Task processing methods, apparatus, computer equipment, storage media and program products

Abstract

This invention relates to the field of knowledge graph technology, and discloses a task processing method, apparatus, computer equipment, storage medium, and program product. The task processing method includes: receiving a task and its corresponding environmental situation information; querying a reasoning graph and a static knowledge graph based on the task to obtain behavior tree element information; wherein, the behavior tree element information includes multiple first nodes used to process the task, the relationships between the multiple first nodes, and the execution logic; constructing a behavior tree in a dynamic decision graph based on the behavior tree element information; filtering the multiple first nodes in the behavior tree based on the environmental situation information to obtain a target behavior tree containing multiple second nodes; and controlling the operation of the multiple second nodes in the target behavior tree to process the task. This invention combines a reasoning graph, a dynamic decision graph, and a static knowledge graph to complete knowledge reasoning and processing of complex tasks, expanding the reasoning scope and improving reasoning efficiency.

Description

Technical Field

[0001] This invention relates to the field of knowledge graph technology, specifically to task processing methods, apparatus, computer equipment, storage media, and program products. Background Technology

[0002] A knowledge graph (KG) is a collection of knowledge that reveals the relationships between entities. It constructs a structured knowledge network by abstracting knowledge elements such as entities, concepts, relationships, and events into nodes and representing the connections between nodes as edges. Images, text, audio, and video can all serve as knowledge sources for knowledge graphs.

[0003] Knowledge graphs in related technologies are collections of various static knowledge. They combine different static knowledge through relationships between them, and are mainly used for storing and retrieving static knowledge. Knowledge graphs in related technologies can also perform simple knowledge reasoning, but only on implicit relations based on the set of relations composed of static knowledge. This simple reasoning has a limited scope. When the reasoning scenario is more complex, the reasoning process becomes more complicated and inefficient because the set of relations based on static knowledge is relatively simple. Summary of the Invention

[0004] In view of this, the present invention provides a task processing method, apparatus, computer equipment, storage medium, and program product to solve the problems of small reasoning scope and low reasoning efficiency.

[0005] In a first aspect, the present invention provides a task processing method, comprising: receiving a task and environmental situation information corresponding to the task; wherein the environmental situation information is information used to describe the task and the specific environmental conditions describing the task; querying a matter graph and a static knowledge graph according to the task to obtain behavior tree element information; wherein the behavior tree element information includes multiple first nodes used to process the task, the relationship between the multiple first nodes, and the execution logic between the multiple first nodes, the matter graph is a graph that represents the execution logic of multiple tasks through multiple nodes, multiple relation edges, and the directionality of the multiple relation edges, and the static knowledge graph is a graph that stores multiple nodes used by multiple tasks and the relationship between the multiple nodes; constructing a behavior tree in a dynamic decision graph according to the multiple first nodes, the relationship between the multiple first nodes, and the execution logic between the multiple first nodes; wherein the dynamic decision graph is a graph that dynamically associates multiple first nodes corresponding to the task; filtering the multiple first nodes in the behavior tree according to the environmental situation information to obtain a target behavior tree containing multiple second nodes; and controlling the operation of multiple second nodes in the target behavior tree to process the task.

[0006] This invention receives a task and its corresponding environmental situation information. Based on the task, it queries a context graph and a static knowledge graph to obtain behavior tree element information, including multiple first nodes used to process the task, the relationships between these first nodes, and the execution logic between them. A behavior tree is constructed in a dynamic decision graph based on this behavior tree element information. Then, multiple first nodes in the behavior tree are filtered according to the environmental situation information to obtain a target behavior tree. This allows control over the operation of multiple second nodes in the target behavior tree to process the task. This invention combines a context graph, a static knowledge graph, and a dynamic decision graph, and introduces behavior tree technology into the dynamic decision graph. Behavior trees can flexibly describe complex behaviors and decision logic, are easy to understand and modify, and are suitable for various application scenarios. This invention forms a multi-layered decision knowledge representation framework, effectively integrating different knowledge types such as static knowledge, dynamic decision knowledge, and context knowledge, with clear hierarchical structure. This invention constructs behavior trees based on event graphs, static knowledge graphs, and dynamic decision graphs, achieving a complementary advantage between traditional knowledge graphs and behavior trees. It incorporates multiple sets of knowledge and relationships from knowledge graphs, as well as the logical relationship representation and execution, enabling interconnection and interoperability between different types of knowledge. Compared to related technologies, this invention uses behavior trees for task processing, allowing for the handling of more complex tasks, expanding the scope of task processing, and improving efficiency.

[0007] In one optional implementation, behavior tree element information is obtained by querying the event graph and the static knowledge graph based on the task, including: obtaining the first functional description information of the task; querying the event graph based on the first functional description information to obtain the execution logic between multiple first nodes; querying the static knowledge graph based on the first functional description information to obtain the multiple first nodes used by the task and the relationship between the multiple first nodes; and the behavior tree element information is composed of the multiple first nodes used by the task, the relationship between the multiple first nodes, and the execution logic between the multiple first nodes.

[0008] In one optional implementation, a behavior tree is constructed based on multiple first nodes, the relationships between the multiple first nodes, and the execution logic between the multiple first nodes, including: connecting the multiple first nodes using relationship edges according to the relationships between the multiple first nodes and the execution logic between the multiple first nodes to obtain the behavior tree.

[0009] In one optional implementation, the relation edges include ordinary edges, conditional edges, and weighted edges. Ordinary edges represent the connection relationships and directions between multiple nodes. Conditional edges are used to perform conditional judgments between multiple nodes and control the connectivity between them based on the judgment results. Weighted edges determine the directionality of multiple nodes based on their weight values. The multiple nodes include behavior tree nodes, feature nodes, and subordinate nodes. Behavior tree nodes include center nodes, control nodes, action nodes, decoration nodes, and reference nodes. Behavior tree nodes are the nodes that constitute the behavior tree. Feature nodes are connected to behavior tree nodes to represent the behavioral elements of the behavior tree nodes. Subordinate nodes are child nodes of feature nodes or other subordinate nodes, used to assist feature nodes or other subordinate nodes in parameter confirmation. The center node is the starting node of the behavior tree nodes, used to represent task completion. Control nodes control the task processing. Action nodes execute the actions corresponding to the behavior tree nodes. Decoration nodes decorate the center nodes, control nodes, action nodes, and reference nodes. Reference nodes reference other behavior trees as execution logic.

[0010] In addition to ordinary edges, this invention designs conditional edges and weighted edges in the design of relational edges. Through structural design, it realizes the dynamic expression of multiple knowledge. Through the configuration of diverse functional nodes, this invention processes tasks more accurately. This invention clarifies the functions of multiple nodes and relational edges, which helps in the development of knowledge graphs, reduces the occurrence of logical errors, and improves the reliability of task processing.

[0011] In one optional implementation, filtering multiple first nodes in the behavior tree based on environmental situation information to obtain a target behavior tree containing multiple second nodes includes: retaining multiple second nodes that match the environmental situation information among the multiple first nodes in the behavior tree to obtain the target behavior tree.

[0012] In one optional implementation, filtering multiple first nodes in the behavior tree based on environmental situation information to obtain a target behavior tree containing multiple second nodes includes: retaining multiple second nodes that match the environmental situation information among the multiple first nodes in the behavior tree to obtain the target behavior tree.

[0013] Secondly, the present invention provides a task processing apparatus, comprising: a task receiving module for receiving a task and corresponding environmental situation information; wherein the environmental situation information is information describing the task and a specific environmental condition describing the task; a query module for querying a matter graph and a static knowledge graph based on the task to obtain behavior tree element information; wherein the behavior tree element information includes multiple first nodes used to process the task, the relationships between the multiple first nodes, and the execution logic between the multiple first nodes, the matter graph being a graph representing the execution logic of multiple tasks through multiple nodes, multiple relation edges, and the directionality of the multiple relation edges, and the static knowledge graph being a graph storing multiple nodes used by multiple tasks and the relationships between the multiple nodes; a behavior tree construction module for constructing a behavior tree in a dynamic decision graph based on multiple first nodes, the relationships between the multiple first nodes, and the execution logic between the multiple first nodes; wherein the dynamic decision graph is a graph that dynamically associates multiple first nodes corresponding to a task; a target behavior tree determination module for filtering multiple first nodes in the behavior tree based on the environmental situation information to obtain a target behavior tree containing multiple second nodes; and a node running module for controlling the running of multiple second nodes in the target behavior tree to process the task.

[0014] Thirdly, the present invention provides a computer device, comprising: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the computer instructions to perform the task processing method described in the first aspect or any corresponding embodiment thereof.

[0015] Fourthly, the present invention provides a computer-readable storage medium storing computer instructions for causing a computer to perform the task processing method described in the first aspect or any corresponding embodiment thereof.

[0016] Fifthly, the present invention provides a computer program product, including computer instructions for causing a computer to execute the task processing method described in the first aspect or any corresponding embodiment thereof. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the specific embodiments or related technologies of the present invention, the drawings used in the description of the specific embodiments or related technologies will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating a task processing method according to an embodiment of the present invention;

[0019] Figure 2 This is a schematic diagram of the knowledge graph set structure according to an embodiment of the present invention;

[0020] Figure 3 This is a schematic diagram of the structure composed of nodes and relational edges according to an embodiment of the present invention;

[0021] Figure 4 This is a schematic diagram of node relationships according to an embodiment of the present invention;

[0022] Figure 5 This is a schematic diagram of the behavior tree node relationship according to an embodiment of the present invention;

[0023] Figure 6 This is a flowchart illustrating another task processing method according to an embodiment of the present invention;

[0024] Figure 7 This is a flowchart illustrating another task processing method according to an embodiment of the present invention;

[0025] Figure 8 This is a schematic diagram of the node relationships corresponding to the travel tasks in location A according to an embodiment of the present invention;

[0026] Figure 9 This is a structural block diagram of a task processing device according to an embodiment of the present invention;

[0027] Figure 10 This is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0029] A knowledge graph (KG) is a collection of knowledge that reveals the relationships between entities. It constructs a structured knowledge network by abstracting knowledge elements such as entities, concepts, relationships, and events into nodes and representing the connections between nodes as edges. Images, text, audio, and video can all serve as knowledge sources for knowledge graphs.

[0030] Behavior trees are programming models that describe the behavior of AI (Artificial Intelligence). They are tree-like structures composed of multiple nodes. The advantage of behavior trees is that they can flexibly describe complex behaviors and decision-making logic, are easy to understand and modify, and are suitable for various types of application scenarios.

[0031] Traditional knowledge graphs are collections of various static knowledge, which can combine different static knowledge and are mainly used for knowledge storage and retrieval. The knowledge reasoning process that traditional knowledge graphs can handle is to reason about implicit relations through existing relation sets. Behavior trees are mostly used to control intelligent agents to achieve autonomous decision-making in the environment, but behavior trees have difficulty in realizing the association between the key static knowledge required for intelligent decision-making.

[0032] This invention provides a task processing method that combines a reasoning graph, a dynamic decision graph, and a static knowledge graph to perform knowledge reasoning and processing of complex tasks, thereby expanding the scope of reasoning and improving reasoning efficiency.

[0033] According to an embodiment of the present invention, a task processing method embodiment is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0034] This embodiment provides a task processing method that can be used in a computer device, on which an artificial intelligence model can be configured. Figure 1 This is a flowchart of a task processing method according to an embodiment of the present invention, such as... Figure 1 As shown, the process includes the following steps:

[0035] Step S101: Receive the task and the corresponding environmental situation information; wherein, the environmental situation information is information used to describe the task and the specific environmental conditions of the task.

[0036] Here, a task is a set of operation instructions input by the user that will be processed by the knowledge graph set. Each task contains multiple events. For example, a task could be "I'm going to travel to place A by plane tomorrow and I need to plan my trip to place A". The multiple events contained in this task could be: flight selection to place A, transportation selection upon arrival, attraction selection, hotel selection, and restaurant selection, etc.

[0037] In this embodiment of the invention, the knowledge graph set is stored in the computer device of this embodiment of the invention, and the knowledge graph set includes a reasoning graph, a static knowledge graph, and a dynamic decision graph.

[0038] In some optional implementations, environmental situation information may include the coordinates of key units, task information, and time requirements. The environmental situation information is updated according to changes in the task and may include time points, seasons, travel purposes, personnel, requirements, etc. For example, if the task is "Tomorrow I will fly to location A and make a travel plan for location A," then the environmental situation information includes: tomorrow, fly, to location A, travel plan for location A. If the user changes the task to "Today I will take a bus to location B and make a travel plan for location B," then the environmental situation information is updated to: today, take a bus, to location B, travel plan for location B.

[0039] Step S102: Based on the task, query the event graph and the static knowledge graph to obtain behavior tree element information; wherein, the behavior tree element information includes multiple first nodes used by the task, the relationship between multiple first nodes, and the execution logic between multiple first nodes; the event graph is a graph that represents the execution logic of multiple tasks through multiple nodes, multiple relation edges, and the directionality of multiple relation edges; the static knowledge graph is a graph that stores multiple nodes used by multiple tasks and the relationship between multiple nodes.

[0040] In this invention, multiple nodes refer to all nodes in the event graph and static knowledge graph, and multiple first nodes refer to multiple first nodes in the event graph and static knowledge graph that correspond to the task. That is, multiple first nodes are nodes selected from multiple nodes that correspond to the task. Some nodes in the multiple nodes have relationships with each other, while some nodes do not have relationships with each other. When multiple nodes are used to process the same task, there is a relationship. When multiple nodes are used to process different tasks, there is no relationship between multiple nodes. Since multiple first nodes in this embodiment of the invention are used to complete the same task, there is a relationship between multiple first nodes. The relationship can be sequential execution, selective execution, parallel execution, etc.

[0041] In some optional implementations, the event graph includes the execution logic of multiple tasks, as well as the execution logic of multiple events within each task, to provide logical relationships between multiple first nodes for the dynamic decision graph. For example, the event graph includes the execution logic of a task to travel to location A, and the execution logic of a task to travel to location B. The task to travel to location A includes multiple events and their execution logic, which are as follows: destination confirmation event, mode of transportation selection event, travel time determination event, travel event, and arrival at destination event. Each event includes multiple nodes. For example, the mode of transportation selection event includes nodes corresponding to taking a plane, taking a bus, driving, etc.

[0042] For example, such as Figure 2 The diagram shows the structure of a knowledge graph set. The event graph is the top-level graph in the knowledge graph set. The event graph includes multiple events, namely: target discovery, target within range, task execution, approaching the target, executing the approaching target plan, detector deployment, plan formulation, and group execution. The relationship edges are represented by dashed lines with arrows.

[0043] In some optional implementations, the static knowledge graph includes a set of static knowledge used in the decision-making process. The static knowledge is related to each other by relational edges. For example, the static knowledge graph includes nodes corresponding to various modes of transportation such as private cars, subways, airplanes, buses, and public buses.

[0044] like Figure 2 As shown, the static knowledge graph is the lowest level graph in the knowledge graph set. It is used to provide nodes corresponding to static knowledge for the event graph and the dynamic decision graph. The nodes corresponding to static knowledge are related by relation edges to represent inclusion relationships. Figure 2 In this context, the nodes corresponding to static knowledge include: ships, power systems, unmanned aerial vehicles, sensors, manned aerial vehicles, payloads, and launch devices. The relationships between these nodes are represented by dashed lines with arrows.

[0045] Step S103: In the dynamic decision graph, a behavior tree is constructed based on multiple first nodes, the relationships between multiple first nodes, and the execution logic between multiple first nodes; wherein, the dynamic decision graph is a graph that dynamically associates multiple first nodes corresponding to a task.

[0046] In some optional implementations, a behavior tree is constructed based on multiple first nodes, the relationships between the multiple first nodes, and the execution logic between the multiple first nodes, including: connecting the multiple first nodes using relationship edges according to the relationships between the multiple first nodes and the execution logic between the multiple first nodes to obtain the behavior tree.

[0047] The dynamic decision graph is used to construct a behavior tree based on behavior tree element information. The behavior tree is used to make decisions by executing the control logic of each node. In the dynamic decision graph, there can be multiple behavior trees and behavior tree nodes, such as... Figure 2 As shown, the dynamic decision graph is the intermediate layer in the knowledge graph set. The behavior tree includes a central node "Task", control nodes, action nodes "Reconnaissance", action nodes "Protection", and element nodes "Sensor, Speed, Power and Task Decomposition". Among them, the element node "Task Decomposition" includes two subordinate nodes "Sensor, Speed, Power and Task Decomposition". In this embodiment of the invention, the event graph, the dynamic decision graph, and the static knowledge graph are connected through key nodes and relation edges. Key nodes are used to realize mutual calling between nodes in the layers.

[0048] In some alternative implementations, such as Figure 3 The diagram illustrates the structure of nodes and relational edges. Multiple nodes in the knowledge graph set are connected by relational edges, which include ordinary edges, conditional edges, and weighted edges. Ordinary edges represent the connection relationships and directions between multiple nodes; conditional edges are used to make conditional judgments between multiple nodes and control the connectivity between them based on the judgment results; weighted edges determine the directionality of multiple nodes based on their weight values. The nodes include behavior tree nodes, element nodes, and subordinate nodes. Behavior tree nodes include center nodes, control nodes, action nodes, decoration nodes, and reference nodes. Behavior tree nodes are the nodes that constitute the behavior tree. Element nodes are connected to behavior tree nodes to represent the behavioral elements of the behavior tree nodes. Subordinate nodes are child nodes of element nodes or other subordinate nodes, used to assist element nodes or other subordinate nodes in parameter confirmation. The center node is the starting node of the behavior tree nodes, representing the completion of a task. Control nodes control the task processing. Action nodes execute the actions corresponding to the behavior tree nodes. Decoration nodes decorate the center nodes, control nodes, action nodes, and reference nodes. Reference nodes reference other behavior trees as execution logic.

[0049] In some alternative implementations, ordinary edges only represent connection relationships and directions, and may have attributes, such as representing the progress relationship of events, the order of dynamic knowledge, or the inclusion relationship of static knowledge through direction.

[0050] In some optional implementations, conditional edges have conditional judgment attributes to control the flow of behavior. They can be used for action control or execution selection. They are precisely matched according to the key value corresponding to the conditional edge. For example, the key value corresponding to the first conditional edge is "visit attractions", and the key value corresponding to the second conditional edge is "eat". If the user's need is to eat, then the node pointed to by the second conditional edge is selected as the next node according to the key value. In actual application, when making a selection based on the conditional edge, the environmental situation information is analyzed to obtain the target key value. The target key value is matched with the key value corresponding to the conditional edge. If the match is successful, the node pointed to by the conditional edge is selected as the next node; if the match fails, no selection is made.

[0051] In some optional implementations, the weighted edges have an execution weight attribute. The optimal path is planned according to the different weights of each path. The selection of the edge is determined by comparing the values ​​corresponding to the weighted edges. Generally, the node corresponding to the weighted edge with the largest value is selected. For example, if the value of the first weighted edge is 0.1 and the value of the second weighted edge is 0.1, then the node pointed to by the second weighted edge is selected as the next node.

[0052] Among them, the condition judgment attributes of the condition edge include three types: execute according to the target, execute according to the purpose, and execute according to the requirements. The target is the object targeted by the current child node task; the purpose is the effect to be achieved in completing the current child node task; and the requirements are the conditions required to complete the current child node task.

[0053] In some optional implementations, the relation edge can also have both a conditional judgment attribute and a weight attribute. The execution logic is as follows: first, determine whether the attribute of the relation edge is unique; if unique, execute that attribute; if not unique, first determine whether the condition is met; if met, determine whether the relation edge and the node it points to are selected based on the weight value; if not selected, the two nodes connected by the relation edge are not connected; if the condition is met, determine whether the relation edge and the node it points to are selected based on the weight value, and then the two nodes connected by the relation edge are connected. Alternatively, the relation edge can also have an adaptive scenario attribute. First, determine whether the node pointed to by the relation edge meets the requirements based on the adaptive scenario attribute, and then make a selection based on the conditional judgment attribute or the weight attribute.

[0054] In some alternative implementations, such as Figure 4 The diagram shows the node relationships. The central node and the feature nodes can be connected through conditional edges and / or weighted edges. The control node and the feature nodes can be connected through conditional edges. The selection of feature nodes is based on the objectives, purposes, and requirements.

[0055] In some alternative implementations, such as Figure 4 As shown, a feature node is connected to multiple subordinate nodes through conditional edges and / or weighted edges. One of the multiple subordinate nodes can be selected based on the applicable scenario, the key value of the conditional edge, and the value value of the weighted edge. The feature node is connected to the central node or control node through relational edges. For example, for the travel mode selection event, the "transportation mode, travel time, comfort, and travel distance" that need to be considered can be used as feature nodes (Keypoints).

[0056] In some alternative implementations, a subordinate node is a subordinate of a feature node, representing multiple choices for the feature node, with one subordinate node representing one choice. For example, if the feature node is "taking a means of transportation", its subordinate nodes are the type of means of transportation, such as "taking a private car", "taking the subway", or "taking an airplane". The sub-subordinate nodes of the subordinate node can be parameters such as "speed, travel time, and power of the airplane".

[0057] When the problem considered by the feature node is too large, it can be decomposed into multiple sub-feature nodes to refine the task in the behavior tree. Feature nodes and subordinate nodes are connected through conditional edges and / or weighted edges, and the key value of the conditional edge or the value value of the weighted edge is used to control which subordinate node (or sub-feature node / sub-subordinate node) the feature node is connected to, so as to determine the selection of feature attribute value or subordinate attribute value.

[0058] In some optional implementations, control nodes include three types: parallel nodes, selection nodes, and sequential nodes. Parallel nodes require multiple events to execute simultaneously. A parallel node has at least two child nodes, and these child nodes execute in parallel. Success is returned if all child nodes complete execution; otherwise, failure is returned. Selection nodes are connected to their child nodes via conditional edges. Selection nodes can be either sequential or random, with sequential selection being more commonly used. Successful execution of a sequentially selected node returns a success value. Figure 4 As shown, a sequential node has multiple child nodes connected in sequence. The execution is carried out in the order of the rank values ​​of the relation edges connecting each child node. For example, if the rank value of the relation edge of the first child node is 1, the rank value of the relation edge of the second child node is 2, and the rank value of the relation edge of the third child node is 3, then the execution order of the sequential node is: the first child node, the second child node, and the third child node.

[0059] In some alternative implementations, such as Figure 4 As shown, action nodes are connected to feature nodes through conditional edges or weighted edges. The corresponding feature node can be selected based on the key value. Action nodes are used to generate instruction action functions for actions, so that the actions corresponding to the action nodes can be executed by calling the instruction action functions.

[0060] In some alternative implementations, decorator nodes are used to embellish nodes and enable them to perform specific actions, such as negation, success, or repetition.

[0061] In some optional implementations, a reference node is used to reference the execution logic of another behavior tree relative to its own behavior tree as the current node; that is, if the exact same execution logic already exists in other behavior trees, in order to simplify the graph and avoid duplication, the reference node can be used to directly reference the execution logic of other behavior trees in the dynamic decision graph.

[0062] In embodiments of the present invention, such as Figure 5The diagram illustrates the relationships between behavior tree nodes. A behavior tree node includes a central node (CentralNode), control nodes (controlNode), and action (behavior) nodes (actionNode). The central node is the starting node (rootNode) of the behavior tree and has one and only one child node (childNode), namely control node 1. The control node controls the decision-making process of events. Control node 1 includes four child nodes: control node 2, control node 3, control node 4, and a referencing node. The four child nodes are executed sequentially according to the rank value of the edges connecting each child node. Figure 5 The execution order is as follows: control node 2, control node 3, control node 4 and reference node, where the reference node references the central node 2; control node 2 includes action (behavior) node 1 and action (behavior) node 2, control node 3 includes action (behavior) node 3 and action (behavior) node 4, and control node 4 includes action (behavior) node 5 and action (behavior) node 6.

[0063] exist Figure 5 In the process, the central node 1 also includes four child nodes, which are four keypoints. The four keypoints are executed in the order of the rank values ​​of the relation edges connecting each child node. One of the keypoints points to three subordinate nodes, and each subordinate node points to two subordinate nodes.

[0064] Step S104: Based on the environmental situation information, filter multiple first nodes in the behavior tree to obtain a target behavior tree containing multiple second nodes.

[0065] In the dynamic decision graph, multiple first nodes, dynamically connected by relation edges, form a behavior tree where multiple first nodes can generate multiple feasible strategies. For example, for each specific event, multiple first nodes can generate multiple feasible strategies. For a travel mode selection event, these strategies include: using public transportation, walking, and driving. Therefore, when environmental situational information contains information about travel modes, a second node can be selected from multiple first nodes. For example, if the environmental situational information includes driving as a travel mode, then the first node corresponding to driving is selected as the second node.

[0066] In the dynamic decision graph of this invention, intelligent decision-making is achieved through agent-based decision-making. Multiple first nodes in the behavior tree are intelligently selected to obtain the target behavior tree. An agent is an entity capable of perceiving the environment, making decisions, and taking actions to achieve a specific goal. An agent can be a software program, a robot, or other entity possessing a certain degree of autonomy and intelligent behavior. For example, an agent can be an AI model.

[0067] It is also used to select and execute decisions from multiple feasible strategies based on current environmental situation data; the dynamic decision graph constructs a decision graph for the key event chains in the event logic graph. For example, in the aforementioned travel process, it is necessary to [ensure that the travel process...].

[0068] Step S105: Control the operation of multiple second nodes in the target behavior tree to process the task.

[0069] In some optional implementations, controlling the operation of multiple second nodes in the target behavior tree includes: obtaining a set of behavior functions; selecting a set of target behavior functions from the set of behavior functions based on environmental situation information; parsing the environmental situation information to obtain function parameters; and calling multiple target behavior functions from the set of target behavior functions based on the function parameters to control the operation of multiple second nodes in the target behavior tree.

[0070] The task processing method provided in this embodiment receives a task and its corresponding environmental situation information. Based on the task, it queries a matter-of-fact graph and a static knowledge graph to obtain behavior tree element information, including multiple first nodes used to process the task, the relationships between these first nodes, and the execution logic between them. A behavior tree is constructed in a dynamic decision graph based on this behavior tree element information. Then, multiple first nodes in the behavior tree are filtered according to the environmental situation information to obtain a target behavior tree. This allows control over the operation of multiple second nodes in the target behavior tree to process the task. This invention combines a matter-of-fact graph, a static knowledge graph, and a dynamic decision graph, and introduces behavior tree technology into the dynamic decision graph. Behavior trees can flexibly describe complex behaviors and decision logic, are easy to understand and modify, and are suitable for various application scenarios. This invention forms a multi-layered decision knowledge representation framework, effectively integrating different knowledge such as static knowledge, dynamic decision knowledge, and matter-of-fact knowledge, resulting in a clear hierarchy. This invention constructs behavior trees based on event graphs, static knowledge graphs, and dynamic decision graphs, achieving a complementary advantage between traditional knowledge graphs and behavior trees. It incorporates multiple sets of knowledge and relationships from knowledge graphs, as well as the logical relationship representation and execution, enabling interconnection and interoperability between different types of knowledge. Compared to related technologies, this invention uses behavior trees for task processing, allowing for the handling of more complex tasks, expanding the scope of task processing, and improving efficiency.

[0071] This embodiment provides a task processing method that can be used in a computer device, on which an artificial intelligence model can be configured. Figure 6 This is a flowchart of another task processing method according to an embodiment of the present invention, such as... Figure 6 As shown, the process includes the following steps:

[0072] Step S601: Receive the task and the corresponding environmental situation information; wherein, the environmental situation information is information used to describe the task and the specific environmental conditions describing the task. For details, please refer to [link to relevant documentation]. Figure 1 Step S101 of the illustrated embodiment will not be described again here.

[0073] Step S602: Based on the task, query the event graph and the static knowledge graph to obtain behavior tree element information; wherein, the behavior tree element information includes multiple first nodes used by the processing task, the relationship between multiple first nodes, and the execution logic between multiple first nodes; the event graph is a graph that represents the execution logic of multiple tasks through multiple nodes, multiple relation edges, and the directionality of multiple relation edges; the static knowledge graph is a graph that stores multiple nodes used by multiple tasks and the relationship between multiple nodes.

[0074] In some optional implementations, step S602 above includes:

[0075] Step S6021: Obtain the first functional description information of the task, and query the event graph according to the first functional description information to obtain the execution logic between multiple first nodes.

[0076] Step S6022: Query the static knowledge graph based on the first function description information to obtain the multiple first nodes used by the task and the relationships between the multiple first nodes.

[0077] Step S6023: The behavior tree element information consists of multiple first nodes used by the task, the relationship between the multiple first nodes, and the execution logic between the multiple first nodes.

[0078] Step S603: In the dynamic decision graph, a behavior tree is constructed based on multiple first nodes, the relationships between them, and the execution logic between them; wherein, the dynamic decision graph is a graph that dynamically associates multiple first nodes corresponding to a task. For details, please refer to... Figure 1 Step S103 of the illustrated embodiment will not be described again here.

[0079] Step S604: Based on the environmental situation information, filter multiple first nodes in the behavior tree to obtain a target behavior tree containing multiple second nodes.

[0080] In some optional implementations, step S604 above includes:

[0081] Step S6041: Based on the environmental situation information, retain the multiple second nodes that match the environmental situation information from the multiple first nodes in the behavior tree to obtain the target behavior tree.

[0082] In some optional implementations, step S6041 above includes:

[0083] Step a1: Obtain the second functional description information corresponding to each first node in the behavior tree, and match the environmental situation information with the second functional description information corresponding to each first node.

[0084] Step a2: Keep the first node corresponding to the successfully matched second function description information as the second node, and remove the first node corresponding to the unmatched second function description information to obtain the target behavior tree.

[0085] Step S605 involves controlling the execution of multiple second nodes in the target behavior tree to process the task. For details, please refer to [link to relevant documentation]. Figure 1 Step S105 of the illustrated embodiment will not be described again here.

[0086] The task processing method provided in this embodiment obtains the target behavior tree by matching environmental situation information with the second functional description information corresponding to each first node. This realizes the structuring of nodes corresponding to process knowledge and enables the automatic construction of behavior trees based on knowledge graph sets. It can automatically construct method functions through behavior tree subgraphs or element nodes and subordinate node subgraphs, thereby realizing decision-making behavior control and improving the flexibility of knowledge graph sets.

[0087] This embodiment provides a task processing method that can be used in a computer device, on which an artificial intelligence model can be configured. Figure 7 This is a flowchart of another task processing method according to an embodiment of the present invention, such as... Figure 7 As shown, the process includes:

[0088] The agent is loaded and initialized. The knowledge graph set is queried to obtain multiple behavior tree nodes for generating the behavior tree and multiple feature nodes connected to the behavior tree nodes. A behavior tree is generated based on the behavior tree nodes. The behavior tree includes a central node. The child nodes of the central node are control nodes, which are used to control the parallel execution of subsequent child nodes. The child nodes of the control nodes include one condition node and two control nodes. The above one condition node and two control nodes are executed through action nodes. A set of behavior functions that may be used in the execution of the behavior tree is generated based on the feature nodes. Multiple target behavior functions in the set of behavior functions are called to control the operation of the nodes in the behavior tree.

[0089] In some optional implementations, embodiments of the present invention load and initialize a decision-making agent, obtain dynamic knowledge, and dynamically connect nodes corresponding to these dynamic knowledge to form a behavior tree. The behavior tree includes multiple executable strategies. When a task needs to be executed intelligently, the agent obtains the behavior tree element information for that task by querying the graph. Based on the obtained behavior tree element information, a behavior tree is generated using the behavior tree nodes, and a set of behavior functions that may be used during the execution of the behavior tree is also generated based on the element nodes. The agent continuously receives environmental situation information about the task, selects an execution decision from multiple feasible strategies based on the environmental situation information, and selects multiple target behavior functions from the set of behavior functions. The agent parses the function parameters of the multiple target behavior functions from the environmental situation information, passes the function parameters to the multiple target behavior functions, runs the multiple target behavior functions, and achieves the agent's optimal decision.

[0090] For example, such as Figure 8 As shown, the top layer is the travel logic graph, the middle layer is the dynamic decision graph, and the bottom layer is the static knowledge graph. The static knowledge graph includes scenic spot knowledge graphs, food knowledge graphs, etc. The scenic spot knowledge graph in the static knowledge graph provides static knowledge for the element nodes corresponding to "scenic spots" in the dynamic decision graph, and the food knowledge graph in the static knowledge graph provides static knowledge for the element nodes corresponding to "food" in the dynamic decision graph. The travel logic graph provides the execution logic for "traveling to location A", "traveling", and "scenic spots" in the dynamic decision graph.

[0091] exist Figure 8In the dynamic decision graph, a travel behavior tree is generated based on the content provided by the travel logic graph and the static knowledge graph. The central node in the travel behavior tree is "Visit Location A". The child nodes of the central node are control nodes. Specifically, the control nodes are selector nodes, used to indicate the selection of one node to execute from multiple subsequent child nodes. The multiple subsequent child nodes are five control nodes, corresponding to "First Attraction", "Second Attraction", "Third Attraction", "Fourth Attraction", and "Fifth Attraction". The control node corresponding to "Third Attraction" is selected. The control node corresponding to "Third Attraction" includes three control nodes, which are executed sequentially. The content of the three control nodes is "Morning Plan", "Afternoon Plan", and "Evening Plan". Among them, the control node corresponding to "Morning Plan" includes three control nodes, which are executed sequentially. The content of the three control nodes is "Transportation", "Breakfast", and "Visit". Among them, the control node corresponding to "Transportation" includes four action nodes, which are "Subway", "Bus", "Cycling", and "Walking". The "Afternoon Plan" includes two control nodes. The system has two control nodes, executed sequentially, with the content corresponding to "lunch" and "playing" respectively. The "playing" control node includes three control nodes, each corresponding to a specific lake, a tower, and a "play method." The "tower" control node includes two action nodes, each corresponding to "ticket" and "guided tour." The "play method" control node includes three action nodes, each corresponding to "boat," "cycling," and "walking." The "evening plan" control node also includes three control nodes, executed sequentially, each corresponding to "dinner," "shopping," and "accommodation." The "shopping" control node includes four action nodes, each corresponding to a specific shopping mall, street A, street B, and night market. In this embodiment, multiple action nodes are selected based on environmental situation information; multiple action nodes corresponding to a single control node can be selected. If the environmental situation information does not mention the content of an action node, the optimal content is selected and executed according to a preset algorithm.

[0092] exist Figure 8In the dynamic decision-making graph, the central node "Travel to Location A" includes six element nodes, corresponding to "Transportation," "Accommodation," "Food," "Travel Methods," "Travel Guides," and "Attractions." Specifically, the "Transportation" element node includes four subordinate nodes, corresponding to "Airplane," "High-Speed ​​Rail," "Self-Driving," and "Bus." The "Food" element node includes four subordinate nodes, corresponding to "Fish in Vinegar Sauce," "Local Cuisine," "Dongpo Pork," and "Beggar's Chicken." The "Accommodation" element node includes four subordinate nodes, corresponding to... The content includes "Youth Hostel", "B&B", "Budget Hotel" and "Star-Rated Hotel"; among them, "B&B" has three subordinate nodes, and the content of the three subordinate nodes is "First B&B", "Second B&B" and "Third B&B"; "Budget Hotel" has five subordinate nodes, and the content of the five subordinate nodes is "First Hotel", "Second Hotel", "Third Hotel", "Fourth Hotel" and "Other"; "Star-Rated Hotel" has five subordinate nodes, and the content of the five subordinate nodes is "Fifth Hotel", "Sixth Hotel", "Seventh Hotel", "Eighth Hotel" and "Other".

[0093] In this embodiment of the invention, selection can be made from multiple subordinate nodes based on environmental situation information. Among the element nodes corresponding to "accommodation," the selection can be based on the applicable scenario, key value, and value, according to the environmental situation information. Specifically, the applicable scenario for the relationship edge between the element node corresponding to "accommodation" and the subordinate node corresponding to "youth hostel" is "budget travel," with a key value of 4 and a value of 0.2 for the conditional judgment attribute. Similarly, the applicable scenario for the relationship edge between the element node corresponding to "accommodation" and the subordinate node corresponding to "homestay" is "budget travel," with a key value of 3 and a value of 0.25 for the conditional judgment attribute. Finally, the applicable scenario for the relationship edge between the element node corresponding to "accommodation" and the subordinate node corresponding to "budget hotel" is "limited budget," with a key value of 3 and a value of 0.25 for the conditional judgment attribute. The attribute key value is 2, the weight attribute value is 0.25, and the applicable scenario for the relationship edge between the element node corresponding to "accommodation" and the subordinate node corresponding to "star-rated hotel" is "sufficient budget". The conditional judgment attribute key value is 1, and the weight attribute value is 0.3. For example, when budget travel information exists in the environmental situation information, the subordinate node "youth hostel" corresponding to the applicable scenario "budget travel" is selected. When no application scenario exists in the environmental situation information, the corresponding subordinate node is selected based on the key value analyzed from the environmental situation information. When the key value analyzed from the environmental situation information is 3, the subordinate node "youth hostel" corresponding to "budget travel" is selected. When the key value cannot be analyzed from the environmental situation information, the subordinate node corresponding to the relationship edge with the largest value is selected. Figure 8 The dependent node "Star-rated Hotel" corresponds to the relation edge with a key value of 0.3.

[0094] This embodiment also provides a task processing device for implementing the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0095] This embodiment provides a task processing device, such as... Figure 9 As shown, it includes:

[0096] The task receiving module 901 is used to receive the task and the corresponding environmental situation information; wherein, the environmental situation information is information used to describe the task and the specific environmental conditions of the task.

[0097] The query module 902 is used to query the event graph and the static knowledge graph based on the task to obtain behavior tree element information. The behavior tree element information includes multiple first nodes used by the task, the relationships between the multiple first nodes, and the execution logic between the multiple first nodes. The event graph is a graph that represents the execution logic of multiple tasks through multiple nodes, multiple relation edges, and the directionality of multiple relation edges. The static knowledge graph is a graph that stores multiple nodes used by multiple tasks and the relationships between multiple nodes.

[0098] The behavior tree construction module 903 is used to construct a behavior tree in a dynamic decision graph based on multiple first nodes, the relationships between multiple first nodes, and the execution logic between multiple first nodes; wherein, the dynamic decision graph is a graph that dynamically associates multiple first nodes corresponding to a task.

[0099] The target behavior tree determination module 904 is used to filter multiple first nodes in the behavior tree based on environmental situation information to obtain a target behavior tree containing multiple second nodes.

[0100] The node execution module 905 is used to control the execution of multiple second nodes in the target behavior tree to process tasks.

[0101] In some alternative implementations, the query module 902 includes:

[0102] The event graph query unit is used to obtain the first functional description information of the task, and to query the event graph based on the first functional description information to obtain the execution logic between multiple first nodes.

[0103] The static knowledge graph query unit is used to query the static knowledge graph based on the first function description information to obtain multiple first nodes used by the task and the relationships between the multiple first nodes.

[0104] The behavior tree element information generation unit is used to generate behavior tree element information, which consists of multiple first nodes used by the task, the relationships between the multiple first nodes, and the execution logic between the multiple first nodes.

[0105] In some alternative implementations, the behavior tree construction module 903 includes:

[0106] The behavior tree construction unit is used to connect multiple first nodes using relation edges according to the relationships between them and the execution logic between them, thus obtaining a behavior tree.

[0107] In some alternative implementations, the target behavior tree determination module 904 includes:

[0108] The target behavior tree determination unit is used to retain multiple second nodes that match the environmental situation information from multiple first nodes in the behavior tree, thereby obtaining the target behavior tree.

[0109] In some optional implementations, the target behavior tree determination unit includes:

[0110] The information matching subunit is used to obtain the second functional description information corresponding to each first node in the behavior tree, and match the environmental situation information with the second functional description information corresponding to each first node.

[0111] The target behavior tree determination sub-unit is used to retain the first node corresponding to the successfully matched second function description information as the second node, and to remove the first node corresponding to the unmatched second function description information, thus obtaining the target behavior tree.

[0112] Further functional descriptions of the above modules and units are the same as those in the corresponding embodiments described above, and will not be repeated here.

[0113] In this embodiment, the task processing device is presented in the form of a functional unit. Here, a unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that execute one or more software or fixed programs, and / or other devices that can provide the above functions.

[0114] This invention also provides a computer device having the above-described features. Figure 9 The task processing device shown.

[0115] Please see Figure 10 , Figure 10 This is a schematic diagram of the structure of a computer device provided in an optional embodiment of the present invention, such as... Figure 10 As shown, the computer device includes one or more processors 1010, memory 1020, and interfaces for connecting the components, including high-speed interfaces and low-speed interfaces. The components communicate with each other via different buses and can be mounted on a common motherboard or otherwise installed as needed. The processors can process instructions executed within the computer device, including instructions stored in or on memory to display graphical information of a GUI on external input / output devices (such as display devices coupled to the interfaces). In some alternative implementations, multiple processors and / or multiple buses can be used with multiple memories and multiple memory modules, if desired. Similarly, multiple computer devices can be connected, each providing some of the necessary operations (e.g., as a server array, a group of blade servers, or a multiprocessor system). Figure 10Take the 1010 processor as an example.

[0116] The processor 1010 may be a central processing unit, a network processor, or a combination thereof. The processor 1010 may further include a hardware chip. This hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The programmable logic device may be a complex programmable logic device (CAMP), a field-programmable gate array (FPGA), a general-purpose array logic (GPRS), or any combination thereof.

[0117] The memory 1020 stores instructions executable by at least one processor 1010 to cause the at least one processor 1010 to perform the method shown in the above embodiments.

[0118] The memory 1020 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the computer device. Furthermore, the memory 1020 may include high-speed random access memory and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some alternative embodiments, the memory 1020 may optionally include memory remotely located relative to the processor 1010, and these remote memories can be connected to the computer device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0119] The memory 1020 may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as flash memory, hard disk or solid-state drive; the memory 1020 may also include a combination of the above types of memory.

[0120] The computer device also includes a communication interface 1030 for communicating with other devices or communication networks.

[0121] This invention also provides a computer-readable storage medium. The methods described above according to embodiments of the invention can be implemented in hardware or firmware, or implemented as computer code that can be recorded on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and then stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code, which, when accessed and executed by the computer, processor, or hardware, implements the methods shown in the above embodiments.

[0122] A portion of this invention can be applied as a computer program product, such as computer program instructions, which, when executed by a computer, can invoke or provide the methods and / or technical solutions according to the invention through the operation of the computer. Those skilled in the art will understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, installation package files, etc. Correspondingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executing the instructions, or the computer compiling the instructions and then executing the corresponding compiled program, or the computer reading and executing the instructions, or the computer reading and installing the instructions and then executing the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to a computer.

[0123] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A task processing method, characterized in that, The method includes: Receive a task and the corresponding environmental situation information; wherein, the environmental situation information is information used to describe the task and the specific environmental conditions of the task; Based on the task, queries are performed in the event graph and the static knowledge graph to obtain behavior tree element information; wherein, the behavior tree element information includes multiple first nodes used to process the task, the relationships between the multiple first nodes, and the execution logic between the multiple first nodes; the event graph is a graph that represents the execution logic of multiple tasks through multiple nodes, multiple relation edges, and the directionality of the multiple relation edges; the static knowledge graph is a graph that stores the multiple nodes used by multiple tasks and the relationships between the multiple nodes. In the dynamic decision graph, a behavior tree is constructed based on the plurality of first nodes, the relationships between the plurality of first nodes, and the execution logic between the plurality of first nodes; wherein, the dynamic decision graph is a graph that dynamically associates the plurality of first nodes corresponding to the task; Based on the environmental situation information, multiple first nodes in the behavior tree are filtered to obtain a target behavior tree containing multiple second nodes; Control the operation of the plurality of second nodes in the target behavior tree to process the task.

2. The method according to claim 1, characterized in that, The step of querying the event graph and static knowledge graph based on the task to obtain behavior tree element information includes: Obtain the first functional description information of the task, and query the event graph according to the first functional description information to obtain the execution logic between the plurality of first nodes; Based on the first functional description information, a query is performed in the static knowledge graph to obtain multiple first nodes used by the task and the relationship between the multiple first nodes; The behavior tree element information consists of multiple first nodes used by the task, the relationships between the multiple first nodes, and the execution logic between the multiple first nodes.

3. The method according to claim 1 or 2, characterized in that, The step of constructing a behavior tree based on the plurality of first nodes, the relationships between the plurality of first nodes, and the execution logic between the plurality of first nodes includes: Based on the relationships between the multiple first nodes and the execution logic between them, the multiple first nodes are connected using relational edges to obtain the behavior tree.

4. The method according to claim 3, characterized in that, The relational edges include ordinary edges, conditional edges, and weighted edges; wherein, the ordinary edges are used to represent the connection relationship and connection direction between the multiple nodes; the conditional edges are used to make conditional judgments between the multiple nodes and to control the connectivity between the multiple nodes according to the judgment results; the weighted edges are used to determine the directionality of the multiple nodes according to their weight values. The plurality of nodes includes behavior tree nodes, element nodes, and subordinate nodes. The behavior tree nodes include a center node, a control node, an action node, a decoration node, and a reference node. The behavior tree nodes are the nodes that constitute the behavior tree. The element nodes are connected to the behavior tree nodes and represent the behavioral elements of the behavior tree nodes. The subordinate nodes are child nodes of the element nodes or other subordinate nodes, used to assist the element nodes or other subordinate nodes in parameter confirmation. The center node is the starting node of the behavior tree nodes and is used to indicate the completion of the task. The control node is used to control the processing of the task. The action node is used to execute the action corresponding to the behavior tree node. The decoration node is used to decorate the center node, the control node, the action node, and the reference node. The reference node is used to reference other behavior trees as execution logic.

5. The method according to claim 1 or 2, characterized in that, The step of filtering multiple first nodes in the behavior tree based on the environmental situation information to obtain a target behavior tree containing multiple second nodes includes: Based on the environmental situation information, the multiple second nodes that match the environmental situation information among the multiple first nodes in the behavior tree are retained to obtain the target behavior tree.

6. The method according to claim 5, characterized in that, The step of retaining the plurality of second nodes that match the environmental situation information from the plurality of first nodes in the behavior tree to obtain the target behavior tree includes: Obtain the second functional description information corresponding to each first node in the behavior tree, and match the environmental situation information with the second functional description information corresponding to each first node; The first node corresponding to the second function description information that is successfully matched is retained as the second node, and the first node corresponding to the second function description information that fails to match is removed to obtain the target behavior tree.

7. A task processing device, characterized in that, The device includes: The task receiving module is used to receive a task and the corresponding environmental situation information; wherein, the environmental situation information is information used to describe the task and the specific environmental conditions of the task. The query module is used to query the event graph and the static knowledge graph according to the task to obtain behavior tree element information; wherein, the behavior tree element information includes multiple first nodes used to process the task, the relationships between the multiple first nodes, and the execution logic between the multiple first nodes; the event graph is a graph that represents the execution logic of multiple tasks through multiple nodes, multiple relationship edges, and the directionality of the multiple relationship edges; and the static knowledge graph is a graph that stores the multiple nodes used by multiple tasks and the relationships between the multiple nodes. The behavior tree construction module is used to construct a behavior tree in a dynamic decision graph based on the plurality of first nodes, the relationships between the plurality of first nodes, and the execution logic between the plurality of first nodes; wherein, the dynamic decision graph is a graph that dynamically associates the plurality of first nodes corresponding to the task; The target behavior tree determination module is used to filter multiple first nodes in the behavior tree based on the environmental situation information to obtain a target behavior tree containing multiple second nodes. The node execution module is used to control the execution of the plurality of second nodes in the target behavior tree to process the task.

8. A computer device, characterized in that, include: A memory and a processor are communicatively connected, the memory stores computer instructions, and the processor executes the computer instructions to perform the task processing method of any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing the computer to perform the task processing method according to any one of claims 1 to 6.

10. A computer program product, characterized in that, Includes computer instructions for causing a computer to perform the task processing method according to any one of claims 1 to 6.

Images