A multi-entity collaborative decision-making method and system and a computer readable storage medium

Abstract

With the development of automation and intelligent technologies, various systems often contain multiple entities capable of autonomous decision-making (such as industrial robots, IoT devices, vehicle control units, and software agents). These entities need to collaborate to complete complex tasks, avoid conflicts, improve efficiency, and optimize resource utilization. However, existing technologies suffer from drawbacks such as decision-making conflicts, limited collaboration methods, lack of adaptive learning, and poor cross-domain applicability. Therefore, a general collaborative decision-making method is needed that can flexibly coordinate the decision-making intentions of multiple entities, dynamically select collaborative strategies, and possess learning capabilities. This application provides a multi-entity collaborative decision-making method, system, and computer-readable storage medium, aiming to enable multiple entities to dynamically select collaborative strategies based on various factors, resolve conflicts through negotiation, and continuously optimize from feedback, making it widely applicable to various multi-entity collaborative scenarios.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence and multi-agent collaborative decision-making, and more specifically, to a multi-entity collaborative decision-making method and system, as well as a computer-readable storage medium. Background Technology

[0002] With the rapid development of automation and intelligent technologies, various systems are generally deploying multiple entities with autonomous decision-making capabilities, such as robot clusters in industrial production lines, intelligent devices in IoT environments, control units in vehicle networks, and virtual intelligent agents in software environments. When dealing with the same event, these entities often generate differentiated decision intentions based on their respective perceived state information, such as disagreements on the subject performing a material handling task, conflicting actions in response to user instructions, or different strategies for dealing with environmental changes. Existing technologies face significant challenges in coordinating multi-entity decision-making in such processes: First, the lack of an effective coordination mechanism for the decision-making intentions of multiple entities regarding the same event easily leads to behavioral conflicts or resource competition, resulting in system instability or even task failure. Second, the coordination strategy selection mechanism is rigid, typically relying on pre-set fixed master-slave relationships or simple majority voting rules, failing to flexibly adapt to diverse strategies such as arbitration, fusion, or negotiation based on dynamic factors such as entity priority, task matching degree, and environmental state. Third, the decision-making process lacks continuous optimization capabilities, failing to learn from historical interaction data and user feedback to improve coordination rules, making it difficult to improve decision quality over time. Furthermore, existing solutions are mostly designed for specific application scenarios, such as only applicable to multi-robot collaboration or single-type device linkage, making it difficult to build a universal decision-making framework across fields such as industrial control, smart homes, and autonomous driving. These shortcomings make it difficult for the system to efficiently coordinate the behavior of multiple entities in complex and dynamic environments, restricting overall task execution efficiency and user experience.

[0003] To address the aforementioned issues, existing technologies urgently need improvement. Summary of the Invention

[0004] The purpose of this application is to provide a multi-entity collaborative decision-making method and system, as well as a computer-readable storage medium, which can effectively coordinate the decision-making intentions of multiple entities, avoid behavioral conflicts, and improve system operation stability and task execution efficiency.

[0005] This application provides a multi-entity collaborative decision-making method, the technical solution of which is as follows: A multi-entity collaborative decision-making method includes the following steps: Obtain the decision intentions of multiple entities regarding the same event; Based on the pre-set collaboration strategy, determine the decision execution method; Events are handled by the execution entity corresponding to the decision execution method.

[0006] Furthermore, this application also proposes that the entity includes, but is not limited to, at least one of a software module, a hardware device, a virtual character, or an intelligent agent.

[0007] Furthermore, this application also proposes that determining the decision execution method includes integrating the decision intentions of multiple entities to generate a comprehensive decision; and / or selecting one entity from multiple entities as the decision execution entity.

[0008] Furthermore, this application proposes that the collaborative strategy makes decisions based on at least one of the following factors, including but not limited to: the entity's priority, the relationship parameters with the target object, the current internal state, the environmental state, the task matching degree, and the historical decision success rate.

[0009] Furthermore, this application also proposes to include: when there is a conflict in the decision-making intentions of multiple entities, to initiate a negotiation process between the entities and reach a consensus through information exchange.

[0010] Furthermore, this application proposes that the consultation process includes: each entity exchanging reasons for its decision, dynamically adjusting its own decision intentions, until a consensus is reached.

[0011] Furthermore, this application also proposes to include recording the decision-making process and storing the communication content between entities in a storage system for user retrieval.

[0012] Furthermore, this application also proposes to include: collecting user feedback after the decision is implemented, and optimizing the collaboration strategy based on the feedback.

[0013] Furthermore, this application also proposes that the decision intention includes, but is not limited to, at least one of actively initiating interaction, generating content, or performing an action.

[0014] Furthermore, this application also proposes a multi-entity collaborative decision-making system, comprising: The intention acquisition module is used to acquire the decision intentions of multiple entities regarding the same event; The collaboration module is used to determine the decision execution method based on the preset collaboration strategy; The execution module is used to process events according to the decision execution method.

[0015] Furthermore, this application also proposes that the collaboration module be configured to merge the decision intentions of multiple entities to generate a comprehensive decision, and / or select one of the multiple entities as the decision execution entity.

[0016] Furthermore, this application proposes that the collaboration module makes decisions based on at least one of the following factors, including but not limited to: the entity's priority, the relationship parameters with the target object, the current internal state, the environmental state, the task matching degree, and the historical decision success rate.

[0017] Furthermore, this application also proposes to include a negotiation module for initiating an inter-entity negotiation process when there is a conflict in the decision-making intentions of multiple entities.

[0018] Furthermore, this application also proposes to include a recording module for storing the communication content between entities into a storage system.

[0019] Furthermore, this application also proposes to include a feedback learning module for collecting user feedback and optimizing collaborative strategies after decision execution.

[0020] Furthermore, this application also proposes a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method.

[0021] As can be seen from the above, this application provides a multi-entity collaborative decision-making method and system, as well as a computer-readable storage medium. The multi-entity collaborative decision-making method includes obtaining the decision intentions of multiple entities for the same event, determining the execution method according to a preset collaborative strategy, and having the corresponding execution entity handle the event. This scheme coordinates the multi-entity decision-making process, avoids intention conflicts, effectively coordinates the decision intentions of multiple entities, avoids behavioral conflicts, and improves system operation stability and task execution efficiency.

[0022] Comparative analysis with existing technologies Regarding decision-making and coordination methods: existing technologies often use a fixed master-slave model or simple voting; while the solution proposed in this application can achieve dynamic selection, and can employ multiple strategies such as arbitration, fusion, and negotiation, and can adaptively adjust according to the situation.

[0023] Regarding the decision-making basis: existing technologies are often based on preset rules or a single factor; while the solution of this application can achieve the integration of multiple factors, such as priority, relationship parameters, internal state, task matching degree, and historical success rate.

[0024] Regarding conflict resolution: existing technologies mostly adopt simple priority ranking; while the solution in this application adopts a negotiation mechanism, in which entities exchange reasons, dynamically adjust their intentions, and form a consensus.

[0025] Regarding learning ability: existing technologies have no or fixed rules; while the solution in this application can achieve feedback learning and optimize collaborative strategies based on user feedback.

[0026] In terms of applicable fields: existing technologies are often limited to specific scenarios; while the solution of this application can be used across fields, including but not limited to industrial robots, the Internet of Things, vehicle systems, and smart agents.

[0027] The comparison shows that this application provides a general and adaptive multi-entity collaborative decision-making framework, which has significant innovation. Attached Figure Description

[0028] Several embodiments of this application are described below with reference to the accompanying drawings. It should be noted that the specific structures, modules, steps, parameters, and connections shown in the drawings are preferred embodiments of this application and not limitations on the scope of protection of this application. Those skilled in the art can make various modifications, substitutions, or combinations to the specific details shown in the drawings based on the teachings of this application, and these modified embodiments should still be considered to fall within the scope of protection of this application.

[0029] Figure 1 This application provides a system architecture diagram, which illustrates an exemplary architecture of the multi-entity collaborative decision-making system of this application. Each module can be adjusted according to actual applications and does not limit the scope of protection.

[0030] Figure 2 This diagram illustrates a decision-making process for this application. It is merely an example and does not constitute a limitation on the claims. Detailed Implementation

[0031] The technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application. Other technologies that may be mentioned in the embodiments can be implemented using existing technology or other patent applications filed by the applicant on the same day, and will not be repeated here. It should be particularly noted that the specific module divisions, process steps, data flow directions, status names, time values, etc., shown in the accompanying drawings are merely illustrative examples and should not constitute a limitation on the scope of protection of the claims of this application. The scope of protection of the claims is determined solely by their wording and should be interpreted in accordance with the overall content of the specification.

[0032] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0033] In multi-entity systems, multiple entities may have mutually exclusive decision intentions regarding the same event, leading to decision conflicts. Existing technologies employing fixed master-slave relationships or simple voting mechanisms cannot dynamically select collaboration strategies based on real-time contexts, resulting in reduced system stability and impacted task completion quality when handling complex tasks. Furthermore, the limited diversity of collaboration methods restricts the system's adaptability to different task requirements, and inflexible mechanisms easily trigger resource contention and behavioral conflicts, thereby affecting key system performance indicators such as task execution continuity and resource utilization.

[0034] For example, in a smart factory's collaborative handling scenario using industrial robots, when a large workpiece arrives at a designated location, each robot generates a decision intention based on its own position coordinates, current load status, and task priority. If multiple robots have similar priority parameters and similar task matching degrees, the existing system cannot effectively coordinate these intentions, causing robots to simultaneously attempt to handle the same workpiece, leading to interference in the robotic arm's motion trajectory or workpiece displacement, resulting in production process interruptions. Furthermore, decision-making conflicts between entities in this scenario manifest as an inability to dynamically adjust strategies. For instance, when the workpiece weight distribution changes, the system still uses preset fixed arbitration rules instead of switching to a fusion or negotiation mode based on the real-time status, thereby exacerbating the risk of behavioral conflicts.

[0035] If the aforementioned problems are not addressed, the system will frequently experience decision conflicts, increasing the probability of equipment failure and reducing overall operational reliability. Furthermore, the lack of a dynamic strategy selection mechanism makes the system ill-suited to changing environmental conditions, leading to a continuous decline in task execution quality and impacting its applicability in cross-domain applications. Consequently, system stability will be significantly weakened, task completion efficiency will be difficult to maintain, and inter-entity coordination capabilities will be limited, further hindering the effective deployment of multi-entity systems in complex scenarios.

[0036] In response, this application proposes a multi-entity collaborative decision-making method, comprising the following steps: Obtain the decision intentions of multiple entities regarding the same event; Based on the pre-set collaboration strategy, determine the decision execution method; Events are handled by the execution entity corresponding to the decision execution method.

[0037] For ease of understanding, the following explains some key terms in this embodiment: An entity is a unit capable of making independent decisions, which can include software modules, hardware devices, virtual characters, or intelligent agents. For example, in industrial automation scenarios, an entity can be an industrial robot; in a smart home environment, an entity can be a control module for a smart light bulb; and in the virtual world, an entity can be a virtual character with autonomous behavior.

[0038] An event is a specific situation or state change that triggers an entity to make a decision. Events can be the perception of the external environment, such as a sensor detecting an obstacle; they can also be changes in internal states, such as a battery level falling below a threshold; or they can be instructions issued by a user, such as voice control commands.

[0039] Decision intention refers to the action or state change that an entity expects to perform in response to a specific event. Each entity will form one or more decision intentions based on its own functions, state, goals, and understanding of the event.

[0040] A collaborative strategy refers to the rules or mechanisms used to coordinate the decision-making intentions of multiple entities. This strategy defines how to determine the final decision-making process when faced with conflicting or collaborative decision-making intentions from multiple entities.

[0041] The decision execution method refers to the specific action plan ultimately determined for handling an event under the influence of a collaborative strategy. This method can be to select one entity from multiple entities as the executor, or to integrate the intentions of multiple entities to form a comprehensive action plan.

[0042] An implementing entity refers to one or more entities that are responsible for actually handling events according to a determined decision-making execution method.

[0043] This embodiment provides a multi-entity collaborative decision-making method, which is characterized by the following steps: obtaining the decision intentions of multiple entities for the same event; determining the decision execution method according to a preset collaborative strategy; and having the execution entity corresponding to the decision execution method process the event.

[0044] Specifically, in the step of acquiring the decision intentions of multiple entities regarding the same event, the system collects the responses of all relevant entities to the same event. For example, each entity can maintain a local state machine, automatically generating a preset decision intention when a specific condition is detected and sending it to a central coordinator. Alternatively, entities can perceive environmental changes through their sensors and generate a simple response instruction as a decision intention based on their internal rule base. For instance, in a simple traffic management system, when a vehicle is detected approaching an intersection, each vehicle entity can generate a decision intention of "go straight," "turn left," or "turn right" and send it out via a wireless communication module. In this way, a comprehensive understanding of the initial judgments and action tendencies of each entity can be obtained, providing a basis for subsequent coordination.

[0045] In the step of determining the decision execution method based on a preset collaboration strategy, the system decides how to handle the event based on the collected decision intentions and referring to pre-set collaboration rules. For example, a simple priority ranking mechanism can be used, where each entity is pre-assigned a fixed priority value. When multiple entities have conflicting decision intentions, the intention of the entity with the highest priority is always selected as the final decision execution method. Alternatively, a round-robin mechanism can be used, selecting entities as decision execution entities in a preset order. For example, in the traffic management system mentioned above, if multiple vehicle entities intend to "go straight," but the intersection only allows one vehicle to pass at a time, the system can simply select the vehicle entity with the smallest number as the execution entity, allowing it to pass first. This avoids direct conflicts between entities and ensures the orderly nature of the decision-making process.

[0046] In the process of handling events by the execution entity corresponding to the decision execution method, once the decision execution method is determined, the corresponding execution entity receives the instruction and directly calls its internal execution module to handle the event. Alternatively, the execution entity can forward the decision instruction to its subordinate execution unit, which will then perform the specific action. For example, in a traffic management system, after a selected vehicle entity receives a "go straight" instruction, its onboard control system will drive the vehicle through the intersection along a predetermined path. This approach ensures that events are handled promptly and clearly, avoiding the chaos caused by multiple entities acting simultaneously.

[0047] The following example will provide a more detailed explanation of the above technical solution: Imagine a smart factory with two industrial robots, Robot A and Robot B, that need to work together to move a large workpiece. Traditional collaborative robot handling systems may suffer from decision conflicts, such as both robots attempting to move the workpiece simultaneously, or using a single, inflexible coordination method that cannot be adjusted according to actual conditions, thus affecting handling efficiency and safety.

[0048] To address this issue, this application proposes a multi-entity collaborative decision-making method. Specifically, when a large workpiece arrives at a designated location, robot A and robot B simultaneously detect the event. Robot A generates a "I'll move it" decision intention based on its current position, load capacity, and task queue. Simultaneously, robot B also generates a "I'll move it" decision intention based on its current battery status, path planning, and available resources. These two decision intentions are then sent to a central coordination system.

[0049] After receiving the decision intentions from robots A and B, the central coordination system determines the execution method based on a pre-defined coordination strategy. For example, this strategy might stipulate that when multiple robots have the same intention for the same handling task, the robot with the smaller number is prioritized as the execution entity. In this case, assuming robot A's number is smaller than robot B's, the system will determine that robot A will be the decision-making execution entity.

[0050] Subsequently, Robot A received instructions from the central coordination system and began performing the workpiece transport task. Robot B, on the other hand, received a standby instruction, remained stationary, and awaited the next task. Thus, by acquiring the decision-making intentions of each entity and arbitrating based on a pre-set collaborative strategy, potential transport conflicts between Robot A and Robot B were avoided, ensuring the smooth execution of the workpiece transport task.

[0051] Based on the above examples, the multi-entity collaborative decision-making method proposed in this embodiment demonstrates significant technical contributions. In traditional solutions, when multiple robots simultaneously detect a handling task and each generates a handling intention, the lack of an effective coordination mechanism may lead to two robots attempting to handle the same workpiece simultaneously, resulting in collisions, equipment damage, or task failure. In contrast, this embodiment effectively avoids such decision conflicts by acquiring the decision intentions of multiple entities and determining a unique decision execution method based on a preset collaborative strategy. For example, in the handling scenario described above, through a preset numbering priority strategy, the system can explicitly designate robot A to execute the task, while robot B remains on standby, thereby eliminating potential conflicts.

[0052] Furthermore, traditional collaborative methods are often quite simplistic, such as being fixed to a single master robot or employing a simple voting mechanism, lacking the flexibility to adjust according to the actual situation. This embodiment provides a general framework that dynamically determines the decision-making execution method through preset collaborative strategies. Even a simple preset strategy is more flexible than having no strategy or a fixed master-slave relationship. This method allows the system to select appropriate execution entities based on different scenarios and needs, thereby improving decision-making efficiency and system robustness. Therefore, this embodiment represents a significant advancement in resolving multi-entity decision-making conflicts and enhancing the flexibility of collaborative decision-making.

[0053] In some of the solutions mentioned above in this application, entities are proposed to participate in collaborative decision-making. However, in this process, the types of entities are not clearly defined, which may lead to a lack of universality and adaptability when the method is applied to different domains.

[0054] In this regard, this application further clarifies that the entity includes, but is not limited to, at least one of software modules, hardware devices, virtual characters, or intelligent agents.

[0055] Specifically, a software module refers to a piece of program code or a set of programs that performs a specific function in a computer system. This concept encompasses various forms, ranging from operating system components and applications to microservices and algorithm models. As a decision-making entity, a software module can be an independent decision-making algorithm, such as a path planning algorithm or a recommendation system algorithm. Its implementation can be through API interfaces to communicate with other modules, receiving input data and outputting decision results; or it can be deployed as a service on a cloud or local server, providing decision-making capabilities through remote calls. Hardware devices refer to entities with physical forms that can perform specific operations, such as sensors, actuators, controllers, and robots. As a decision-making entity, a hardware device can be an intelligent sensor with computing and communication capabilities, capable of preliminary analysis of environmental data and forming decision intentions; or it can be an industrial robot that generates action decisions based on received instructions and its own state. Its implementation can be through embedded systems integrating decision logic, making decisions directly on the device; or it can be connected to a central control system via a network, where the central system makes decisions and then issues instructions to the hardware device for execution. Virtual characters refer to non-physical entities that exist in virtual environments or simulation systems and possess certain behavioral logic and interactive capabilities. Examples include NPCs (non-player characters) in games, simulated agents in simulation environments, or digital humans in the metaverse. As decision-making entities, virtual characters can generate action decisions or interaction intentions in the virtual world based on their preset AI logic, environmental perception, and goals. This can be achieved by defining their decision logic through scripting languages ​​or behavior trees, or by using machine learning models to enable more complex adaptive decision-making capabilities. An intelligent agent refers to an autonomous entity capable of perceiving its environment, reasoning, learning, and taking actions to achieve specific goals. An intelligent agent is a broader concept that can include software modules, hardware devices, virtual characters, etc. Examples include a reinforcement learning-based AI agent or a single decision-making unit in a multi-agent system. As decision-making entities, intelligent agents can autonomously generate decision intentions based on their internal state, environmental information, and learned strategies. This can be achieved through rule-based expert systems, model-based planning algorithms, or deep learning-based decision networks. Furthermore, the phrase "at least one" emphasizes the diversity and flexibility of entity types. This means that in practical applications, a single type of entity can be selected to participate in collaborative decision-making, such as a decision-making system composed solely of software modules, depending on the specific needs of the scenario. Alternatively, it can be a combination of multiple types of entities, such as hardware devices and software modules jointly participating in decision-making. This flexibility ensures that this method can adapt to a wide range of application scenarios, avoiding strict restrictions on specific entity types, thereby improving the method's versatility.

[0056] Based on the aforementioned multi-entity collaborative decision-making method, by clearly defining entity types, the proposed solution can more effectively acquire decision intentions, determine collaborative strategies, and handle events. When multiple entities generate decision intentions for the same event, these entities can be diverse, such as software modules responsible for data processing, hardware devices performing physical operations, virtual characters simulating complex behaviors, or intelligent agents with autonomous learning capabilities. This diversity allows the method to cover a wide range of application scenarios, from pure software environments to the physical world and virtual spaces. After acquiring the decision intentions of these different types of entities, the system determines the decision execution method according to the preset collaborative strategy. For example, in an industrial automation scenario, an industrial robot may propose the intention of "moving item A," while a scheduling system may propose the intention of "optimizing the transportation path." By taking into account the intentions of these heterogeneous entities, the collaborative module can select one entity as the executor based on factors such as entity priority and task matching, or merge the intentions of multiple entities to generate a comprehensive decision. This clear and inclusive approach to entity types makes the formulation of collaborative strategies more precise and flexible, fully utilizing the advantages of different types of entities. Ultimately, the event is handled by the execution entity corresponding to the determined decision execution method. For example, if it is determined that the handling will be performed by an industrial robot, the hardware will operate according to the comprehensive decision or the decision intention of the selected entity. If it is determined that the content will be generated by a software module, the software module will generate the corresponding content based on the fused intention. This mechanism ensures that regardless of the entity type, all entities can effectively participate in the collaborative decision-making process and ultimately promote the effective handling of the event. In this way, the solution of this application solves the problem of insufficient versatility caused by the ambiguity of entity types in traditional methods, enabling multi-entity collaborative decision-making methods to adapt to a wider range of fields and application scenarios.

[0057] The following example illustrates this. Consider an intelligent warehousing system. In this system, various types of entities need to collaborate to complete order picking tasks. The software module could be an order management system whose decision intention is "prioritize urgent order A." The hardware device could be an Automated Guided Vehicle (AGV) whose decision intention is "go to the nearest available charging station to charge." The agent could be a path planning agent based on reinforcement learning, whose decision intention is "plan a path for the AGV to avoid congested areas." The virtual role could be a virtual operator in a simulator simulating the warehousing environment, whose decision intention is "test new picking strategies in the simulation environment." When the system receives a new order, the order management system (software module) generates a decision intention to process the order. Simultaneously, the AGV (hardware device) generates its own intention based on its battery level and task status. The path planning agent generates a path optimization intention based on the current warehouse layout and the AGV's position. The system acquires these decision intentions from different types of entities. For example, the order management system submits order processing requests via an API interface, the AGV reports its status and intentions via a wireless communication module, and the path planning agent generates path suggestions using internal algorithms. The coordination module comprehensively considers these intentions. For instance, if the AGV has low battery and an urgent order, the coordination module might combine the priority intentions of the order management system and the optimization intentions of the path planning agent to generate a comprehensive decision: "The AGV will first go to charge, and after charging, it will immediately execute the urgent order A, using the optimized path planned by the agent." Ultimately, the AGV, as the execution entity, will perform charging and picking operations according to this comprehensive decision.

[0058] By explicitly defining entities as including, but not limited to, at least one of software modules, hardware devices, virtual characters, or intelligent agents, the solution presented in this application significantly enhances the versatility and applicability of multi-entity collaborative decision-making methods. This allows the method to be flexibly applied to various heterogeneous systems and scenarios. For example, in industrial automation, it can coordinate robots and scheduling algorithms; in smart homes, it can coordinate smart speakers and virtual assistants; and in virtual simulation, it can coordinate different virtual characters. This broad compatibility with entity types effectively solves the problem of cross-domain application limitations caused by traditional methods due to entity type restrictions. This enables the method to be seamlessly integrated into diverse scenarios such as industry, home, automotive, and virtual environments, thereby improving the overall system's collaborative efficiency and decision-making quality.

[0059] In some of the solutions mentioned above in this application, a decision-making execution method is proposed to coordinate the decision intentions of multiple entities. However, in this process, the coordination strategy may lack a specific implementation mechanism to flexibly handle different situations. For example, it may be unable to effectively integrate multiple intentions to generate a comprehensive decision or select the best execution entity, resulting in low decision-making efficiency, incomplete conflict resolution, or unreasonable resource allocation.

[0060] In this regard, this application further proposes that determining the decision execution method includes integrating the decision intentions of multiple entities to generate a comprehensive decision; and / or selecting one entity from multiple entities as the decision execution entity.

[0061] Specifically, fusing the decision-making intentions of multiple entities to generate a comprehensive decision refers to integrating the individual intentions of different entities regarding the same event to form a unified and comprehensive decision-making scheme. This fusion process can be achieved through various technical means. For example, it can be based on voting or consensus mechanisms, using majority voting, weighted voting, or more complex distributed consensus protocols to determine the final comprehensive decision. Alternatively, it can be based on rule-based or model-based fusion, pre-setting a set of rules to logically judge and combine different intentions based on their type, priority, or entity attributes, or utilizing machine learning models to learn how to generate the optimal comprehensive decision from multiple intentions through training.

[0062] Simultaneously, selecting one entity from the multiple entities as the decision execution entity means that, under specific circumstances, the system identifies and assigns the most suitable entity to specifically execute the decision. This selection process can also be based on various considerations. For example, selection can be based on priority or role allocation, specifying the entity with the highest authority or most relevant role according to preset entity priorities, role permissions, or scope of responsibility. Alternatively, selection can be based on capability or status assessment, evaluating the current capabilities (such as processing power, resource consumption, geographical location) or internal status (such as battery power, health status, task load) of each entity to select the most suitable entity to execute the task.

[0063] This application's solution significantly enhances the flexibility and adaptability of multi-entity collaborative decision-making methods by introducing two specific mechanisms: merging the decision intentions of multiple entities to generate a comprehensive decision, and selecting one entity from among multiple entities as the decision-making execution entity. After acquiring the decision intentions of multiple entities regarding the same event, the system is no longer limited to a single decision-making execution mode but can dynamically select the most suitable execution method for the current situation based on preset collaborative strategies. When a task requires collaboration among multiple entities and their intentions need to be unified and coordinated, the system can employ a fusion mechanism to integrate the intentions of each entity into a unified comprehensive decision, ensuring consistency and overall optimality of action. Conversely, when a task is more suitable for efficient independent completion by a single entity, the system can use a selection mechanism to choose the entity with the best execution capability or the most suitable conditions from among many entities to undertake the task, thereby avoiding resource waste and decision redundancy. This dynamic selection and flexible switching capability enables the above method to effectively cope with complex and ever-changing application scenarios, ensuring the efficiency and accuracy of decision execution.

[0064] The following is a concrete example to illustrate this. In a smart warehousing system, there are multiple handling robots and a central control system. When the system receives a new goods inbound instruction, this instruction is treated as an event. Each handling robot generates its own decision intention based on factors such as its current location, battery level, load, and distance from the goods storage area. For example, "I intend to move goods A to area X," "I am charging and cannot move immediately," or "I am available and closest to goods B." At this time, the central control system determines the decision execution method according to a preset coordination strategy. If the inbound instruction involves multiple different types of goods that need to be moved to different areas simultaneously and requires multiple robots to work together, the system may choose to merge the intentions of each robot to generate a comprehensive decision that includes collaborative path planning and task allocation for multiple robots. For example, "Robot 1 moves goods A to area X, and Robot 2 moves goods B to area Y." Conversely, if the inbound instruction only involves a single item that needs to be moved by a single robot, the system will select the most suitable robot from all the robots that have expressed their intentions, based on factors such as priority, current load, and distance from the goods. For example, "Select Robot 3 to move goods C to area Z."

[0065] Through the above technical solution, this application effectively addresses the problem of the lack of specific implementation mechanisms for collaborative strategies to flexibly handle different scenarios. By introducing two specific and switchable execution methods—decision intention fusion and execution entity selection—the system can dynamically select the most suitable decision execution strategy based on the nature of the task and the state of the entity. This not only avoids inefficient decision-making and unreasonable resource allocation caused by conflicting intentions, but also ensures the generation of comprehensive decisions when unified action is needed, and the selection of the best execution entity when efficient delegation is required, thereby significantly improving the efficiency, accuracy, and adaptability of multi-entity collaborative decision-making.

[0066] In some of the solutions mentioned above in this application, a collaborative strategy is proposed to determine the decision execution method. However, in this process, the collaborative strategy may lack dynamism and adaptability, and cannot optimize the decision based on multiple real-time factors, resulting in low decision quality, low resource utilization efficiency, and inefficient conflict resolution.

[0067] In response, this application further proposes a collaborative strategy that makes decisions based on at least one of the following factors: the entity's priority, the relationship parameters with the target object, the current internal state, the environmental state, the task matching degree, and the historical decision success rate.

[0068] In multi-entity collaborative decision-making, entity priority refers to the relative importance or weight assigned to different entities in decision-making and execution. This priority can be pre-set, for example, statically configured based on an entity's role, function, or permission level; or dynamically adjusted according to real-time context, such as temporarily increasing the priority of an entity when it is in a critical task execution phase, or dynamically evaluated based on the entity's resources and processing capabilities. Relationship parameters with the target object refer to the degree of association or characteristics between the entity and the target object of the event or task to be processed. For example, in a smart home scenario, the intimacy between a smart speaker and the user, or the trust between a smart lock and family members, can all serve as relationship parameters. These parameters can be quantified based on historical interaction data, user configuration, or preset rules. Current internal state refers to the real-time operating status or intrinsic attributes of an entity when making decisions. This includes, but is not limited to, the entity's battery level, load, operating mode, health status, and emotional state (for virtual intelligent agents). For example, a mobile robot with insufficient battery power may not be suitable for long-distance transport tasks, and an overloaded server module should avoid receiving new computing tasks. Environmental state refers to the real-time information of the external environment in which the entity exists. This includes, but is not limited to, time, weather, light intensity, network bandwidth, traffic conditions, and ambient temperature. For example, in autonomous driving scenarios, rain and snow can affect the sensor's perception capabilities, and insufficient light at night can affect the decision-making of the vision module. Task matching degree refers to the suitability or ability level of an entity to complete a specific task. This can be evaluated based on the entity's functional characteristics, skill set, resource allocation, and historical task completion performance. For example, a software module with image recognition capabilities has a higher matching degree when handling visual tasks, and an industrial robot equipped with a heavy robotic arm has a high matching degree when handling heavy objects. Historical decision success rate refers to the success rate of an entity or collaborative strategy in similar decisions in the past. This can be obtained by recording and analyzing historical decision data, calculating the proportion of successful execution, or evaluating the satisfaction with the decision results. For example, if an entity has a high historical success rate in handling a specific type of conflict, its decision intention may be given higher weight in similar future situations.

[0069] This application proposes a multi-entity collaborative decision-making method, the core of which lies in dynamically considering multiple factors to determine the decision execution method, thereby overcoming the limitations of traditional collaborative strategies. Specifically, after obtaining the decision intentions of multiple entities regarding the same event, the system no longer relies solely on a preset single rule or fixed pattern to determine how to execute, but elevates the formulation process of the collaborative strategy to a multi-dimensional, adaptive, and intelligent judgment process. The collaborative strategy comprehensively evaluates multiple factors such as the entity's priority, relationship parameters with the target object, current internal state, environmental state, task matching degree, and historical decision success rate. These factors work together to provide the collaborative strategy with rich and real-time contextual information, enabling the collaborative strategy to flexibly select the most appropriate decision execution method based on the complexity and dynamism of the current situation. For example, when multiple entity intentions conflict, the collaborative strategy can quickly arbitrate and select the optimal execution entity based on the entity's priority and task matching degree; when multiple intentions need to be merged, the strategy can intelligently weight and merge them based on the relationship parameters between each entity and the target object and the historical decision success rate to generate a better comprehensive decision. This multi-factor-based dynamic decision-making mechanism makes the entire collaborative decision-making process more intelligent and adaptable, effectively avoiding decision conflicts, improving resource utilization efficiency, and ensuring high-quality task completion.

[0070] The following example illustrates a multi-entity collaborative decision-making scenario within a smart home system. When a user issues the voice command "Prepare dinner," multiple smart entities (such as a smart refrigerator, smart oven, smart speaker, and smart lighting system) will generate their own decision intentions. The smart refrigerator might intend to "recommend a recipe and check ingredients," the smart oven might intend to "preheat," the smart speaker might intend to "play cooking music," and the smart lighting system might intend to "adjust the kitchen lights." In this situation, the collaborative strategy needs to determine how to execute these intentions. The system makes decisions based on multiple factors: smart refrigerators and smart ovens typically have high priority in the "preparing dinner" task because they are directly related to the cooking process; smart speakers may have a high affinity with the user because they frequently play the user's favorite music; smart ovens may check if they have completed cleaning or are malfunctioning, while smart refrigerators check food inventory; the system detects the kitchen's light intensity, and if the light is insufficient, the smart lighting system's intention to adjust brightness is strengthened; smart refrigerators have a high degree of compatibility in recipe recommendations and food management, while smart ovens have a high degree of compatibility in heating; if the user has previously been satisfied with the cooking music recommended by the smart speaker, the success rate of playing that music will be even higher. After comprehensively evaluating these factors, the collaborative strategy may decide to adopt a "fusion" decision-making execution method: the smart refrigerator immediately recommends a recipe and checks the ingredients, the smart oven begins preheating, the smart speaker plays the user's preferred cooking music, and the smart lighting system adjusts the kitchen lights to a suitable brightness according to the ambient light. This multi-factor consideration makes the decision-making execution method more intelligent and tailored to user needs.

[0071] Through the above technical solution, this application effectively addresses the lack of dynamism and adaptability in determining decision execution methods in traditional collaborative strategies. By incorporating multiple dynamic factors, such as entity priority, relationship parameters with the target object, current internal state, environmental state, task matching degree, and historical decision success rate, into the decision-making basis of the collaborative strategy, the system can intelligently select or generate the optimal decision execution method based on real-time context and entity state. This significantly improves the accuracy and context adaptability of decision-making, avoiding decision-making errors and resource waste caused by single or fixed rules. For example, high-priority entities can prioritize the execution of critical tasks, ensuring the smooth operation of the system's core functions; considering the internal state of entities avoids assigning tasks to entities that do not meet the conditions, improving execution efficiency; and the learning mechanism based on historical success rate enables the collaborative strategy to be continuously optimized, thereby achieving higher-quality decisions and effectively reducing the probability of conflict, ultimately improving the collaborative efficiency and robustness of the entire multi-entity system.

[0072] In some of the embodiments described above in this application, the decision-making intentions of multiple entities are obtained and the decision-making execution method is determined. However, when there are conflicts in decision-making intentions, the lack of a dynamic negotiation mechanism may lead to the inability to effectively resolve conflicts, resulting in decision-making deadlock, waste of resources, or system instability, thus affecting overall decision-making efficiency and collaborative effectiveness. Therefore, this application further proposes that when there are conflicts in the decision-making intentions of multiple entities, a negotiation process should be initiated between the entities to reach a consensus through information exchange.

[0073] The phrase "when the decision-making intentions of multiple entities conflict" refers to situations in multi-entity collaborative decision-making where different entities propose inconsistent, mutually exclusive, or resource-competitive decision-making plans or action suggestions for the same event. This conflict may manifest as different execution methods for the same task, different allocation demands for shared resources, or different expectations of the event's outcome. Conflict detection can be based on pre-defined conflict rules. For example, when two entities simultaneously request exclusive access to the same resource, or when the action plans proposed by two entities are logically mutually exclusive, the system determines it as a conflict. Alternatively, the system can analyze the decision-making intentions of each entity and their potential consequences, using predictive models to assess whether these consequences will interfere with each other or cause negative impacts, thereby identifying potential conflicts. This condition is a prerequisite for initiating the negotiation process, ensuring that the negotiation mechanism is activated only when necessary, avoiding unnecessary computational and communication overhead, and improving system efficiency.

[0074] "Initiating an inter-entity negotiation process" refers to the process by which the system guides or facilitates interaction and communication among relevant entities to seek solutions after detecting a conflict of decision-making intentions. This process can be managed by a central coordinator or negotiation module, including collecting information from each entity, distributing negotiation requests, coordinating information exchange, and ultimately adjudicating or guiding a consensus; alternatively, entities can communicate directly, point-to-point or multi-point-to-multi-point, autonomously exchanging information and adjusting their intentions until an agreement is reached. The negotiation process provides a structured framework that enables conflicting entities to communicate and interact in an orderly manner, and is a key step in resolving conflicts.

[0075] "Achieving consensus through information exchange" refers to the exchange of data, reasons, preferences, constraints, and other information related to decision-making intentions among conflicting entities during the negotiation process. Through this information exchange, the entities ultimately reach a mutually acceptable and conflict-free decision-making solution. Information exchange can include each entity explaining the basis, objectives, and expected effects of its decision-making intentions to other entities, and providing relevant data or models as support. After receiving information from other entities, each entity reassesses and modifies its own decision-making intentions based on the new information and overall objectives to adapt to collaborative needs. Alternatively, entities can proactively propose compromise solutions that balance the interests of all parties, or provide multiple alternatives for other entities to choose from. Information exchange is the core mechanism of negotiation, enabling entities to understand each other's positions and constraints, thus laying the foundation for finding a mutually acceptable solution. Achieving consensus is the ultimate goal of the negotiation process, ensuring the effectiveness of collaborative decision-making and the smoothness of its implementation.

[0076] The proposed solution, within a basic multi-entity collaborative decision-making process, first acquires the decision intentions of multiple entities regarding the same event and preliminarily determines the decision execution method based on a preset collaborative strategy. However, before or during the determination of the decision execution method, the system continuously monitors or evaluates whether these decision intentions conflict. Once a conflict is detected among the decision intentions of multiple entities, the system immediately initiates a negotiation process between the entities. In this negotiation process, conflicting entities no longer simply submit their respective intentions but enter an interactive phase, resolving differences through information exchange. This information exchange can include each entity explaining its decision rationale, providing supporting data, or expressing its priorities and constraints. Based on this exchanged information, each entity can dynamically adjust its decision intentions to seek a mutually acceptable solution. The goal of the negotiation process is to facilitate consensus among all relevant entities, forming a comprehensive decision intention or execution plan that is recognized by all entities and free from conflict. Once consensus is reached, the system can proceed with the subsequent steps of determining the decision execution method based on this consensus, and ultimately, the corresponding execution entity handles the event. In this way, the proposed solution embeds a dynamic conflict resolution mechanism into the basic collaborative decision-making process, ensuring that entities can effectively coordinate even in complex and ever-changing scenarios, avoiding decision stagnation or erroneous execution due to conflicting intentions, thereby significantly improving the robustness and efficiency of collaborative decision-making.

[0077] The following example illustrates this. In a smart factory, when multiple industrial robots need to collaboratively move a large workpiece, conflicting decision intentions may arise. For instance, robot A and robot B simultaneously detect the workpiece and, based on their respective internal states (such as battery level, load, and distance) and task matching, independently generate a decision intention of "I will move it." At this point, the system detects a conflict between the two robots' decision intentions regarding the same workpiece. The system then initiates a negotiation process between the entities. During the negotiation, robot A and robot B exchange information. Robot A might send reasons to the system and robot B, such as having sufficient battery power, being closest to the workpiece, and possessing higher load capacity. Robot B, on the other hand, might provide feedback that its current task queue is idle, its path planning is better, and it has a more flexible gripping mechanism. After receiving this information, the system can guide the two robots to dynamically adjust their decision intentions according to preset negotiation rules or collaborative strategies (e.g., prioritizing the optimal solution combining load capacity and path efficiency). For example, after learning about the path advantages of robot B, robot A might adjust its intention to "dominate the transport but request robot B's assistance in path planning," while robot B would adjust to "assist in path planning and assist in transport." Through this information exchange and dynamic adjustment, robot A and robot B ultimately reach an agreement, forming a comprehensive decision for collaborative transport, such as robot A being responsible for the main transport, and robot B being responsible for assisting in positioning or providing path optimization suggestions.

[0078] Through the aforementioned technical solution, when the decision-making intentions of multiple entities conflict, the system can promptly identify and initiate a negotiation process between the entities, effectively avoiding decision-making deadlock, resource waste, or system instability caused by conflicting intentions. The negotiation process allows for thorough information exchange between entities; for example, each entity can explain its decision-making rationale, provide supporting data, or express its priorities, enabling all parties to understand each other's positions and constraints. Based on this, entities can dynamically adjust their decision-making intentions until a consensus is reached, ensuring the consistency and rationality of the final decision. This not only enhances the flexibility and reliability of the multi-entity collaborative decision-making process but also significantly improves overall decision-making efficiency and collaborative effectiveness, enabling the system to operate more robustly and efficiently in the face of complex and ever-changing environments and tasks.

[0079] In some of the solutions mentioned above in this application, a consultation process is proposed to resolve conflicts of decision-making intentions. However, in this process, the consultation process may lack specific operational mechanisms, such as how entities exchange information, how to adjust their intentions based on new information, and how to ensure that the consultation continues until an agreement is reached. This can lead to low consultation efficiency, failure to effectively eliminate conflicts, or potential deadlock, affecting the reliability and stability of the system's decision-making.

[0080] In this regard, this application further proposes a consultation process that includes: each entity exchanging reasons for its decision, dynamically adjusting its own decision-making intentions, until a consensus is reached.

[0081] "Exchange of decision-making rationale among entities" refers to the proactive sharing of the basis, logic, or supporting data behind each entity's decision with other entities during the negotiation process. This helps reveal the underlying reasons and considerations for each entity's decision, promoting mutual understanding. Specifically, entities can use predefined communication protocols to package their decision intentions, supporting real-time data (e.g., sensor readings, internal state parameters, task priorities, resource usage, etc.), and a brief description of their reasoning process into a message, which can then be broadcast or sent to other relevant entities in a peer-to-peer manner. Alternatively, entities can provide a summary of their decision model or rules, or publicly disclose their current state and objective function, enabling other entities to understand the underlying reasons for their decisions and thus providing a common basis for conflict resolution. "Dynamically adjusting one's own decision-making intentions" refers to the ability of entities to flexibly modify their own decision-making direction or strategy after receiving and analyzing the decision-making rationale of other entities. This avoids rigidity and inflexibility in decision-making intentions, allowing entities to adapt in real time based on new information or the perspectives of others. For example, after receiving decision-making reasons from other entities, the decision-making logic module within an entity can reassess its own decision-making intentions. This might involve updating its internal state model, recalculating its utility function, adjusting its priority weights, or employing game theory or reinforcement learning-based strategies to iteratively adjust its own strategy based on the behavior and reasons of other entities, aiming to reach a Nash equilibrium or Pareto optimal solution. "Until consensus is reached" means that the negotiation process will continue until all participating entities reach a consensus on a common decision-making intention. This ensures the thoroughness of the negotiation process, prevents negotiation from being interrupted or ending without results, and guarantees the integrity of conflict resolution and the reliability of system decisions. The system can set a consensus judgment mechanism, for example, when the decision-making intentions of all entities tend to be consistent within a preset tolerance range, or when all entities explicitly send an "agree" signal, consensus is considered to have been reached. Furthermore, an arbitrator or coordinator entity can be introduced to collect the adjusted intentions of each entity and determine whether consensus has been reached; if not, the next round of negotiation continues.

[0082] This application's solution clarifies the specific operational steps of the negotiation process, enabling the initiation of a structured and efficient negotiation mechanism when conflicting decision-making intentions exist among multiple entities. When a conflict in decision-making intentions is detected, the system triggers the negotiation process. In this process, entities no longer simply express their intentions but further "exchange decision-making reasons," allowing each entity to access the deeper information and considerations behind the decisions of other entities. Based on these shared reasons, each entity can "dynamically adjust its own decision-making intentions," meaning they can flexibly modify their initial decisions based on new information and a more comprehensive understanding of the overall situation. This process of "exchanging reasons" and "dynamic adjustment" is iterative until all entities "reach a consensus." This mechanism concretizes the abstract concept of "information exchange to reach consensus" into actionable steps, ensuring the effectiveness and thoroughness of conflict resolution, thereby significantly improving the efficiency and robustness of multi-entity collaborative decision-making and avoiding the ambiguity and inefficiency of the negotiation process.

[0083] The following example illustrates this concept. In a smart factory, suppose there are two industrial robots, R1 and R2. Both detect a large workpiece that needs to be moved and each decides to "move it," leading to a decision conflict. At this point, the system initiates a negotiation process. First, each entity exchanges its reasons for its decision: Robot R1 sends a message explaining that its current load is light, it is closer to the workpiece, and its battery is fully charged; Robot R2 sends a message explaining that it has higher moving accuracy and has been designated as the priority executor for this type of workpiece in the task scheduling system. Next, each entity dynamically adjusts its decision intention: After receiving R2's reasons, Robot R1 analyzes and finds that R2 has advantages in accuracy and task assignment, so R1 reassesses and adjusts its decision intention from "move it" to "assist in moving it" or "wait." After receiving R1's reasons, Robot R2 acknowledges R1's advantages in battery and distance but still insists on its accuracy and assigned role. Subsequently, the two may conduct multiple rounds of information exchange and intention adjustments until a consensus is reached. For example, R1 eventually agrees that R2 will lead the transport and expresses its willingness to provide assistance (such as clearing the transport path). R2 accepts R1's assistance proposal, thus reaching a consensus: R2 will lead the transport and R1 will provide assistance.

[0084] Through the above technical solution, this application effectively solves the problem of the lack of specific operational mechanisms in traditional negotiation processes, ensuring the depth and effectiveness of information exchange between entities. By enabling entities to share their decision-making rationale and dynamically adjust their intentions, the efficiency and success rate of negotiation are significantly improved, avoiding deadlock and thus ensuring the reliability and stability of multi-entity collaborative decision-making.

[0085] In some of the solutions mentioned above in this application, a multi-entity collaborative decision-making method is proposed to coordinate the decision intentions of multiple entities and perform event processing. However, in this process, the decision-making process is not recorded and the communication content between entities is not stored, which makes it impossible for users to query the decision history, audit the decision quality, or optimize based on historical data, resulting in a lack of system traceability and insufficient decision transparency.

[0086] In this regard, this application further proposes to include recording the decision-making process and storing the communication content between entities in a storage system for users to query.

[0087] Recording the decision-making process refers to the system automatically capturing and saving all important information related to the decision at each key stage of multi-entity collaborative decision-making. This can include the decision initiation time, participating entities, the initial decision intentions of each entity, the selected collaboration strategy, the final decision execution method, and the results of event handling. In terms of implementation, this can be achieved by setting log points at each key node of the decision-making process, automatically capturing and structurally storing this information through a programming interface; alternatively, an event-driven architecture can be adopted, where a dedicated recording service is triggered to write relevant data to storage when the decision state changes.

[0088] Inter-entity communication refers to all data exchanged between different entities during the decision-making process. This may include decision rationale exchanged during the negotiation phase, dynamic adjustments to their intentions, status updates, task assignment notifications, etc. To ensure data integrity, this communication content is recorded in detail. In terms of implementation, message interception or replication functions can be integrated into the entity communication module to synchronously record the entire message content, sender, receiver, timestamp, and other metadata when a message is sent or received; alternatively, message queues or middleware can be deployed to capture and store data as messages flow through the system.

[0089] Storing data in a storage system refers to persistently saving the aforementioned decision-making process information and inter-entity communication content to a reliable data storage medium. This storage system aims to ensure data integrity, reliability, and long-term availability for subsequent queries and analysis. In terms of implementation, relational databases (such as MySQL and PostgreSQL) can be used to store structured decision logs and communication records, managing data through clearly defined table structures; alternatively, non-relational databases (such as MongoDB and Cassandra) can be used to store semi-structured or unstructured communication data to accommodate more flexible data models and high-concurrency write requirements; or, for large-scale log data, a distributed file system (such as HDFS) can be used for storage, coupled with indexing techniques for efficient retrieval.

[0090] Providing user-query capabilities refers to offering corresponding interfaces or tools that allow users to actively retrieve and access decision-making processes and communication content within the storage system. This enables users to obtain the historical data they need based on specific query criteria (such as time range, entity ID, event type, etc.), thereby supporting decision analysis, feedback collection, and strategy optimization. In terms of implementation, a user-friendly web interface or client application can be developed, providing search boxes, filters, and results display functionality; alternatively, an API can be provided, allowing other systems or data analysis tools to access and extract data programmatically.

[0091] In multi-entity collaborative decision-making methods, when the decision intentions of multiple entities regarding the same event are acquired, collaborative strategies are determined, and decision execution methods are selected and the event is handled by the executing entity, this application further introduces a recording mechanism for the entire decision-making process. Specifically, at each key stage of the decision-making process, the system captures and records important events and state changes in real time, such as the initial decision intentions of each entity, the basis for selecting collaborative strategies, the determination result of the decision execution method, and the final event handling action. Simultaneously, to comprehensively reflect the interaction between entities, all communication content between entities during the decision-making process, including but not limited to information exchange during negotiation, notification of intention adjustments, and status updates, is also recorded synchronously and completely. This recorded decision-making process information and inter-entity communication content are then uniformly stored in a persistent storage system. This storage system ensures the integrity, reliability, and long-term availability of the data. In this way, users can access this stored data at any time through a query interface to understand the background of the decision, evaluate the rationality of the decision, analyze the collaboration patterns between entities, and provide data support for subsequent strategy optimization. This recording and query mechanism is closely integrated with the basic collaborative decision-making process, making the previously opaque decision-making process traceable and auditable, greatly improving the transparency and management capabilities of the entire multi-entity collaborative decision-making system.

[0092] For example, in a smart factory, when multiple industrial robots collaboratively move large workpieces, the system records the entire decision-making process. When a workpiece arrives, each robot, based on its position, load capacity, and current task, generates a decision intention (e.g., "Robot A intends to move," "Robot B intends to wait"). This intention information, along with metadata such as the time of generation and robot ID, is recorded. Subsequently, the system determines that Robot A will be the execution entity to move the workpiece based on a preset collaboration strategy (e.g., robot priority, task matching). The process and result of determining this decision-making method are also recorded. If, during the decision-making process, multiple robots have similar priorities and initiate a negotiation process, the communication content regarding the decision reasons exchanged between the robots (e.g., "Robot A reports sufficient load capacity and shortest path," "Robot B reports its current task is full") and the dynamic adjustment of their intentions (e.g., "Robot B agrees to wait") will be stored in detail in the storage system. This recorded data can be stored in multiple tables in a relational database, such as a "decision log table" to record decision events and a "communication log table" to record messages between entities. When users need to understand the decision-making process of a certain transportation task, they can enter the task ID or time range through the management interface to query. The system will display the complete chain from the generation of intention to the final execution, including the communication details of all participating entities, so as to clearly understand the decision-making process.

[0093] Building upon the aforementioned multi-entity collaborative decision-making methods, this system introduces techniques to record the decision-making process and inter-entity communication content, storing it in a storage system for user retrieval. This transforms what might have been an opaque decision-making process into one that is fully traceable. Users no longer face the problems of unsearchable decision history and difficulty in auditing decision quality, effectively solving the technical issues of lack of system traceability and insufficient decision transparency. Specifically, comprehensive recording of the decision-making process ensures that key information such as the background, strategy selection basis, and executing entities for each collaborative decision is verifiable, providing a solid foundation for subsequent auditing and problem diagnosis. Simultaneously, the storage of inter-entity communication content fully preserves the details of negotiation and information exchange between entities, which is crucial for understanding the logic behind decisions and evaluating the efficiency of entity collaboration. Users can easily access this historical data through the query function, which can be used not only for decision analysis and evaluating the effectiveness of collaborative strategies but also to provide valuable data support for the continuous optimization of collaborative strategies, thereby improving the intelligence and reliability of the entire multi-entity collaborative decision-making system.

[0094] In some of the solutions mentioned above in this application, a collaborative strategy is proposed to determine the decision execution method. However, in its implementation, the collaborative strategy is statically preset and cannot be dynamically adjusted according to the actual execution results and user feedback. This results in the decision system lacking adaptive learning ability, which limits the continuous improvement and optimization of decision quality.

[0095] In this regard, this application further proposes to collect user feedback after the decision is implemented, and to optimize the collaboration strategy based on the feedback.

[0096] Here, "after decision execution" refers to the point in time after the execution entity corresponding to the decision execution method completes the process of handling the event in the multi-entity collaborative decision-making method. This can be achieved by the execution entity sending a completion signal or status update to the decision system after completing the task, or by the system monitoring the event processing results, such as task completion or goal achievement. Furthermore, a time window can be preset to collect feedback within a certain period after the event processing is completed. This aims to ensure that the feedback is based on actual execution results rather than preset or assumed information, providing a real and effective basis for subsequent optimization.

[0097] "Collecting user feedback" refers to obtaining users' evaluations, opinions, or satisfaction information regarding the results of decision implementation. This can be achieved through user interfaces providing ratings, reviews, or questionnaires, or by analyzing users' related comments on social media and in customer service conversations using natural language processing technology. Furthermore, user feedback can be indirectly inferred by monitoring user behavior data, such as the frequency and duration of user interactions with the implementation results, and subsequent actions. This approach aims to directly obtain users' subjective evaluations of the decision's effectiveness, serving as a crucial indicator for assessing the merits of collaborative strategies and a vital input for strategy optimization.

[0098] "Optimizing the collaborative strategy based on the feedback" refers to adjusting and improving the collaborative strategy used to determine the decision-making execution method based on collected user feedback information. This can be achieved through machine learning algorithms, such as reinforcement learning and supervised learning, adjusting parameter weights, rule priorities, or model structures in the collaborative strategy based on feedback data. Alternatively, it can be achieved through expert systems or rule engines, triggering preset strategy adjustment rules based on user feedback; for example, automatically switching to another collaborative strategy or adjusting relevant parameters when user satisfaction falls below a threshold. Furthermore, manual intervention is possible, with system administrators or domain experts manually adjusting the collaborative strategy based on feedback data. This aims to enable the collaborative strategy to be adaptive and learnable, continuously improving based on actual results, thereby enhancing the quality of decision-making and user satisfaction.

[0099] This application's solution constructs a closed-loop learning mechanism by collecting user feedback after decision execution and optimizing the collaborative strategy based on that feedback. In the multi-entity collaborative decision-making method, the decision intentions of multiple entities regarding the same event are first obtained. Then, the decision execution method is determined according to a preset collaborative strategy, and the corresponding execution entity handles the event. After the event is processed, the system does not stop at the execution result but actively collects user feedback on the decision execution. This feedback data, whether explicit satisfaction evaluations or implicit behavioral patterns, serves as valuable experience input for evaluating the effectiveness of the current collaborative strategy. Subsequently, the system uses this feedback data to adjust and optimize the collaborative strategy, such as updating strategy parameters, modifying rule weights, or adjusting the decision model. This continuous feedback-learning-optimization cycle enables the collaborative strategy to dynamically adapt to constantly changing environments, task requirements, and user preferences, thereby overcoming the limitations of traditional static preset strategies and significantly improving the accuracy, efficiency, and user satisfaction of decision-making. In this way, the entire multi-entity collaborative decision-making system can continuously learn and evolve from experience, achieving true self-adaptation and intelligence.

[0100] The following example illustrates this. In a smart home system, when a user issues the voice command "I'm going to sleep," multiple smart devices (such as lights, air conditioners, curtains, and security systems) will generate their own decision intentions (such as dimming lights, switching the air conditioner to sleep mode, closing curtains, and arming the security system). The system determines a comprehensive decision execution method based on preset coordination strategies, such as a fusion strategy, and coordinates the devices to perform the corresponding actions simultaneously. After these actions are completed, the system collects user feedback. For example, the next morning, the user might express "I slept very well last night" via voice or a smart panel, or the system might infer positive feedback by monitoring that the user did not manually adjust any devices overnight. Based on this positive feedback, the system optimizes its coordination strategy. Specifically, the system might increase the priority or weight of the current fusion strategy when processing the "I'm going to sleep" command, or fine-tune the parameters of each device's actions, such as setting the dimming level of the lights or the temperature of the air conditioner's sleep mode to better suit the user's habits. Conversely, if the user reports "the lights are too bright," the system will adjust the dimming parameters accordingly to avoid similar problems in the future. Through this continuous feedback learning mechanism, smart home systems can continuously improve the intelligence and personalization of their decision-making, better meeting user needs.

[0101] Through the aforementioned technical solutions, the multi-entity collaborative decision-making method can continuously learn from actual execution results and user feedback, thereby overcoming the limitations of statically preset collaborative strategies. This endows the decision-making system with adaptive learning capabilities, enabling it to dynamically adjust collaborative strategies according to environmental changes and user needs, significantly improving decision quality and user satisfaction. Simultaneously, this mechanism enhances the system's robustness and adaptability, allowing it to make better decisions in complex and ever-changing situations, and achieving personalized and intelligent strategies while reducing manual maintenance costs.

[0102] In some of the solutions mentioned above in this application, decision intentions are proposed to obtain the decision-making intentions of entities as the basis for collaborative decision-making. However, in this process, the scope of decision intentions may not be clear or comprehensive enough, resulting in vague expression of entity intentions, affecting the accurate formulation of collaborative strategies and the effective resolution of conflicts, thereby reducing decision-making efficiency.

[0103] In this regard, this application further clarifies the scope of decision intention, which includes, but is not limited to, at least one of actively initiating interaction, generating content, or performing an action.

[0104] Among these, decision intention refers to the expectations, goals, or action tendencies expressed by an entity regarding a specific event or task. It can be generated in various ways, such as automatically generating it based on the entity's current state and environmental information through a pre-set rule engine or expert system; predicting entity behavior based on historical data and real-time perception information through machine learning models; or parsing the voice or text commands of entities (such as intelligent agents) through natural language processing technology. Actively initiating interaction refers to an entity's intention to actively establish connections with other entities or the environment. For example, an entity sends requests, queries, or notification messages to other entities, or actively triggers interaction with the user interface, such as displaying prompts or requesting user input. Generating content refers to an entity's intention to create information or data. For example, an entity generates reports, analysis results, or recommendation lists based on input data or internal logic, or an entity creates multimedia content such as text, images, audio, or video. Performing actions refers to an entity's intention to perform physical or logical operations. For example, an entity controls hardware devices to perform physical operations, such as moving a robotic arm or adjusting the temperature, or an entity performs logical operations in a software system, such as modifying database records or calling API services.

[0105] This application's solution, by clearly categorizing decision intentions, enables a more accurate and comprehensive capture of an entity's true intent when acquiring decision intentions from multiple entities regarding the same event. When an entity expresses its intention—whether it's actively communicating with other entities, generating new information, or directly performing an operation—these intentions can be clearly identified and categorized by the system. This clear expression of intent significantly reduces ambiguity in intent understanding, thus providing a solid foundation for subsequently determining the decision execution method based on preset collaborative strategies. For example, when the system identifies an entity's intent as "performing an action," it can prioritize its impact on the physical world and security; when the intent is "generating content," it may focus on information integration and output quality. This refined understanding of intent types makes the selection and application of collaborative strategies more targeted, avoiding strategy misjudgments or execution deviations caused by ambiguous intents, thereby ensuring that the executing entity corresponding to the decision execution method can efficiently and accurately handle the event.

[0106] The following is a concrete example. On a smart factory production line, when an anomaly is detected at a workstation, multiple entities may generate decision intentions. For example, a robot entity may have the intention to "execute an action," i.e., stop its current operation and move to a safe area; a monitoring system entity may have the intention to "generate content," i.e., generate a report containing the time, type, and location of the anomaly and send it to maintenance personnel; and a scheduling system entity may have the intention to "initiate interaction," i.e., send instructions to other relevant robot entities, requesting them to adjust their production plans or provide assistance. The solution proposed in this application allows the system to clearly identify and distinguish these different types of decision intentions. For example, the system will recognize that the robot entity's intention is to "execute an action," the monitoring system's intention is to "generate content," and the scheduling system's intention is to "initiate interaction." Based on these clearly defined intention types, the system can, according to preset coordination strategies, such as prioritizing safety-related "execution actions," simultaneously initiating "content generation" to record the anomaly, and triggering "initiated interaction" to coordinate with other entities. This clear intention recognition enables the entire anomaly handling process to proceed efficiently and orderly.

[0107] Through the above technical solution, this application effectively solves the problem of ambiguous entity intent expression caused by unclear or incomplete decision-making intent scope in traditional solutions. By clarifying the specific types of decision-making intent, including proactively initiating interactions, generating content, or performing actions, the system can more accurately understand and capture the true intentions of entities, thereby significantly improving the accuracy and effectiveness of collaborative strategy formulation. This clear expression of intent helps reduce misunderstandings and conflicts between entities, enabling more efficient determination of decision execution methods in multi-entity collaborative decision-making processes, with the corresponding executing entity handling the events, ultimately improving overall decision-making efficiency and system stability.

[0108] In traditional multi-entity systems, multiple entities may have conflicting decision-making intentions regarding the same event during the decision-making process. If there is a lack of effective coordination mechanisms, it can lead to behavioral conflicts or resource competition, affecting system stability and task completion quality. At the same time, existing systems usually have fixed master-slave relationships or simple voting mechanisms, which cannot dynamically select the optimal collaboration strategy based on real-time situations and lack the ability to learn from historical decisions, thus limiting cross-domain applications.

[0109] To address this issue, this application proposes a multi-entity collaborative decision-making system, comprising an intention acquisition module, a collaboration module, and an execution module. The intention acquisition module is configured to acquire the decision intentions of multiple entities regarding the same event, ensuring that the intentions of all entities are collected uniformly, thus avoiding the risk of conflict due to dispersed intentions. The collaboration module is configured to determine the decision execution method based on a preset collaboration strategy. This strategy can dynamically select execution methods such as arbitration, fusion, or negotiation, thereby overcoming the limitations of fixed rules or single voting mechanisms and achieving flexible decision coordination. The execution module is configured to process the event according to the determined decision execution method, ensuring the effective implementation of the collaborative results.

[0110] The solution presented in this application achieves unified collection, dynamic coordination, and efficient execution of multi-entity decisions through the organic cooperation of the aforementioned modules. Specifically, the intention acquisition module provides the basic input for collaboration, the collaboration module flexibly determines the execution method based on preset strategies, and the execution module ensures the implementation of decisions. The collaborative work of these three modules effectively resolves decision-making conflicts and enhances the system's flexibility and cross-domain adaptability. For example, in a collaborative handling scenario using industrial robots, when multiple robots intend to handle the same workpiece, the system collects intentions through the intention acquisition module, the collaboration module selects the execution entity according to preset strategies, and the execution module controls the selected robot to complete the handling, avoiding conflicts and improving efficiency. Through the above technical solution, the system can not only adapt to different fields such as industrial automation and smart homes, but also maintain robustness and efficiency in dynamic environments.

[0111] In some of the embodiments described above in this application, a collaboration module is proposed to determine the decision execution method according to a preset collaboration strategy. However, in this process, the specific implementation of the collaboration strategy may not be flexible enough and may not be able to dynamically select the fusion decision or the execution entity according to the real-time situation, resulting in low decision efficiency or incomplete conflict resolution.

[0112] In response, this application further proposes a specific configuration method for the collaboration module, namely, the collaboration module is configured to merge the decision intentions of multiple entities to generate a comprehensive decision, and / or select one of the multiple entities as the decision execution entity.

[0113] The collaboration module is configured to merge the decision intentions of multiple entities to generate a comprehensive decision. Fusion decision refers to integrating decision intentions from different entities regarding the same event to form a unified and comprehensive decision result. This fusion can be achieved in various ways. For example, a weighted average method can be used, assigning different weights to each entity based on its priority, confidence level, or historical performance, and then summing the intentions of each entity using weighted averages. Alternatively, it can be based on a predefined set of rules or an expert system, making logical judgments and combinations based on the type and context of the intentions. Another approach is to utilize machine learning models, which learn how to extract key information from the intentions of multiple entities and generate the optimal comprehensive decision through training. The collaboration module is also configured to select one entity from the multiple entities as the decision execution entity. Selecting the decision execution entity means that, under specific circumstances, when the system determines that it is more efficient or reasonable for a single entity to execute the decision, it selects the most suitable entity from among many entities to execute the decision. This selection can be based on specific attributes of entities. For example, it can be based on entity priority, prioritizing entities with higher privileges or more important roles; it can also be based on entity capabilities or resource status, such as selecting entities with the lowest current load, the strongest processing capacity, or specific execution capabilities; or it can be based on the matching degree between entities and tasks, evaluating the suitability of each entity to perform the current task, and selecting the entity with the highest matching degree.

[0114] The solution proposed in this application enables a multi-entity collaborative decision-making system to dynamically adjust its decision execution method according to actual needs through flexible configuration of the collaboration module. Specifically, when the intention acquisition module obtains the decision intentions of multiple entities regarding the same event, the collaboration module evaluates the current situation and task characteristics based on a preset collaboration strategy. If it determines that it is necessary to integrate multiple opinions to form a more comprehensive and robust decision, the collaboration module will activate a fusion mechanism to merge these decision intentions and generate a comprehensive decision. This comprehensive decision is then passed to the execution module, which processes the event according to the comprehensive decision. Conversely, if it is determined that it is more efficient for a single entity to execute the decision or that a specific entity has an irreplaceable advantage, the collaboration module will activate a selection mechanism to select the most suitable entity from multiple entities as the decision execution entity. This selected entity will be responsible for processing the event, and its decision intention or execution capability will directly guide the operation of the execution module. This "AND / OR" configuration method allows the collaboration module to intelligently switch between fusion and selection based on different collaboration strategies and real-time situations, thereby avoiding the limitations of traditional fixed collaboration methods and significantly improving the system's ability to cope with complex and changing environments. Through this dynamic selection mechanism, based on the decision intentions of multiple entities, this system can more accurately and efficiently determine the decision execution method, and then the execution module handles the event, ensuring the adaptability and effectiveness of the decision.

[0115] The following example illustrates this. In a smart home system, when a user issues the command "Prepare dinner," the intention acquisition module obtains the decision intentions of multiple entities (e.g., smart refrigerator, smart oven, smart lighting system, smart speaker). The smart refrigerator might intend to "check the ingredient list," the smart oven might intend to "preheat to 180 degrees Celsius," the smart lighting system might intend to "adjust the kitchen lights," and the smart speaker might intend to "play cooking background music." At this point, the coordination module makes a judgment based on a preset coordination strategy. As a specific implementation, if the system determines that these intentions need to be coordinated to create an overall atmosphere, the coordination module can be configured to merge the decision intentions of the smart refrigerator, smart oven, smart lighting system, and smart speaker to generate a "comprehensive dinner preparation decision." This comprehensive decision might include a series of coordinated actions such as "checking ingredients and recommending recipes, preheating the oven, adjusting the kitchen lights to a warm tone, and playing soft music," which are then uniformly scheduled and executed by the execution module. As another specific implementation, if the user subsequently issues the command "turn on the range hood", the collaboration module can be configured to select the smart range hood as the decision execution entity from all entities, because this task is best performed by a single entity with specific functions, and the execution module directly triggers the range hood to turn on.

[0116] Through the above technical solution, the collaboration module of this application can flexibly choose to merge the decision intentions of multiple entities to generate a comprehensive decision, or select one entity from multiple entities as the decision execution entity, based on the actual situation. This dynamic selection mechanism effectively solves the problem of insufficient flexibility in traditional collaboration strategies, enabling the system to adaptively adopt the most appropriate decision execution method based on factors such as the complexity of the event, the correlation between entities, and the urgency of the task. For example, in scenarios requiring multi-party collaboration and information complementarity, fusion decision-making can gather the wisdom of various entities to form a more comprehensive and optimized solution; while in scenarios where the task is clear and can be efficiently completed by a single entity, selecting a single execution entity can avoid unnecessary coordination overhead and improve decision-making efficiency. Therefore, this solution significantly improves the adaptability, decision-making efficiency, and conflict resolution capabilities of the multi-entity collaborative decision-making system, ensuring the effectiveness and robustness of the system behavior under various complex situations.

[0117] In some of the solutions mentioned above in this application, a collaboration module is proposed to determine the decision execution method according to a preset collaboration strategy. However, in this process, since the formulation of the collaboration strategy may only rely on a single or fixed factor, it cannot fully reflect the real-time status of the entity, environmental changes and task requirements, resulting in low decision efficiency and insufficient adaptability, and it is impossible to achieve optimal collaboration in complex and ever-changing scenarios.

[0118] In this regard, this application further proposes that the collaborative module makes decisions based on at least one of the following factors, including but not limited to: the entity's priority, the relationship parameters with the target object, the current internal state, the environmental state, the task matching degree, and the historical decision success rate.

[0119] Specifically, entity priority refers to the relative importance or decision weight assigned to different entities in a multi-entity collaborative decision-making system. For example, it can be set based on the entity's role in the system, functional criticality, or preset permission level, such as setting the priority of the core control unit higher than that of auxiliary sensor units, or dynamically adjusting the priority according to the task type. Another approach is to use machine learning models to automatically learn and adjust priorities based on the entity's past performance and system objectives. Relationship parameters with the target object refer to the degree of association or characteristics between the entity and the target object involved in the decision. For example, in a smart home scenario, the intimacy, trust, or historical interaction frequency between the smart speaker and the user can serve as relationship parameters; in industrial control, the physical distance, connection status, or task relevance between the robot and the workpiece can also serve as relationship parameters. Current internal state refers to the real-time operating status or internal attributes of an entity when making a decision. For example, for hardware devices, its battery level, CPU load, memory usage, temperature, or operating mode (such as energy-saving mode or high-performance mode) can all serve as internal states; for software agents, their emotional state, task queue length, or resource usage can also serve as internal states. Environmental state refers to the real-time information of the external environment in which the entity exists. For example, this could be ambient temperature, light intensity, humidity, network latency, traffic conditions, weather information, or external events (such as alarm triggering). This information can be acquired in real time via sensors or received through external system interfaces. Task matching refers to the degree to which an entity is suited or capable of performing a specific task. For example, it can be assessed based on an entity's skill set, available resources, area of ​​expertise, or historical task completion efficiency. In robot collaboration, the degree of matching between a robot arm's grasping ability and the weight and shape of the object to be moved, or the ability of a software module to handle specific data types, can both serve as examples of task matching. Historical decision success rate refers to the record of successes an entity or system has achieved when making similar decisions in the past. For example, it can be calculated by statistically analyzing the ratio of the number of times an entity achieves its expected goal after making a decision and executing it in a specific situation to the total number of decisions, or by using reinforcement learning algorithms to update the success rate based on reward and punishment mechanisms.

[0120] The aforementioned multi-entity collaborative decision-making system acquires the decision intentions of multiple entities regarding the same event through an intention acquisition module. Based on this, the collaboration module does not simply determine the decision execution method according to a single or fixed rule, but rather comprehensively considers multi-dimensional information. Specifically, the collaboration module dynamically evaluates the priority of entities to ensure that the decision weight of key entities is reflected; simultaneously, it combines the relationship parameters between entities and target objects to make the decision more consistent with the actual situation and the collaborative relationship between entities. Furthermore, the collaboration module also perceives the current internal state of entities and the state of the external environment in real time to ensure the feasibility of the decision and its adaptability to environmental changes. During task allocation, the collaboration module calculates the task matching degree and prioritizes the entity most suitable for executing the task. Further, the system also utilizes historical decision success rates to learn from past experience and optimize current decision-making strategies. Through this multi-factor fusion decision-making mechanism, the collaboration module can generate more accurate, efficient, and adaptable decision execution methods, thereby effectively guiding the execution module in handling the event. This mechanism, which comprehensively considers multiple dynamic factors, enables the entire system to avoid inefficiency and insufficient adaptability caused by a single decision-making basis when facing complex and ever-changing tasks and environments, and significantly improves the intelligence level and robustness of multi-entity collaborative decision-making.

[0121] The following example illustrates this. In a multi-sensor fusion scenario for autonomous vehicles, when a potential obstacle appears in front of the vehicle, multiple sensors, such as cameras, radar, and lidar, act as entities and each generate its own decision intention regarding the obstacle (e.g., the camera identifies a "pedestrian," the radar detects a "moving object," and the lidar measures "distance and contour"). At this point, the collaborative module comprehensively utilizes these decision factors to determine the final decision-making method. For example, the collaborative module might assign higher weight to lidar for distance measurement based on entity priority, while the camera might have higher priority for object classification. Simultaneously, the collaborative module considers relationship parameters with the target object; for example, if the target object is identified as a pedestrian, the system tends to adopt a more conservative avoidance strategy. Furthermore, the collaborative module evaluates the current internal state of each sensor; for example, if a sensor malfunctions or its performance degrades, its decision weight will be reduced accordingly. Environmental conditions, such as the impact of rain or fog on camera visual recognition, are also taken into consideration, thereby adjusting the confidence levels of each sensor. Task matching reflects the proficiency of different sensors in specific tasks (such as ranging, classification, and speed estimation). Finally, the coordination module also considers historical decision success rates; for example, under similar weather conditions, which sensor combination's decision-making in the past prevented accidents, thus optimizing the current decision. Through comprehensive analysis of these factors, the coordination module can generate a highly reliable overall judgment (e.g., determining that there is a "pedestrian" ahead and emergency braking is required), and the execution module (such as the vehicle's braking system) will then carry out the corresponding action.

[0122] Through the above technical solution, this application effectively solves the problem of low decision-making efficiency and insufficient adaptability caused by the reliance on single or fixed factors in traditional collaborative strategy formulation. The collaborative module, by comprehensively considering multiple dynamic factors such as entity priority, relationship parameters with the target object, current internal state, environmental state, task matching degree, and historical decision success rate, can more comprehensively and in real-time reflect the complexity of the system and environment. This makes the determination of decision execution methods more accurate and flexible, significantly improving the intelligence level of multi-entity collaborative decision-making and its adaptability to complex and ever-changing scenarios. Ultimately, the solution of this application can effectively avoid decision conflicts, optimize resource utilization, and continuously improve the quality and efficiency of task completion.

[0123] In some of the solutions mentioned above in this application, a multi-entity collaborative decision-making system is proposed to coordinate the decisions of multiple entities. However, in this process, when the decision intentions of multiple entities conflict, there is a lack of an effective negotiation mechanism to dynamically adjust and reach a consensus.

[0124] In this regard, this application further proposes that the system also includes a negotiation module, which is used to initiate a negotiation process between entities when there is a conflict in the decision-making intentions of multiple entities.

[0125] The negotiation module is a dedicated component within the system for handling decision-making conflicts between entities. This module can be a standalone software service or process responsible for receiving conflict signals and coordinating communication between entities, or it can be a sub-module integrated within the collaboration module, activated upon detecting a conflict. Conflicts arise when the decision intentions of multiple entities conflict, meaning the system detects that the submitted decision intentions are mutually exclusive, competing for resources, or cannot be executed simultaneously. Conflict detection can be identified by comparing whether the submitted decision intentions are mutually exclusive or competing for resources—for example, two entities requesting exclusive access to the same resource, or two entities issuing contradictory action instructions; it can also be determined using a pre-defined set of rules or a machine learning model, for example, when multiple entities offer different operational suggestions for the same target object, and these suggestions cannot be executed simultaneously. Once a conflict is detected, the system initiates a negotiation process between entities. This process can be achieved by sending negotiation request messages to all relevant entities and establishing a temporary communication channel; or by activating a predefined negotiation protocol, such as one based on auctions, voting, or reason exchange.

[0126] This application's solution effectively resolves decision-making conflicts by introducing a negotiation module into a multi-entity collaborative decision-making system and initiating a negotiation process when decision intentions conflict. Specifically, when the intention acquisition module obtains the decision intentions of multiple entities regarding the same event, the collaboration module attempts to determine the execution method based on a preset collaboration strategy. If, during this process, the collaboration module detects a conflict in the decision intentions of multiple entities—for example, multiple entities making mutually exclusive requests for the same resource or task—the negotiation module is activated. The negotiation module then initiates a negotiation process between entities, prompting conflicting entities to exchange information, such as their decision reasons, priorities, or current states. Through this information exchange, each entity can dynamically adjust its decision intentions until a consensus is reached or a feasible, conflict-free comprehensive decision is formed. Once the negotiation is successful, the revised decision intention is fed back to the collaboration module, which then determines the final decision execution method, and the execution module handles the event. This mechanism ensures that even in complex and dynamically changing scenarios, the system can effectively resolve conflicts, avoiding system chaos or task failures caused by inconsistent decisions, thereby significantly improving the system's robustness and decision quality.

[0127] The following is a concrete example to illustrate this. In a smart factory, suppose the intention acquisition module obtains the decision intentions of two industrial robots (entities) regarding the handling of the same large workpiece; that is, both robot A and robot B indicate "I will handle it." When the collaboration module attempts to determine the execution method, it detects a conflict in the decision intentions of robot A and robot B because the same workpiece cannot be handled by two robots simultaneously. At this point, the negotiation module is activated. The negotiation module initiates a negotiation process between robot A and robot B. Robot A might send its decision reasons to robot B, such as "I am closer to the workpiece and have sufficient battery power"; robot B might also respond with its decision reasons, such as "I am equipped with a more suitable special fixture for this workpiece, and have a higher historical handling success rate." Through the information exchange coordinated by the negotiation module, robot A and robot B dynamically adjust their respective decision intentions. For example, after learning that robot B has the advantage of a special fixture, robot A might adjust its intention to "I wait, and robot B will handle it," or the two parties might negotiate a collaborative solution, such as "robot B handles the workpiece to a designated area, and robot A assists in positioning." Once a consensus is reached, the negotiation module will feed back the final, conflict-free decision intention to the collaboration module. The collaboration module can then determine whether robot B will perform the handling task, or whether robots A and B will work together to complete it. Finally, the execution module will handle the handling event.

[0128] Through the aforementioned technical solution, the system possesses specialized capabilities for handling decision-making conflicts. When a conflict is detected between the decision-making intentions of multiple entities, a negotiation process can be initiated promptly, preventing escalation of the conflict or system behavioral chaos. The negotiation process facilitates information exchange and intention adjustment among entities, enabling them to dynamically reach a consensus, thereby improving the robustness of decision-making and the stability of the system. Combined with the basic solution, the negotiation module ensures that the multi-entity collaborative decision-making system can operate efficiently and reliably even in complex and ever-changing environments, significantly improving the system's adaptability and decision quality.

[0129] In some of the solutions mentioned above in this application, a multi-entity collaborative decision-making system is proposed to handle event decisions. However, in this process, the communication content between entities is not recorded, resulting in the decision-making process being untraceable and unable to be queried or audited by users, thereby limiting the transparency and optimization capabilities of the system.

[0130] In response, this application proposes a multi-entity collaborative decision-making system, which also includes a recording module for storing the communication content between entities into a storage system.

[0131] The recording module's role is to capture, process, and store all communication data between entities within the multi-entity collaborative decision-making system. This module can be a standalone software service, capturing inter-entity communication data in real time by monitoring the system's internal message bus or API calls; alternatively, it can be a built-in function of the collaboration or execution module, directly calling the storage interface to record data when communication occurs. The communication content between entities refers to the data generated by information exchange between different entities during the multi-entity collaborative decision-making process. This content is crucial for understanding the decision-making process, analyzing conflict resolution mechanisms, and evaluating the effectiveness of collaborative strategies. Specifically, the communication content can include, but is not limited to, structured data such as message bodies, sender identifiers, receiver identifiers, timestamps, message types, decision intentions, negotiation reasons, and execution instructions; it can also encompass raw data streams, log files, API call records, and status update information. The storage system refers to the hardware and / or software infrastructure used for persistently storing data. Its function is to ensure that recorded communication content is not lost and can be accessed, queried, and analyzed long-term. The storage system can use a relational database (such as MySQL or PostgreSQL) to store structured communication logs; it can also use a non-relational database (such as MongoDB or Cassandra) to store semi-structured or unstructured communication content; or it can be a distributed file system (such as HDFS) to store a large number of log files or raw data.

[0132] In a multi-entity collaborative decision-making system, the intention acquisition module is responsible for acquiring the decision intentions of multiple entities regarding the same event. The transmission of these intentions themselves constitutes the communication content between entities. Subsequently, the collaboration module, according to a preset collaboration strategy, merges, arbitrates, or negotiates these intentions to determine the decision execution method. The strategy selection, intention fusion logic, and information exchange between entities during this process are all important communication content. Finally, the execution module processes the event according to the determined decision execution method. The instructions it issues to the executing entities and the execution feedback it receives also fall under the category of communication content. Based on this, the recording module is configured to capture and collect the communication content between these entities in real-time or near real-time at each of the above stages. This captured communication content is then persistently stored in a storage system. Through this mechanism, key information throughout the entire decision-making chain, from the generation of entity intentions, the formulation of collaboration strategies, conflict resolution to the execution of the final decision, is completely recorded. This gives the system, which was originally only able to execute decisions, the ability to fully trace and audit the decision-making process, greatly improving the system's transparency and providing a solid data foundation for subsequent decision analysis, strategy optimization, and problem investigation.

[0133] As a specific implementation method, let's take the collaborative handling of large workpieces by industrial robots in a smart factory as an example. When a large workpiece needs to be moved, multiple industrial robots will send their respective decision intentions to the intention acquisition module based on their own status (such as position, load capacity, and current task), such as "I will move it" or "I will wait." This intention information, including robot ID, intention type, and status parameters, is captured by the recording module. If the collaboration module determines that negotiation is necessary, for example, if multiple robots have similar priorities, then the robots will exchange negotiation reasons, such as "My battery is fully charged, and the path is unobstructed" or "I am currently performing an urgent task." These messages during the negotiation process, including the sender, receiver, negotiation content, and timestamp, are also captured by the recording module in real time. Finally, the collaboration module determines which robot will act as the execution entity and sends an execution instruction to it, such as "Robot A, please move the workpiece to the designated location." This instruction and Robot A's execution feedback (such as "Start moving" or "Moving complete") are also captured by the recording module. All captured inter-entity communication content, including intentions, negotiation messages, instructions, and feedback, is structured and stored in a distributed storage system, such as a log storage solution based on Kafka message queues and Elasticsearch. Kafka is responsible for receiving and buffering communication logs in real time, while Elasticsearch is used for persistent storage and providing efficient query services.

[0134] By introducing a recording module and storing the communication content between entities in a storage system, this application effectively solves the problem of untraceable decision-making processes. This enables the system to completely record every step from the generation of entity intentions to the execution of decisions, thereby providing users with the ability to query and audit the decision-making process and greatly improving the system's transparency. Furthermore, this recorded communication content constitutes valuable data assets that can be used for in-depth analysis of the effectiveness of collaborative strategies, identification of the root causes of decision conflicts, and evaluation of the rationality of entity behavior. This provides data support for the continuous optimization of collaborative strategies, ensuring that the multi-entity collaborative decision-making system can continuously learn and improve, ultimately enhancing the overall decision-making quality and efficiency.

[0135] In some of the solutions mentioned above in this application, a multi-entity collaborative decision-making system is proposed to obtain decision intentions, determine execution methods, and handle events. However, in this process, there is a lack of a mechanism to learn from user feedback to optimize collaborative strategies, which results in the inability to adaptively improve decision quality and make it difficult to dynamically adjust collaborative strategies according to actual results.

[0136] In this regard, this application further proposes to include a feedback learning module, which is used to collect user feedback and optimize the collaboration strategy after the decision is implemented.

[0137] The feedback learning module is a component specifically designed to receive and process feedback information on decision execution results, and adjust and optimize the system's collaborative strategies accordingly. Its role is to endow the system with adaptive capabilities, enabling it to learn from historical experience and continuously improve decision quality. One implementation approach is as a standalone software service deployed on a server, communicating asynchronously with the collaborative and execution modules via message queues or Remote Procedure Calls (RPCs), receiving decision results and user feedback, and sending strategy update instructions to the collaborative module. Alternatively, the feedback learning module can be a subsystem embedded within the collaborative module, utilizing the collaborative module's computing resources and data interfaces to process feedback data in real time and directly modify the internal parameters or rules of the collaborative strategy. Collecting user feedback serves to obtain quantitative or qualitative evaluations of decision execution effectiveness, providing real-world data for strategy optimization. Specifically, explicit feedback mechanisms can be provided through user interfaces (such as mobile applications, web pages, or intelligent voice assistants), such as satisfaction ratings, comment input, and problem reports. Furthermore, implicit feedback, such as task completion time, resource consumption, error rate, and number of repeated user operations, can be automatically extracted by analyzing system logs, sensor data, or user behavior patterns. The role of optimizing collaboration strategies is to adjust or improve the decision-making logic and parameters adopted by the collaboration module based on the collected feedback information, thereby improving future decision performance and user experience. For example, rule-based optimization methods can be used to adjust the priority weights of specific entities in specific situations based on negative feedback, or to modify the triggering conditions of conflict resolution rules. As another example, machine learning models can be used for optimization. Through reinforcement learning algorithms, user feedback can be used as a reward signal to train the model to learn better collaboration strategies; or methods such as Bayesian optimization can be used to iteratively update strategy parameters to maximize user satisfaction or task success rate.

[0138] The system of this application first obtains the decision intentions of multiple entities regarding the same event through an intention acquisition module. Then, the collaboration module, based on a preset collaboration strategy, integrates these intentions to determine the optimal decision execution method. Next, the execution module processes the event according to this execution method. To ensure continuous improvement in decision quality, this application further introduces a feedback learning module. After the execution module completes event processing, the feedback learning module actively or passively collects user feedback. This feedback data, whether explicit user evaluations or implicit system performance indicators, serves as an important basis for evaluating the effectiveness of the current collaboration strategy. Based on this feedback information, the feedback learning module conducts in-depth analysis and evaluation of the collaboration strategy used in the collaboration module. For example, if the feedback shows poor decision-making results or low user satisfaction, the feedback learning module will identify weaknesses in the strategy. Based on this, the feedback learning module will optimize the collaboration strategy, such as adjusting the weights of various factors in the strategy, modifying decision rules, or updating machine learning model parameters. Through this closed-loop learning mechanism, the results of each decision will become valuable experience for the next decision optimization, enabling the entire multi-entity collaborative decision-making system to continuously adapt to environmental changes and user needs. This effectively solves the problems of traditional systems having a single collaborative mode and lacking adaptive learning, and achieves adaptive improvement of decision quality and dynamic adjustment of collaborative strategies.

[0139] As a specific implementation, in a smart home system, when a user issues the voice command "I want to sleep," the intention acquisition module receives decision intentions from multiple smart devices, such as lighting control entities, air conditioning control entities, curtain control entities, and security system entities. For example, the lighting entity's intention might be "dimming," the air conditioning entity's intention might be "switching to sleep mode," the curtain entity's intention might be "closing," and the security entity's intention might be "arming." The collaboration module, based on a preset fusion collaboration strategy, integrates these intentions into a comprehensive decision, which is then executed synchronously by the execution module. After the decision is executed, the feedback learning module intervenes. For example, the system can ask the user "Did you sleep well last night?" the next morning via a smart speaker or push a short satisfaction survey to the smart home app. If the user reports "very comfortable, the temperature is suitable," the feedback learning module records this positive feedback and strengthens the weight of the device parameter configurations under "sleep mode" in the current collaboration strategy, for example, increasing the priority of the current air conditioning temperature setting under "sleep mode." Conversely, if a user reports that "the air conditioner is too cold," the feedback learning module will identify the shortcomings of the current strategy and adjust the temperature setting parameter of the air conditioner in "sleep mode" in the collaborative strategy accordingly, or reduce the weight of this parameter in the overall strategy, so as to ensure that future decisions can better meet the user's personalized needs.

[0140] By introducing a feedback learning module, the multi-entity collaborative decision-making system of this application effectively solves the problem of traditional systems lacking an adaptive learning mechanism after decision execution. After decision execution, this module can collect user feedback promptly and accurately. This feedback data directly reflects the actual effect of the decision and the user experience. Based on this authentic and effective feedback information, the feedback learning module can continuously analyze, evaluate, and optimize the collaborative strategy, such as dynamically adjusting strategy parameters, updating decision rules, or improving the machine learning model. This allows the system to move beyond preset fixed strategies and dynamically adjust and improve according to actual operating conditions and user needs, thereby significantly improving the accuracy, efficiency, and user satisfaction of decisions. This mechanism ensures that the quality of collaborative decision-making continuously improves over time and with the accumulation of experience, achieving adaptive adjustment of collaborative strategies and continuous optimization of decision-making capabilities.

[0141] Traditional multi-entity collaborative decision-making methods lack specific computer-executable implementations when coordinating the decision-making intentions of multiple entities, dynamically selecting collaborative strategies, and resolving conflicts through negotiation. This results in the inability to store, load, and automatically execute the methods in computer systems, limiting their practical application and deployment efficiency.

[0142] In response, this application proposes a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method.

[0143] Specifically, the computer-readable storage medium is a physical or virtual carrier capable of storing digital data and readable by a computer system. This medium can be a non-transitory storage medium, such as a hard disk drive (HDD), solid-state drive (SSD), optical disc (CD-ROM, DVD, Blu-ray disc), flash memory (USB flash drive, SD card), read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), or electrically erasable programmable read-only memory (EEPROM), etc. These media provide persistent storage, ensuring that programs and data are not lost after power failure. Alternatively, the medium can also be a transient storage medium, such as random access memory (RAM). In specific application scenarios where a program is loaded into RAM and executed, RAM itself can also be considered a storage medium. The function of the computer-readable storage medium is to provide persistent storage, ensuring that programs and data are not lost after power failure and can be accessed and executed by the processor at any time.

[0144] The computer program is a collection of instructions designed to guide a computer in performing specific tasks or operations. This program can be a compiled program, such as one written in high-level languages ​​like C / C++ or Java, which is compiled into machine code or bytecode and can run directly on the operating system. Alternatively, it can be an interpreted program, such as one written in languages ​​like Python or JavaScript, which is translated and executed line by line by an interpreter at runtime. The function of the computer program is to transform abstract collaborative decision-making methods into a sequence of instructions that the computer can understand and execute, thereby automating the execution of the methods.

[0145] The processor is the core component of a computer system, responsible for executing instructions, processing data, and controlling other hardware devices. This processor can be a central processing unit (CPU), such as an Intel Core or AMD Ryzen series, serving as the core of general-purpose computing. Alternatively, it can be a graphics processing unit (GPU), playing a crucial role in parallel computing and specific algorithms (such as machine learning), or it can be an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), a microcontroller (MCU), etc., which are typically optimized for specific tasks. The processor's role is to read and execute computer programs stored on computer-readable storage media, thereby driving the operation of multi-entity collaborative decision-making methods.

[0146] The solution presented in this application encapsulates a multi-entity collaborative decision-making method into a computer program and stores it on a computer-readable storage medium, which is then executed by a processor. This achieves the automation, deployability, and repeatability of the method. Specifically, the computer-readable storage medium, serving as the carrier of the method, ensures the persistent storage and readily accessible nature of the program, overcoming the limitation of methods that cannot be persistently stored. The computer program stored on it transforms abstract decision-making logic into executable code, enabling the computer system to load and run the method, thus improving deployment flexibility. When the processor executes the program, it automatically follows the aforementioned method flow to obtain the decision intentions of multiple entities regarding the same event, determines the decision execution method based on a preset collaborative strategy, and has the corresponding execution entity handle the event. This collaborative working mechanism allows the originally theoretical multi-entity collaborative decision-making method to run efficiently in a real computing environment, thereby elevating the capabilities of dynamically selecting collaborative strategies, resolving conflicts through negotiation, and continuously optimizing from feedback from the conceptual level to an operational and deployable automated execution level.

[0147] As a specific implementation, a solid-state drive (SSD) deployed on a cloud server can be used as the computer-readable storage medium. A microservice application written in Java is stored on this SSD. This application is packaged into a JAR file and deployed on the server as the computer program. The JAR file contains all the business logic and algorithms implementing the aforementioned multi-entity collaborative decision-making method, such as functions like obtaining decision intentions, fusing or selecting decision execution methods, initiating negotiation processes, recording the decision-making process, and optimizing collaborative strategies based on feedback. The processor can specifically be an Intel Xeon series CPU configured in the cloud server. When the cloud server starts, the operating system loads and runs the Java microservice application. This application runs continuously as a background service, listening for decision intention requests from various entities. When a request is received, the CPU executes the instructions in the JAR file, processes these intentions according to the collaborative strategies implemented in the program (such as fusion, arbitration, or negotiation), and ultimately determines the decision execution method, then instructs the corresponding execution entity to handle the event.

[0148] Through the above technical solution, this application addresses the problem of the lack of concrete computer-executable implementations for multi-entity collaborative decision-making methods. This solution enables the abstract collaborative decision-making method to be persistently stored, loaded, and automatically executed by the processor, thereby significantly improving the practical application and deployment efficiency of the method. This not only ensures the automation of the collaborative decision-making process, eliminating the need for manual intervention, but also allows the advanced functions of the above method, such as dynamic strategy selection, conflict negotiation, and feedback learning, to run stably and efficiently in large-scale, complex computing environments, significantly improving the overall decision-making efficiency and robustness of multi-entity systems.

[0149] Other application scenarios The following simplified embodiments illustrate the application of the present invention in other fields. Specific implementations of these embodiments can be found in the examples described in the detailed embodiments above, and will not be repeated here.

[0150] Simplified Example 1: Multi-control unit collaboration in vehicle systems In the smart cockpit, when a user says "open the sunroof," the voice recognition unit, seat control unit, and air conditioning unit may generate an intention. The system selects the voice recognition unit as the executor based on the matching degree between each unit and the current task, and executes the command directly. If the user is satisfied with the result, the system records positive feedback to reinforce the matching rule.

[0151] Simplified Example 2: Multi-Agent Family Collaboration In a family AI agent system, the master agent wants to recommend movies, the child wants to play games, and the elderly want to listen to music. Based on the agents' role priorities, their relationship with the user, and their current emotional state, the system selects a fusion strategy and generates a comprehensive decision: "Watch the movie first, then play the game, and finally listen to music." After execution, user satisfaction feedback is collected to optimize the weighting of the coordination rules.

[0152] The above descriptions are merely embodiments of this application and are not intended to limit the scope of protection of this application. It should be noted that the term "user" in this application is a preferred embodiment of the feedback source, but not a limitation on the scope of protection. Those skilled in the art should understand that the feedback source can be any entity capable of interacting with the interactive system, such as a human user, other systems, devices, or applications. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

[0153] This solution is applicable not only to local devices but can also be deployed on cloud servers. Those skilled in the art should understand that the technical solution of this invention is not limited to a specific deployment environment, and any implementation based on the technical concept of this application should be considered to fall within the protection scope of this invention.

Claims

1. A multi-entity collaborative decision-making method, characterized in that, Includes the following steps: Obtain the decision intentions of multiple entities regarding the same event; Based on the pre-set collaboration strategy, determine the decision execution method; The event is handled by the execution entity corresponding to the decision execution method.

2. The method according to claim 1, characterized in that, The entity includes, but is not limited to, at least one of software modules, hardware devices, virtual characters, or intelligent agents.

3. The method according to claim 1, characterized in that, The method for determining the decision execution includes merging the decision intentions of multiple entities to generate a comprehensive decision; and / or selecting one of the multiple entities as the decision execution entity.

4. The method according to claim 1, characterized in that, The collaborative strategy makes decisions based on at least one of the following factors, including but not limited to: entity priority, relationship parameters with the target object, current internal state, environmental state, task matching degree, and historical decision success rate.

5. The method according to claim 1, characterized in that, Also includes: When the decision-making intentions of multiple entities conflict, a negotiation process is initiated between the entities to reach a consensus through information exchange.

6. The method according to claim 5, characterized in that, The negotiation process includes: each entity exchanging reasons for its decision, dynamically adjusting its own decision intentions, until a consensus is reached.

7. The method according to claim 1, characterized in that, It also includes recording the decision-making process and storing the communication content between entities in a storage system for users to query.

8. The method according to claim 1, characterized in that, Also includes: After the decision is implemented, user feedback is collected, and the collaboration strategy is optimized based on the feedback.

9. The method according to claim 1, characterized in that, The decision intention includes, but is not limited to, at least one of the following: actively initiating interaction, generating content, or performing an action.

10. A multi-entity collaborative decision-making system, characterized in that, include: The intention acquisition module is used to acquire the decision intentions of multiple entities regarding the same event; The collaboration module is used to determine the decision execution method based on the preset collaboration strategy; An execution module is used to process the event according to the decision execution method.

11. The system according to claim 10, characterized in that, The collaboration module is configured to merge the decision intentions of multiple entities to generate a comprehensive decision, and / or select one of the multiple entities as the decision execution entity.

12. The system according to claim 10, characterized in that, The collaboration module makes decisions based on at least one of the following factors, including but not limited to: entity priority, relationship parameters with the target object, current internal state, environmental state, task matching degree, and historical decision success rate.

13. The system according to claim 10, characterized in that, It also includes a negotiation module, which is used to initiate a negotiation process between entities when there is a conflict in the decision-making intentions of multiple entities.

14. The system according to claim 10, characterized in that, It also includes a recording module for storing the communication content between entities into the storage system.

15. The system according to claim 10, characterized in that, It also includes a feedback learning module, which is used to collect user feedback and optimize collaboration strategies after the decision is implemented.

16. A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method of any one of claims 1 to 9.

Images