Task execution method and device of intelligent agent, equipment, medium and product

Abstract

The embodiment of the invention provides a task execution method of an agent, which can be applied to the technical field of artificial intelligence. The task execution method of the intelligent agent comprises the following steps: generating dictionary programming data corresponding to user task information through a preset logic form analysis rule; analyzing the dictionary programming data to analyze an operator library according to a preset logic form to generate a logic form operator graph; and generating an AND-OR matrix sequence according to the logic form operator graph, wherein the AND-OR matrix sequence is used for realizing a task execution process corresponding to the user task information. The embodiment of the invention further provides an agent task execution device and equipment, a storage medium and a program product.

Description

Technical Field

[0001] The present invention relates to the technical field of artificial intelligence, specifically to the technical field of image processing, and more specifically to a method, device, equipment, medium and product for an intelligent agent to execute tasks. Background Art

[0002] Artificial Intelligence (AI for short) is an important driving force for the new round of scientific and technological revolution and industrial transformation. It is a new key technical science that studies, develops theories, methods, technologies and application systems for simulating, extending and expanding human intelligence. As an important part of intelligent science, artificial intelligence attempts to understand the essence of intelligence and produce a new intelligent machine (i.e., an intelligent agent) that can respond in a way similar to human intelligence.

[0003] In order for an intelligent agent to execute corresponding action tasks, various complex task plans need to be carried out specifically. Among them, currently, when an intelligent agent based on a generative model (such as the Large Language Model, i.e., the LLM language model) conducts task planning, it is necessary to map the unstructured text it outputs to instructions (such as robotic arm movements, API calls, and HTTP access, etc.). At this time, it is usually required that the generative model can output structured parameters to facilitate correct calls and calculations by the processor.

[0004] On the one hand, although existing intelligent agents provide their own action spaces, enabling them to execute specific atomic actions and advanced skills, these actions and skills also require structured parameters for invocation. However, the language understanding module provided by current intelligent agents cannot correctly generate structured invocation parameters. On the other hand, relying on the task planning and decomposition process of the above-mentioned generative model usually has good common sense and generalization. As the length of the output characters increases, both the training cost and the inference cost are continuously increasing, which poses relatively high requirements for engineering.

[0005] Therefore, existing intelligent agents cannot perform high-efficiency action planning in large quantities and at low cost based on existing generative models, and existing technologies also cannot provide detailed information feedback on the task planning and execution process, resulting in users being unable to accurately understand the task execution process of intelligent agents, causing poor interaction experiences. At the same time, intelligent agents cannot correct their own execution states according to task execution. Summary of the Invention

[0006] In view of at least one of the above-mentioned technical problems existing in the prior art, embodiments of the present invention aim to be able to achieve an effective connection between flexible natural language expression and precise task execution, so as to achieve efficient task operator invocation, and to implement a task execution method, device, equipment, medium and product of an intelligent agent for efficient task planning and execution at low cost. Thus, based on low-cost special programming rules friendly to language processing, natural language processing does not need to generate very long structured parameters, thereby ensuring the accuracy rate and inference speed of task planning and execution; in addition, through grammar checking and parsing execution, the ability to provide call parameters mapped to the action skills of the intelligent agent can not only deconstruct the dependence of the intelligent agent on natural language parsing, but also provide additional interpretable information feedback to the generation model, so as to be able to correct its own execution state before executing the task, thereby greatly improving the user interaction experience and reaching a higher level of intelligence.

[0007] One aspect of an embodiment of the present invention provides a task execution method for an intelligent agent, which includes: generating dictionary programming data corresponding to user task information through a preset logical form parsing rule; parsing the dictionary programming data to generate a logical form operator graph according to a preset logical form parsing operator library; and generating an AND-OR matrix sequence according to the logical form operator graph, and the AND-OR matrix sequence is used to implement the task execution process corresponding to the user task information.

[0008] According to an embodiment of the present invention, before generating dictionary programming data corresponding to user task information through a preset logical form parsing rule, it further includes: generating user task information according to the interaction application scenario between the intelligent agent and the user; generating class extensible markup data corresponding to the user task information through a preset natural language processing model.

[0009] According to an embodiment of the present invention, in generating dictionary programming data corresponding to user task information through a preset logical form parsing rule, it includes: parsing the class extensible markup data through a preset logical form parsing rule to generate dictionary programming data.

[0010] According to an embodiment of the present invention, in parsing the dictionary programming data to generate a logical form operator graph according to a preset logical form parsing operator library, it includes: sequentially calling at least one execution operator required for task execution through a preset logical form parsing operator library according to the parsing information of the dictionary programming data; generating a logical form operator graph according to the simulated execution results of each execution operator in the at least one execution operator.

[0011] According to an embodiment of the present invention, the task execution method of the intelligent agent further includes: obtaining task atomic skills and / or task atomic actions required for task execution according to the AND-OR matrix sequence and a preset skill action mapping relationship; performing task execution of the intelligent agent according to the task atomic skills and / or task atomic actions.

[0012] According to an embodiment of the present invention, in obtaining the task atomic skills and / or task atomic actions required for task execution according to the AND / OR matrix sequence and the preset skill action mapping relationship, it further includes: calling a preset natural language generation model to convert the logical form parsing exception description information corresponding to the AND / OR matrix sequence and the user task information, and generating the interpretable information required for task execution.

[0013] Another aspect of the embodiments of the present invention provides a task execution device for an intelligent agent, which includes a logical parsing module, an operator graph generation module, and a sequence generation module. The logical parsing module is used to generate dictionary programming data corresponding to the user task information through a preset logical form parsing rule; the operator graph generation module is used to parse the dictionary programming data to generate a logical form operator graph according to a preset logical form parsing operator library; and the sequence generation module is used to generate an AND / OR matrix sequence according to the logical form operator graph, and the AND / OR matrix sequence is used to implement the task execution process corresponding to the user task information.

[0014] Another aspect of the embodiments of the present invention provides an electronic device, including one or more processors and a memory, and the memory is used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors are caused to execute the above-mentioned task execution method of the intelligent agent.

[0015] Another aspect of the embodiments of the present invention provides a computer-readable storage medium, on which executable instructions are stored, and when the instructions are executed by a processor, the processor is caused to execute the above-mentioned task execution method of the intelligent agent.

[0016] Another aspect of the embodiments of the present invention provides a computer program product, including a computer program, and when the computer program is executed by a processor, the above-mentioned task execution method of the intelligent agent is implemented.

[0017] The task execution method of the intelligent agent provided by the embodiments of the present invention can at least partially solve the problem of relatively low intelligence level in the task execution process of the intelligent agent in the related art, and thus can at least achieve one of the following technical effects: Based on low-cost (simple and streamlined) language processing-friendly dedicated programming rules, natural language processing does not need to generate very long structured parameters, thereby ensuring the accuracy rate and reasoning speed of task planning and execution; in addition, through grammar checking and parsing execution, the ability to provide call parameters mapped to the action skills of the intelligent agent can not only deconstruct the dependence of the intelligent agent on natural language parsing, but also provide additional interpretable information feedback to the generation model at the same time, so as to be able to correct its own execution status before executing the task, thereby greatly improving the user interaction experience and reaching a higher intelligence level.

[0018] Therefore, the task execution method of the above-mentioned agent in the embodiments of the present invention can achieve an effective connection between flexible natural language expression and precise task execution, thereby enabling efficient task operator invocation to achieve efficient task planning and execution at low cost. At the same time, it can provide corresponding interpretability information as the basis for correcting the state of the agent itself, realizing a good interaction experience with users.

[0019] It should be understood that the above general description and the following specific embodiments are only exemplary and explanatory, and they do not limit the scope of what the present invention intends to claim. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Through the following description of the embodiments of the present invention with reference to the drawings, the above content and other objects, features, and advantages of the present invention will become clearer. In the drawings: Figure 1 Schematically shows an application scenario diagram of a task execution method, device, equipment, medium, and program product of an agent according to an embodiment of the present invention; Figure 2 Schematically shows a flowchart of a task execution method of an agent according to an embodiment of the present invention; Figure 3A Schematically shows a graphical expression of a logical form operator diagram of a task execution method of an agent according to an embodiment of the present invention; Figure 3B Schematically shows a data architecture diagram of an AND-OR matrix sequence of a task execution method of an agent according to an embodiment of the present invention; Figure 4 Schematically shows another application scenario flowchart of a task execution method of an agent according to an embodiment of the present invention; Figure 5 Schematically shows a structural block diagram of a task execution device of an agent according to an embodiment of the present invention; and Figure 6 Schematically shows a block diagram of an electronic device suitable for implementing the task execution method of an agent according to an embodiment of the present invention.

[0021] The above-mentioned drawings are part of the specification of the embodiments of the present invention, which illustrate the exemplary embodiments of the present invention. The attached drawings and the description of the specification are used together to explain the principles of the embodiments of the present invention. It should be understood that the above general description of the drawings and the following specific embodiments are only exemplary and explanatory, and they do not limit the scope of what the present invention intends to claim. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0022] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clearly understood, the following will clearly explain the spirit of the content disclosed by the present invention with reference to the accompanying drawings and detailed descriptions. After any person skilled in the relevant technical field understands the embodiments of the content of the present invention, they can make changes and modifications based on the technologies taught by the content of the present invention, which do not depart from the spirit and scope of the content of the present invention.

[0023] The exemplary embodiments of the present invention and their descriptions are used to explain the present invention, but not to limit the present invention. Additionally, elements / components with the same or similar reference numerals used in the accompanying drawings and embodiments are used to represent the same or similar parts.

[0024] Regarding the use of "first", "second",... etc. in the present invention, they do not particularly refer to the order or sequence, nor are they used to limit the present invention. They are only used to distinguish elements or operations described with the same technical terms.

[0025] Regarding the directional terms used in the present invention, such as: up, down, left, right, front, or back, etc., they are only references to the directions in the accompanying drawings. Therefore, the directional terms used are for explanation and not for limiting this creation.

[0026] Regarding the use of "comprising", "including", "having", "containing", etc. in the present invention, they are all open-ended terms, meaning including but not limited to.

[0027] Regarding the use of "and / or" in the present invention, it includes any one or all combinations of the described things.

[0028] Regarding "multiple" in the present invention, it includes "two" and "more than two"; regarding "multiple groups" in the present invention, it includes "two groups" and "more than two groups".

[0029] Regarding the terms "substantially", "about", etc. used in the present invention, they are used to modify any quantity or error that can vary slightly, but these slight variations or errors will not change their essence. Generally, the range of such slight variations or errors modified by such terms can be 20% in some embodiments, 10% in some embodiments, 5% in some embodiments, or other values. Those skilled in the art should understand that the aforementioned values can be adjusted according to actual needs and are not limited thereto.

[0030] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used here should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0031] In the case of using expressions such as "at least one of A, B, and C, etc.", generally, it should be interpreted according to the meaning that those skilled in the art usually understand this expression (for example, "a system having at least one of A, B, and C" should include, but not be limited to, a system having only A, only B, only C, having A and B, having A and C, having B and C, and / or having A, B, and C, etc.). In the case of using expressions such as "at least one of A, B, or C, etc.", generally, it should be interpreted according to the meaning that those skilled in the art usually understand this expression (for example, "a system having at least one of A, B, or C" should include, but not be limited to, a system having only A, only B, only C, having A and B, having A and C, having B and C, and / or having A, B, and C, etc.). Those skilled in the art should also understand that substantially any disjunctive conjunction and / or phrase representing two or more alternative items, whether in the specification, claims, or drawings, should be understood to give the possibility of including one of these items, either side of these items, or both items. For example, the phrase "A or B" should be understood to include the possibility of "A" or "B", or "A and B".

[0032] In view of at least one of the above-mentioned technical problems existing in the prior art, embodiments of the present invention aim to be able to achieve an effective connection between flexible natural language expression and precise task execution, so as to achieve efficient task operator invocation, and to provide a task execution method, device, equipment, medium, and product for an intelligent agent that can achieve efficient task planning and execution at low cost. Thus, based on low-cost programming rules friendly to language processing, natural language processing does not need to generate long structured parameters, thereby ensuring the accuracy rate and inference speed of task planning and execution; in addition, through grammar checking and parsing execution, the ability to provide call parameters mapped to the action skills of the intelligent agent can not only deconstruct the dependence of the intelligent agent on natural language parsing, but also provide additional interpretable information feedback to the generation model, so as to be able to correct its own execution state before executing the task, thereby greatly improving the user interaction experience and reaching a higher level of intelligence.

[0033] One aspect of the embodiments of the present invention provides a task execution method for an intelligent agent, which includes: generating dictionary programming data corresponding to user task information through a preset logical form parsing rule; parsing the dictionary programming data to generate a logical form operator graph according to a preset logical form parsing operator library; and generating an AND / OR matrix sequence according to the logical form operator graph, and the AND / OR matrix sequence is used to implement the task execution process corresponding to the user task information.

[0034] Figure 1 Schematically shows an application scenario diagram of the task execution method, device, equipment, medium, and program product of an intelligent agent according to an embodiment of the present invention.

[0035] As Figure 1 shown, the application scenario 100 according to this embodiment may include terminal devices 101, 102, 103, a network 104, and a server 105. The network 104 is used to provide a medium for communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links, or fiber optic cables, etc.

[0036] Users can use the terminal devices 101, 102, 103 to interact with the server 105 through the network 104 to receive or send messages, etc. Various communication client applications may be installed on the terminal devices 101, 102, 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc. (only as examples).

[0037] The terminal devices 101, 102, 103 may be various electronic devices with a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop portable computers, and desktop computers, etc.

[0038] The server 105 may be a server providing various services, such as a background management server that supports the websites browsed by users using the terminal devices 101, 102, 103 (only as an example). The background management server may analyze and process data such as user requests received, and feedback the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal devices.

[0039] It should be noted that the task execution method of the agent provided by the embodiments of the present invention can generally be executed by the server 105. Correspondingly, the task execution device of the agent provided by the embodiments of the present invention can generally be set in the server 105. The task execution method of the agent provided by the embodiments of the present invention can also be executed by a server or a server cluster different from the server 105 and capable of communicating with the terminal devices 101, 102, 103 and / or the server 105. Correspondingly, the task execution device of the agent provided by the embodiments of the present invention can also be set in a server or a server cluster different from the server 105 and capable of communicating with the terminal devices 101, 102, 103 and / or the server 105.

[0040] It should be understood that Figure 1 the numbers of the terminal devices, the network, and the server in

[0041] are merely illustrative. According to the implementation requirements, there may be any number of terminal devices, networks, and servers. Figure 1 the scenario described below, throughFigures 2 - 4 A detailed description is given of the task execution method of the agent in the disclosed embodiments.

[0042] As Figure 2 shown, one aspect of the embodiments of the present invention provides a task execution method for an agent, which includes operations S201 to S203.

[0043] In operation S201, dictionary programming data corresponding to user task information is generated through a preset logical form parsing rule; In operation S202, the dictionary programming data is parsed to generate a logical form operator graph according to a preset logical form parsing operator library; and In operation S203, an AND-OR matrix sequence is generated according to the logical form operator graph, and the AND-OR matrix sequence is used to implement the task execution process corresponding to the user task information.

[0044] The agent can be the execution subject of the above task execution method of the embodiments of the present invention, or an execution party controlled by the task execution method. Specifically, it can be a humanoid intelligent robot or other AI devices, and usually has an actuator itself to complete specific action tasks. For example, a humanoid robot can use a mechanical manipulator to complete the action task of picking up an object, and even perform action tasks such as product packaging on an industrial production line. Among them, these action tasks can be used as the target execution tasks of the agent. To complete these target execution tasks, the agent needs to complete a series of execution actions in sequence.

[0045] The preset logical form parsing rule can be understood as a preset rule of a dedicated programming language based on the logical form (Logic-Form), which is used to implement the mapping relationship between natural language and the task sequence executable by the agent. Specifically, the preset logical form parsing rule can be based on a preset logical form parsing (Logic-Form Parser, abbreviated as LFP) module to implement the parsing and execution of data conforming to the logical form programming language. By customizing operators and syntax, a small computational graph is constructed to realize the mapping from natural language to action sequence.

[0046] Among them, LFP can be developed based on the Python language. By providing an abstract layer syntax parsing and parser for the generation model of natural language processing (Natural Language Processing, abbreviated as NLP), the decoupling of action planning at the natural language level and action execution at the agent execution level is realized, so as to achieve the flexibility and independence on the NLP side, and by further providing a lexical checker and a syntax checker, the robustness of the agent's execution instruction actions is ensured.

[0047] User task information can be the task information generated by the agent in the interaction application scenario with the user. This task information can usually be presented in the form of natural language, and specifically can include the target task that the agent believes the user needs the agent to execute and its related task execution data, status data, etc. For example, when the user points to the apple on the left table and asks the agent: "How many apples are there on the table?", the corresponding user task information of the agent can be only brief task requirement information such as "who the user is", "user's pointing", "user's question 'How many apples are there on the table?'", and of course it can also include the agent's own state execution information (such as the action it is currently performing). Ultimately, it can be used to implement a series of tasks such as instructing the agent to observe the table and the objects on the table along the user's pointing, distinguishing the apples on the table and confirming the quantity, and converting the confirmation result into natural language and playing it to the user through the speaker, and finally providing an interactive response to the user. Therefore, user task information can serve as the basis for the agent to respond to the user's needs. Among them, user task information can be embodied in natural language instructions or statements that can be recognized and processed by the NLP side.

[0048] It should be particularly noted that the agent can detect its own current environment and its own state information in real time, and at the same time confirm the execution progress of each task. Specifically, it can judge the space environment and its own state in each iteration process. The current execution state can be at least one of the state information of the agent's own execution action and its own current environment state detected at the current moment. Among them, the state information of the agent's own execution action can include the name of the currently executed action and the execution information corresponding to the action name (such as the torque, rotation angle of the wrist, expected execution time, and the spatial positions of each key point of the fingers, etc.), and the state of the agent's own current environment can include the information of the people and objects around the space where the agent is currently located, such as the size, position (distance), and even shape and color of the obstacles on the travel route, or it can also be the dynamic change information of the people in the surrounding environment, such as the moving speed of the people and the emotional information on the face, etc.

[0049] Based on the preset logical form parsing rules, the user task information is preliminarily processed to achieve data conversion, that is, the corresponding dictionary programming data can be generated. The dictionary programming data can be data in the form of a dictionary (Dictionary, abbreviated as Dic) structure based on the Python language, usually composed of key (Key) and value (Value) pairs, and the format content is similar to dic = {key1 :value1, key2 : value2}. Therefore, with the help of dictionary programming data, the connection requirement between natural language expression and precise action execution can be effectively achieved.

[0050] Furthermore, the preset logical form parsing operator library can be a database of custom execution operators designed based on preset logical form parsing rules, which includes numerous action execution operators and skill execution operators that can be executed or called by the agent execution module. Each execution operator includes basic execution information such as execution definition, category, execution description, and the supported LFP version. For example, for the execution operator of "counting quantity", its execution definition is "count(*individuals)", the category is "operator", the execution description is "counting quantity", and the supported LFP version is 0.6.14. Among them, each execution operator can be used to define corresponding execution actions or skills. When called by the execution module or executed by itself, it can provide corresponding execution basis for the actuator during the task execution stage. Since the preset logical form parsing operator library is a dedicated operator library in the embodiments of the present invention, it can be flexibly added, deleted, and modified through a custom form, enabling it to meet the new addition of agent actions downward. At the same time, it can also provide protocol layers, encapsulation details, and sample data to meet the training and generation of language models, etc.

[0051] During the process of parsing dictionary programming data, the matching execution operators in the preset logical form parsing operator library can be continuously called, and the execution results of the execution operators can be predicted through the simulated execution of the execution operators in the simulated interaction environment. Then, the execution order of the execution operators is sorted according to the execution results of these execution operators to form the required logical form operator graph. Among them, the parsing of dictionary programming data can specifically refer to the processing process of Python dictionary format data, which will not be elaborated here.

[0052] The logical form operator graph can specifically be an operator calculation graph formed based on multiple execution operators, rather than truly displayable graphical content. It is mainly used to define the execution order or execution relationship of each execution operator retrieved according to the dictionary programming data parsing process. Since the logical form operator graph has a more direct calculation graph expression relative to the agent execution module, it realizes the simple conversion of complex natural language expressed user task information based on logical form LFP programming, laying a good foundation for the subsequent efficient and correct task execution process of the agent. Therefore, it can effectively eliminate the situation of using structured parameters for invocation in traditional technologies.

[0053] The AND-OR matrix sequence (OAMatrix) can be a unified structured data for data transfer between execution operators. Specifically, it can be an execution sequence including AND relationship data, OR relationship data, and corresponding atomic action or skill information, which can be specifically referred to later Figure 3BThe content shown. The AND-OR matrix sequence can be generated according to the execution operators and their corresponding execution relationships in the logical form operator graph, and can then be used as the input data for the agent execution module (such as Pibaby) to realize the action skill parsing of task execution, so that it can be directly called for execution by the execution module.

[0054] Among them, the AND-OR matrix sequence can usually be used as the transfer data between execution operators. If it is required that the calculation processes between each execution operator maintain a random state, then the output of the previous execution operator and the input of the next execution operator both need to maintain the data format of the OAmatrix of the AND-OR matrix sequence. Therefore, based on this flexible data structure of the AND-OR matrix sequence, different possibilities in natural language expressions can be expressed as AND-OR graphs, and the best skill action sequence can be freely selected in the subsequent planning and execution stage.

[0055] Therefore, with the help of this AND-OR matrix sequence, the task execution of the agent can support the processing of the ambiguity of natural language. For example, "go to the table and get a banana", but there are 3 bananas and any one of them is okay, this is an or relationship. However, for "get a banana + wipe the table", if there are 3 bananas and 2 tables, then getting a banana + wiping the table is an and relationship, and there are 6 possibilities. With the help of this AND-OR matrix sequence, it can effectively support the execution ambiguity generated by such natural language instructions, helping the agent to make the most appropriate execution response to real humans during the task execution process.

[0056] In summary, it can be seen that based on the preset logical form parsing rules, a bridge can be built between unstructured natural language and institutionalized parameter calls, so as to give the execution sequence data of the AND-OR matrix architecture (AND-OR matrix sequence). Therefore, based on a low-cost programming language (LFP) friendly to language processing, natural language processing does not need to generate very long structured parameters, thus ensuring the accuracy rate and reasoning speed of task planning and execution; in addition, through syntax checking and parsing execution, the ability to provide call parameters mapped to the agent's action skills can not only deconstruct the agent's dependence on natural language parsing, but also provide additional interpretable information feedback to the generation model so that it can correct its own execution state before executing the task, thus greatly improving the user interaction experience and reaching a higher level of intelligence.

[0057] It can be seen that the task execution method of the above-mentioned agent in the embodiment of the present invention can effectively connect the flexible expression of natural language and the precise task execution, so as to achieve an efficient task execution operator call, realize low-cost and efficient task planning and execution, and achieve a good interaction experience with users.

[0058] To enable those skilled in the art to have a clearer understanding of the above-mentioned task execution method of the intelligent agent in the embodiments of the present invention, the following is further provided Figures 3A - 4 description.

[0059] As Figures 2 - 4 shown, according to an embodiment of the present invention, before generating the dictionary programming data corresponding to the user task information through the preset logical form parsing rule in operation 201, it further includes: Generating user task information according to the interaction application scenario between the intelligent agent and the user; Generating the class extensible markup data corresponding to the user task information through the preset natural language processing model.

[0060] The interaction application scenario may be a scenario where there is an interaction between the intelligent agent and the user, such as a scenario where the user and the intelligent agent are having a conversation, a scenario where the intelligent agent shakes hands with the user, winks at the user, etc., or a scenario where corresponding actions and expression reactions are provided according to the user's color and expression. Specifically, it may represent the application scenario generated during the interaction process between the two.

[0061] As Figure 4 shown, the user task information 401 is the task requirement information generated by the intelligent agent according to this interaction application scenario, and is used to define the target task for the task execution of the intelligent agent. For example, in a conversation scenario, the user points at the apple on the table to the intelligent agent and asks "How many apples are there on the table?", at this time, the intelligent agent can generate the user natural language data "How many apples are there on the table?" according to this interaction application scenario as the required instructions or statements for subsequent defining multiple target tasks such as "turn the face, look at the table, distinguish apples from other objects, confirm the number of apples", which constitute the components of the user task information. This user task information can be input into the preset natural language processing model in the form of a string (abbreviated as str).

[0062] As Figure 4As shown in the figure, the preset natural language processing model 402 is a preset generation model that processes the user task information 401 in the form of a string based on natural language processing technology (NLP) to generate corresponding class extensible markup data. This language model can be pre-trained with data to obtain the corresponding information processing capabilities. Among them, the preset natural language processing model 402 can understand and analyze the instructions or statements of the user task information 401, and then output the above-mentioned class extensible markup data. Among them, the class extensible markup data is usually markup language data in the format of class XML (Extensible Markup Language). This type of data can achieve format consistency during data transmission, and can also be effectively stored in a structured manner, facilitating subsequent parsing and call processing. In addition, it can provide readability, facilitate understanding and editing, and has high flexibility and scalability.

[0063] For example, for the user task information corresponding to the above-mentioned question "How many apples are there on the table?", the final output class extensible markup data of the preset natural language processing model 402 can be: 'describe(<arg>question = "the number of apples on the table", arg> individuals = count( <arg>individuals< / arg> = grounding( <arg>text< / arg> = "the apples on the table")))'.

[0064] Furthermore, as Figure 4 shown, the definition of the logic form parsing intermediate layer (Logic Form) 403 can be implemented by based on the preset logic form parsing rules, and the parsing of natural language instructions can be achieved with the help of LFP. Among them, the logic form parsing intermediate layer 403 can be based on the unified and independent U abstraction layer protocol, and is completely defined based on Python class functions and parameter interfaces. It provides data conversion in formats such as json data format and xml format on the basis of pydantic, and can also provide lexical and syntactic analysis.

[0065] As Figures 2 - 4 shown, according to an embodiment of the present invention, in operation 201, when generating dictionary programming data corresponding to user task information through preset logic form parsing rules, it includes: Parsing the class extensible markup data through the preset logic form parsing rules to generate dictionary programming data.

[0066] As Figure 4As shown, according to the logic parsing module 431 (lf_dict) based on the preset logic form parsing rules, the parsing of class extensible markup data can be realized, and the corresponding dictionary programming data is generated by parsing this data. For example, for the final output class extensible markup data of the above-mentioned preset natural language processing model 402, the dictionary programming data in its python dictionary format can be: { 'u_name':'describe' , 'u_args':{ 'question':'The number of apples on the table', 'individuals': { 'u_name': 'count', 'u_args': { 'individuals':{ 'u_name': "grounding', 'u_args': {'text':'The apples on the table'} }}}}} Specifically, the logic parsing module 431 can provide a logic form conversion tool (such as lf_converter) to parse the class extensible markup data into dictionary programming data in the form of a dictionary or json.

[0067] Therefore, based on the unified and independent U abstraction layer protocol, the data conversion for json and yml can be provided entirely based on the function and parameter interfaces of Python classes, so as to effectively connect the requirements between flexible natural language expressions and precise action executions according to the custom logic form parsing intermediate layer (Logic Form) 403. Moreover, as described later, through the logic form parsing intermediate layer 403, a lexical analyzer, a syntax analyzer, and an operator system can be specifically implemented for the logic form programming language, better meeting flexibility and scalability.

[0068] As Figures 2 - 4 shown, according to an embodiment of the present invention, in operation 202 of parsing the dictionary programming data to generate a logic form operator graph according to the preset logic form parsing operator library, it includes: According to the parsing information of the dictionary programming data, at least one execution operator required for task execution is sequentially called through the preset logic form parsing operator library; Generate a logic form operator graph according to the simulated execution results of each execution operator in the at least one execution operator.

[0069] As Figure 4As shown, the operator graph generation module 432 can receive the dictionary programming data output by the logic parsing module 431, perform parsing on the dictionary programming data, and extract the information related to operator calls therein as the parsing information of the dictionary programming data, where the information related to operator calls may include the names of specific execution actions or execution skills.

[0070] According to the parsing information of the dictionary programming data, corresponding execution operators are called in sequence in the preset logical form parsing operator library. Among them, the preset logical form parsing operator library can be an extensible operator library based on a unified interface and data structure, and specifically can include numerous execution operators. Each execution operator can be divided into multiple categories such as motion and operator according to the supported version of LFP. For example, the execution operator for looking at an object can be defined as look at and belongs to the motion category, while the execution operator for calculating the quantity can be defined as count and belongs to the operator category. Correspondingly, in the embodiments of the present invention, some execution operators of the preset logical form parsing operator library and their related information are shown in Table 1 below: Operator Definition Category Description LFP Minimum Supported Version count(*individuals) operator Calculate Quantity 0.6.14 calc_volume(*individuals) operator Calculate the Volume and Size of an Object 0.6.14 intersect(*individuals) operator Calculate the Intersection of Multiple Object Lists 0.6.14 look_at(individuals) motion Look at an Object 0.6.14 stand_up() motion Stand Up 0.6.14 point_at(individuals) motion Point at an Object 0.6.14 wave_hand() motion Wave Hand 0.6.14 shake_head() motion Shake Head 0.6.14 nod() motion Nod 0.6.14 turn_on(individuals) motion Turn on Electrical Appliances 0.6.14 look_in_mirror() motion Look in the Mirror 0.6.14 turn(direction = ""back"") motion Turn Around (Left / Right / Back) 0.6.14 seek(goal) motion Seek 0.6.14 track_grab(individuals) motion Snatch 0.6.15 Table 1 As Figure 4 shown, the operator graph generation module 432 can call at least three execution operators of describe, count, and grounding in the preset logical form parsing operator library according to the parsing information of the dictionary programming data, such as the skill parsing content 'u_name': 'describe', 'u_name': 'count', and 'u_name': "grounding".

[0071] For each called execution operator, the operator graph generation module 432 can perform virtual execution according to the virtual execution environment (simulate) of the intelligent agent to obtain the corresponding simulated execution result. Therefore, the simulated execution result can be the virtual execution result of each corresponding execution operator in the virtual execution environment. Among them, the virtual execution environment is usually a simulation environment of the interaction application scenario between the intelligent agent and the user. For example, the living room environment where the intelligent agent and the user are located is scanned and modeled to form a virtual complete living room environment. By executing the corresponding execution operator in this complete living room environment, according to the execution of the execution operator in this virtual environment, the corresponding simulated execution result is generated. Among them, the simulated execution result can correspond to the execution order of each execution operator, so that an execution relationship can be established between the multiple called execution operators to form a computational graph, that is, a logical form operator graph.

[0072] As Figure 3AAs shown, the logical form operator graph can include a first execution operator 301, a second execution operator 302, and a third execution operator 303. Among them, when the simulation execution result of the first execution operator 301 can serve as the condition required for the execution of the second execution operator 302, the execution of the first execution operator 301 is defined to be before the execution of the second execution operator 302. For example, in response to the aforementioned user's question "How many apples are on the table", the agent can call the grounding execution operator to perform an observation task, specifically by using the agent's visual sensor (such as a camera) to perform environmental detection (such as looking at the table according to the user's pointing and parsing all the images generated during the "looking" process to obtain all the observed content); further, call the count execution operator to perform a counting task, specifically to identify and distinguish objects from the observed content generated during the grounding process, distinguish the table and the apples, and count the number of apples; finally, call the describe execution operator to perform a description task, specifically to perform a natural language description of the observed content during the grounding process and the objects and the number of objects distinguished during the count process. At this time, correspondingly, the first execution operator 301 can be grounding, the second execution operator 302 can be count, and the third execution operator 303 can be describe, and the three can form the above-mentioned logical form operator graph.

[0073] Specifically, the operator graph generation module 432 can parse the above dictionary programming data and call the execution operator through a parsing tool such as lf_parser, and finally generate the required logical form operator graph (Logic Form Graph). Therefore, the logical form operator graph can be an AND-OR computational graph of skill action sequences.

[0074] It can be seen that through the above operator graph generation module 432, corresponding command line and GUI tools can be provided to parse and process the dictionary programming data based on the logical form programming language, generate the corresponding computational graph, so as to convert the natural language instruction information into a computational graph that is convenient for execution. In addition, by customizing a set of dedicated operator libraries as the preset logical form parsing operator libraries, flexible operator adjustment operations such as addition and deletion can be performed, which can meet the new actions of the agent downward, and at the same time provide a protocol layer to encapsulate details and provide sample data to meet the training and generation of language models (such as LLM language models). Finally, by providing the implementation of the computational graph based on Logic Form and the operator library, complex instructions can be expressed as AND-OR computational graphs, which can significantly improve the efficiency and accuracy of task planning and execution. At the same time, it can also be used to provide semantic interpretable information when an error occurs during the parsing and execution of the computational graph.

[0075] Such as Figure 4As shown, the sequence generation module 433 can further generate an AND-OR matrix sequence according to the above logical form operator graph. The AND-OR matrix sequence can be data for task execution that can be implemented by the execution module (such as Pibaby) of the agent. For example, Figure 3B As shown, the AND-OR matrix sequence can include AND-relation data and OR-relation data, where the AND-relation data can include corresponding atomic actions. Through the AND-OR matrix array, a unified data structure for passing between operators can be provided, avoiding the need for the preset natural language processing model 402 to focus on parameter adaptation details.

[0076] For example, Figures 2 - 4 As shown, according to an embodiment of the present invention, the method for the agent to execute tasks further includes: Obtaining the task atomic skills and / or task atomic actions required for task execution according to the AND-OR matrix sequence and the preset skill-action mapping relationship; Performing task execution of the agent according to the task atomic skills and / or task atomic actions.

[0077] For example, Figure 4 As shown, the execution sequence interface 404 can be used as the data interface of the execution module of the agent, and processes the AND-OR matrix sequence output by the received sequence generation module 433 through the preset skill-action mapping relationship. The AND-OR matrix sequence is an executable sequence for the above execution sequence interface 404 (such as ObeyU).

[0078] The preset skill-action mapping relationship can include the mapping relationship between atomic skills and the AND-OR matrix sequence, and the mapping relationship between atomic actions and the AND-OR matrix sequence. The so-called atomic skill can be understood as the smallest skill unit U that cannot be further divided, and the so-called atomic action can be understood as the smallest action unit Action that cannot be further divided. Through the above preset skill-action mapping relationship, the corresponding atomic actions and atomic skills can be extracted according to the AND-relation data, OR-relation data, etc. of the AND-OR matrix sequence, and the corresponding task atomic skills and / or task atomic actions, that is, U and / or Action, can be generated.

[0079] The task atomic skill can be the smallest unit skill required for task execution in the embodiment of the present invention. Correspondingly, the task atomic action can be the smallest action required for task execution in the embodiment of the present invention. The task atomic skills and / or task atomic actions can be directly called by the execution module of the agent to achieve sequential execution of the task.

[0080] Therefore, by implementing the mapping between the U abstraction layer (UAbs) and uvlib in the execution module (such as Pibaby), such as one-to-one or one-to-many, an adaptation scheme for coordinating natural language and the specific execution skill actions of the agent based on the preset logical form parsing rules is provided, and detailed error reporting and execution description information can also be provided.

[0081] As Figures 2 - 4 shown, according to an embodiment of the present invention, in obtaining the task atomic skills and / or task atomic actions required for task execution according to the AND-OR matrix sequence and the preset skill action mapping relationship, it further includes: Invoking a preset natural language generation model to convert the logical form parsing exception description information corresponding to the AND-OR matrix sequence and the user task information, and generating interpretable information required for task execution.

[0082] As Figure 4 shown, the preset natural language generation model 405 can be a preset model based on natural language generation technology (abbreviated as NLG). It can be trained with language data in advance and has the ability to generate natural language text, and is used to process the input data in natural language and generate natural language text information that conforms to human expression habits.

[0083] The logical form parsing exception description information (exception) can be the exception information generated during the generation process of the AND-OR matrix sequence of the above-mentioned logical form parsing intermediate layer 403. Specifically, it can be information related to the execution failure information and its reasons generated when the corresponding execution process fails to achieve the expected goal.

[0084] As Figure 4 shown, in the process of the logical parsing module 431 parsing the class extension marked data to generate dictionary programming data, if the corresponding dictionary programming data cannot be generated, the corresponding logical parsing exception information E1 is generated as the exception information for the dictionary programming data generation process, that is, token exception; correspondingly, in the operator graph generation module 432, if the corresponding execution operator cannot be effectively called, the corresponding operator graph exception information E2 is generated as the exception information for the process that cannot achieve the logical form operator graph, that is, parse exception; similarly, in the sequence generation module 433, if a valid AND-OR matrix sequence cannot be generated, the corresponding sequence generation exception information E3 is generated as the exception information for the failure of the AND-OR matrix sequence to be achieved. These logical parsing exception information E1, operator graph exception information E2, and sequence generation exception information E3 are respectively used to describe the adverse results and corresponding adverse reasons generated during their respective execution processes, and finally can be uniformly described to generate the corresponding logical form parsing exception description information E4.

[0085] Further, by invoking the preset natural language generation model 405, natural language polishing on the above logical form parsing exception description information E4 can be performed from the NLG side, so that the finally generated natural language text can be input as a meaningful message to the execution sequence interface 404, thereby completing the conversion of the logical form parsing exception description information corresponding to the user task information. Further, during the process of the execution sequence interface 404 receiving the corresponding AND-OR matrix sequence and generating task atomic skills and / or task atomic actions, the preset natural language generation model 405 can also be further invoked to perform natural language description on the corresponding information generated during this process, generating the corresponding natural language text, thereby completing the conversion of the corresponding AND-OR matrix sequence.

[0086] The natural language text converted by invoking the preset natural language generation model 405 above, as interpretability information, can retain the detailed natural language description text, describe in detail the problems and other contents generated during the task execution method process of the above embodiments of the present invention, thereby enabling the self-awareness of the intelligent agent, being able to explain its own problems, illustrate its own execution situation, explain its own execution process, provide detailed explanations and specific natural language responses that conform to human expression habits, so that the intelligent agent can reflect a higher level of intelligence.

[0087] Among them, during the operator graph generation process of the above operator graph generation module 432, a natural language description of the calculation result of each operator is also provided, and a calculation graph description in the form of a running account is formed, saving all calculation details and results, and providing them to the downstream preset natural language generation model for natural language text generation.

[0088] The interpretability information can be a natural language text that details the execution process of the above task execution method of the intelligent agent after being processed by natural language generation (NLG), including execution exception information (exception), etc. Therefore, the task execution method of the embodiments of the present invention can provide an interpretability function during the operation of the operator for the language model, enabling the intelligent agent to have a higher level of intelligence and being more in line with the reactions of real humans.

[0089] In summary, the task execution method of the intelligent agent provided by the embodiments of the present invention can achieve at least one of the following technical effects: (1) Realize better encapsulation of the action details of the intelligent agent, facilitating the language model to generate action plans.

[0090] (2) The preset logical form parsing operator library based on LFP is an operator library that is easy to expand, can support the intelligent agent to easily add various actions, the operator adjustment is more flexible, and can adapt to various task executions.

[0091] (3) Provide the ability to align natural language with objects in the scene, and be able to map the concepts in the instruction to the objects in the scene.

[0092] (4) Be able to check the correctness of the logic-form format and grammar of the language model output, and be used as a pre- and post-processing tool for the language model.

[0093] (5) Be able to provide semantic natural language information to support the interpretability of the agent.

[0094] Based on the above task execution method of the agent, the present invention also provides a task execution device 500 for the agent. The following will be combined with Figure 5 to describe the device 500 in detail.

[0095] Figure 5 The structural block diagram of the task execution device 500 for the agent according to an embodiment of the present invention is schematically shown.

[0096] As Figure 5 shown, the task execution device 500 for the agent in this embodiment includes a logic parsing module 510, an operator graph generation module 520, and a sequence generation module 530.

[0097] The logic parsing module 510 is used to generate dictionary programming data corresponding to the user task information through preset logic form parsing rules. In one embodiment, the logic parsing module 510 can be used to perform the operation S201 described above, which will not be elaborated here.

[0098] The operator graph generation module 520 is used to parse the dictionary programming data to generate a logic form operator graph according to the preset logic form parsing operator library. In one embodiment, the operator graph generation module 520 can be used to perform the operation S202 described above, which will not be elaborated here.

[0099] The sequence generation module 530 is used to generate an AND-OR matrix sequence according to the logic form operator graph, and the AND-OR matrix sequence is used to implement the task execution process corresponding to the user task information. In one embodiment, the sequence generation module 530 can be used to perform the operation S203 described above, which will not be elaborated here.

[0100] According to an embodiment of the present invention, any multiple of the logical parsing module 510, the operator graph generation module 520, and the sequence generation module 530 may be combined and implemented in one module, or any one of them may be split into multiple modules. Alternatively, at least part of the functions of one or more of these modules may be combined with at least part of the functions of other modules and implemented in one module. According to an embodiment of the present invention, at least one of the logical parsing module 510, the operator graph generation module 520, and the sequence generation module 530 may be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on chip, a system on substrate, a system on package, an application specific integrated circuit (ASIC), or any other reasonable manner of integrating or packaging circuits, etc., implemented by hardware or firmware, or implemented in any one of the three implementation manners of software, hardware, and firmware, or in any appropriate combination of several of them. Alternatively, at least one of the logical parsing module 510, the operator graph generation module 520, and the sequence generation module 530 may be at least partially implemented as a computer program module, which can execute corresponding functions when the computer program module is run.

[0101] Figure 6 FIG. schematically shows a block diagram of an electronic device suitable for implementing the task execution method of an agent according to an embodiment of the present invention.

[0102] The above-mentioned electronic device provided by an embodiment of the present invention includes one or more processors and a memory, and the memory is used to store one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors are caused to execute the above-mentioned task execution method of the agent.

[0103] As Figure 6 shown, the electronic device 600 according to an embodiment of the present invention includes a processor 601, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 602 or the program loaded from the storage part 608 into the random access memory (RAM) 603. The processor 601 may include, for example, a general microprocessor (such as a CPU), an instruction set processor and / or a related chipset and / or a dedicated microprocessor (such as an application specific integrated circuit (ASIC)), etc. The processor 601 may also include on-board memory for caching purposes. The processor 601 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present invention.

[0104] In the RAM 603, various programs and data required for the operation of the electronic device 600 are stored. The processor 601, the ROM 602, and the RAM 603 are connected to each other via the bus 604. The processor 601 performs various operations of the method flow according to the embodiments of the present invention by executing the programs in the ROM 602 and / or the RAM 603. It should be noted that the programs may also be stored in one or more memories other than the ROM 602 and the RAM 603. The processor 601 may also perform various operations of the method flow according to the embodiments of the present invention by executing the programs stored in the one or more memories.

[0105] According to an embodiment of the present invention, the electronic device 600 may further include an input / output (I / O) interface 605, and the input / output (I / O) interface 605 is also connected to the bus 604. The electronic device 600 may further include one or more of the following components connected to the I / O interface 605: an input part 606 including a keyboard, a mouse, etc.; an output part 607 including, for example, a cathode ray tube (CRT), a liquid crystal display (LCD), etc. and a speaker, etc.; a storage part 608 including a hard disk, etc.; and a communication part 609 including a network interface card such as a LAN card, a modem, etc. The communication part 609 performs communication processing via a network such as the Internet. The drive 610 is also connected to the I / O interface 605 as needed. A removable medium 611, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 610 as needed so that a computer program read from it can be installed into the storage part 608 as needed.

[0106] The present invention also provides a computer-readable storage medium on which executable instructions are stored, and when the instructions are executed by a processor, the processor is caused to execute the task execution method of the above-mentioned agent.

[0107] Among them, the computer-readable storage medium may be included in the device / apparatus / system described in the above embodiments; or it may exist separately without being assembled into the device / apparatus / system. The above computer-readable storage medium carries one or more programs, and when the one or more programs are executed, the method according to the embodiments of the present invention is implemented.

[0108] According to an embodiment of the present invention, the computer-readable storage medium may be a non-volatile computer-readable storage medium, which may include, for example, but is not limited to: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above. In the present invention, the computer-readable storage medium may be any tangible medium that contains or stores a program, and this program can be used by or in conjunction with an instruction execution system, apparatus, or device. For example, according to an embodiment of the present invention, the computer-readable storage medium may include one or more memories other than the above-described ROM 602 and / or RAM 603 and / or ROM 602 and RAM 603.

[0109] An embodiment of the present invention further includes a computer program product, which includes a computer program that, when executed by a processor, implements the above-described task execution method of the intelligent agent.

[0110] Wherein, the computer program contains program code for executing the method shown in the flowchart. When the computer program product runs in a computer system, the program code is used to enable the computer system to implement the method provided by the embodiment of the present invention.

[0111] When the computer program is executed by the processor 601, it executes the above-described functions defined in the system / apparatus of the embodiment of the present invention. According to an embodiment of the present invention, the above-described systems, apparatuses, modules, units, etc. can be implemented by computer program modules.

[0112] In one embodiment, the computer program may rely on tangible storage media such as optical storage devices and magnetic storage devices. In another embodiment, the computer program may also be transmitted and distributed in the form of a signal on a network medium, and be downloaded and installed through the communication part 609, and / or be installed from the removable medium 611. The program code included in the computer program can be transmitted by any suitable network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0113] In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 609, and / or be installed from the removable medium 611. When the computer program is executed by the processor 601, it executes the above-described functions defined in the system of the embodiment of the present invention. According to an embodiment of the present invention, the above-described systems, devices, apparatuses, modules, units, etc. can be implemented by computer program modules.

[0114] In accordance with embodiments of the present invention, program code for executing the computer programs provided by the embodiments of the present invention may be written in any combination of one or more programming languages. Specifically, these computing programs may be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. The programming languages include, but are not limited to, programming languages such as Java, C++, Python, the "C" language, or similar programming languages. The program code may be executed entirely on the user's computing device, partially on the user's device, partially on a remote computing device, or entirely on a remote computing device or server. In cases involving a remote computing device, the remote computing device may be connected to the user's computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or it may be connected to an external computing device (e.g., by connecting through the Internet using an Internet service provider).

[0115] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagram may represent a module, a program segment, or a part of code that contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than that marked in the accompanying drawings. For example, two consecutive blocks shown may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram or flowchart, and combinations of blocks in the block diagram or flowchart, may be implemented by a dedicated hardware-based system for performing the specified functions or operations, or may be implemented by a combination of dedicated hardware and computer instructions.

[0116] In addition, all actions of obtaining information, signals, or data in the present invention are carried out on the premise of complying with the corresponding data protection laws, regulations, and policies of the country where it is located, and obtaining authorization from the owner of the corresponding device.

[0117] Those skilled in the art can understand that the features described in the various embodiments and / or claims of the present invention can be combined or combined in various ways, even if such combinations or combinations are not explicitly described in the present invention. In particular, without departing from the spirit and teachings of the present invention, the features described in the various embodiments and / or claims of the present invention can be combined and combined in various ways. All such combinations and / or combinations fall within the scope of the present invention.

[0118] The embodiments of the present invention have been described above. However, these embodiments are for illustrative purposes only and are not intended to limit the scope of the present invention. Although the embodiments have been described separately above, this does not mean that the measures in each embodiment cannot be used advantageously in combination. The scope of the present invention is defined by the appended claims and their equivalents. Without departing from the scope of the present invention, those skilled in the art can make various substitutions and modifications, and all such substitutions and modifications should fall within the scope of the present invention.

Claims

1. A method for task execution of an agent, characterized in that, Including: Generating dictionary programming data corresponding to user task information through a preset logical form parsing rule; Parsing the dictionary programming data to generate a logical form operator graph according to a preset logical form parsing operator library; And Generating an AND / OR matrix sequence according to the logical form operator graph, where the AND / OR matrix sequence is used to implement the task execution process corresponding to the user task information.

2. The method according to claim 1, characterized in that Before generating the dictionary programming data corresponding to the user task information through the preset logical form parsing rule, it further includes: Generating user task information according to the interaction application scenario between the intelligent agent and the user; Generating class extensible markup data corresponding to the user task information through a preset natural language processing model.

3. The method according to claim 2, characterized in that, In generating the dictionary programming data corresponding to the user task information through the preset logical form parsing rule, it includes: Parsing the class extensible markup data through a preset logical form parsing rule to generate the dictionary programming data.

4. The method according to claim 1, wherein In parsing the dictionary programming data to generate a logical form operator graph according to a preset logical form parsing operator library, it includes: According to the parsing information of the dictionary programming data, sequentially calling at least one execution operator required for the task execution through a preset logical form parsing operator library; Generating the logical form operator graph according to the simulation execution results of each execution operator in the at least one execution operator.

5. The method according to claim 1, characterized in that, It further includes: Obtaining the task atomic skills and / or task atomic actions required for the task execution according to the AND / OR matrix sequence and a preset skill-action mapping relationship; Performing the task execution of the intelligent agent according to the task atomic skills and / or task atomic actions.

6. The method according to claim 5, characterized in that, In obtaining the task atomic skills and / or task atomic actions required for the task execution according to the AND / OR matrix sequence and a preset skill-action mapping relationship, it further includes: Invoking a preset natural language generation model to convert the AND / OR matrix sequence and the logical form parsing exception description information corresponding to the user task information to generate the interpretability information required for the task execution.

7. A task execution device for an intelligent agent, characterized in that, Including: A logical parsing module for generating dictionary programming data corresponding to user task information through a preset logical form parsing rule; An operator graph generation module for parsing the dictionary programming data to generate a logical form operator graph according to a preset logical form parsing operator library; And A sequence generation module for generating an AND / OR matrix sequence according to the logical form operator graph, where the AND / OR matrix sequence is used to implement the task execution process corresponding to the user task information.

8. An electronic device, including: One or more processors; A memory for storing one or more programs, wherein, when the one or more programs are executed by the one or more processors, the one or more processors execute the method according to any one of claims 1 to 6.

9. A computer-readable storage medium, on which executable instructions are stored, and when the instructions are executed by a processor, the processor executes the method according to any one of claims 1 to 6.

10. A computer program product, including a computer program, and when the computer program is executed by a processor, it implements the method according to any one of claims 1 to 6.

Images