AI-driven intelligent self-adaptive learning method and system

Abstract

The invention relates to the technical field of artificial intelligence, in particular to an AI-driven intelligent self-adaptive learning method and system, and the method comprises the steps: confirming to receive a self-adaptive learning instruction, obtaining a learning task type, retrieving a target reference template from a prior knowledge base through external environment data and the learning task type, and obtaining a target reference template; performing parameter initialization on external environment data by utilizing the target reference template to obtain an initialized parameter set, obtaining an optimized parameter node set by utilizing an optimization algorithm and initialized parameters, performing parameter adjustment on the optimized parameter set by utilizing a priori knowledge base to obtain a global parameter set, generating a control instruction set by utilizing the global parameter set, and storing the control instruction set in a database; and driving the intelligent agent with the body by using a control instruction set, collecting a plurality of execution result data by using a perception driving unit, generating a detection report, and obtaining an optimized priori knowledge base. According to the method, the decision-making ability and the task execution efficiency of the self-adaptive intelligent agent in the adaptive learning process can be improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to an AI-driven embodied intelligent adaptive learning method and system. Background Art

[0002] AI-driven embodied intelligence adaptive learning uses AI technology to enable intelligent agents to autonomously learn and adapt to environmental changes. This approach uses the embodied agent's perception-driven, computation-driven, data-driven, and interaction-driven units to collect and process external environmental data in real time, and utilizes a pre-built prior knowledge base for parameter initialization and optimization. This significantly improves the embodied agent's decision-making capabilities and task execution efficiency in complex and dynamic environments, enabling it to continuously adjust and optimize its behavior based on real-time data to better complete tasks.

[0003] Existing adaptive learning methods often suffer from latency and data loss when processing large amounts of real-time data, resulting in embodied agents being unable to respond promptly to environmental changes. Therefore, improving the decision-making capabilities and task execution efficiency of embodied agents during adaptive learning is an urgent technical challenge. Summary of the Invention

[0004] The present invention provides an AI-driven embodied intelligent adaptive learning method and system, the main purpose of which is to improve the decision-making ability and task execution efficiency of embodied intelligent agents during the adaptive learning process.

[0005] To achieve the above objectives, the present invention provides an AI-driven embodied intelligent adaptive learning method, comprising: Confirming receipt of an adaptive learning instruction, parsing the adaptive learning instruction, obtaining a learning task type, and confirming an embodied intelligent agent and a 6G network communication link, wherein the embodied intelligent agent includes: a perception drive unit, a computation drive unit, a data drive unit, and an interaction drive unit; Using a preset sampling time interval, a perception drive unit, and a 6G network communication link to collect external environmental data including multiple environmental parameters, using the external environmental data and the learning task type to retrieve a target reference template from a pre-built prior knowledge base, using the target reference template to initialize each of the multiple environmental parameters corresponding to the external environmental data to obtain an initialized parameter set, and transmitting the initialized parameter set to the calculation drive unit, wherein the prior knowledge base includes multiple reference templates; Acquire an optimization algorithm based on the initialization parameters in the initialization parameter set, optimize and calculate the initialization parameters using the optimization algorithm to obtain optimization parameter nodes, wherein the optimization parameter nodes include optimization parameters and confidence evaluation parameters, and summarize the optimization parameter nodes to obtain an optimization parameter node set; transmitting the optimized parameter node set to the data-driven unit; Extracting an optimization parameter set and a confidence evaluation parameter set from the optimization parameter node set of the data-driven unit, adjusting the optimization parameter set using the prior knowledge base and the confidence evaluation parameter set to obtain a global parameter set, and transmitting the global parameter set to the interactive driving unit; generating a control instruction set using a global parameter set, driving the embodied intelligent agent using the control instruction set, collecting multiple execution result data of the embodied intelligent agent using a perception drive unit, transmitting the multiple execution result data to a data drive unit based on a 6G network communication link, and generating a detection report based on the multiple execution result data in the data drive unit; The priori knowledge base is updated according to the detection report to obtain an optimized priori knowledge base.

[0006] Optionally, the collecting of external environmental data including multiple environmental parameters using a preset sampling time interval, a perception drive unit, and a 6G network communication link includes: Utilize 6G network communication links and perception drive units to build a distributed perception network, and use the distributed perception network and sampling time intervals to acquire visual image data and LiDAR point cloud data; The perception-driven unit is adjusted based on the learning task type, and the adjusted perception-driven unit is used to perform real-time enhancement processing on the visual image data to obtain a preprocessed image set. The LiDAR point cloud data is then subjected to noise reduction and motion compensation processing to obtain optimized point cloud data. The preprocessed image set and optimized point cloud data are fused into multi-source data to obtain external environment data.

[0007] Optionally, the method uses the external environment data and the learning task type to retrieve a target reference template from a pre-built prior knowledge base, and uses the target reference template to initialize each of the multiple environment parameters corresponding to the external environment data to obtain an initialization parameter set, including: The pre-confirmed edge computing module is used to analyze the environmental parameters in the external environment data to obtain the current task encoding vector, current environment feature fingerprint, and device information corresponding to the environmental parameters. The following operations are performed on each of the multiple reference templates: Extracting a task encoding vector of the reference template, and calculating task type similarity using the current task encoding vector, the task encoding vector, and a pre-established task type similarity calculation formula; Extracting the environmental feature fingerprint of the reference template, and calculating the environmental similarity using the current environmental feature fingerprint, the environmental feature fingerprint, and a pre-built environmental similarity calculation formula; Compare device information with the pre-built device compatibility table to obtain hardware compatibility; The overall matching degree is calculated using task type similarity, environment similarity, and hardware compatibility. The calculation formula is as follows: in, is the total matching degree, is the task type similarity, Indicates the environmental similarity, For hardware compatibility, All are weight coefficients; Summarize the total matching degrees to obtain a total matching degree set, sort the total matching degrees in the total matching degree set in descending order to obtain a total matching degree sequence, identify five candidate reference templates in the total matching degree sequence, wherein the five candidate reference templates are reference templates corresponding to the first five total matching degrees in the total matching degree sequence, input each of the five candidate reference templates into a pre-built digital twin platform for verification, and obtain five verification parameters, and take the candidate reference template corresponding to the largest verification parameter among the five verification parameters as the target reference template; The verification parameters corresponding to the target reference template are used to initialize the environmental parameters to obtain the initialization parameters, and the initialization parameters are summarized to obtain an initialization parameter set.

[0008] Optionally, the obtaining of an optimization algorithm based on the initialization parameters in the initialization parameter set, and optimizing the initialization parameters using the optimization algorithm to obtain an optimized parameter node, includes: Obtaining an optimization algorithm using the initialization parameters in the initialization parameter set and a pre-built algorithm mapping table; Using the optimization algorithm to perform hierarchical optimization calculations on the initialization parameters to obtain multiple edge node optimization parameters; Performing a fixed-step gradient descent optimization on the plurality of edge node optimization parameters to obtain a plurality of preliminary optimization parameters; Verifying and adjusting a plurality of preliminary optimization parameters using the digital twin platform to obtain a plurality of optimization parameters; Utilizing a pre-built confidence evaluation method, each of the multiple optimization parameters is optimized and evaluated to obtain a confidence evaluation parameter; The optimization parameters and their corresponding confidence assessment parameters are integrated to generate an optimization parameter node.

[0009] Optionally, the use of the prior knowledge base and the confidence assessment parameter set to adjust the optimization parameter set to obtain a global parameter set includes: According to the confidence evaluation parameters in the confidence evaluation parameter set, the optimization parameters in the optimization parameter set are divided into three division parameters, and the division parameters are high confidence parameters, medium confidence parameters or low confidence parameters, and the low confidence parameters are reinitialized using the prior knowledge base to obtain the corrected parameters; Fine-tune the confidence parameters using the prior knowledge base to obtain fine-tuning parameters; Obtaining a global optimization strategy based on the correction parameters, fine-tuning parameters, high-confidence parameters and a priori knowledge base; The correction parameters, fine-tuning parameters and high-confidence parameters are integrated using the global optimization strategy to obtain a global parameter set.

[0010] Optionally, generating a control instruction set by using a global parameter set includes: generating a preliminary control instruction set based on the global parameter set; Performing syntax verification and semantic verification on each of the preliminary control instructions in the preliminary control instruction set using an interactive driving unit; If the grammatical verification of the preliminary control instruction fails, revising the preliminary control instruction according to the pre-established grammatical rules to obtain a grammatical revision instruction; If the syntax verification of the preliminary control instruction passes, the preliminary control instruction is regarded as a syntax qualified instruction; If the semantic verification of the preliminary control instruction is unclear, semantically optimizing the preliminary control instruction according to the pre-built task execution rule to obtain a semantically optimized instruction; If the semantic verification of the preliminary control instruction is clear, the preliminary control instruction is regarded as a semantically qualified instruction; Integrate the grammatically qualified instructions, the semantically qualified instructions, the grammatically revised instructions, and the semantically optimized instructions to obtain a verified instruction set; Perform similarity comparison using the verified instruction set, the pre-confirmed historical successful instruction set, and the pre-built structural similarity algorithm to obtain a similarity evaluation value; If the similarity evaluation value is less than a preset evaluation threshold, the verified instruction set is adjusted based on the learning task type to obtain an adjusted instruction set; Using the adjusted instruction set as the verified instruction set, and returning to the step of performing similarity comparison using the verified instruction set, the pre-confirmed historical successful instruction set, and the pre-built structural similarity algorithm to obtain a similarity evaluation value, until the similarity evaluation value is greater than or equal to an evaluation threshold; If the similarity evaluation value is greater than or equal to the evaluation threshold, the verified instruction set is used as a candidate control instruction set; The candidate control instruction set is pre-executed using a pre-built simulation execution environment, and after confirming that the effect of the pre-execution is the preset expected effect, the candidate control instruction set is determined as the control instruction set.

[0011] Optionally, the using of a control instruction set to drive the embodied intelligent agent and using a perception driving unit to collect a plurality of execution result data of the embodied intelligent agent includes: Using a perception drive unit and a preset collection frequency to collect data when driving the embodied intelligent body, to obtain a plurality of raw execution data; performing data preprocessing on each of the plurality of original execution data to obtain a plurality of preprocessed execution data; Performing data cleaning on each of the plurality of pre-processed execution data to obtain a plurality of accurate execution data; Feature extraction is performed on each of the multiple accurate execution data to obtain multiple execution result data.

[0012] Optionally, transmitting the plurality of execution result data to the data driving unit based on the 6G network communication link, and generating a detection report according to the plurality of execution result data in the data driving unit, includes: Transmitting the plurality of execution result data to a data driving unit using a 6G network communication link to obtain a plurality of transmitted execution result data; Performing data accuracy verification on the multiple execution result data after transmission to obtain a check result, wherein the check result includes a check feasibility and a check infeasibility; After confirming that the inspection result is feasible, the plurality of execution result data are compared and analyzed with a plurality of preset qualified performance data to obtain an analysis result report; The analysis result report is diagnosed for abnormality using preset abnormality diagnosis rules to obtain a detection report.

[0013] Optionally, updating the prior knowledge base according to the test report to obtain an optimized prior knowledge base includes: Extract feedback data from the test report; Using the feedback data to update the reference template, an updated reference template is obtained; The updated benchmark template is integrated into the prior knowledge base to obtain an optimized prior knowledge base.

[0014] To achieve the above objectives, the present invention further provides an AI-driven embodied intelligent adaptive learning system, comprising: A learning task parsing module is used to confirm receipt of an adaptive learning instruction, parse the adaptive learning instruction, obtain the learning task type, and identify the embodied intelligent agent and the 6G network communication link. The embodied intelligent agent includes a perception drive unit, a computation drive unit, a data drive unit, and an interaction drive unit. a parameter optimization module for collecting external environmental data including multiple environmental parameters using a preset sampling time interval, a perception drive unit, and a 6G network communication link, retrieving a target reference template from a pre-built prior knowledge base using the external environmental data and the learning task type, initializing each of the multiple environmental parameters corresponding to the external environmental data using the target reference template to obtain an initialized parameter set, and transmitting the initialized parameter set to the calculation drive unit, wherein the prior knowledge base includes multiple reference templates; Acquire an optimization algorithm based on the initialization parameters in the initialization parameter set, optimize and calculate the initialization parameters using the optimization algorithm to obtain optimization parameter nodes, wherein the optimization parameter nodes include optimization parameters and confidence evaluation parameters, and summarize the optimization parameter nodes to obtain an optimization parameter node set; A parameter adjustment module is used to transmit the optimization parameter node set to the data driving unit, extract the optimization parameter set and the confidence evaluation parameter set from the optimization parameter node set of the data driving unit, adjust the optimization parameter set using the prior knowledge base and the confidence evaluation parameter set to obtain a global parameter set, and transmit the global parameter set to the interactive driving unit; An interactive feedback module is used to generate a control instruction set using a global parameter set, drive the embodied intelligent body using the control instruction set, and collect multiple execution result data of the embodied intelligent body using a perception drive unit, transmit the multiple execution result data to a data drive unit based on a 6G network communication link, generate a detection report based on the multiple execution result data in the data drive unit, and update the prior knowledge base according to the detection report to obtain an optimized prior knowledge base.

[0015] In order to solve the above problem, the present invention further provides an electronic device, comprising: a memory storing at least one instruction; and A processor executes instructions stored in the memory to implement the above-mentioned AI-driven embodied intelligent adaptive learning method.

[0016] In order to solve the above problems, the present invention also provides a computer-readable storage medium, which stores at least one instruction, and the at least one instruction is executed by a processor in an electronic device to implement the above-mentioned AI-driven embodied intelligent adaptive learning method.

[0017] The present invention solves the problems described in the background technology. The present invention confirms the receipt of adaptive learning instructions, parses the adaptive learning instructions, obtains the learning task type, and confirms the embodied intelligent body and the 6G network communication link. The embodied intelligent body includes: a perception drive unit, a calculation drive unit, a data drive unit, and an interaction drive unit. It can be seen that the embodiment of the present invention can realize real-time and efficient data interaction between the embodied intelligent body and the management platform through the high speed and low latency characteristics of the 6G network communication link, ensuring the rapid transmission and execution of task instructions. On this basis, the present invention uses a preset sampling time interval, a perception drive unit, and a 6G network communication link to collect external environmental data including multiple environmental parameters, uses the external environmental data and the learning task type to retrieve a target reference template from a pre-built prior knowledge base, uses the target reference template to initialize each of the multiple environmental parameters corresponding to the external environmental data, obtains an initialization parameter set, and transmits the initialization parameter set to the calculation drive unit. The prior knowledge base includes multiple reference templates. It can be seen that the embodiment of the present invention can ensure the timeliness and accuracy of the external environmental data and provide a reliable data foundation for subsequent processing. Furthermore, the present invention obtains an optimization algorithm based on the initialization parameters in the initialization parameter set, optimizes and calculates the initialization parameters using the optimization algorithm, and obtains optimization parameter nodes, wherein the optimization parameter nodes include optimization parameters and confidence assessment parameters, and summarizes the optimization parameter nodes to obtain an optimization parameter node set. It can be seen that the embodiment of the present invention uses the optimization algorithm to efficiently optimize the initialization parameters, generates an optimization parameter node set including optimization parameters and confidence assessment parameters, and significantly improves the efficiency and accuracy of parameter optimization. Next, the present invention transmits the optimization parameter node set to the data-driven unit, extracts the optimization parameter set and the confidence assessment parameter set from the optimization parameter node set of the data-driven unit, and uses the prior knowledge base and the confidence assessment parameter set to adjust the parameters of the optimization parameter set to obtain a global parameter set, and transmits the global parameter set to the interactive driving unit. It can be seen that the embodiment of the present invention adjusts the optimization parameters, integrates the prior knowledge base and the confidence assessment parameter set, and generates a global parameter set, which significantly improves the accuracy and reliability of the parameters. Furthermore, the present invention uses a global parameter set to generate a control instruction set, uses the control instruction set to drive the embodied intelligent body, and uses a perception drive unit to collect multiple execution result data of the embodied intelligent body, transmits the multiple execution result data to the data drive unit based on the 6G network communication link, generates a detection report based on the multiple execution result data in the data drive unit, updates the prior knowledge base based on the detection report, and obtains an optimized prior knowledge base. It can be seen that the embodiment of the present invention uses a feedback mechanism to update the prior knowledge base based on the execution result data, realizes adaptive optimization, and significantly improves the accuracy and efficiency of task execution. Therefore, the present invention can improve the decision-making ability and task execution efficiency of the embodied intelligent body in the adaptive learning process. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 A schematic diagram of a flow chart of an AI-driven embodied intelligent adaptive learning method provided in one embodiment of the present invention; Figure 2 A functional module diagram of an AI-driven embodied intelligent adaptive learning system provided by one embodiment of the present invention; Figure 3 A schematic diagram of the structure of an electronic device for implementing the AI-driven embodied intelligent adaptive learning method provided by one embodiment of the present invention.

[0019] Description of reference numerals: 1. Electronic device; 10. Processor; 11. Memory; 12. Bus.

[0020] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0021] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0022] The embodiments of the present application provide an AI-driven embodied intelligent adaptive learning method. The execution subject of the AI-driven embodied intelligent adaptive learning method includes but is not limited to at least one of the electronic devices such as a server and a terminal that can be configured to execute the method provided by the embodiments of the present application. In other words, the AI-driven embodied intelligent adaptive learning method can be executed by software or hardware installed on a terminal device or a server device, and the software can be a blockchain platform. The server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, etc.

[0023] Reference Figure 1 FIG2 is a flow chart of an AI-driven embodied intelligent adaptive learning method according to an embodiment of the present invention. In this embodiment, the AI-driven embodied intelligent adaptive learning method includes: S1. Confirm receipt of adaptive learning instructions, parse the adaptive learning instructions, obtain the learning task type, and confirm the embodied intelligent body and 6G network communication link, wherein the embodied intelligent body includes: a perception drive unit, a computing drive unit, a data drive unit, and an interaction drive unit.

[0024] It should be explained that the adaptive learning instructions are instructions issued by the embodied intelligent agent manager, which are used to start and configure the adaptive learning process of the embodied intelligent agent to adapt to different task requirements and environmental conditions. The learning task type refers to a specific intelligent task type, such as image recognition, path planning or data analysis. The 6G network communication link refers to a high-speed, low-latency communication connection link based on the sixth generation of mobile communication technology, that is, a 6G network, which is used to transmit data and instructions between the embodied intelligent agent and its management platform or other intelligent agents to ensure the efficiency and real-time nature of information interaction. The embodied intelligent agent is an intelligent system with a physical form and perception ability, which can perform learning tasks in the environment, and realize adaptive learning and intelligent decision-making through the collaborative work of its various units. The embodied intelligent agent includes a perception drive unit, a computing drive unit, a data drive unit and an interaction drive unit. For the specific application of the units, please refer to the subsequent embodiments.

[0025] S2. Use the preset sampling time interval, the perception drive unit and the 6G network communication link to collect external environmental data including multiple environmental parameters.

[0026] It should be explained that the use of a preset sampling time interval, a sensing drive unit, and a 6G network communication link to collect external environmental data including multiple environmental parameters includes: Utilize 6G network communication links and perception drive units to build a distributed perception network, and use the distributed perception network and sampling time intervals to acquire visual image data and LiDAR point cloud data; The perception-driven unit is adjusted based on the learning task type, and the adjusted perception-driven unit is used to perform real-time enhancement processing on the visual image data to obtain a preprocessed image set. The LiDAR point cloud data is then subjected to noise reduction and motion compensation processing to obtain optimized point cloud data. The preprocessed image set and optimized point cloud data are fused into multi-source data to obtain external environment data.

[0027] It should be explained that the distributed perception network is composed of multiple visual sensors, LiDAR, millimeter-wave radar and ultrasonic sensors in the perception drive unit connected through a 6G network communication link, which can realize multi-point, synchronous data collection and transmission links. Among them, the perception drive unit includes multiple environmental detection units, and the environmental detection unit includes: visual sensors, LiDAR, millimeter-wave radar and ultrasonic sensors. Adjusting the perception drive unit based on the learning task type means adjusting the parameters of the perception drive unit based on the current learning task type, for example, adjusting the exposure time and frame rate of the visual sensor, the scanning frequency of the LiDAR, and the detection range of the millimeter-wave radar, so that the perception drive unit can adapt to the requirements of the current learning task type and improve the accuracy and efficiency of external environment data collection.

[0028] Furthermore, the preset sampling time interval is a fixed time interval set according to the requirements of the learning task type and the data processing capability, for example, 10 sampling times per second.

[0029] It is understood that visual image data is environmental information collected by visual sensors and represented in the form of two-dimensional or three-dimensional images, including scene details and object appearance features. Real-time enhancement processing refers to the operation of instantly optimizing and improving the visual image data obtained by the perception-driven unit during the external environment data collection process, including adjusting the image brightness, contrast, and sharpness, and removing noise to obtain clearer, more accurate, and easier to subsequently process and analyze image data. The preprocessed image set refers to the collection of visual image data that has undergone real-time enhancement processing.

[0030] Furthermore, LiDAR point cloud data is collected by a laser radar (LiDAR) sensor and represents environmental distance and shape information in the form of a three-dimensional point cloud. Optimized point cloud data refers to LiDAR point cloud data that has undergone noise reduction and motion compensation. These noise reduction and motion compensation processes improve the accuracy and reliability of LiDAR point cloud data. Multi-source data fusion utilizes a convolutional neural network to extract image features from a preprocessed image set, then uses a point cloud processing algorithm to extract LiDAR features from the optimized point cloud data. These image and LiDAR features are then fused at the feature level to generate data representing a more comprehensive and accurate representation of the environment. Optionally, the VoxelNet algorithm is used as the point cloud processing algorithm. External environmental data, collected by the perception-drive unit and used to describe and analyze the environment surrounding the embodied intelligent agent, includes visual image data and LiDAR point cloud data. Optionally, high-pass filtering is used to reduce noise on the LiDAR point cloud data, and a motion compensation algorithm modified using a Gaussian mixture model is used to compensate for motion on the LiDAR point cloud data.

[0031] For example, in intelligent driving scenarios, multi-source data fusion typically involves combining visual image data and LiDAR point cloud data, aligning, correcting, and fusing these data through algorithms to improve perception of the surrounding environment. The purpose of multi-source data fusion is to compensate for the shortcomings of a single data source, such as the disadvantages of visual data in low light conditions or the limitations of LiDAR data in object recognition, thereby providing a more reliable information foundation for subsequent decision-making and control. S3. Use the external environment data and the learning task type to retrieve the target benchmark template from the pre-built prior knowledge base, use the target benchmark template to initialize each of the multiple environmental parameters corresponding to the external environment data, obtain an initialized parameter set, and pass the initialized parameter set to the computing drive unit, wherein the prior knowledge base includes multiple benchmark templates.

[0032] Furthermore, the target reference template is retrieved from a pre-built prior knowledge base using the external environment data and the learning task type, and each of the multiple environment parameters corresponding to the external environment data is initialized using the target reference template to obtain an initialization parameter set, including: The pre-confirmed edge computing module is used to analyze the environmental parameters in the external environment data to obtain the current task encoding vector, current environment feature fingerprint, and device information corresponding to the environmental parameters. The following operations are performed on each of the multiple reference templates: Extracting a task encoding vector of the reference template, and calculating task type similarity using the current task encoding vector, the task encoding vector, and a pre-established task type similarity calculation formula; Extracting the environmental feature fingerprint of the reference template, and calculating the environmental similarity using the current environmental feature fingerprint, the environmental feature fingerprint, and a pre-built environmental similarity calculation formula; Compare device information with the pre-built device compatibility table to obtain hardware compatibility; The overall matching degree is calculated using task type similarity, environment similarity, and hardware compatibility. The calculation formula is as follows: in, is the total matching degree, is the task type similarity, Indicates the environmental similarity, For hardware compatibility, All are weight coefficients; Summarize the total matching degrees to obtain a total matching degree set, sort the total matching degrees in the total matching degree set in descending order to obtain a total matching degree sequence, identify five candidate reference templates in the total matching degree sequence, wherein the five candidate reference templates are reference templates corresponding to the first five total matching degrees in the total matching degree sequence, input each of the five candidate reference templates into a pre-built digital twin platform for verification, and obtain five verification parameters, and take the candidate reference template corresponding to the largest verification parameter among the five verification parameters as the target reference template; The verification parameters corresponding to the target reference template are used to initialize the environmental parameters to obtain the initialization parameters, and the initialization parameters are summarized to obtain an initialization parameter set.

[0033] It should be explained that the pre-confirmed edge computing module refers to a computing module deployed near the perception-drive unit. It performs preliminary data processing and analysis at the data source, reducing data transmission latency and improving the real-time and responsiveness of the embodied agent's adaptive learning. Optionally, the edge computing module is an NVIDIA Jetson AGX Xavier. Other technologies can achieve the same effect and are not described here. Environmental parameters refer to various information used to describe and analyze the embodied agent's surroundings, building the embodied agent's understanding of the surrounding environment. These include temperature, humidity, light intensity, object position, velocity, acceleration, etc., and are primarily derived from visual image data and LiDAR point cloud data. The current task encoding vector is a high-dimensional vector representation generated by the edge computing module after analyzing the environmental parameters based on the current learning task type. The edge computing module first extracts key features related to the learning task from the environmental parameters, such as object position and velocity. These key features serve as input conditions or influencing factors for completing the learning task. Target information is also defined based on the learning task's objectives, such as object classification or path planning. The key features and target information are integrated to form a vector describing the current learning task type, i.e., the current task encoding vector. Optionally, the vector is formed using a convolutional neural network algorithm.

[0034] For example, when the learning task type is image classification, key features include image size (e.g., 224×224 pixels), color space (e.g., RGB), texture features, and edge information. Target information includes classification categories (e.g., animals, plants, vehicles, etc.) or specific labels (e.g., cats, dogs, cars, airplanes, etc.). When the learning task is autonomous driving, key features include vehicle speed, acceleration, distance to the vehicle ahead, lane position, and traffic sign type. Target information includes safe navigation (e.g., collision avoidance, lane maintenance), path planning (e.g., optimal path from start to finish), and traffic signal recognition (e.g., red, green, and yellow lights).

[0035] It should be understood that the current environmental feature fingerprint is a vector representation of the embodied agent's environment based on environmental parameters. It is generated by extracting and fusing task features from multiple environmental parameters. It can reflect the unique properties and state of the environment. Optionally, the vector is formed using a convolutional neural network algorithm. Task features refer to environmental attributes or conditions related to a specific learning task and can reflect the specific requirements of the learning task on the environment. For example, in autonomous driving tasks, these include road curvature, lane clarity, and traffic sign visibility. Device information refers to the detailed specifications and status information of various hardware devices involved in data collection and processing, including device model, sensor parameters, computing power, storage capacity, etc., reflecting the capabilities and limitations of the devices. The pre-built prior knowledge base is a knowledge base that has been pre-established and stored with multiple benchmark templates. A benchmark template is a knowledge unit template within the pre-built prior knowledge base and represents benchmark parameter configurations for different tasks and environments. It contains parameter configurations and behavior patterns for specific task types, environmental characteristics, and device requirements. Each benchmark template has a corresponding task encoding vector, environmental feature fingerprint, and other attributes. The purpose of the benchmark template is to provide a standard for parameter initialization of the embodied agent in different tasks and environments. By matching the current task and environment, the most appropriate benchmark template is selected and its parameter configuration is used to initialize the embodied agent's behavior, enabling it to quickly adapt to new tasks and environments. Task type similarity measures the similarity between the current learning task type and the corresponding learning task type of the benchmark template. By quantifying the similarity between tasks, the best matching template for the current task is selected from multiple benchmark templates. Combining environment similarity and hardware compatibility ensures that the selected benchmark template meets current requirements in terms of task type, environment adaptability, and device compatibility, thereby providing the optimal parameter initialization solution for the agent. Environment similarity is a metric that quantifies the degree of similarity between the current environment and the benchmark template environment in terms of features or conditions. Hardware compatibility is a metric that assesses the hardware compatibility between the current device and the benchmark template. High hardware compatibility indicates that the device's hardware specifications (such as computing power, sensor type, and storage capacity) meet the requirements of the learning task, ensuring smooth task execution. For example, if the task requires high-performance computing but the current device's computing power is insufficient, low hardware compatibility may indicate that the user needs to upgrade the device or adjust the task parameters.

[0036] Furthermore, the task type similarity calculation formula is as follows: in, is the task type similarity, Encode the vector for the current task, is the task encoding vector, Represents the modulo operation; The formula for calculating the environmental similarity is as follows: in, Indicates the environmental similarity, Indicates the current environment feature fingerprint, Represents the environmental feature fingerprint, is the fingerprint length, It represents the Hamming distance calculation function that takes the current environment feature fingerprint and the environment feature fingerprint as calculation parameters; It should be explained that the task encoding vector is a high-dimensional vector representation used to uniquely identify and describe the learning task corresponding to the benchmark template. The environmental feature fingerprint of the benchmark template is a vector representation of the environment to which the benchmark template is applicable. Optionally, the vector is formed using a convolutional neural network algorithm. The pre-built device compatibility comparison table is a pre-established and stored table that lists the compatibility information of different devices with each benchmark template. It records in detail the degree of adaptation of each device model, sensor parameters, computing power, storage capacity and other hardware specifications to each benchmark template. The device compatibility comparison table is as follows: It should be noted that a full device model match means that the current device's hardware model is exactly the same as the hardware model specified in the baseline template. For example, if the baseline template requires the use of an NVIDIA Jetson AGX Xavier and the current device is also an NVIDIA Jetson AGX Xavier, the hardware compatibility is 1. Same model, different version means that the current device's hardware model and the hardware model specified in the baseline template are from the same series, but different versions. For example, if the baseline template requires the use of an NVIDIA Jetson AGX Xavier and the current device is an NVIDIA Jetson TX2 (same series but different version), the hardware compatibility is 0.8. Cross-platform compatibility mode means that the current device's hardware model is not from the same series as the hardware model specified in the baseline template, but the device is cross-platform compatible. In this case, the device's hardware architecture and performance characteristics differ significantly from those required by the baseline template, but through software optimization and adaptation, it can still perform the task. For example, if the baseline template requires the use of an NVIDIA Jetson AGX Xavier and the current device is a Raspberry Pi 4B (a completely different platform), the hardware compatibility is 0.3.

[0037] Furthermore, the fingerprint length is determined during the task feature generation process by extracting and fusing multiple environmental parameters. It is equal to the dimension of the environmental feature fingerprint vector. The dimension of an environmental feature fingerprint vector refers to the number of features contained in the vector, each of which represents a quantitative representation of the environment. The dimension of the environmental feature fingerprint vector is determined by the selected feature set, which is pre-set based on the requirements of the learning task and the environmental data.

[0038] For example, in the autonomous driving task, the following five features are selected to construct the environmental feature fingerprint vector: road curvature, lane line clarity, traffic sign visibility, distance to the preceding vehicle, and vehicle speed. The five features are integrated into a five-dimensional vector, that is, the dimension of the environmental feature fingerprint vector is 5, that is, the fingerprint length is 5.

[0039] Furthermore, the total matching degree is an indicator used to comprehensively evaluate the degree of matching between the current task and the benchmark template in terms of task type, environmental characteristics and hardware compatibility. It is calculated by combining task type similarity, environmental similarity and hardware compatibility in a weighted summation manner. The total matching degree dataset is a collection of the total matching degrees of all benchmark templates. It contains the total matching degree value of each benchmark template, which is used for subsequent sorting and filtering operations. The total matching degree sequence is a sequence formed by arranging the total matching degrees in the total matching degree dataset in order from large to small. The candidate benchmark templates are several benchmark templates with the highest total matching degrees selected from the total matching degree sequence. The pre-built digital twin platform is a virtual simulation environment used to simulate and verify the performance of the benchmark template in the actual environment. Optionally, NVIDIA Omniverse is used as the digital twin platform. Other technologies can achieve the same effect and will not be described here.

[0040] Furthermore, inputting each of the five candidate benchmark templates into a pre-built digital twin platform for verification refers to loading each candidate benchmark template into the pre-built digital twin platform, setting up a test environment within the digital twin platform that resembles the actual application scenario, including but not limited to environmental parameters, task type, and hardware configuration. Each candidate benchmark template is simulated and its performance in the simulated environment is recorded, including task completion status and performance indicators. Based on the simulation results, verification parameters for each candidate benchmark template are calculated. Optionally, a performance evaluation model and data analysis algorithm are used to calculate the verification parameters. Verification parameters are evaluation parameters obtained by verifying the candidate benchmark templates within the digital twin platform. During the parameter initialization phase, the verification parameters are used to initially configure the environmental parameters. The target benchmark template is the template that best matches the current task and environment, retrieved from a priori knowledge base based on the current learning task type and external environment data. Parameter initialization is the process of initially configuring various parameters during the external environment data processing process using the parameter values ​​in the target benchmark template. Initialization parameters are parameters obtained through the initialization process. Parameter values ​​are set based on the configuration information in the target benchmark template. The initialization parameter set is a collection of all initialization parameters.

[0041] For example, it is assumed that the order of parameters defined in the target reference template is {temperature, humidity, speed}, and the corresponding parameter values ​​are {25 , 60%, 50km / h}, in the parameter initialization process, according to the configuration information of the target benchmark template, the parameters in the external environment data processing process are initially configured to obtain the initialization parameter set. The order and value of the parameters in the initialization parameter set are consistent with the target benchmark template, that is, the initialization parameter set is {temperature: 25 , humidity: 60%, speed: 50 km / h}. The purpose of parameter initialization is to ensure that the external environment data processing process can be initialized and configured according to the requirements of the target benchmark template, providing a basis for subsequent processing and optimization.

[0042] S4. Obtain an optimization algorithm based on the initialization parameters in the initialization parameter set, use the optimization algorithm to optimize the initialization parameters to obtain optimization parameter nodes, wherein the optimization parameter nodes include optimization parameters and confidence assessment parameters, and summarize the optimization parameter nodes to obtain an optimization parameter node set.

[0043] It should be explained that the initialization parameter acquisition optimization algorithm based on the initialization parameter set, and the optimization calculation of the initialization parameter using the optimization algorithm to obtain the optimization parameter node include: Obtaining an optimization algorithm using the initialization parameters in the initialization parameter set and a pre-built algorithm mapping table; Using the optimization algorithm to perform hierarchical optimization calculations on the initialization parameters to obtain multiple edge node optimization parameters; Performing a fixed-step gradient descent optimization on the plurality of edge node optimization parameters to obtain a plurality of preliminary optimization parameters; Verifying and adjusting a plurality of preliminary optimization parameters using the digital twin platform to obtain a plurality of optimization parameters; Utilizing a pre-built confidence evaluation method, each of the multiple optimization parameters is optimized and evaluated to obtain a confidence evaluation parameter; The optimization parameters and their corresponding confidence assessment parameters are integrated to generate an optimization parameter node.

[0044] It is understood that the pre-built algorithm mapping table is a pre-established and stored lookup table of algorithms used to determine the most suitable optimization algorithm based on different initialization parameter characteristics. The algorithm mapping table is constructed based on historical data and covers the correspondence between various parameter patterns and optimization algorithms. Optionally, a decision tree algorithm is used to identify the optimization algorithm from the algorithm mapping table. Optimization calculation is a computational method that uses mathematical models to adjust initialization parameters to improve parameter performance and better meet the requirements of a specific task. It generally involves defining an objective function and constraints and finding the parameter values ​​that optimize the objective function through iterative calculations. Optimization algorithms are specific methods used for optimization calculations, such as gradient descent, genetic algorithms, and particle swarm optimization. Hierarchical optimization calculation decomposes complex optimization tasks into multiple levels or stages, with each level focusing on a specific parameter subset or optimization objective, and optimizing them step by step. For example, key parameters are optimized first to achieve the primary objective, followed by optimization of secondary parameters to refine and improve the overall effect. The purpose of using hierarchical optimization calculation is to help simplify complex optimization tasks and improve the efficiency and effectiveness of optimizing initialization parameters. Edge node optimization parameters are parameters obtained by running an optimization algorithm on an edge node in a distributed computing environment. Edge nodes are typically computing units located close to data sources or end devices, enabling local data processing and optimization to reduce data transmission latency. NVIDIA Jetson series modules can optionally be used as edge nodes.

[0045] It should be understood that fixed-step gradient descent optimization is an iterative optimization method that updates parameters in the opposite direction of the objective function's gradient over a fixed number of iterations to minimize the objective function. This method is simple and straightforward and applicable to many machine learning and engineering optimization problems. For example, in machine learning, stochastic gradient descent (SGD) is used as an iterative optimization method to update parameters by computing the gradient using a single training example or a small batch of examples in each iteration. The optimization can be performed with a fixed number of iterations (e.g., 1000). The initial optimized parameters are those obtained after the fixed-step gradient descent optimization.

[0046] It is understood that the verification and adjustment of multiple preliminary optimization parameters using the pre-built digital twin platform refers to simulating the actual application effects of these parameters in a virtual simulation environment. The effectiveness of the parameters is evaluated through simulation results and adjusted as needed to ensure that they perform well in the real environment. Verification is to evaluate the effect of the parameters by running them in a simulation environment, and adjustment is to optimize the parameters using the genetic algorithm in the digital twin platform based on the verification results to improve actual performance. The optimized parameters are the parameters obtained after verification and adjustment.

[0047] For example, in an autonomous driving scenario, initially optimized parameters perform well on straight roads, but the vehicle frequently deviates from its lane on curves. Verification on the digital twin platform reveals that the vehicle's recognition and reaction time on curves are insufficient. Based on this finding, relevant parameters are adjusted, increasing the weighting for curve recognition and optimizing the vehicle's steering control parameters to improve its performance on curves.

[0048] It should be understood that the pre-built confidence assessment method is a method for quantitatively assessing the credibility and reliability of optimization parameters. These methods are based on statistics, machine learning or other evaluation techniques, and can provide uncertainty measurements of parameter performance. Optionally, a Bayesian analysis method is used as the confidence assessment method. Optimization assessment refers to applying a confidence assessment method (such as a Bayesian analysis method) to calculate the confidence assessment parameters of the optimization parameters and quantify their uncertainty. The confidence assessment parameter is a measurement value obtained by calculating the confidence assessment of the optimization parameter, which is used to measure the credibility of the optimization parameter in actual application, and the measurement value is a value between 0 and 1. The credibility of the optimization parameter is quantified through the confidence assessment parameter, the subsequent optimization direction is clarified, and the optimization efficiency and effect are improved.

[0049] It should be explained that integrating the optimization parameters and their corresponding confidence assessment parameters to generate an optimization parameter node refers to creating a data structure containing multiple optimization parameters and their corresponding confidence assessment parameters. This structure can uniformly represent the optimization parameters and their corresponding confidence assessment parameters, facilitating subsequent processing and analysis. An optimization parameter node set is the collection of all optimization parameter nodes.

[0050] S5. Extracting an optimization parameter set and a confidence evaluation parameter set from the optimization parameter node set of the data-driven unit, adjusting the optimization parameter set using the prior knowledge base and the confidence evaluation parameter set, and obtaining a global parameter set.

[0051] It should be explained that the optimization parameter set is adjusted using the prior knowledge base and the confidence evaluation parameter set to obtain the global parameter set, including: According to the confidence evaluation parameters in the confidence evaluation parameter set, the optimization parameters in the optimization parameter set are divided into three division parameters, and the division parameters are high confidence parameters, medium confidence parameters or low confidence parameters, and the low confidence parameters are reinitialized using the prior knowledge base to obtain the corrected parameters; Fine-tune the confidence parameters using the prior knowledge base to obtain fine-tuning parameters; Obtaining a global optimization strategy based on the correction parameters, fine-tuning parameters, high-confidence parameters and a priori knowledge base; The correction parameters, fine-tuning parameters and high-confidence parameters are integrated using the global optimization strategy to obtain a global parameter set.

[0052] It should be explained that the SQL query method is used as a method for extracting the optimization parameter set and the confidence assessment parameter set from the optimization parameter node set of the data-driven unit. The confidence assessment parameter set is a set containing multiple confidence assessment parameters extracted from the optimization parameter node set. The optimization parameter set is a set containing multiple optimization parameters extracted from the optimization parameter node set.

[0053] It is understandable that the parameter division is to divide the corresponding optimization parameters into high confidence parameters, medium confidence parameters or low confidence parameters according to the specific values ​​of the confidence assessment parameters. Among them, the high confidence parameter is the optimization parameter whose corresponding confidence assessment parameter is greater than or equal to 0.9, the medium confidence parameter is the optimization parameter whose corresponding confidence assessment parameter is between 0.7 and 0.9, and the low confidence parameter is the optimization parameter whose corresponding confidence assessment parameter is less than or equal to 0.7. Reinitializing the low confidence parameters using the prior knowledge base to obtain the corrected parameters means that a large amount of historical data and empirical models stored in the prior knowledge base will be called, and by matching the characteristics of the low confidence parameters, the successful parameter configurations in similar scenarios are extracted from the prior knowledge base, and the low confidence parameters are reinitialized based on this. Reinitialization refers to the operation of re-initializing the low confidence parameters based on the successful parameter configurations obtained in similar scenarios. Fine-tuning the medium-confidence parameters using the prior knowledge base. Fine-tuning parameters refers to the prior knowledge base providing the direction and range for fine-tuning. Based on experience adjusting parameters in similar tasks and environments, the medium-confidence parameters are fine-tuned. Correction parameters are those obtained by reinitializing the low-confidence parameters. Fine-tuning parameters are those obtained by fine-tuning the medium-confidence parameters. The purpose of reinitializing the low-confidence parameters is to improve their fit with the current task and environment. By incorporating successful experiences from the prior knowledge base, more reasonable initial values ​​are provided for the low-confidence parameters. Fine-tuning the medium-confidence parameters aims to further optimize the medium-confidence parameters based on the existing foundation, making them more precisely adapted to the current task requirements and improving performance.

[0054] For example, in an autonomous driving task, the optimization parameters include speed, steering angle, and acceleration. After confidence evaluation, the confidence evaluation parameter corresponding to speed is 0.65, the confidence evaluation parameter corresponding to steering angle is 0.75, and the confidence evaluation parameter corresponding to throttle opening is 0.92. Based on the confidence evaluation parameters, the optimization parameters are divided into three categories: high-confidence parameter: throttle opening (0.92), medium-confidence parameter: steering angle (0.75), and low-confidence parameter: speed (0.65). Assume that successful parameter configurations for similar scenarios (in urban road environments) stored in the prior knowledge base indicate that the speed is typically between 30 and 50 km / h. The current low-confidence speed parameter is 45 km / h (confidence level 0.65). Based on the prior knowledge base, the speed is reinitialized to 40 km / h. In similar scenarios, the steering angle typically requires fine-tuning based on road curvature. The current steering angle is 30 degrees (confidence level 0.75). Based on the prior knowledge base, the steering angle is fine-tuned to 32 degrees. The throttle opening is a high-confidence parameter (confidence 0.92) and is directly included in the global parameter set without adjustment.

[0055] Furthermore, obtaining a global optimization strategy based on the correction parameters, fine-tuning parameters, high-confidence parameters, and a priori knowledge base includes utilizing historical data from the priori knowledge base and a pre-built learning model. Optionally, a convolutional neural network learning model is employed as the learning model. Parameters from the historical data serve as inputs to the learning model, and their corresponding system performance indicators serve as outputs. The relationship between the learning parameters and the optimization effect is then learned to train the learning model. The correction parameters, fine-tuning parameters, and high-confidence parameters are input into the trained learning model, and the optimal parameter combination and its corresponding system performance indicator are output. This serves as the basis for forming a global optimization strategy. The system performance indicator is a quantitative metric derived from the evaluation results of the parameters input into the learning model. Historical data refers to data collected during past related learning tasks. The parameter combination refers to the combination obtained by inputting the correction parameters, fine-tuning parameters, and high-confidence parameters into the trained learning model, where the learning model weights the correction parameters, fine-tuning parameters, and high-confidence parameters according to different weights. The global optimization strategy is an optimization strategy formed by determining the weights for each input parameter type and its corresponding system performance indicator. The input parameters are correction parameters, fine-tuning parameters, and high-confidence parameters that are input into the learning model. By rationally utilizing parameters of different confidence levels, the global optimization strategy can optimize resource allocation, avoid resource waste, and improve resource utilization efficiency.

[0056] Furthermore, the global optimization strategy is used to integrate the correction parameters, fine-tuning parameters, and high-confidence parameters to obtain a global parameter set. This means weighting the correction parameters, fine-tuning parameters, and high-confidence parameters according to their weights determined in the global optimization strategy, and summarizing the weighted correction parameters, fine-tuning parameters, and high-confidence parameters to obtain a global parameter set. The global parameter set is a set of multiple parameters obtained by summarizing the weighted correction parameters, fine-tuning parameters, and high-confidence parameters. By integrating parameters of different confidence levels, the embodied intelligent agent can better cope with various complex environments and task requirements, thereby enhancing the robustness and adaptability of the embodied intelligent agent.

[0057] S6. Generate a control instruction set using the global parameter set.

[0058] It should be explained that the generation of the control instruction set using the global parameter set includes: generating a preliminary control instruction set based on the global parameter set; Performing syntax verification and semantic verification on each of the preliminary control instructions in the preliminary control instruction set using an interactive driving unit; If the grammatical verification of the preliminary control instruction fails, revising the preliminary control instruction according to the pre-established grammatical rules to obtain a grammatical revision instruction; If the syntax verification of the preliminary control instruction passes, the preliminary control instruction is regarded as a syntax qualified instruction; If the semantic verification of the preliminary control instruction is unclear, semantically optimizing the preliminary control instruction according to the pre-built task execution rule to obtain a semantically optimized instruction; If the semantic verification of the preliminary control instruction is clear, the preliminary control instruction is regarded as a semantically qualified instruction; Integrate the grammatically qualified instructions, the semantically qualified instructions, the grammatically revised instructions, and the semantically optimized instructions to obtain a verified instruction set; Perform similarity comparison using the verified instruction set, the pre-confirmed historical successful instruction set, and the pre-built structural similarity algorithm to obtain a similarity evaluation value; If the similarity evaluation value is less than a preset evaluation threshold, the verified instruction set is adjusted based on the learning task type to obtain an adjusted instruction set; Using the adjusted instruction set as the verified instruction set, and returning to the step of performing similarity comparison using the verified instruction set, the pre-confirmed historical successful instruction set, and the pre-built structural similarity algorithm to obtain a similarity evaluation value, until the similarity evaluation value is greater than or equal to an evaluation threshold; If the similarity evaluation value is greater than or equal to the evaluation threshold, the verified instruction set is used as a candidate control instruction set; The candidate control instruction set is pre-executed using a pre-built simulation execution environment, and after confirming that the effect of the pre-execution is the preset expected effect, the candidate control instruction set is determined as the control instruction set.

[0059] It should be explained that generating a preliminary control instruction set based on a global parameter set refers to generating the preliminary control instruction set based on the global parameter set using a pre-built parameter-instruction mapping model in the interactive driving unit. Optionally, an LSTM sequence generation model can be used as the parameter-instruction mapping model to convert the global parameter set into a preliminary control instruction set executable by the embodied intelligent agent.

[0060] Furthermore, the syntax verification is based on pre-built syntax rules, and optionally, the instruction structure rules defined by the BNF paradigm are used as the syntax rules to check the compliance of the preliminary control instructions. If there is a syntax error, such as missing operation objects or action parameters, the syntax reviser is called to automatically complete the completion and generate syntax revision instructions. If the syntax is compliant, it is marked as a syntax qualified instruction. Among them, the syntax reviser is an automated tool for fixing syntax errors in preliminary control instructions. It analyzes the preliminary control instructions based on pre-built syntax rules (such as instruction structure rules defined by the BNF paradigm). If a syntax error is found (such as missing operation objects or action parameters), the syntax reviser will automatically complete the missing part or correct the erroneous structure to generate a syntax revision instruction. Syntax revision instructions are instructions that have been corrected by the syntax reviser, while syntax qualified instructions are instructions that have not been revised and directly passed the syntax verification.

[0061] It can be understood that semantic verification is to check the executableness of instructions based on pre-built task execution rules. If the semantics are ambiguous, semantic optimization is performed to generate semantically optimized instructions; if the semantics are clear, they are marked as semantically qualified instructions. Semantic optimization is the operation of using the perception-driven unit to refine instructions when the semantics of instructions are ambiguous. For example, when the execution target is not specified in the preliminary control instructions, the perception-driven unit is used to clarify the target. Semantically optimized instructions are instructions obtained after semantic optimization of preliminary control instructions that have been semantically verified as semantically ambiguous. Semantically qualified instructions are preliminary control instructions that have been confirmed to be qualified after semantic verification. The verified instruction set is a set of multiple instructions obtained by summarizing syntactically qualified instructions, semantically qualified instructions, syntactically revised instructions, and semantically optimized instructions. Syntactic verification and semantic verification can effectively reduce errors in preliminary control instructions, improve the accuracy and reliability of preliminary control instructions, and ensure that preliminary control instructions can correctly guide the behavior of the embodied intelligent agent.

[0062] Furthermore, task execution rules are a set of predefined conditions and constraints based on the specific learning task type, used to determine whether instructions are sufficiently clear for execution. Task execution rules include the clarity of the operation object, the completeness of the operation conditions, the logic of the operation sequence, and the adaptability to the environment.

[0063] It should be explained that the pre-verified historically successful instruction set is a set of instructions extracted from a priori knowledge base and verified in similar learning tasks. Optionally, a cosine similarity algorithm is used as a structural similarity algorithm to quantify the similarity between the verified instruction set and the historically successful instruction set. Similarity comparison is a method used to evaluate the similarity between the verified instruction set and the historically successful instruction set. The similarity evaluation value is a number between 0 and 1, indicating the degree of similarity between the two, with higher values ​​indicating greater similarity. The preset evaluation threshold is a pre-set similarity evaluation standard. Optionally, the evaluation threshold is set to 0.8 here to determine whether the verified instruction set is sufficiently similar to the historically successful instruction set. Targeted adjustment involves revising the verified instruction set based on the learning task type, such as adjusting the instruction order, to improve the similarity evaluation value. An adjusted instruction set is a set obtained by targeted adjustment of a verified instruction set whose similarity evaluation value is less than the evaluation threshold. A candidate control instruction set is a verified instruction set whose similarity evaluation value is greater than the evaluation threshold.

[0064] Furthermore, the ROS simulation platform can be used as a simulation execution environment to simulate the execution environment of the embodied intelligent body to verify the effect of the candidate control instruction set. Pre-execution refers to the simulation of the candidate control instruction set in the model execution environment. The preset expected effect is the expected performance preset according to the learning task type, such as image recognition accuracy ≥ 95%, and robotic arm grasping success rate ≥ 90%. If the effect of the pre-execution meets the expected effect, the candidate control instruction set is determined as the control instruction set. The control instruction set is a collection of multiple control instructions that drive the embodied intelligent body. Pre-executing the candidate control instruction set in the simulated execution environment can detect potential problems and errors in advance, avoid executing erroneous instructions in the actual environment, and reduce the risk of system crashes or task failures.

[0065] S7. Use the control instruction set to drive the embodied intelligent body, and use the perception drive unit to collect multiple execution result data of the embodied intelligent body.

[0066] It should be explained that the use of the control instruction set to drive the embodied intelligent body and the use of the perception drive unit to collect multiple execution result data of the embodied intelligent body include: Using a perception drive unit and a preset collection frequency to collect data when driving the embodied intelligent body, to obtain a plurality of raw execution data; performing data preprocessing on each of the plurality of original execution data to obtain a plurality of preprocessed execution data; Performing data cleaning on each of the plurality of pre-processed execution data to obtain a plurality of accurate execution data; Feature extraction is performed on each of the multiple accurate execution data to obtain multiple execution result data.

[0067] It should be explained that the preset acquisition frequency is a fixed frequency set based on the type of learning task. For example, in an image recognition task, it can be set to 30 frames per second, and in a robotic arm control task, it can be set to 100 frames per second. The multiple raw execution data are generated when the embodied intelligent agent executes the control instruction set, including sensor data and actuator feedback data. Sensor data includes visual sensor data, distance sensor data, and environmental sensor data. Visual sensor data comes from visual sensors such as cameras and can be two-dimensional or three-dimensional image data, containing environmental scene details and object appearance features. For example, it is used for environmental visual information in image recognition or visual navigation tasks. Distance sensor data comes from sensors such as LiDAR, millimeter-wave radar, or ultrasonic sensors and is presented in the form of point clouds or distance values. It provides distance and shape information of environmental objects and is used for tasks such as ranging, obstacle avoidance, and spatial perception. Environmental sensor data comes from temperature and humidity sensors, air pressure sensors, and light sensors. It provides information on the physical state of the environment, such as temperature, humidity, air pressure, and light intensity, for environmental perception. The preset acquisition frequency ensures data continuity and integrity, preventing data loss or incomplete acquisition.

[0068] Furthermore, actuator feedback data includes position feedback data and velocity feedback data. Position feedback data originates from position sensors and provides information about the embodied agent's current position, such as the joint angles of a robotic arm or the displacement of a mobile platform, for precise control and trajectory planning. Velocity feedback data, calculated by position sensors, reflects the embodied agent's current motion speed and is used for speed control and motion coordination.

[0069] It should be explained that data preprocessing is the preliminary processing of the original execution data, for example, removing invalid data, filling missing data, etc. Preprocessed execution data is data that has undergone preliminary processing, which improves data quality and availability. Optionally, Trifacta is used to perform data preprocessing on the original execution data to identify and filter out original execution data that does not conform to the format or is obviously erroneous, thereby obtaining preprocessed execution data. Data cleaning is to further remove noise and outliers from the preprocessed execution data. Accurate execution data is high-quality data after data cleaning. Optionally, OpenRefine tools are used for data cleaning to improve the accuracy and reliability of the data by removing or correcting erroneous data, missing values, and outliers, thereby obtaining accurate execution data. Feature extraction refers to extracting core features from accurate execution data, such as image features or motion features. Optionally, a convolutional neural network is used as the feature extraction method. Among them, core features refer to attributes or patterns that have significant discrimination and representativeness in the data, which can effectively characterize the essence and main information of the data. For example, in image data, core features can be edges, textures, shapes, and colors; in motion data, core features can be position, velocity, acceleration, and angular velocity. Execution result data is derived by extracting features from a single accurate execution data set and is used for subsequent analysis and decision-making. Feature extraction on multiple accurate execution data sets will yield multiple execution result data sets. By extracting key features, data dimensionality and complexity are reduced, improving processing speed.

[0070] S8. Transmit multiple execution result data to the data driving unit based on the 6G network communication link, and generate a detection report according to the multiple execution result data in the data driving unit.

[0071] It should be explained that the transmitting of multiple execution result data to the data driving unit based on the 6G network communication link and the generating of the detection report according to the multiple execution result data in the data driving unit include: Transmitting the plurality of execution result data to a data driving unit using a 6G network communication link to obtain a plurality of transmitted execution result data; Performing data accuracy verification on the multiple execution result data after transmission to obtain a check result, wherein the check result includes a check feasibility and a check infeasibility; After confirming that the inspection result is feasible, the plurality of execution result data are compared and analyzed with a plurality of preset qualified performance data to obtain an analysis result report; The analysis result report is diagnosed for abnormality using preset abnormality diagnosis rules to obtain a detection report.

[0072] Furthermore, the data accuracy verification check is to verify the execution result data after transmission to ensure that the data is not damaged or lost during the transmission process. Optionally, CRC verification technology is used for verification and inspection. The inspection results are divided into two types: if the data is complete and accurate, the inspection is feasible; if the data is damaged or lost, the inspection is not feasible. Through the data accuracy verification check, it can be confirmed whether the data after transmission is complete. During the data transmission process, data may be lost or damaged due to network problems or storage medium failures. The data accuracy verification check can detect these problems in time and ensure the integrity of the data. After confirming that the data accuracy verification check has passed, the multiple execution result data are compared and analyzed with the preset multiple qualified performance data. Optionally, a statistical analysis method is used to compare and analyze the multiple execution result data with the preset multiple qualified performance data.

[0073] It should be explained that the preset qualified performance data are based on the performance standards set for the learning task, such as image recognition accuracy ≥ 95%, robotic arm grasping success rate ≥ 90%, etc. The comparison results are used to form an analysis report. The analysis report includes a comparison of the actual performance indicators with the qualified performance data.

[0074] It is understood that the preset anomaly diagnosis rules are logical rules used to identify anomalies in the analysis result report. Based on these logical rules, the analysis result report is evaluated to determine whether the embodied intelligent agent has performance anomalies or faults. For example, if the image recognition accuracy rate falls below 95%, an anomaly alarm is triggered; if the robotic arm's grasping success rate falls below 90% multiple times in a row, it is determined that the robotic arm may have a fault. The detection report is the result of an anomaly diagnosis of the analysis result report using the anomaly diagnosis rules.

[0075] S9. Update the prior knowledge base according to the detection report to obtain an optimized prior knowledge base.

[0076] It should be explained that the updating of the prior knowledge base according to the test report to obtain an optimized prior knowledge base includes: Extract feedback data from the test report; Using the feedback data to update the reference template, an updated reference template is obtained; The updated benchmark template is integrated into the prior knowledge base to obtain an optimized prior knowledge base.

[0077] Furthermore, optionally, an SQL query method is used as the method for extracting feedback data. Feedback data is key information extracted from the test report, wherein the key information includes the accuracy or success rate of the embodied intelligent agent when executing control instructions, execution errors, and targeted improvement suggestions.

[0078] It should be explained that using feedback data to update the baseline template means using feedback data to update the parameter configuration and behavior pattern in the baseline template to obtain an updated baseline template, so that it is more in line with actual task requirements and can provide more accurate initial parameter configuration for subsequent tasks.

[0079] Furthermore, integrating the updated benchmark template into the prior knowledge base means adding the updated benchmark template to the prior knowledge base to obtain an optimized prior knowledge base, making the knowledge system of the prior knowledge base richer and more complete, and being able to provide stronger support for the continuous optimization and performance improvement of the embodied intelligent body.

[0080] The present invention solves the problems described in the background technology. The present invention confirms the receipt of adaptive learning instructions, parses the adaptive learning instructions, obtains the learning task type, and confirms the embodied intelligent body and the 6G network communication link. The embodied intelligent body includes: a perception drive unit, a calculation drive unit, a data drive unit, and an interaction drive unit. It can be seen that the embodiment of the present invention can realize real-time and efficient data interaction between the embodied intelligent body and the management platform through the high speed and low latency characteristics of the 6G network communication link, ensuring the rapid transmission and execution of task instructions. On this basis, the present invention uses a preset sampling time interval, a perception drive unit, and a 6G network communication link to collect external environmental data including multiple environmental parameters, uses the external environmental data and the learning task type to retrieve a target reference template from a pre-built prior knowledge base, uses the target reference template to initialize each of the multiple environmental parameters corresponding to the external environmental data, obtains an initialization parameter set, and transmits the initialization parameter set to the calculation drive unit. The prior knowledge base includes multiple reference templates. It can be seen that the embodiment of the present invention can ensure the timeliness and accuracy of the external environmental data and provide a reliable data foundation for subsequent processing. Furthermore, the present invention obtains an optimization algorithm based on the initialization parameters in the initialization parameter set, optimizes and calculates the initialization parameters using the optimization algorithm, and obtains optimization parameter nodes, wherein the optimization parameter nodes include optimization parameters and confidence assessment parameters, and summarizes the optimization parameter nodes to obtain an optimization parameter node set. It can be seen that the embodiment of the present invention uses the optimization algorithm to efficiently optimize the initialization parameters, generates an optimization parameter node set including optimization parameters and confidence assessment parameters, and significantly improves the efficiency and accuracy of parameter optimization. Next, the present invention transmits the optimization parameter node set to the data-driven unit, extracts the optimization parameter set and the confidence assessment parameter set from the optimization parameter node set of the data-driven unit, and uses the prior knowledge base and the confidence assessment parameter set to adjust the parameters of the optimization parameter set to obtain a global parameter set, and transmits the global parameter set to the interactive driving unit. It can be seen that the embodiment of the present invention adjusts the optimization parameters, integrates the prior knowledge base and the confidence assessment parameter set, and generates a global parameter set, which significantly improves the accuracy and reliability of the parameters. Furthermore, the present invention uses a global parameter set to generate a control instruction set, uses the control instruction set to drive the embodied intelligent body, and uses a perception drive unit to collect multiple execution result data of the embodied intelligent body, transmits the multiple execution result data to the data drive unit based on the 6G network communication link, generates a detection report based on the multiple execution result data in the data drive unit, updates the prior knowledge base based on the detection report, and obtains an optimized prior knowledge base. It can be seen that the embodiment of the present invention uses a feedback mechanism to update the prior knowledge base based on the execution result data, realizes adaptive optimization, and significantly improves the accuracy and efficiency of task execution. Therefore, the present invention can improve the decision-making ability and task execution efficiency of the embodied intelligent body in the adaptive learning process.

[0081] like Figure 2 , which is a functional module diagram of an AI-driven embodied intelligent adaptive learning system provided by one embodiment of the present invention.

[0082] The AI-driven embodied intelligent adaptive learning system 100 of the present invention can be installed in an electronic device. Depending on the functions implemented, the AI-driven embodied intelligent adaptive learning system 100 can include a learning task analysis module 101, a parameter optimization module 102, a parameter adjustment module 103, and an interactive feedback module 104. The module of the present invention can also be referred to as a unit, which refers to a series of computer program segments that can be executed by an electronic device processor and can perform fixed functions, which are stored in the memory of the electronic device.

[0083] The learning task parsing module 101 is used to confirm the receipt of the adaptive learning instruction, parse the adaptive learning instruction, obtain the learning task type, and confirm the embodied intelligent agent and the 6G network communication link, wherein the embodied intelligent agent includes: a perception drive unit, a calculation drive unit, a data drive unit, and an interaction drive unit; The parameter optimization module 102 is configured to collect external environmental data including multiple environmental parameters using a preset sampling time interval, a perception drive unit, and a 6G network communication link, retrieve a target reference template from a pre-built priori knowledge base using the external environmental data and a learning task type, initialize each of the multiple environmental parameters corresponding to the external environmental data using the target reference template to obtain an initialized parameter set, and transmit the initialized parameter set to the calculation drive unit, wherein the priori knowledge base includes multiple reference templates; Acquire an optimization algorithm based on the initialization parameters in the initialization parameter set, optimize and calculate the initialization parameters using the optimization algorithm to obtain optimization parameter nodes, wherein the optimization parameter nodes include optimization parameters and confidence evaluation parameters, and summarize the optimization parameter nodes to obtain an optimization parameter node set; The parameter adjustment module 103 is used to transmit the optimization parameter node set to the data driving unit, extract the optimization parameter set and the confidence evaluation parameter set from the optimization parameter node set of the data driving unit, adjust the optimization parameter set using the prior knowledge base and the confidence evaluation parameter set to obtain a global parameter set, and transmit the global parameter set to the interactive driving unit; The interactive feedback module 104 is used to generate a control instruction set using a global parameter set, drive the embodied intelligent body using the control instruction set, and use the perception drive unit to collect multiple execution result data of the embodied intelligent body, transmit the multiple execution result data to the data drive unit based on the 6G network communication link, generate a detection report based on the multiple execution result data in the data drive unit, and update the prior knowledge base based on the detection report to obtain an optimized prior knowledge base.

[0084] In detail, the modules in the AI-driven embodied intelligent adaptive learning system 100 in the embodiment of the present invention are used in the same manner as above. Figure 1 The technical means are the same as the AI-driven embodied intelligent adaptive learning method described in , and can produce the same technical effects, so I will not go into details here.

[0085] like Figure 3 , which is a structural diagram of an electronic device for implementing an AI-driven embodied intelligent adaptive learning method provided by one embodiment of the present invention.

[0086] The electronic device 1 may include a processor 10, a memory 11 and a bus 12, and may also include a computer program stored in the memory 11 and executable on the processor 10, such as an AI-driven embodied intelligent adaptive learning method program.

[0087] The memory 11 includes at least one type of readable storage medium, including flash memory, a mobile hard disk, a multimedia card, a card-type memory (e.g., SD or DX memory), a magnetic memory, a magnetic disk, an optical disk, etc. In some embodiments, the memory 11 may be an internal storage unit of the electronic device 1, such as a mobile hard disk of the electronic device 1. In other embodiments, the memory 11 may also be an external storage device of the electronic device 1, such as a plug-in mobile hard disk, a smart media card (SMC), a secure digital (SD) card, a flash memory card, etc. equipped on the electronic device 1. Furthermore, the memory 11 includes both the internal storage unit of the electronic device 1 and an external storage device. The memory 11 can be used not only to store application software installed in the electronic device 1 and various types of data, such as the code of the AI-driven embodied intelligent adaptive learning method program, but also to temporarily store data that has been output or is about to be output.

[0088] In some embodiments, the processor 10 may be comprised of an integrated circuit, such as a single packaged integrated circuit or a plurality of packaged integrated circuits with the same or different functions, including a combination of one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and various control chips. The processor 10 is the control core (control unit) of the electronic device, connecting the various components of the entire electronic device using various interfaces and circuits. It executes or executes programs or modules stored in the memory 11 (such as AI-driven embodied intelligent adaptive learning method programs) and calls data stored in the memory 11 to perform various functions of the electronic device 1 and process data.

[0089] The bus 12 may be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus. The bus 12 may be divided into an address bus, a data bus, a control bus, etc. The bus 12 is configured to enable communication between the memory 11 and at least one processor 10, etc.

[0090] Figure 3 Only the electronic device with components is shown, and it can be understood by those skilled in the art that Figure 3 The structure shown does not constitute a limitation on the electronic device 1 , and may include fewer or more components than shown in the figure, or combine certain components, or arrange the components differently.

[0091] For example, although not shown, the electronic device 1 may further include a power supply (e.g., a battery) to power various components. Preferably, the power supply may be logically connected to the at least one processor 10 via a power management device, thereby enabling functions such as charge management, discharge management, and power consumption management via the power management device. The power supply may further include any components such as one or more DC or AC power supplies, a recharging device, a power failure detection circuit, a power converter or inverter, and a power status indicator. The electronic device 1 may also include various sensors, Bluetooth modules, Wi-Fi modules, etc., which are not further described here.

[0092] Furthermore, the electronic device 1 may also include a network interface. Optionally, the network interface may include a wired interface and / or a wireless interface (such as a WI-FI interface, a Bluetooth interface, etc.), which is generally used to establish a communication connection between the electronic device 1 and other electronic devices.

[0093] Optionally, the electronic device 1 may further include a user interface, which may be a display or an input unit (such as a keyboard). Optionally, the user interface may also be a standard wired interface or a wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen. The display may also be appropriately referred to as a display screen or a display unit, and is used to display information processed by the electronic device 1 and to display a visual user interface.

[0094] The AI-driven embodied intelligent adaptive learning method program stored in the memory 11 of the electronic device 1 is a combination of multiple instructions. When running in the processor 10, it can achieve: Confirming receipt of an adaptive learning instruction, parsing the adaptive learning instruction, obtaining a learning task type, and confirming an embodied intelligent agent and a 6G network communication link, wherein the embodied intelligent agent includes: a perception drive unit, a computation drive unit, a data drive unit, and an interaction drive unit; Using a preset sampling time interval, a perception drive unit, and a 6G network communication link to collect external environmental data including multiple environmental parameters, using the external environmental data and the learning task type to retrieve a target reference template from a pre-built prior knowledge base, using the target reference template to initialize each of the multiple environmental parameters corresponding to the external environmental data to obtain an initialized parameter set, and transmitting the initialized parameter set to the calculation drive unit, wherein the prior knowledge base includes multiple reference templates; Acquire an optimization algorithm based on the initialization parameters in the initialization parameter set, optimize and calculate the initialization parameters using the optimization algorithm to obtain optimization parameter nodes, wherein the optimization parameter nodes include optimization parameters and confidence evaluation parameters, and summarize the optimization parameter nodes to obtain an optimization parameter node set; transmitting the optimized parameter node set to the data-driven unit; Extracting an optimization parameter set and a confidence evaluation parameter set from the optimization parameter node set of the data-driven unit, adjusting the optimization parameter set using the prior knowledge base and the confidence evaluation parameter set to obtain a global parameter set, and transmitting the global parameter set to the interactive driving unit; generating a control instruction set using a global parameter set, driving the embodied intelligent agent using the control instruction set, collecting multiple execution result data of the embodied intelligent agent using a perception drive unit, transmitting the multiple execution result data to a data drive unit based on a 6G network communication link, and generating a detection report based on the multiple execution result data in the data drive unit; The priori knowledge base is updated according to the detection report to obtain an optimized priori knowledge base.

[0095] Specifically, the specific implementation method of the processor 10 for the above instructions can refer to Figures 1 to 3 The description of the relevant steps in the corresponding embodiments will not be repeated here.

[0096] Furthermore, if the modules / units integrated into the electronic device 1 are implemented as software functional units and sold or used as independent products, they may be stored in a computer-readable storage medium. The computer-readable storage medium may be volatile or non-volatile. For example, the computer-readable medium may include any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a mobile hard drive, a magnetic disk, an optical disk, a computer memory, or a read-only memory (ROM).

[0097] The present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program. When the computer program is executed by a processor of an electronic device, the computer program can implement: Confirming receipt of an adaptive learning instruction, parsing the adaptive learning instruction, obtaining a learning task type, and confirming an embodied intelligent agent and a 6G network communication link, wherein the embodied intelligent agent includes: a perception drive unit, a computation drive unit, a data drive unit, and an interaction drive unit; Using a preset sampling time interval, a perception drive unit, and a 6G network communication link to collect external environmental data including multiple environmental parameters, using the external environmental data and the learning task type to retrieve a target reference template from a pre-built prior knowledge base, using the target reference template to initialize each of the multiple environmental parameters corresponding to the external environmental data to obtain an initialized parameter set, and transmitting the initialized parameter set to the calculation drive unit, wherein the prior knowledge base includes multiple reference templates; Acquire an optimization algorithm based on the initialization parameters in the initialization parameter set, optimize and calculate the initialization parameters using the optimization algorithm to obtain optimization parameter nodes, wherein the optimization parameter nodes include optimization parameters and confidence evaluation parameters, and summarize the optimization parameter nodes to obtain an optimization parameter node set; transmitting the optimized parameter node set to the data-driven unit; Extracting an optimization parameter set and a confidence evaluation parameter set from the optimization parameter node set of the data-driven unit, adjusting the optimization parameter set using the prior knowledge base and the confidence evaluation parameter set to obtain a global parameter set, and transmitting the global parameter set to the interactive driving unit; generating a control instruction set using a global parameter set, driving the embodied intelligent agent using the control instruction set, collecting multiple execution result data of the embodied intelligent agent using a perception drive unit, transmitting the multiple execution result data to a data drive unit based on a 6G network communication link, and generating a detection report based on the multiple execution result data in the data drive unit; The priori knowledge base is updated according to the detection report to obtain an optimized priori knowledge base.

[0098] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, systems and methods can be implemented in other ways. For example, the system embodiments described above are only exemplary, and actual implementations may have other division methods.

[0099] The modules described as separate components may or may not be physically separate, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed across multiple network elements. Some or all of the modules may be selected to achieve the purpose of the solution of this embodiment according to actual needs.

[0100] In addition, the functional modules in various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or hardware plus software functional modules.

[0101] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0102] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not limiting. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. AI-driven embodied intelligent adaptive learning method, characterized by: The method comprises: Confirming receipt of an adaptive learning instruction, parsing the adaptive learning instruction, obtaining a learning task type, and confirming an embodied intelligent agent and a 6G network communication link, wherein the embodied intelligent agent includes: a perception drive unit, a computation drive unit, a data drive unit, and an interaction drive unit; Using a preset sampling time interval, a perception drive unit, and a 6G network communication link to collect external environmental data including multiple environmental parameters, using the external environmental data and the learning task type to retrieve a target reference template from a pre-built prior knowledge base, using the target reference template to initialize each of the multiple environmental parameters corresponding to the external environmental data to obtain an initialized parameter set, and transmitting the initialized parameter set to the calculation drive unit, wherein the prior knowledge base includes multiple reference templates; Acquire an optimization algorithm based on the initialization parameters in the initialization parameter set, optimize and calculate the initialization parameters using the optimization algorithm to obtain optimization parameter nodes, wherein the optimization parameter nodes include optimization parameters and confidence evaluation parameters, and summarize the optimization parameter nodes to obtain an optimization parameter node set; transmitting the optimized parameter node set to the data-driven unit; Extracting an optimization parameter set and a confidence evaluation parameter set from the optimization parameter node set of the data-driven unit, adjusting the optimization parameter set using the prior knowledge base and the confidence evaluation parameter set to obtain a global parameter set, and transmitting the global parameter set to the interactive driving unit; generating a control instruction set using a global parameter set, driving the embodied intelligent agent using the control instruction set, collecting multiple execution result data of the embodied intelligent agent using a perception drive unit, transmitting the multiple execution result data to a data drive unit based on a 6G network communication link, and generating a detection report based on the multiple execution result data in the data drive unit; The priori knowledge base is updated according to the detection report to obtain an optimized priori knowledge base.

2. The AI-driven embodied intelligent adaptive learning method according to claim 1, wherein: The method of collecting external environmental data including multiple environmental parameters using a preset sampling time interval, a sensing drive unit, and a 6G network communication link includes: Utilize 6G network communication links and perception drive units to build a distributed perception network, and use the distributed perception network and sampling time intervals to acquire visual image data and LiDAR point cloud data; The perception-driven unit is adjusted based on the learning task type, and the adjusted perception-driven unit is used to perform real-time enhancement processing on the visual image data to obtain a preprocessed image set. The LiDAR point cloud data is then subjected to noise reduction and motion compensation processing to obtain optimized point cloud data. The preprocessed image set and optimized point cloud data are fused into multi-source data to obtain external environment data.

3. The AI-driven embodied intelligent adaptive learning method according to claim 2, wherein: The method uses the external environment data and the learning task type to retrieve a target reference template from a pre-built prior knowledge base, and uses the target reference template to initialize each of the multiple environment parameters corresponding to the external environment data to obtain an initialization parameter set, including: The pre-confirmed edge computing module is used to analyze the environmental parameters in the external environment data to obtain the current task encoding vector, current environment feature fingerprint, and device information corresponding to the environmental parameters. The following operations are performed on each of the multiple reference templates: Extracting a task encoding vector of the reference template, and calculating task type similarity using the current task encoding vector, the task encoding vector, and a pre-established task type similarity calculation formula; Extracting the environmental feature fingerprint of the reference template, and calculating the environmental similarity using the current environmental feature fingerprint, the environmental feature fingerprint, and a pre-built environmental similarity calculation formula; Compare device information with the pre-built device compatibility table to obtain hardware compatibility; The overall matching degree is calculated using task type similarity, environment similarity, and hardware compatibility. The calculation formula is as follows: in, is the total matching degree, is the task type similarity, Indicates the environmental similarity, For hardware compatibility, All are weight coefficients; Summarize the total matching degrees to obtain a total matching degree set, sort the total matching degrees in the total matching degree set in descending order to obtain a total matching degree sequence, identify five candidate reference templates in the total matching degree sequence, wherein the five candidate reference templates are reference templates corresponding to the first five total matching degrees in the total matching degree sequence, input each of the five candidate reference templates into a pre-built digital twin platform for verification, and obtain five verification parameters, and take the candidate reference template corresponding to the largest verification parameter among the five verification parameters as the target reference template; The verification parameters corresponding to the target reference template are used to initialize the environmental parameters to obtain the initialization parameters, and the initialization parameters are summarized to obtain an initialization parameter set.

4. The AI-driven embodied intelligent adaptive learning method according to claim 3, wherein: The method of obtaining an optimization algorithm based on the initialization parameters in the initialization parameter set and optimizing the initialization parameters using the optimization algorithm to obtain an optimization parameter node includes: Obtaining an optimization algorithm using the initialization parameters in the initialization parameter set and a pre-built algorithm mapping table; Using the optimization algorithm to perform hierarchical optimization calculations on the initialization parameters to obtain multiple edge node optimization parameters; Performing a fixed-step gradient descent optimization on the plurality of edge node optimization parameters to obtain a plurality of preliminary optimization parameters; Verifying and adjusting a plurality of preliminary optimization parameters using the digital twin platform to obtain a plurality of optimization parameters; Utilizing a pre-built confidence evaluation method, each of the multiple optimization parameters is optimized and evaluated to obtain a confidence evaluation parameter; The optimization parameters and their corresponding confidence assessment parameters are integrated to generate an optimization parameter node.

5. The AI-driven embodied intelligent adaptive learning method according to claim 4, wherein: The prior knowledge base and the confidence evaluation parameter set are used to adjust the parameters of the optimization parameter set to obtain a global parameter set, including: According to the confidence evaluation parameters in the confidence evaluation parameter set, the optimization parameters in the optimization parameter set are divided into three division parameters, and the division parameters are high confidence parameters, medium confidence parameters or low confidence parameters, and the low confidence parameters are reinitialized using the prior knowledge base to obtain the corrected parameters; Fine-tune the confidence parameters using the prior knowledge base to obtain fine-tuning parameters; Obtaining a global optimization strategy based on the correction parameters, fine-tuning parameters, high-confidence parameters and a priori knowledge base; The correction parameters, fine-tuning parameters and high-confidence parameters are integrated using the global optimization strategy to obtain a global parameter set.

6. The AI-driven embodied intelligent adaptive learning method according to claim 5, wherein: The generating of the control instruction set by using the global parameter set includes: generating a preliminary control instruction set based on the global parameter set; Performing syntax verification and semantic verification on each of the preliminary control instructions in the preliminary control instruction set using an interactive driving unit; If the grammatical verification of the preliminary control instruction fails, revising the preliminary control instruction according to the pre-established grammatical rules to obtain a grammatical revision instruction; If the syntax verification of the preliminary control instruction passes, the preliminary control instruction is regarded as a syntax qualified instruction; If the semantic verification of the preliminary control instruction is unclear, semantically optimizing the preliminary control instruction according to the pre-built task execution rule to obtain a semantically optimized instruction; If the semantic verification of the preliminary control instruction is clear, the preliminary control instruction is regarded as a semantically qualified instruction; Integrate the grammatically qualified instructions, the semantically qualified instructions, the grammatically revised instructions, and the semantically optimized instructions to obtain a verified instruction set; Perform similarity comparison using the verified instruction set, the pre-confirmed historical successful instruction set, and the pre-built structural similarity algorithm to obtain a similarity evaluation value; If the similarity evaluation value is less than a preset evaluation threshold, the verified instruction set is adjusted based on the learning task type to obtain an adjusted instruction set; Using the adjusted instruction set as the verified instruction set, and returning to the step of performing similarity comparison using the verified instruction set, the pre-confirmed historical successful instruction set, and the pre-built structural similarity algorithm to obtain a similarity evaluation value, until the similarity evaluation value is greater than or equal to an evaluation threshold; If the similarity evaluation value is greater than or equal to the evaluation threshold, the verified instruction set is used as a candidate control instruction set; The candidate control instruction set is pre-executed using a pre-built simulation execution environment, and after confirming that the effect of the pre-execution is the preset expected effect, the candidate control instruction set is determined as the control instruction set.

7. The AI-driven embodied intelligent adaptive learning method according to claim 6, wherein: The method of driving the embodied intelligent body by using a control instruction set and collecting a plurality of execution result data of the embodied intelligent body by using a perception driving unit includes: Using a perception drive unit and a preset collection frequency to collect data when driving the embodied intelligent body, to obtain a plurality of raw execution data; performing data preprocessing on each of the plurality of original execution data to obtain a plurality of preprocessed execution data; Performing data cleaning on each of the plurality of pre-processed execution data to obtain a plurality of accurate execution data; Feature extraction is performed on each of the multiple accurate execution data to obtain multiple execution result data.

8. The AI-driven embodied intelligent adaptive learning method according to claim 7, wherein: The method of transmitting the plurality of execution result data to the data driving unit based on the 6G network communication link, and generating a detection report according to the plurality of execution result data in the data driving unit, includes: Transmitting the plurality of execution result data to a data driving unit using a 6G network communication link to obtain a plurality of transmitted execution result data; Performing data accuracy verification on the multiple execution result data after transmission to obtain a check result, wherein the check result includes a check feasibility and a check infeasibility; After confirming that the inspection result is feasible, the plurality of execution result data are compared and analyzed with a plurality of preset qualified performance data to obtain an analysis result report; The analysis result report is diagnosed for abnormality using preset abnormality diagnosis rules to obtain a detection report.

9. The AI-driven embodied intelligent adaptive learning method according to claim 8, wherein: The updating of the prior knowledge base according to the test report to obtain an optimized prior knowledge base includes: Extract feedback data from the test report; Using the feedback data to update the reference template, an updated reference template is obtained; The updated benchmark template is integrated into the prior knowledge base to obtain an optimized prior knowledge base.

10. AI-driven embodied intelligent adaptive learning system, characterized by: The system comprises: A learning task parsing module is used to confirm receipt of an adaptive learning instruction, parse the adaptive learning instruction, obtain the learning task type, and identify the embodied intelligent agent and the 6G network communication link. The embodied intelligent agent includes a perception drive unit, a computation drive unit, a data drive unit, and an interaction drive unit. a parameter optimization module for collecting external environmental data including multiple environmental parameters using a preset sampling time interval, a perception drive unit, and a 6G network communication link, retrieving a target reference template from a pre-built prior knowledge base using the external environmental data and the learning task type, initializing each of the multiple environmental parameters corresponding to the external environmental data using the target reference template to obtain an initialized parameter set, and transmitting the initialized parameter set to the calculation drive unit, wherein the prior knowledge base includes multiple reference templates; Acquire an optimization algorithm based on the initialization parameters in the initialization parameter set, optimize and calculate the initialization parameters using the optimization algorithm to obtain optimization parameter nodes, wherein the optimization parameter nodes include optimization parameters and confidence evaluation parameters, and summarize the optimization parameter nodes to obtain an optimization parameter node set; A parameter adjustment module is used to transmit the optimization parameter node set to the data driving unit, extract the optimization parameter set and the confidence evaluation parameter set from the optimization parameter node set of the data driving unit, adjust the optimization parameter set using the prior knowledge base and the confidence evaluation parameter set to obtain a global parameter set, and transmit the global parameter set to the interactive driving unit; An interactive feedback module is used to generate a control instruction set using a global parameter set, drive the embodied intelligent body using the control instruction set, and collect multiple execution result data of the embodied intelligent body using a perception drive unit, transmit the multiple execution result data to a data drive unit based on a 6G network communication link, generate a detection report based on the multiple execution result data in the data drive unit, and update the prior knowledge base according to the detection report to obtain an optimized prior knowledge base.

Images