleshooting, task re-allocation or strategy adjustment according to the sub-schedule, task conclusion information and data reference information, greatly improving the fault tolerance and overall reliability of the system.

[0090] In some implementable embodiments, in step 120, the first agent can generate task description information corresponding to the sub-tasks according to the context information of the target task, and then send the sub-schedule corresponding to the sub-tasks and the task description information to the corresponding second agent according to the main schedule, so that the second agent executes the sub-tasks based on the sub-schedule and the task description information.

[0091] Here, the context information includes at least one of the main schedule, task conclusion information and data reference information corresponding to the sub-tasks. It should be understood that the explanations of the task conclusion information and the data reference information have been described in detail in the above embodiments, and will not be repeated here.

[0092] In the above embodiments, since each second agent will send the sub-schedule maintained in the second agent, the task conclusion information and the data reference information generated by the second agent when executing the sub-tasks to the first agent after executing the sub-tasks, the first agent can generate the task description information corresponding to the sub-tasks by taking at least one of the main schedule, the task conclusion information and the data reference information as the context information of the target task. The task description information can be used to inform the second agent what action to take and which task objectives have been completed and which task objectives have not been completed.

[0093] Through the task description information, the second agent's updated sub-schedule information, the task conclusion information and the data reference information generated by the second agent can be carried in the task description information, so as to realize cross-agent progress and achievement sharing, avoid misunderstanding of the task state between different agents, reduce unnecessary communication cost and redundant execution caused by information asymmetry between agents, and greatly improve the cooperation efficiency between multiple agents.

[0094] Figure 7 FIG. 1 is a structural schematic diagram of a multi-agent task progress management device according to some embodiments. As shown in FIG. 1, the multi-agent task progress management device 100 includes a first agent 110 and a second agent 120. Figure 7 As shown in FIG. 1, the multi-agent task progress management device 700 includes a first agent 710 and a second agent 720. The task decomposition module 701 is configured to decompose, by the first agent, a target task into a plurality of subtasks, and construct a main schedule corresponding to the target task, the main schedule including a task target corresponding to each of the subtasks and a completion state of the task target; The task sending module 702 is configured to send, by the first agent, a sub-schedule corresponding to each of the subtasks to a corresponding second agent according to the main schedule; the sub-schedule including a task target corresponding to the subtask and a completion state of the task target; The first updating module 703 is configured to execute the subtask according to the sub-schedule by the second agent, and update the completion state of the task target in the sub-schedule based on a task execution result of the subtask; The second updating module 704 is configured to update, in response to completion of the subtask, the completion state of the task target corresponding to the subtask in the main schedule according to the sub-schedule corresponding to the subtask.

[0095] Optionally, the first updating module 703 is specifically configured to: determine, for each task target included in the sub-schedule, a decision logic corresponding to the task target according to the task target and a current state of the second agent; update, in response to the decision logic being to update the sub-schedule, the completion state of the task target in the sub-schedule based on the decision logic; in response to the decision logic being to execute an action corresponding to the task target, execute the action to update the current state of the second agent, and return to the step of determining the decision logic corresponding to the task target according to the task target and the current state of the second agent.

[0096] Optionally, the first updating module 703 is specifically configured to: in a case where the decision logic indicates that the task target has been completed, update the completion state of the corresponding task target in the sub-schedule to a completed state; in a case where the decision logic indicates that the second agent is unable to complete the task target, add a note for the task target or add a new task target in the sub-schedule.

[0097] Optionally, the multi-agent task schedule management apparatus 700 further includes: The recording module is configured to record a thinking process of the second agent obtaining the decision logic in an internal part of the second agent, the thinking process being used to show a process of the second agent thinking to obtain the decision logic.

[0098] Optionally, the second updating module 704 is specifically configured to: in response to completion of the subtask, send the first agent the subschedule corresponding to the subtask; update, by the first agent, the completion state of the task target in the master schedule corresponding to the subschedule according to the received subschedule.

[0099] Optionally, the second updating module 704 is further configured to: in response to completion of the subtask, generate, by the second agent, task conclusion information and / or data reference information corresponding to the subtask; the task conclusion information is used to summarize the task execution of the subtask, and the data reference information is used to describe intermediate data generated by the second agent in the process of executing the subtask; send the first agent the task conclusion information and / or the data reference information, which are used as context information of the target task to be passed to other agents.

[0100] Optionally, the second updating module 704 is further configured to: associate, by the first agent, the task conclusion information and / or the data reference information with the corresponding task target in the master schedule.

[0101] Optionally, the task sending module 702 is specifically configured to: generate, by the first agent, task description information corresponding to the subtask according to context information of the target task, the context information including at least one of the master schedule, task conclusion information and data reference information corresponding to the subtask, the task conclusion information being used to summarize the task execution of the subtask, and the data reference information being used to describe intermediate data generated by the second agent in the process of executing the subtask; send the subschedule corresponding to the subtask and the task description information to the corresponding second agent according to the master schedule, so that the second agent executes the subtask based on the subschedule and the task description information.

[0102] The function logic performed by each functional module in the multi-agent task progress management apparatus 700 described above has been described in detail in the method part, and will not be repeated here.

[0103] Reference will be made to Figure 8 which shows a structural schematic diagram of an electronic device (e.g., a server) 800 suitable for implementing the embodiments of the present disclosure. Figure 8The electronic device shown is only an example and should not limit the functions and scope of use of the embodiments of the present disclosure.

[0104] like Figure 8 As shown, electronic device 800 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 801, which can perform various appropriate actions and processes according to programs stored in a read-only memory (ROM) 802 or programs loaded from a storage device 808 into a random access memory (RAM) 803. RAM 803 also stores various programs and data required for the operation of electronic device 800. Processing device 801, ROM 802, and RAM 803 are interconnected via a bus 804. An input / output (I / O) interface 805 is also connected to bus 804.

[0105] Typically, the following devices may be connected to the I / O interface 805: an input device 806 including, for example, a touch screen, a touchpad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 807 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 808 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 809. The communication device 809 may allow the electronic device 800 to communicate with other devices wirelessly or by wire to exchange data. Figure 8 The electronic device 800 is shown with various devices, but it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed instead.

[0106] In particular, according to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a non-transitory computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication device 809, or installed from the storage device 808, or installed from the ROM 802. When the computer program is executed by the processing device 801, the above-mentioned functions defined in the method of the embodiment of the present disclosure are performed.

[0107] It is noted that the aforementioned computer-readable medium of the present disclosure can be a computer-readable signal medium or a computer-readable storage medium or any combination thereof. The computer-readable storage medium can be, for example and without limitation, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of the computer-readable storage medium can include, but are not limited to, an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the present disclosure, the computer-readable storage medium can be any tangible medium that contains or stores a program used by or in connection with an instruction execution system, apparatus, or device. In the present disclosure, the computer-readable signal medium can include a computer-readable program code transmitted by a computer-readable medium or a carrier wave in a baseband or as part of a carrier wave. Such a propagated computer-readable signal medium can take many forms, including but not limited to, an electromagnetic signal, an optical signal, or any suitable combination of the foregoing. The computer-readable signal medium can also be any computer-readable medium that is not a computer-readable storage medium and that can be used to carry or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained in the computer-readable medium can be transmitted by any suitable medium, including but not limited to, wire, cable, RF (radio frequency), or the like, or any suitable combination of the foregoing.

[0108] In some embodiments, the first and second agents can communicate using any currently known or future developed network protocol, such as HTTP (HyperText Transfer Protocol), and can be interconnected with digital data communications (e.g., a communications network) of any form or medium, such as the Internet or World Wide Web. Examples of communications networks include local area networks ("LANs"), wide area networks ("WANs"), internetworks (e.g., the Internet), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future developed networks.

[0109] The aforementioned computer-readable medium can be included in the aforementioned electronic device; or can exist independently of the electronic device.

[0110] The computer readable medium described above carries one or more programs, when the one or more programs are executed by the electronic device, cause the electronic device to: decompose, by a first agent, a target task into a plurality of subtasks, and construct a main schedule table corresponding to the target task, the main schedule table comprising a task target corresponding to each of the subtasks and a completion state of the task target; send, by the first agent, a sub-schedule table corresponding to each of the subtasks to a corresponding second agent according to the main schedule table; the sub-schedule table comprising a task target corresponding to the subtask and a completion state of the task target; execute, by the second agent, the subtask according to the sub-schedule table, and update the completion state of the task target in the sub-schedule table based on a task execution result of the subtask; and in response to completion of the subtask, update the completion state of the task target corresponding to the subtask in the main schedule table according to the sub-schedule table corresponding to the subtask.

[0111] Computer program code for carrying out operations of the present disclosure can be written in any of one or more programming languages, including object oriented programming languages such as Java, Smalltalk, C++, as well as conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code can execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer, or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computer (for example, through the Internet using an Internet Service Provider).

[0112] The flow and block diagrams in the drawings show architectural, functional, and operational architectures of possible implementations of systems, methods, and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flow and block diagrams can represent a module, a segment, or a portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that in some alternative implementations, the functions noted in the blocks can occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently or the blocks may be executed in the reverse order, depending on the functionality involved. It will also be noted that each block of the block diagrams and / or flowchart illustrations, and combinations thereof, can be implemented by special purpose hardware-based systems that perform the specified functions or operations, or combinations of special purpose hardware and computer instructions.

[0113] The modules involved in the embodiments of the present disclosure can be implemented in the form of software or in the form of hardware. In some cases, the name of the module does not constitute a limitation on the module itself.

[0114] The functions described above in this document can be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that can be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system on a chip (SOCs), complex programmable logic devices (CPLDs), etc.

[0115] In the context of the present disclosure, a machine-readable medium can be a tangible medium that contains or stores a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable storage media can include, without limitation, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media can include, without limitation, one or more lines of electrical wire, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or Flash memory), optical fiber, portable compact disc read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0116] The above description is merely preferred embodiments of the present disclosure and a description of principles of applied technologies. It should be understood by those skilled in the art that the disclosed scope of the present disclosure is not limited to the technical solutions formed by the specific combinations of the above technical features, and also covers other technical solutions formed by any combinations of the above technical features or equivalent features without departing from the above disclosed concept. For example, the above technical features can be replaced with technical features disclosed in the present disclosure (but not limited to) having similar functions to form technical solutions.

[0117] Moreover, while operations have been depicted in a particular order, this should not be understood as requiring such an order nor limiting it to only those operations shown and described. One of ordinary skill in the art will recognize that many of the operations can be performed in a differing order, or be performed concurrently, that some operations can be performed in any order or omitted, and that some operations can be performed in parallel. Similarly, while several specific implementation details have been discussed in the context of the above discussion, these should not be construed as limitations on the scope of the present disclosure. Certain features described in the context of separate embodiments can also be implemented in combination in a single embodiment. Conversely, various features described in the context of a single embodiment can also be implemented in multiple embodiments separately or in any suitable sub-combination.

[0118] Although the subject matter has been described in language specific to structural features and / or methodological acts, it is to be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or acts described above. Rather, the specific features and acts described above are disclosed as example forms of implementing the claims. With respect to the devices in the above-described embodiments, in which various modules perform operations, the specific manner in which the various modules perform the operations has been described in detail in the embodiments relating to the method. Here, no detailed explanation will be given.

Claims

1. A multi-agent task progress management method, characterized in that: include: Decomposing the target task into a plurality of subtasks by the first agent, and constructing a master schedule corresponding to the target task, wherein the master schedule includes the task objectives corresponding to each of the subtasks and the completion status corresponding to the task objectives; Sending, by the first agent, a sub-schedule corresponding to each sub-task to a corresponding second agent according to the main schedule; the sub-schedule including a task goal corresponding to the sub-task and a completion status of the task goal; executing the subtask according to the subschedule by the second agent, and updating the completion status of the task target in the subschedule based on the task execution result of the subtask; In response to the subtask being completed, the completion status of the task target corresponding to the subtask in the main schedule is updated according to the sub-schedule corresponding to the subtask.

2. The method according to claim 1, characterized in that The executing the subtask according to the subschedule by the second agent and updating the completion status of the task target in the subschedule based on the task execution result of the subtask includes: For each task objective included in the sub-schedule, determining a decision logic corresponding to the task objective based on the task objective and the current state of the second agent; In response to the decision logic being to update the sub-schedule, updating the completion status of the task objectives in the sub-schedule based on the decision logic; In response to the decision logic executing the action corresponding to the task goal, the action is executed to update the current state of the second intelligent agent, and the step of determining the decision logic corresponding to the task goal based on the task goal and the current state of the second intelligent agent is returned.

3. The method according to claim 2, characterized in that The updating of the completion status of the task objectives in the sub-schedule based on the decision logic includes: When the decision logic indicates that the task goal has been completed, updating the completion status of the corresponding task goal in the sub-schedule to a completed status; In the case where the decision logic indicates that the second agent is unable to achieve the task goal, a note regarding the task goal is added to the sub-schedule or a new task goal is added.

4. The method according to claim 2, characterized in that The method further comprises: The thinking process of the second intelligent agent in obtaining the decision logic is recorded inside the second intelligent agent, and the thinking process is used to show the process of the second intelligent agent thinking and obtaining the decision logic.

5. The method according to claim 1, wherein In response to the subtask being completed, updating the completion status of the task target corresponding to the subtask in the main schedule according to the sub-schedule corresponding to the subtask includes: In response to the subtask being completed, sending the sub-schedule corresponding to the subtask to the first agent; The first agent updates the completion status of the task target corresponding to the sub-schedule in the main schedule according to the received sub-schedule.

6. The method according to claim 5, characterized in that The method further comprises: In response to completing the subtask, generating task conclusion information and / or data reference information corresponding to the subtask through the second agent; the task conclusion information is used to summarize the task execution status of the subtask, and the data reference information is used to describe the intermediate data generated by the second agent in the process of executing the subtask; The task conclusion information and / or the data reference information are sent to the first agent, and the task conclusion information and / or the data reference information are used to be transmitted to other agents as context information of the target task.

7. The method according to claim 6, characterized in that The method further comprises: The task conclusion information and / or the data reference information are associated with the corresponding task objectives in the master schedule through the first agent.

8. The method according to any one of claims 1 to 7, characterized in that The step of sending, by the first agent, the sub-schedule corresponding to each sub-task to the corresponding second agent according to the main schedule includes: generating, by the first agent, task description information corresponding to the subtask based on context information of the target task, the context information including at least one of the master schedule, task conclusion information corresponding to the subtask, and data reference information, the task conclusion information being used to summarize the task execution status of the subtask, and the data reference information being used to describe intermediate data generated by the second agent in the process of executing the subtask; According to the main schedule, the sub-schedule corresponding to the sub-task and the task description information are sent to the corresponding second agent, so that the second agent performs the sub-task based on the sub-schedule and the task description information.

9. A multi-agent task progress management device, characterized in that: include: The task decomposition module is configured to decompose the target task into a plurality of subtasks through the first agent, and construct a master schedule corresponding to the target task, wherein the master schedule includes the task objectives corresponding to each of the subtasks and the completion status corresponding to the task objectives; a task sending module configured to send, through the first agent, a sub-schedule corresponding to each sub-task to the corresponding second agent according to the main schedule; the sub-schedule including the task goal corresponding to the sub-task and the completion status of the task goal; a first updating module configured to execute the subtask according to the subschedule through the second agent, and update the completion status of the task target in the subschedule based on the task execution result of the subtask; The second updating module is configured to update the completion status of the task target corresponding to the subtask in the main schedule according to the sub-schedule corresponding to the subtask in response to the completion of the subtask.

10. A computer-readable medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processing device, the steps of the method according to any one of claims 1 to 8 are implemented.

11. An electronic device, characterized in that: include: a storage device having a computer program stored thereon; A processing device, configured to execute the computer program in the storage device to implement the steps of the method according to any one of claims 1 to 8.

12. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.

Images