Capability identification method and system of robot body, terminal and medium

Abstract

The invention discloses a robot body capability identification method and system, a terminal and a medium, and the method comprises the steps: generating a robot analysis instruction when a target robot system is accessed, obtaining robot body information according to the robot analysis instruction, and carrying out the information processing of the robot body information, and obtaining target machine information; obtaining demand problem information of the user, performing intention analysis on the demand problem information to obtain a target intention, and performing retrieval processing on the target machine information according to the target intention to obtain a machine information retrieval result; and generating corresponding demand reply information according to the machine information retrieval result, and sending the demand reply information to the user. According to the method, the large language model is deployed on the edge device, and the challenge of resource limitation is overcome; a complex reasoning process is realized under limited computing resources through an innovative multi-round question and answer mechanism; in addition, deep analysis of robot functions is achieved, and strategy making is more accurate and efficient.

Description

Technical Field

[0001] The present invention relates to the technical field of robot functions, and in particular, to a method, system, terminal, and computer-readable storage medium for identifying the capabilities of a robot body. Background Art

[0002] Nowadays, the application scenarios and tasks of robots are becoming more and more extensive and complex. For example, industrial manufacturing robots and domestic service robots, etc. In these scenarios, quickly understanding the functions of robots and managing robots are the basis for using robots safely and effectively, and are also the key to the intelligent upgrade of robots. In addition, large models, as key technologies for general intelligence, are important cornerstones for connecting the virtual and physical worlds. How to introduce high-intelligence large model technologies into low-power devices on the edge side is a necessary way for the current intelligent upgrade of robots.

[0003] Currently, deploying large language models on edge devices under resource-constrained conditions is a challenge, usually achieved through model compression technologies, which include knowledge distillation, multi-layer pruning, quantization, distributed processing, caching mechanisms, and task-specific fine-tuning, etc.; by combining these technologies, the model can be effectively compressed within a specific knowledge domain. However, the deployment limitations of large models on edge devices, although existing model compression technologies can reduce the model size, often lead to a significant decline in model performance and cannot maintain the original understanding ability and reasoning depth; for complex or non-standard named ROS systems (Robot Operating System, an operating system based on robots), it is impossible to accurately understand the functional structure of robots, and when implementing multi-round conversations, it cannot fundamentally solve the problem of insufficient understanding and reasoning ability of the model in complex conversations, and lacks sufficient generality and adaptability to handle various different types and complexities of ROS systems.

[0004] Therefore, the existing technology still needs to be improved and developed. Summary of the Invention

[0005] The main purpose of the present invention is to provide a method for identifying the capabilities of a robot body, aiming to solve the problem that although the existing model compression technology can reduce the model size, it cannot maintain the original understanding ability and reasoning depth, and for complex or non-standard named ROS systems, it is impossible to accurately understand the functional structure of robots, and lacks sufficient generality and adaptability to handle various different types and complexities of ROS systems.

[0006] To achieve the above purpose, the present invention provides a method, system, terminal, and medium for identifying the capabilities of a robot body. The method for identifying the capabilities of the robot body includes the following steps:

[0007] When accessing the target robot system, generate a robot analysis instruction, obtain the robot body information of the target robot system according to the robot analysis instruction, and perform information processing on the robot body information to obtain target machine information;

[0008] Obtain the demand problem information of the user, perform intent analysis on the demand problem information to obtain a target intent, and perform a retrieval process on the target machine information according to the target intent to obtain a machine information retrieval result;

[0009] Generate a corresponding demand response message according to the machine information retrieval result and send the demand response message to the user.

[0010] Optionally, for the method for identifying the capabilities of the robot body, before the step of "when accessing the target robot system, generate a robot analysis instruction, obtain the robot body information of the target robot system according to the robot analysis instruction, and perform information processing on the robot body information to obtain target machine information", it further includes:

[0011] Obtain the feedforward layer weight values and activation values of the large language model, perform range statistics on the feedforward layer weight values and the activation values to obtain a floating-point value distribution range, and perform integer quantization on the feedforward layer weight values and the activation values according to the floating-point value distribution range to obtain a target language model;

[0012] Perform range statistics on the feedforward layer weight values and the activation values to obtain a floating-point value distribution range, and perform integer quantization on the feedforward layer weight values and the activation values according to the floating-point value distribution range to obtain a target language model, and deploy the target language model.

[0013] Optionally, for the method for identifying the capabilities of the robot body, the step of "when accessing the target robot system, generate a robot analysis instruction, obtain the robot body information of the target robot system according to the robot analysis instruction, and perform information processing on the robot body information to obtain target machine information" specifically includes:

[0014] When accessing the target robot system, generate a robot analysis instruction, send the robot analysis instruction to the target robot system, and receive the robot body information sent by the target robot system, where the robot body information includes a robot module, a robot topic, and robot parameters;

[0015] Perform structured processing on the robot body information to obtain a structured prompt, and input the structured prompt into the target language model to obtain corresponding target machine information;

[0016] The target machine information is obtained by the target language model performing text understanding and information inference on the structured prompt words, and the target machine information includes a robot structure type and a corresponding sensor type.

[0017] Optionally, the method for identifying the capability of the robot body, wherein the step of obtaining the user's demand question information, performing intent analysis on the demand question information to obtain the target intent, and performing retrieval processing on the target machine information according to the target intent to obtain the machine information retrieval result, specifically includes:

[0018] Obtaining user demand question information, performing text analysis on the demand question information to obtain valid information text and irrelevant information text, and removing the irrelevant information text;

[0019] Perform intent analysis on the valid information text to obtain the target intent, perform category analysis on the target intent to obtain the target machine category, generate a corresponding search path based on the target machine category, perform information retrieval on the target machine information based on the search path, and obtain a machine information retrieval result.

[0020] Optionally, the method for identifying the capability of the robot body, wherein the step of obtaining the user's demand question information, performing intent analysis on the demand question information to obtain a target intent, and performing retrieval processing on the target machine information according to the target intent to obtain a machine information retrieval result, further comprises:

[0021] When it is detected that the user inputs target document information, the target document information is acquired, and the target document information is associated with the demand question information to obtain a judgment result;

[0022] If the judgment result is that the target document information is associated with the demand question information, updating the demand question information;

[0023] If the judgment result is that the target document information and the demand question information are not associated, the target document information and the demand question information are synchronously processed.

[0024] Optionally, the method for identifying the capability of the robot body, wherein the step of generating corresponding demand response information according to the machine information retrieval result and sending the demand response information to the user, specifically comprises:

[0025] Generate corresponding demand response information according to the machine information retrieval result, and make a qualification judgment on the demand response information according to the target language model to obtain a corresponding judgment result;

[0026] If the judgment result is that the demand reply information has been fully replied, then send the demand reply information to the user.

[0027] Optionally, for the method for identifying the capabilities of the robot body, wherein generating corresponding demand reply information according to the machine information retrieval result, and performing a qualification judgment on the demand reply information according to the target language model to obtain a corresponding judgment result, and then further comprising:

[0028] If the judgment result is that the demand reply information has not been fully replied, then optimize the retrieval path according to the target language model to obtain a target retrieval path;

[0029] Perform information retrieval on the target machine information according to the target retrieval path to obtain optimized demand reply information, and send the optimized demand reply information to the user.

[0030] Optionally, for the method for identifying the capabilities of the robot body, wherein the system for identifying the capabilities of the robot body includes:

[0031] An information processing module, configured to generate a robot analysis instruction when accessing a target robot system, obtain the robot body information of the target robot system according to the robot analysis instruction, and perform information processing on the robot body information to obtain target machine information;

[0032] An information retrieval module, configured to obtain the demand problem information of the user, perform intention analysis on the demand problem information to obtain a target intention, and perform retrieval processing on the target machine information according to the target intention to obtain a machine information retrieval result;

[0033] A reply generation module, configured to generate corresponding demand reply information according to the machine information retrieval result, and send the demand reply information to the user.

[0034] In addition, to achieve the above object, the present invention further provides a terminal, wherein the terminal includes: a memory, a processor, and a program for identifying the capabilities of the robot body stored on the memory and executable on the processor, and when the program for identifying the capabilities of the robot body is executed by the processor, the steps of the method for identifying the capabilities of the robot body as described above are implemented.

[0035] In addition, to achieve the above object, the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a program for identifying the capabilities of the robot body, and when the program for identifying the capabilities of the robot body is executed by a processor, the steps of the method for identifying the capabilities of the robot body as described above are implemented.

[0036] In the present invention, when accessing a target robot system, a robot analysis instruction is generated. According to the robot analysis instruction, the robot body information of the target robot system is obtained, and the robot body information is processed to obtain target machine information; the demand problem information of the user is obtained, the intention of the demand problem information is analyzed to obtain a target intention, and the target machine information is retrieved according to the target intention to obtain a machine information retrieval result; a corresponding demand response information is generated according to the machine information retrieval result, and the demand response information is sent to the user. The present invention realizes the deployment of a large language model on an edge device, overcomes the challenge of resource limitations, enables complex robot function analysis to be performed locally, improves the response speed and security of the system; can also quickly obtain and analyze the interface information of unknown robots, greatly improving the adaptability to new or unfamiliar robot systems; can also simulate complex reasoning processes under limited computing resources, effectively making up for the deficiencies of the large language model on edge devices, and realizing in-depth understanding and analysis of robot functions, making task allocation and strategy formulation more accurate and efficient, and optimizing the operation effect of the entire robot system; reduces the need for manual intervention through automated analysis, reduces the operation complexity and the risk of human errors, and improves the reliability of the overall system. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 is a flowchart of a preferred embodiment of the method for identifying the capabilities of the robot body in the present invention;

[0038] Figure 2 is a schematic diagram of the overall process of a preferred embodiment of the method for identifying the capabilities of the robot body in the present invention;

[0039] Figure 3 is a structural diagram of a preferred embodiment of the system for identifying the capabilities of the robot body in the present invention;

[0040] Figure 4 is a schematic diagram of the operating environment of a preferred embodiment of the terminal of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0041] To make the objectives, technical solutions and advantages of the present invention clearer and more definite, the following further describes the present invention in detail with reference to the accompanying drawings and by way of examples. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0042] It should be noted that if there are directional indications (such as up, down, left, right, front, back...) involved in the embodiments of the present invention, the directional indications are only used to explain the relative positional relationship, movement conditions, etc. between components in a specific posture (as shown in the attached drawings). If the specific posture changes, the directional indications will also change accordingly.

[0043] In addition, if there are descriptions such as "first", "second", etc. involved in the embodiments of the present invention, the descriptions of "first", "second", etc. are only for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include at least one such feature. In addition, the technical solutions between various embodiments can be combined with each other, but it must be based on the ability of those of ordinary skill in the art to implement. When the combination of technical solutions results in contradictions or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0044] The method for identifying the capabilities of the robot body according to the preferred embodiment of the present invention, as Figure 1 shown, the method for identifying the capabilities of the robot body includes the following steps:

[0045] Step S10: When accessing the target robot system, generate a robot analysis instruction, obtain the robot body information of the target robot system according to the robot analysis instruction, and perform information processing on the robot body information to obtain target machine information.

[0046] The step S10 includes:

[0047] Step S11: When accessing the target robot system, generate a robot analysis instruction, send the robot analysis instruction to the target robot system, and receive the robot body information sent by the target robot system, where the robot body information includes a robot module, a robot topic, and robot parameters;

[0048] Step S12: Perform structured processing on the robot body information to obtain a structured prompt word, and input the structured prompt word into the target language model to obtain corresponding target machine information; where the target machine information is obtained by the target language model through text understanding and information inference of the structured prompt word, and the target machine information includes the robot structure type and the corresponding sensor type.

[0049] Specifically, before accessing the target robot system, it is necessary to deploy the corresponding large language model on the edge device. When deploying the large language model on the resource-constrained edge device, in the embodiments of the present invention, the MPQ (Mixed-Precision Quantization) method is adopted. By reducing the storage precision of the large language model, while reducing the computing resource requirements, the model performance is maintained as much as possible, thereby improving the inference speed and efficiency, which is suitable for the applications of real-time and offline services and the deployment of edge devices. The specific deployment process is as follows: it is necessary to perform 4-bit integer quantization on the feedforward layer weight values and activation values of the large language model. Through the w4a4 quantization method in mixed-precision quantization, 4-bit integer quantization is performed. During the quantization process, first, the range statistics of the feedforward layer weight values and the activation values in the large language model are performed to obtain the floating-point value distribution range, aiming to ensure that the quantized values can effectively represent the original information, obtain the corresponding minimum and maximum values, and thus determine the scaling factor to ensure that when converting to integer values, the information loss can be minimized as much as possible. After obtaining the floating-point value distribution range, a linear mapping method is used to map the 32-bit floating-point values to 4-bit integers to obtain the target language model. Finally, obtain the model deployment instruction of the target user, and deploy the target language model according to the model deployment instruction. The model using w4a4 quantization in the present invention not only greatly reduces the storage space requirements but also significantly reduces the computational complexity. Especially in matrix multiplication and forward propagation calculations, the inference speed is greatly accelerated. In addition, the corresponding model can run in a more efficient and lower-power manner, thereby improving the inference efficiency.

[0050] The present invention enables the system to join any number of multi-robot systems by setting an independent identifier, i.e., a domain ID, on the system node. Different nodes under the same domain ID can achieve data sharing and communication without direct connection through IP addresses. When accessing the target robot system, the target robot system includes multiple robot systems, for example, a machine vehicle operating system, a drone operating system, and a robot operating system, etc. Generate a robot analysis instruction, send the robot analysis instruction to the target robot system, and receive the robot body information sent by the target robot system. The robot body information includes a robot module, a robot topic, and robot parameters. The robot module usually consists of specific hardware components (such as sensors, actuators, etc.) and corresponding software nodes that control these hardware components. The robot topic is the core mechanism for communication between robots, responsible for transmitting sensor data (such as lidar, camera, etc.), control commands (such as motion instructions), and status information, etc. The robot parameters are used to store global or local system configurations, usually used to control the behavior of nodes, and can be dynamically adjusted during operation, thereby affecting the behavior and system configuration of the robot.

[0051] Afterwards, based on this information, a hierarchical analysis method is used to classify topics into corresponding modules, so as to achieve a clearer organization and understanding of the robot's functional structure, specifically, the robot's body information is structured to obtain structured prompt words, and the structured prompt words are used as the input of the target language model; afterward, the structured prompt words are input into the target language model, and the target language model can recognize and understand the structured prompt words through the knowledge accumulated by pre-training, so as to infer the corresponding target machine information, wherein the target machine information robot structure type (for example, wheeled robot, drone, mechanical arm, etc.) and the corresponding sensor type (for example, laser radar, camera, etc.). The present invention realizes the efficient deployment of large language models on edge devices, overcomes the challenge of resource limitations, enables complex robot function analysis to be performed locally, improves the response speed and security of the system, and can also quickly obtain and parse the interface information of unknown ROS robots, greatly improves the adaptability to new or unfamiliar robot systems, and has good versatility and scalability, can be applied to various types of ROS robot systems, without the need for a large number of customized development for specific robots.

[0052] Step S20, obtaining the user's demand question information, performing intent analysis on the demand question information to obtain the target intent, and performing retrieval processing on the target machine information according to the target intent to obtain a machine information retrieval result.

[0053] The step S20 comprises:

[0054] Step S21, obtaining the user's demand question information, performing text analysis on the demand question information, obtaining valid information text and irrelevant information text, and removing the irrelevant information text;

[0055] Step S22: perform intent analysis on the valid information text to obtain the target intent, perform category analysis on the target intent to obtain the target machine category, generate a corresponding search path based on the target machine category, perform information search on the target machine information based on the search path, and obtain a machine information retrieval result.

[0056] Specifically, in the embodiments of the present invention, the user can directly put forward the implementation requirements of specific functions in the form of natural language based on needs, and the requirements input by the user will be inferred and relevant answers will be returned. Among them, the input methods of the user include but are not limited to voice input and text input. Taking text input as an example, when the user inputs the corresponding requirement problem information (for example, are there robots that can deliver goods by flying low in the group), the requirement problem information of the user is obtained. After that, it is necessary to infer the requirement problem information. During the inference process, a hierarchical design information retrieval method is adopted, and then a large language model is combined for progressive multi-round question and answer. Specifically, starting from the module information at the top layer, the module and its topic most relevant to the requirement problem information are retrieved and selected as the enhanced information for the next round of inference. That is, the requirement problem information is text-analyzed to obtain valid information text and irrelevant information text, and the irrelevant information text is removed; the valid information text is intention-analyzed to obtain the target intention, the target intention is category-analyzed to obtain the target machine category (for example, drone), and a corresponding retrieval path is generated according to the target machine category. Information retrieval is performed on the target machine information according to the retrieval path to obtain a machine information retrieval result (for example, flight control and attitude control). The present invention provides strong support for multi-type or heterogeneous robot collaboration scenarios. By quickly analyzing the capabilities of each robot, better decision-making basis is provided, the collaboration efficiency of the overall system is improved, and in-depth understanding and analysis of robot functions are realized, making task allocation and strategy formulation more accurate and efficient, and optimizing the operation effect of the entire robot system.

[0057] Further, if the target document information is detected to be input by the user during the inference process, the target document information is obtained, and the target document information and the requirement problem information are associated and judged to obtain a judgment result; if the judgment result is that the target document information is associated with the requirement problem information, the requirement problem information is updated and processed, and the updated requirement problem information is inferred; if the judgment result is that the target document information is not associated with the requirement problem information, the target document information and the requirement problem information are synchronized, that is, while the requirement problem information is inferred, the target document information is inferred, so as to improve the information processing efficiency.

[0058] Step S30: Generate a corresponding requirement reply message according to the machine information retrieval result, and send the requirement reply message to the user.

[0059] The step S30 includes:

[0060] Step S31: Generate corresponding demand response information according to the machine information retrieval result, and perform a qualification judgment on the demand response information according to the target language model to obtain a corresponding judgment result;

[0061] Step S32: If the judgment result is that the demand response information has been fully replied, send the demand response information to the user.

[0062] Specifically, after obtaining the machine information retrieval result, it is necessary to combine with a large language model, that is, the target language model, for progressive multi-round Q&A. In the embodiment of the present invention, a qualification judgment mechanism will be embedded in each round of Q&A process, and the semantic understanding ability of the target language model will be combined to judge whether the information at the current level can fully answer the questions raised by the user. Specifically, corresponding demand response information is generated according to the machine information retrieval result (for example, it is learned through consultation that the xxtopic of the drone can be realized, so this robot group can be used for delivery), and a qualification judgment is performed on the demand response information according to the target language model to obtain a corresponding judgment result; if the judgment result is that the demand response information has been fully replied, the demand response information is sent to the user. Through the innovative multi-round Q&A mechanism, the present invention simulates a complex reasoning process with limited computing resources, effectively making up for the deficiency of the large language model ability on edge devices.

[0063] Furthermore, if the judgment result is that the demand response information has not been fully replied, the target speech model will be automatically scheduled to propose an optimal in-depth retrieval path based on the existing information and hierarchical structure, thereby avoiding the efficiency problems that may be brought by traditional breadth or depth traversal, and through progressive multi-round reasoning and hierarchical information retrieval, gradually analyze the specific topics and parameter details in different modules until a clear guidance path can be proposed, that is, optimize the retrieval path according to the target language model to obtain a target retrieval path; then, perform information retrieval on the target machine information according to the target retrieval path to obtain an optimized demand response information, and send the optimized demand response information to the user.

[0064] Furthermore, the overall process of the ability recognition method of the robot body in the present invention is as Figure 2 shown. First, before accessing the target robot system, it is necessary to deploy the corresponding large language model on the edge device. The specific deployment process is to obtain the model deployment instruction of the target user, access the corresponding large language model according to the model deployment instruction, and then perform 4-bit integer quantization on the feedforward layer weight value and activation value of the large language model by using the w4a4 quantization method in mixed precision quantization to obtain the target language model, and deploy the target language model.

[0065] Secondly, by setting independent identifiers, i.e., domain IDs, on system nodes, the system can be joined into any number of multi-robot systems. Different nodes under the same domain ID can achieve data sharing and communication without direct connection via IP addresses. When accessing the target robot system, where the target robot system includes multiple robot systems, such as a machine vehicle operating system, a drone operating system, and a robot operating system, etc., a robot analysis instruction is generated, the robot analysis instruction is sent to the target robot system, and the robot body information sent by the target robot system is received. Among them, the robot body information includes robot modules, robot topics, and robot parameters. For example, the robot body information of the machine vehicle operating system, i.e., the machine vehicle interface data, includes wheel group control, camera, and servo; the robot body information of the drone operating system, i.e., the drone interface data, includes flight control, camera, and attitude control; the robot body information of the robot operating system, i.e., the robot interface data, includes motion control, robotic arm, and camera.

[0066] Finally, when the user inputs the corresponding demand problem information (such as, are there any robots in the group that can deliver goods by flying low), the demand problem information of the user is obtained. After that, the demand problem information needs to be inferred. Specifically, the demand problem information is text-analyzed to obtain valid information text and irrelevant information text, and the irrelevant information text is removed; the valid information text is intention-analyzed to obtain the target intention, the target intention is category-analyzed to obtain the target machine category (such as a drone), and a corresponding retrieval path is generated according to the target machine category. Information retrieval is performed on the injected target machine information according to the retrieval path to obtain a machine information retrieval result (such as flight control and attitude control). After obtaining the machine information retrieval result, a corresponding demand reply information is generated according to the machine information retrieval result (such as, through inquiry, it is known that the xxtopic of the drone can achieve this, so this robot group can be used for delivering goods).

[0067] Furthermore, as Figure 3 shown, based on the above-mentioned method for identifying the capabilities of a robot body, the present invention also correspondingly provides a system for identifying the capabilities of a robot body. The system for identifying the capabilities of a robot body includes:

[0068] An information processing module 51, configured to generate a robot analysis instruction when accessing the target robot system, obtain the robot body information of the target robot system according to the robot analysis instruction, and perform information processing on the robot body information to obtain target machine information;

[0069] An information retrieval module 52 is configured to obtain the requirement problem information of a user, perform intention analysis on the requirement problem information to obtain a target intention, and perform retrieval processing on the target machine information according to the target intention to obtain a machine information retrieval result;

[0070] A reply generation module 53 is configured to generate a corresponding requirement reply information according to the machine information retrieval result, and send the requirement reply information to the user.

[0071] Further, as Figure 4 shown, based on the above-mentioned ability recognition method of the robot body, the present invention also correspondingly provides a terminal, and the terminal includes a processor 10, a memory 20, and a display 30. Figure 4 Only some components of the terminal are shown, but it should be understood that it is not required to implement all the shown components, and more or fewer components can be alternatively implemented.

[0072] In some embodiments, the memory 20 may be an internal storage unit of the terminal, such as a hard disk or a memory of the terminal. In some other embodiments, the memory 20 may also be an external storage device of the terminal, such as a plug-in hard disk equipped on the terminal, a Smart Media Card (SMC), a Secure Digital (SD) card, a Flash Card, etc. Further, the memory 20 may also include both the internal storage unit and the external storage device of the terminal. The memory 20 is used to store application software installed on the terminal and various types of data, such as program codes for installing the terminal. The memory 20 may also be used to temporarily store data that has been output or will be output. In one embodiment, a robot body ability recognition program 40 is stored on the memory 20, and the robot body ability recognition program 40 can be executed by the processor 10, so as to implement the robot body ability recognition method in the present application.

[0073] In some embodiments, the processor 10 may be a Central Processing Unit (CPU), a microprocessor, or other data processing chips, and is used to run the program codes stored in the memory 20 or process data, such as executing the robot body ability recognition method, etc.

[0074] In some embodiments, the display 30 may be an LED display, a liquid crystal display, a touch liquid crystal display, an OLED (Organic Light-Emitting Diode) touch device, etc. The display 30 is used to display information on the terminal and to display a visual user interface. Components of the terminal communicate with each other via a system bus.

[0075] In one embodiment, when the processor 10 executes the program 40 for identifying the capabilities of the robot body in the memory 20, the following steps are implemented:

[0076] When accessing the target robot system, generate a robot analysis instruction, obtain the robot body information of the target robot system according to the robot analysis instruction, and perform information processing on the robot body information to obtain target machine information;

[0077] Obtain the demand problem information of the user, perform intention analysis on the demand problem information to obtain a target intention, and perform retrieval processing on the target machine information according to the target intention to obtain a machine information retrieval result;

[0078] Generate a corresponding demand response message according to the machine information retrieval result and send the demand response message to the user.

[0079] Among them, before the step of "when accessing the target robot system, generate a robot analysis instruction, obtain the robot body information of the target robot system according to the robot analysis instruction, and perform information processing on the robot body information to obtain target machine information", it further includes:

[0080] Obtain the feedforward layer weight values and activation values of the large language model, perform range statistics on the feedforward layer weight values and the activation values to obtain a floating-point value distribution range, and perform integer quantization on the feedforward layer weight values and the activation values according to the floating-point value distribution range to obtain a target language model;

[0081] Perform range statistics on the feedforward layer weight values and the activation values to obtain a floating-point value distribution range, and perform integer quantization on the feedforward layer weight values and the activation values according to the floating-point value distribution range to obtain a target language model, and deploy the target language model.

[0082] Among them, the step of "when accessing the target robot system, generate a robot analysis instruction, obtain the robot body information of the target robot system according to the robot analysis instruction, and perform information processing on the robot body information to obtain target machine information" specifically includes:

[0083] When connected to the target robot system, a robot analysis instruction is generated, the robot analysis instruction is sent to the target robot system, and robot body information sent by the target robot system is received, wherein the robot body information includes a robot module, a robot topic, and a robot parameter;

[0084] Structuring the robot body information to obtain structured prompt words, and inputting the structured prompt words into the target language model to obtain corresponding target machine information;

[0085] The target machine information is obtained by the target language model performing text understanding and information inference on the structured prompt words, and the target machine information includes a robot structure type and a corresponding sensor type.

[0086] The step of obtaining the user's demand information, performing intent analysis on the demand information to obtain the target intent, and performing retrieval processing on the target machine information according to the target intent to obtain the machine information retrieval result specifically includes:

[0087] Obtaining user demand question information, performing text analysis on the demand question information to obtain valid information text and irrelevant information text, and removing the irrelevant information text;

[0088] Perform intent analysis on the valid information text to obtain the target intent, perform category analysis on the target intent to obtain the target machine category, generate a corresponding search path based on the target machine category, perform information retrieval on the target machine information based on the search path, and obtain a machine information retrieval result.

[0089] The method further comprises: obtaining the user's demand information, performing intent analysis on the demand information to obtain the target intent, and performing retrieval processing on the target machine information according to the target intent to obtain the machine information retrieval result, and then further comprising:

[0090] When it is detected that the user inputs target document information, the target document information is acquired, and the target document information is associated with the demand question information to obtain a judgment result;

[0091] If the judgment result is that the target document information is associated with the demand question information, updating the demand question information;

[0092] If the judgment result is that the target document information and the demand question information are not associated, the target document information and the demand question information are synchronously processed.

[0093] Among them, generating corresponding demand response information according to the machine information retrieval result and sending the demand response information to the user specifically includes:

[0094] Generating corresponding demand response information according to the machine information retrieval result, and making a qualification judgment on the demand response information according to the target language model to obtain a corresponding judgment result;

[0095] If the judgment result is that the demand response information has been fully replied, then send the demand response information to the user.

[0096] Among them, after generating corresponding demand response information according to the machine information retrieval result and making a qualification judgment on the demand response information according to the target language model to obtain a corresponding judgment result, it further includes:

[0097] If the judgment result is that the demand response information has not been fully replied, then optimize the retrieval path according to the target language model to obtain a target retrieval path;

[0098] Retrieve information from the target machine information according to the target retrieval path to obtain optimized demand response information, and send the optimized demand response information to the user.

[0099] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores an ability recognition program of a robot body, and when the ability recognition program of the robot body is executed by a processor, the steps of the above-mentioned ability recognition method of the robot body are implemented.

[0100] It should be noted that in this article, the term "including", "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device. Without more limitations, the element defined by the statement "including a..." does not exclude the existence of additional identical elements in the process, method, article or device including the element.

[0101] Of course, those of ordinary skill in the art can understand that all or part of the processes of implementing the above-mentioned embodiment methods can be completed by a computer program instructing related hardware (such as a processor, a controller, etc.), and the program can be stored in a computer-readable computer-readable storage medium. When the program is executed, it can include the processes of the above-mentioned method embodiments. The computer-readable storage medium can be a memory, a magnetic disk, an optical disk, etc.

[0102] It should be understood that the application of the present invention is not limited to the above examples. For those of ordinary skill in the art, improvements or modifications can be made according to the above description, and all such improvements and modifications shall fall within the protection scope of the appended claims of the present invention.

Claims

1. A method for identifying the capabilities of a robot, characterized in that: The capability identification method of the robot body comprises: When accessing to the target robot system, a robot analysis instruction is generated, robot body information of the target robot system is acquired according to the robot analysis instruction, and information processing is performed on the robot body information to obtain target machine information; Acquire user's demand question information, perform intent analysis on the demand question information to obtain target intent, and perform retrieval processing on the target machine information according to the target intent to obtain machine information retrieval results; Generate corresponding demand reply information according to the machine information retrieval result, and send the demand reply information to the user.

2. The method for identifying the capability of a robot according to claim 1, characterized in that: When the target robot system is connected, a robot analysis instruction is generated, robot body information of the target robot system is obtained according to the robot analysis instruction, and information processing is performed on the robot body information to obtain target machine information, which also includes: Obtaining feedforward layer weight values ​​and activation values ​​of a large language model, performing range statistics on the feedforward layer weight values ​​and the activation values ​​to obtain a floating-point value distribution range, and performing integer quantization on the feedforward layer weight values ​​and the activation values ​​according to the floating-point value distribution range to obtain a target language model; Perform range statistics on the feedforward layer weight value and the activation value to obtain a floating-point value distribution range, perform integer quantization on the feedforward layer weight value and the activation value according to the floating-point value distribution range to obtain a target language model, and deploy the target language model.

3. The method for identifying the capability of a robot according to claim 2, characterized in that: When the target robot system is connected, a robot analysis instruction is generated, robot body information of the target robot system is obtained according to the robot analysis instruction, and information processing is performed on the robot body information to obtain target machine information, specifically including: When connected to the target robot system, a robot analysis instruction is generated, the robot analysis instruction is sent to the target robot system, and robot body information sent by the target robot system is received, wherein the robot body information includes a robot module, a robot topic, and a robot parameter; Structuring the robot body information to obtain structured prompt words, and inputting the structured prompt words into the target language model to obtain corresponding target machine information; The target machine information is obtained by the target language model performing text understanding and information inference on the structured prompt words, and the target machine information includes a robot structure type and a corresponding sensor type.

4. The method for identifying the capability of a robot according to claim 2, characterized in that: The obtaining of user demand information, performing intent analysis on the demand information to obtain a target intent, and performing retrieval processing on the target machine information according to the target intent to obtain a machine information retrieval result specifically includes: Obtaining user demand question information, performing text analysis on the demand question information to obtain valid information text and irrelevant information text, and removing the irrelevant information text; Perform intent analysis on the valid information text to obtain the target intent, perform category analysis on the target intent to obtain the target machine category, generate a corresponding search path based on the target machine category, perform information retrieval on the target machine information based on the search path, and obtain a machine information retrieval result.

5. The method for identifying the capability of a robot according to claim 4, characterized in that: The step of obtaining the user's demand information, performing intent analysis on the demand information to obtain a target intent, and performing retrieval processing on the target machine information according to the target intent to obtain a machine information retrieval result, further includes: When it is detected that the user inputs target document information, the target document information is acquired, and the target document information is associated with the demand question information to obtain a judgment result; If the judgment result is that the target document information is associated with the demand question information, updating the demand question information; If the judgment result is that the target document information and the demand question information are not associated, the target document information and the demand question information are synchronously processed.

6. The method for identifying the capability of a robot according to claim 4, characterized in that: The generating corresponding demand reply information according to the machine information retrieval result and sending the demand reply information to the user specifically includes: Generate corresponding demand response information according to the machine information retrieval result, and make a qualification judgment on the demand response information according to the target language model to obtain a corresponding judgment result; If the judgment result is that the demand reply information has been completely replied, the demand reply information is sent to the user.

7. The method for identifying the capability of a robot according to claim 6, characterized in that: The step of generating corresponding demand reply information according to the machine information retrieval result, and performing qualification judgment on the demand reply information according to the target language model to obtain a corresponding judgment result, further comprising: If the judgment result is that the demand reply information is not fully replied, optimizing the search path according to the target language model to obtain a target search path; The target machine information is retrieved according to the target retrieval path to obtain optimized demand response information, and the optimized demand response information is sent to the user.

8. A robot body capability recognition system, characterized in that: The capability identification system of the robot body includes: An information processing module, for generating a robot analysis instruction when connected to a target robot system, obtaining robot body information of the target robot system according to the robot analysis instruction, and performing information processing on the robot body information to obtain target machine information; An information retrieval module is used to obtain the user's demand question information, perform intent analysis on the demand question information to obtain the target intent, and perform retrieval processing on the target machine information according to the target intent to obtain the machine information retrieval result; The reply generation module is used to generate corresponding demand reply information according to the machine information retrieval result, and send the demand reply information to the user.

9. A terminal, characterized in that: The terminal includes a memory, a processor, and a robot body capability identification program stored in the memory and executable on the processor. When the robot body capability identification program is executed by the processor, the steps of the robot body capability identification method as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and the computer-readable storage medium stores a capability identification program of a robot body. When the capability identification program of the robot body is executed by a processor, the steps of the capability identification method of the robot body as described in any one of claims 1-7 are implemented.

Images