AI-driven medical and invasive classroom interaction assisting method, system and device

Abstract

The invention discloses an AI-driven medical and invasive classroom interaction assisting method, system and device. The method comprises the following steps: constructing a medical and invasive classroom interaction scene library; configuring a classroom interaction agent based on the large model platform; real-time interaction guidance is executed through the intelligent agent; collecting and analyzing multi-modal interaction data; generating a multi-dimensional interaction effect evaluation report based on the data analysis result; the system comprises a scene construction module, an agent configuration module, an interaction execution module, a data acquisition and analysis module and an evaluation report module. The apparatus includes a memory, a processor, and a computer program stored on the memory. According to the technology provided by the invention, a personalized, safe, controllable and multi-dimensional evaluation classroom interaction effect is realized through deep fusion of a large model technology and a department and wound teaching scene.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of artificial intelligence education, in particular to an AI-driven science and technology classroom interaction assistance method, system and device. BACKGROUND

[0002] With the popularization of science and technology education, the quality of classroom interaction directly affects the cultivation of students' inquiry ability. However, the existing science and technology classroom interaction mode has the following significant defects:

[0003] Insufficient personalized guidance: traditional interaction tools mostly use unified teaching guidance strategies, which cannot provide adaptive support according to students' learning stages, knowledge reserves, and inquiry ability differences, and are difficult to meet the individualized learning needs of different students.

[0004] Real-time feedback lag and high data security risk: existing technologies mostly rely on cloud servers to process interaction data, resulting in delays in data transmission and response, affecting the coherence of classroom interaction; at the same time, the cloud stores students' experimental data, interaction records and other sensitive information, which has the risk of data leakage, and does not meet the data privacy protection requirements in the education scenario.

[0005] Difficulties in integrating multi-modal data: science and technology classroom interaction involves multi-type data such as voice questioning, experimental operation, and text feedback. Existing tools lack effective data fusion mechanisms and cannot correlate and analyze these data to comprehensively assess student status.

[0006] Single evaluation dimension: existing interaction effect evaluation mostly focuses on the number of student participation or experimental completion, ignoring the evaluation of key dimensions such as "problem-solving innovation" and "interdisciplinary knowledge application ability" in the inquiry process, making it difficult for teachers to fully grasp the classroom effect and optimize teaching strategies.

[0007] Low integration with science and technology experiment scenarios: existing classroom interaction tools are mostly suitable for theoretical knowledge explanation and lack deep integration with science and technology experiment scenarios, making it difficult to provide intelligent guidance for complex experimental inquiry processes.

[0008] Although some technologies have attempted to improve classroom interaction experience, for example, the Chinese patent with publication number CN113095969A provides an immersive flipped classroom teaching system based on multiple virtual avatars, which constructs a virtual teaching environment with VR technology, supports teachers and students to interact in the form of virtual avatars, and enhances the immersion and interactivity of teaching. However, this technology still has the following limitations: first, it relies on cloud rendering and resource pushing, making it difficult to achieve low-latency real-time interaction; second, although virtual avatars improve participation, they do not provide personalized guidance based on individual student differences; third, the system focuses on scenario construction and visual presentation, lacks deep integration and intelligent analysis capabilities for multi-modal data, and cannot provide accurate academic assessment and teaching optimization suggestions.

[0009] Furthermore, US Patent No. US20190114543A1 discloses a local learning system in an artificial intelligence device. By deploying a lightweight neural network on a local device and performing online learning and model pruning, it achieves localized data processing and privacy protection, avoiding the latency and risks of cloud transmission. This technology provides a feasible path for localized AI applications, but it focuses on single-modal processing such as voice and image processing for general-purpose devices. It is not optimized for the special needs of multimodal interaction, interdisciplinary knowledge integration, and real-time teaching guidance in educational scenarios, and lacks deep integration with science and technology innovation teaching content, making it unsuitable for direct application in classroom interaction assistance.

[0010] Therefore, there is an urgent need for an interactive assistance solution that can achieve personalized guidance, real-time secure data processing, multi-dimensional evaluation, and deep integration with science and technology innovation scenarios to solve the above-mentioned technical deficiencies. Summary of the Invention

[0011] Terminology Definition

[0012] 1. Inquiry ability: refers to students' ability to analyze the rationality of experimental plans, propose innovative solutions to problems, and apply interdisciplinary knowledge. It is assessed through three dimensions: complexity of the operation trajectory, innovativeness of the questions asked, and frequency of interdisciplinary keywords.

[0013] 2. Scenario Adaptability: This refers to the degree of matching between scenario resources and teaching objectives, which is quantified by two indicators: the completeness of knowledge point coverage and the matching degree between the difficulty of experimental resources and the grade level.

[0014] In view of the above-mentioned shortcomings of current artificial intelligence education technology, the purpose of this invention is to provide an AI-driven interactive auxiliary technology for science and technology innovation classrooms. This technology achieves personalized, safe and controllable, and multi-dimensional evaluation classroom interaction effects through the deep integration of large-scale model technology with science and technology innovation teaching scenarios.

[0015] To achieve the above objectives, a first aspect of the present invention provides an AI-driven interactive assistance method for science and technology innovation classrooms, the AI-driven interactive assistance method for science and technology innovation classrooms comprising the following steps:

[0016] Construct an interactive scenario library for science and technology innovation classrooms. The scenario library is divided into scenario levels according to academic stage, and each scenario is associated with experimental resource packages and knowledge point mapping relationships.

[0017] Classroom interactive intelligent agents are configured based on a large model platform, and these agents support lesson preparation assistance and interdisciplinary joint guidance.

[0018] The intelligent agent performs real-time interactive guidance, which includes identifying the student's intent and generating guidance content.

[0019] Collect multimodal interaction data and store it locally, and conduct real-time learning analysis through a large locally deployed model;

[0020] The student performance was assessed based on participation, inquiry ability, and knowledge mastery. Suggestions for optimizing the intelligent agent's language, functions, and logic were proposed. The classroom effect was summarized by combining scenario adaptability and interaction efficiency, and a visual report was generated.

[0021] The classroom interactive intelligent agent includes:

[0022] The AI-powered teaching assistant for science and technology can automatically extract the core elements of an experiment to generate lesson plans that include objectives, steps, and precautions, and link them to relevant scenario resources.

[0023] Collaborative rules for interdisciplinary intelligent agents include triggering mechanisms based on the depth of interdisciplinary intersection of problems and priority response rules based on scenario matching degree;

[0024] Personalized guidance strategies and dynamic adjustment mechanisms based on student profile tags.

[0025] In some embodiments of the first aspect of this application, hierarchical scenarios are divided according to the educational stage, students' inquiry ability and the type of science and technology innovation project. The educational stage includes primary school, junior high school and senior high school, and supports the construction of interdisciplinary composite scenarios.

[0026] Each level of scenario is associated with an experimental resource package and scenario-specific safety specifications that match its corresponding learning stage and ability level. The experimental resource package includes tools, simulation resources and experimental procedure breakdown documents, and the difficulty and presentation format of the resources can be dynamically adjusted based on classroom interaction data.

[0027] We construct a dynamic mapping relationship between scenario levels and multidisciplinary knowledge points, link it with the ability development goals such as scientific inquiry and interdisciplinary application, and adjust the mapping weights based on student interaction data to achieve gradient coverage of the same knowledge point in different scenarios.

[0028] In some embodiments of the first aspect of this application, a basic intelligent agent framework is deployed, and a natural language processing interface, an experimental scenario association interface, and a local data interaction interface are configured.

[0029] Import the science and technology innovation curriculum outline and experimental standard procedures, automatically extract the core links of the experiment to generate lesson plans, support personalized modification of lesson plans and synchronously update related resources and knowledge points;

[0030] Register multidisciplinary intelligent agents and assign domain labels, set up a dynamic collaborative triggering mechanism based on the depth of cross-disciplinary intersection of problems, and set up multi-agent response priority rules based on scenario matching degree;

[0031] Based on students' historical experiment records and interactive data, profile tags are generated, guidance difficulty and feedback style parameters are set, and a dynamic strategy adjustment mechanism is configured.

[0032] In some embodiments of the first aspect of this application, the collaboration rules include agent collaboration triggering conditions and multi-agent collaboration response priorities, and the configuration of personalized guidance strategies includes:

[0033] Generate student profile tags;

[0034] Set guidance strategy parameters based on student profile tags;

[0035] Configuration strategy dynamic adjustment mechanism.

[0036] In some embodiments of the first aspect of this application, the real-time interactive guidance performed by the intelligent agent includes:

[0037] Analyze student multimodal input, which includes voice input, text input, and operation input;

[0038] Generate targeted guidance content, which includes Q&A guidance, operation prompts, and process advancement;

[0039] Display real-time feedback, including text feedback, visual feedback, and dynamic effect feedback.

[0040] In some embodiments of the first aspect of this application, the acquisition and analysis of multimodal interaction data includes:

[0041] Multimodal data is collected, including interaction data, operation data, and feedback data. The interaction data includes the voice / text dialogue content between students and the intelligent agent, the frequency of questions, and keywords. The operation data includes experimental parameter adjustment records, operation duration, step success rate, and operation trajectory. The feedback data includes students' satisfaction ratings for the guidance content and self-annotated difficulty tags. All data is stored locally with encryption to ensure the privacy and security of educational data.

[0042] The collected multimodal data is analyzed in real time using a local large model. The real-time analysis includes learning analysis, agent performance analysis, and anomaly detection. The learning analysis includes identifying weak knowledge points through word frequency analysis and judging inquiry ability by combining operation trajectory. The agent performance analysis includes calculating the matching degree between student questions and guidance content and evaluating response accuracy. The anomaly detection includes identifying dangerous parameter settings and equipment operation errors in experimental operations and triggering real-time warnings.

[0043] Generate a data dashboard that displays student participation metrics, experiment progress statistics, hot topic rankings, and abnormal operation alerts in real time, supporting teachers to dynamically intervene in the classroom.

[0044] In some embodiments of the first aspect of this application, generating a multi-dimensional interaction effect evaluation report based on the learning analysis results includes:

[0045] The system generates student competency assessment content based on participation, inquiry ability, and knowledge mastery. Participation is quantified based on the number of active interactions, experimental operation time, number of independent inquiry attempts, and the proportion of questions asked. Inquiry ability includes analyzing the rationality of experimental plans, the innovation of problem-solving, and the frequency of interdisciplinary knowledge application. Knowledge mastery includes assessing the depth of understanding of scenario-related knowledge points and generating personalized practice recommendations.

[0046] The paper proposes suggestions for agent optimization, which include speech optimization, function supplementation, and guidance logic adjustment. Speech optimization includes simplifying responses to general questions and improving the accuracy of responses to domain-specific questions. Function supplementation includes adding a knowledge base for scarce domains and optimizing the accuracy of multimodal input parsing. Guidance logic adjustment includes reducing step prompts and adding open-ended questions for innovative and inquiry-based students, and strengthening guidance on operational details for students with weak foundations.

[0047] The overall classroom effect is summarized and evaluated by combining scenario adaptability and interaction efficiency. Scenario adaptability includes evaluating the matching degree between scenario resources and teaching objectives and the completeness of cross-disciplinary knowledge coverage. Interaction efficiency includes statistical analysis of the average response time of the intelligent agent, student problem-solving rate, and the percentage of experimental downtime.

[0048] The student ability assessment content, agent optimization suggestions, and overall classroom effectiveness are integrated into a complete report in the form of text and visual charts.

[0049] To achieve the above objectives, a second aspect of the present invention provides an AI-driven interactive assistance system for science and technology innovation classrooms, the AI-driven interactive assistance system for science and technology innovation classrooms comprising:

[0050] The scenario construction module divides the interactive scenario levels of science and technology innovation classrooms according to the learning stage, associates and adapts experimental resource packages and safety specifications for each level of scenario, establishes a dynamic mapping relationship between scenarios and multi-disciplinary knowledge points, and supports dynamic adjustment of scenario levels, resources and knowledge point mappings based on classroom interaction data.

[0051] The intelligent agent configuration module is based on a large model platform to deploy the basic framework of classroom interactive intelligent agents, and configures the lesson preparation assistance and interactive speech customization functions of science and technology innovation AI teaching assistant, the joint guidance rules of interdisciplinary intelligent agents, and the personalized guidance strategies based on students' historical data and real-time performance.

[0052] The interactive execution module receives and parses students' multimodal inputs, including voice, text, operations, facial expressions, and body movements. Based on the scene library and intelligent agent configuration, it generates targeted guidance content such as question-and-answer guidance, operation prompts, and process advancement, and displays real-time feedback in the form of text, visualization, and dynamic effects.

[0053] The data acquisition and analysis module collects and stores the interaction data, experimental operation data and feedback data between students and the intelligent agent in a localized manner. Through a large model deployed locally, it performs real-time learning analysis, intelligent agent performance analysis and anomaly detection on this multimodal data, and generates a real-time visualization interface to display key classroom data.

[0054] The evaluation report module assesses student performance from the dimensions of participation, inquiry ability, and knowledge mastery. Based on interactive data, it provides suggestions for optimizing the agent's language, functions, and logic. By evaluating scenario adaptability and interaction efficiency, it summarizes the overall classroom effect, proposes improvement directions, and finally integrates them into a multi-dimensional interactive effect evaluation report.

[0055] In some embodiments of the second aspect of this application, the scene construction module includes:

[0056] The hierarchical division unit is used to divide the scene hierarchy according to the learning stage, student ability and project type, supports the construction of cross-disciplinary composite scenes, and dynamically adjusts the hierarchical structure and related resources based on classroom data;

[0057] The resource association unit is used to associate and adapt experimental resource packages and safety specifications for each scenario, and dynamically adjust the difficulty and presentation of resources.

[0058] The knowledge point mapping unit is used to establish a dynamic mapping relationship between scenarios and multidisciplinary knowledge points, associate it with ability development goals, and adjust the mapping weights based on interactive data to achieve gradient mapping of the same knowledge point in different scenarios.

[0059] The agent configuration module includes:

[0060] The teaching assistant configuration unit is used to configure lesson preparation assistance and customized interactive scripts for science and technology innovation AI teaching assistants;

[0061] The collaborative rule unit is used to set joint guidance rules for interdisciplinary intelligent agents. The joint guidance rules include a dynamic collaborative triggering mechanism based on the depth of cross-disciplinary intersection of the problem and an integration rule for the differential responses of multiple intelligent agents.

[0062] Personalized strategy unit, used to configure guidance difficulty and feedback style based on students' historical data and real-time performance;

[0063] The interactive execution module includes:

[0064] A multimodal input parsing unit is used to receive and parse students' multimodal input information and combine it with semantic analysis to improve the accuracy of intent recognition;

[0065] The guidance logic generation unit is used to generate guidance content that adapts to the scenario and student needs based on the scenario library and intelligent agent configuration.

[0066] The real-time feedback display unit is used to display the intelligent agent's guidance content in multiple formats;

[0067] The data acquisition and analysis module includes:

[0068] The multimodal data acquisition unit is used to collect interaction data between students and intelligent agents, experimental operation data, and student feedback data on guidance, and uses local storage to ensure data security.

[0069] The real-time analysis unit is used to perform learning analysis, agent performance analysis, and abnormal operation detection on the collected multimodal data based on a locally deployed large model, thereby realizing real-time data processing.

[0070] The data dashboard unit is used to generate a real-time visual interface to centrally display key classroom data.

[0071] The evaluation report module includes:

[0072] The learning ability assessment unit is used to comprehensively evaluate students' performance from three dimensions: participation, inquiry ability, and knowledge mastery, and generate personalized learning suggestions.

[0073] The agent optimization unit is used to provide direction for agent improvement based on classroom interaction data, including speech optimization, function supplementation, and guidance logic adjustment.

[0074] The lesson effectiveness summary unit is used to output an overall evaluation of the lesson by assessing the suitability of the learning environment and the efficiency of interaction, and to propose directions for improvement.

[0075] To achieve the above objectives, a third aspect of the present invention provides a computer device comprising: a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the AI-driven interactive assistance method for science and technology innovation classrooms.

[0076] The advantages of this invention are as follows: First, by using a segmented scenario library and personalized strategies for intelligent agents, it adapts the guidance logic to different student abilities, achieving personalized interactive guidance and enhancing students' interest in exploration and learning efficiency. Second, it employs a locally deployed large model to process multimodal data, achieving real-time and secure data processing, avoiding cloud transmission delays, ensuring real-time early warning and problem response for experimental operations, and reducing the risk of privacy leaks through local data storage, complying with educational data security standards. Third, it generates evaluation reports from multiple dimensions, including student abilities, intelligent agent performance, and overall classroom effectiveness, focusing not only on experimental results but also on the innovation and knowledge application abilities during the exploration process, helping teachers to accurately optimize teaching strategies. Fourth, it designs exclusive guidance logic and resource packages for science and technology innovation experiments, which can be deeply integrated with science and technology innovation scenarios, achieving linked teaching of experimental operations and theoretical knowledge, and enhancing students' understanding of complex science and technology innovation concepts. Fifth, through a multi-agent collaborative mechanism, it supports joint guidance of interdisciplinary projects, cultivating students' interdisciplinary thinking and comprehensive application abilities. Attached Figure Description

[0077] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0078] Figure 1 This is a flowchart illustrating an AI-driven interactive assistance method for science and technology innovation classrooms as described in this invention.

[0079] Figure 2 This is a schematic diagram of the structure of an AI-driven interactive auxiliary system for science and technology innovation classrooms according to the present invention;

[0080] Figure 3 This is a schematic block diagram of a computer device according to an embodiment of this application. Detailed Implementation

[0081] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0082] Figure 1 This document illustrates a flowchart of an AI-driven interactive assistance method for science and technology innovation classrooms, as shown in one embodiment of this application. The method includes the following steps:

[0083] Step S1: Construct a library of interactive scenarios for science and technology innovation classrooms.

[0084] In this embodiment of the invention, the constructed interactive scenario library for science and technology innovation classrooms is divided into hierarchical scenarios according to academic stages, and each hierarchical scenario is associated with a corresponding experimental resource package and knowledge point mapping relationship;

[0085] The specific construction process of the science and technology innovation classroom interactive scenario library includes the following steps:

[0086] Step S1.1: Divide the scenarios into levels according to the learning stage.

[0087] The hierarchical scenario serves as an important basis for subsequent resource allocation and knowledge point association. In this embodiment of the invention, the hierarchical division by educational stage includes three levels: primary school, junior high school, and senior high school. Of course, a finer division can be made according to the actual situation, and this invention does not impose any special limitations.

[0088] Step S1.2: Associate an experimental resource package that matches the corresponding learning stage with each level of scenario.

[0089] In this embodiment of the invention, the experimental resource package associated with each level of scenario includes tools and simulation resources that match its corresponding learning stage.

[0090] For example, in the "Image Recognition Basics" level scenario at the primary school level, the resource package associated with this level scenario includes an animal image dataset and a simple annotation tool; while in the "Unmanned Surface Vessel Trajectory Optimization" level scenario at the high school level, the resource package associated with propeller parameter adjustment tools and fluid dynamics simulation resources is provided.

[0091] Step S1.3: Construct the mapping relationship between scene levels and knowledge points.

[0092] In this embodiment of the invention, the so-called mapping relationship between scene level and knowledge point is the dynamic association relationship between scene level and core knowledge point of course.

[0093] For example, in the "Image Recognition Fundamentals" scenario level at the primary school level, this scenario level is mapped to the "Feature Extraction" knowledge point; while in the "Unmanned Surface Vessel Trajectory Optimization" scenario level at the high school level, it is mapped to the "Control Algorithm" knowledge point.

[0094] Step S2: Configure the classroom interactive intelligent agent based on the large model platform.

[0095] In this embodiment of the invention, the configured intelligent agent has the functions of lesson preparation assistance and interdisciplinary joint guidance.

[0096] The specific process of configuring classroom interactive intelligent agents based on the large model platform includes the following steps:

[0097] Step S2.1: Deploy the basic intelligent agent framework.

[0098] In this embodiment of the invention, the basic operating framework of the classroom interactive intelligent agent can be built based on a general large model or a special model adapted to educational scenarios.

[0099] The core functional module interfaces of the intelligent agent framework include a natural language processing interface (supporting speech-to-text and intent recognition), an experimental scenario association interface (connecting to the resource packages and knowledge point mapping relationships of the scenario library), and a data interaction interface (connecting to the local data storage module), ensuring that the intelligent agent can call scenario library resources in real time and process multimodal input data.

[0100] Step S2.2: Configure the lesson preparation assistance function of the science and technology innovation AI teaching assistant.

[0101] Specifically, the module loads a lesson preparation assistance module into the science and technology innovation AI teaching assistant, imports the science and technology innovation curriculum outlines and experimental standard procedures for each grade level, and sets lesson plan generation rules: for example, for the "speech recognition" experiment, the module automatically extracts the core steps of the experiment (sample collection, feature extraction, model training, accuracy testing), generates a lesson plan according to the "objective-steps-precautions" structure, and associates the corresponding tool resources.

[0102] It also allows for personalized lesson plan adjustments, supporting modifications to lesson plan steps via an interactive interface. For example, by adding a "noise interference test" step, the agent automatically updates associated resource packages and knowledge point mappings, such as adding "signal-to-noise ratio" related knowledge points to the agent.

[0103] Step S2.3: Configure the collaborative rules for interdisciplinary intelligent agents.

[0104] Based on the interdisciplinary nature of science and technology innovation projects, multidisciplinary intelligent agents can be registered in the large model platform, and each intelligent agent can be assigned a unique domain label for later use.

[0105] Then, the conditions for triggering agent collaboration are set. For example, when a student asks a question or performs an operation involving interdisciplinary knowledge, the AI ​​teaching assistant automatically calls upon the chemistry agent. In this way, the AI ​​teaching assistant is responsible for identifying the problem scenario, while the chemistry agent is responsible for interpreting the specific meaning of the data, thereby realizing the response rule of "master-slave division of labor" among multiple agents.

[0106] To address the issue of multiple agents being able to respond to a certain type of problem, it is necessary to configure collaborative response priority rules for multiple agents. In this embodiment of the invention, the calling priority of agents can be sorted according to "scenario matching degree." For example, in the "unmanned surface vessel trajectory optimization" scenario, the control algorithm agent is triggered first, rather than the general scientific agent.

[0107] Step S2.4: Configure personalized boot strategy.

[0108] The specific steps for configuring a personalized bootstrapping strategy include:

[0109] Step S2.4.1: Generate student profile tags.

[0110] In this embodiment of the invention, student profile tags can be generated through large-scale model analysis by importing student historical data. These tags include key information such as grade level (primary / middle / high school), knowledge weaknesses, and inquiry ability level. The student historical data includes past experimental operation records, interactive question content, and ability assessment results.

[0111] Step S2.4.2: Set guidance strategy parameters based on student profile tags.

[0112] For elementary school students or those with a basic operational focus, a high-prompt strategy is used, which provides detailed step-by-step guidance during experiments and gives specific answers directly when asked questions. For high school students or those with an innovative and inquiry-based focus, a low-prompt strategy is used, which only prompts key principles during experiments and guides thinking with counter-questions when asked questions.

[0113] Step S2.4.3: Configure the dynamic adjustment mechanism of the strategy.

[0114] By monitoring students' current interaction data in real time (such as three consecutive incorrect operations), the agent automatically switches guidance strategies temporarily. For example, it may switch from low prompting level to high prompting level and provide specific suggestions for correcting the operation.

[0115] Step S3: Perform real-time interactive guidance through the intelligent agent.

[0116] In this embodiment of the invention, real-time interactive guidance includes identifying the student's intent and generating guidance content, as well as providing real-time assistance for experimental operations.

[0117] The specific process of executing real-time interactive guidance through an intelligent agent includes the following steps:

[0118] Step S3.1: Parse the student's multimodal input.

[0119] The intelligent agent receives students' voice, text, or operational input through a multimodal input parsing module. Voice input is converted into text through speech recognition and the intent is identified; text input is parsed for keywords (such as "rudder angle") and associated with scene parameters; operational input captures experimental interface operations in real time (such as adjusting the unmanned surface vessel's rudder angle to 30°) and records the operation trajectory.

[0120] Step S3.2: Generate targeted guidance content.

[0121] Based on the scene library and agent configuration, the guidance logic generation module generates corresponding guidance content. In this embodiment of the invention, the guidance content can be specifically divided into the following three categories:

[0122] Question-and-answer guidance: Promptly ask follow-up questions to students' inquiries. For example, when a student asks, "What should I do if the unmanned surface vessel is yawing?", reply, "Check if the rudder angles are symmetrical? Have you tried calibrating the coordinate system?"

[0123] Operation prompts: When students set unreasonable parameters, such as excessively large rudder angles on the unmanned surface vessel (USV), a real-time warning will be issued and adjustments will be suggested. For example, if the rudder angle of the USV is too large, the prompt will read, "Current rudder angle of 30° may cause the vessel to capsize; it is recommended to adjust it to within 15°."

[0124] Process progression: When an experiment stalls, push out the next step guidance.

[0125] Step S3.3: Display real-time feedback.

[0126] The real-time feedback display module presents the guided content in multiple formats. In this embodiment of the invention, the presentation methods can be divided into the following three types:

[0127] Text feedback: The guiding message is displayed in a pop-up window on the interface.

[0128] Visual feedback: For example, arrows can be used to indicate the yaw direction of the unmanned surface vessel, and waveform graphs can be used to show the speech-to-text effect.

[0129] Dynamic effect feedback: For example, triggering a "flower shower effect" when the experiment is successful.

[0130] Step S4: Collect and analyze multimodal interaction data.

[0131] In this embodiment of the invention, the multimodal interaction data includes interaction data, operation data, and feedback data, all of which are processed through a large model deployed locally.

[0132] The specific process of collecting and analyzing multimodal interaction data includes the following steps:

[0133] Step S4.1: Collect multimodal data.

[0134] In this embodiment of the invention, the multimodal data collected by the multimodal data acquisition module includes the following three core types of data:

[0135] Interaction data: Records the voice / text dialogue content between students and the intelligent agent, including question frequency, keywords, etc.

[0136] Operational data: Records of experimental parameter adjustments (such as unmanned surface vessel rotation speed and image recognition threshold), operation time, and success rate.

[0137] Feedback data: Collect students' satisfaction ratings for the intelligent agent guidance, and self-labeled difficulty tags, etc.

[0138] It is important to note that all collected data is stored locally to ensure information security.

[0139] Step S4.2: Real-time analysis of the local large model.

[0140] In this embodiment of the invention, the collected multimodal data can be processed using a locally deployed large model. The processing specifically includes the following parts:

[0141] Learning Situation Analysis: High-frequency questions are identified through word frequency analysis, and students' inquiry abilities are assessed by combining this with their operational trajectory. For example, if the word "neural network" appears too frequently in a student's question, it is marked as a weak point that requires focused explanation.

[0142] Agent performance analysis: Calculate the response accuracy of the guiding dialogue. Specifically, the response accuracy of relevant questions can be evaluated by the matching degree between student questions and agent responses.

[0143] Anomaly detection: Identifies dangerous operations and triggers warnings. For example, it will issue an alarm when the parameters of the unmanned surface vessel (USV) exceed safe limits.

[0144] Step S4.3: Generate a data dashboard.

[0145] In this embodiment of the invention, the data dashboard module generates a real-time visualization interface for teachers, through which key information such as student participation indicators, experiment progress statistics, and hot topic rankings can be displayed.

[0146] Student participation metrics can be evaluated using data such as the percentage of students who actively ask questions and the number of interactive actions.

[0147] Step S5: Generate a multi-dimensional interaction effect evaluation report based on the data analysis results.

[0148] In this embodiment of the invention, the generated multi-dimensional interactive effect evaluation report includes student ability assessment, agent optimization suggestions, and classroom effect summary.

[0149] The specific process of generating a multi-dimensional interaction effect evaluation report based on data analysis results includes the following steps:

[0150] Step S5.1: Generate student ability assessment content.

[0151] This invention comprehensively evaluates student performance and generates student ability assessment content from three dimensions: participation, inquiry ability, and knowledge mastery.

[0152] Among them, participation can be measured by indicators such as the number of times of active interaction, the duration of experimental operation, the number of times of independent exploration attempts, and the proportion of active questioning; exploration ability can be measured by indicators such as the innovativeness of problem-solving and the rationality of experimental plans; knowledge mastery can be measured by indicators such as assessing the ability to apply knowledge points related to the scenario and generating personalized learning suggestions.

[0153] Step S5.2: Propose suggestions for agent optimization.

[0154] This invention provides direction for improving intelligent agents based on interactive data. Aspects of intelligent agent optimization and improvement include speech optimization, feature addition, and adjustment of guidance logic.

[0155] The optimization of the dialogue includes making the agent's responses to general questions more accessible and its responses to domain-specific questions more professional; the addition of functions includes adding relevant knowledge bases for specific domains to the agent; and the adjustment of the guidance logic includes reducing step prompts and adding open-ended questions for innovative and inquiry-based students.

[0156] Step S5.3: Summarize the overall effectiveness of the class.

[0157] In this embodiment of the invention, the output of the comprehensive classroom evaluation needs to consider scenario adaptability and interaction efficiency to comprehensively evaluate the overall effect of the classroom.

[0158] Among them, scenario adaptability can be evaluated by analyzing the matching degree between the current scenario and the teaching objectives; interaction efficiency can be evaluated by statistically analyzing data such as the average response time of the intelligent agent and the student problem-solving rate.

[0159] Step S5.4: Integrate to form a complete report.

[0160] The report integrates student ability assessments, agent optimization suggestions, and classroom effectiveness summaries into a multi-dimensional report, presented in text and visual charts (such as ability radar charts and problem distribution pie charts) to help teachers intuitively grasp the effectiveness of classroom interaction.

[0161] Figure 2 A schematic diagram of the structure of an AI-driven interactive auxiliary system for science and technology innovation classrooms, as shown in one embodiment of this application, is illustrated. Figure 2 As shown, the system 100 of this application includes a scene construction module 101, an intelligent agent configuration module 102, an interaction execution module 103, a data acquisition and analysis module 104, and an evaluation report module 105.

[0162] The function of Module 101 is to divide the interactive scenarios in science and technology innovation classrooms into levels according to academic stages, associate and adapt experimental resource packages and safety specifications for each level of scenario, establish a dynamic mapping relationship between scenarios and multidisciplinary knowledge points, and support dynamic adjustment of scenario levels, resource and knowledge point mappings based on classroom interaction data. Module 101 includes the following units:

[0163] Hierarchical Unit Division: This unit divides scenarios into levels according to learning stage, student ability, and project type, supports the construction of interdisciplinary composite scenarios, and dynamically adjusts the hierarchical structure and related resources based on classroom data.

[0164] Specifically, based on the division of academic stages, the scenario levels are further subdivided according to students' ability levels and the types of science and technology innovation projects, thereby refining the level subdivision; this is used to construct interdisciplinary composite scenario levels, thereby achieving the integration and division of interdisciplinary scenarios.

[0165] Meanwhile, to adapt to diverse and flexible learning scenarios, this module can also dynamically adjust the scenario hierarchy based on classroom interaction data. For example, if a student's completion rate in a certain scenario consistently falls below 60%, it will automatically be downgraded from "Advanced" to "Basic," triggering a synchronized adjustment of resource packages and related knowledge points.

[0166] Resource Association Unit: This unit associates and adapts experimental resource packages and safety specifications for each scenario, and dynamically adjusts the difficulty and presentation of resources.

[0167] Specifically, the difficulty and presentation format of associated resources are dynamically adjusted based on students' real-time operation data; and experimental resource packages are linked with scenario-specific safety specifications and early warning thresholds.

[0168] In addition, this unit also supports version recording and update management of experimental resource packages, allowing teachers to review historical versions and mark high-quality resource versions based on teaching feedback.

[0169] Knowledge Point Mapping Unit: This unit establishes a dynamic mapping relationship between scenarios and multidisciplinary knowledge points, links them to ability development goals, and adjusts the mapping weights based on interactive data to achieve gradient mapping of the same knowledge point in different scenarios.

[0170] Specifically, it enables the mapping of a single scenario with knowledge points from multiple disciplines; it associates the mapping of course knowledge points with corresponding competency development goals; and it dynamically adjusts the mapping weight of knowledge points based on multimodal interactive data. For example, when the "control algorithm" accounts for more than 40% of the frequency of student questions, it automatically increases its mapping weight in the relevant scenario and adds example resources.

[0171] Meanwhile, in order to achieve spiral-progressive teaching of knowledge points, the same knowledge point is mapped to different difficulty levels in different scenario levels. For example, the knowledge point of "variable control" corresponds to basic, advanced and innovative difficulties in different scenarios in elementary school, junior high school and high school.

[0172] The intelligent agent configuration module 102 is designed to deploy a basic framework for classroom interactive intelligent agents based on a large model platform. It configures functions such as lesson preparation assistance and customized interactive scripts for science and technology innovation AI teaching assistants, joint guidance rules for interdisciplinary intelligent agents, and personalized guidance strategies based on students' historical data and real-time performance. The intelligent agent configuration module 102 includes the following units:

[0173] Teaching Assistant Configuration Unit: This unit provides AI teaching assistants for science and technology innovation with functions such as lesson preparation assistance, customized interactive scripts, and adaptation to experimental characteristics. The granularity of lesson plan steps can be adjusted according to the complexity of the experiment, and the script style can be customized according to the teaching objectives.

[0174] Specifically, in addition to basic lesson preparation assistance and customized scripts, a further configuration of experimental characteristic adaptation function is provided. This function can automatically adjust the granularity of the steps in the lesson plan according to the operational complexity of the experiment.

[0175] At the same time, the style of the script is customized for different teaching objectives. If the objective is to standardize operations, the script focuses on prompting the accuracy of the steps. If the objective is to cultivate innovation, the script focuses on guiding openness.

[0176] Collaborative Rule Unit: This unit sets up joint guidance rules for interdisciplinary agents, including a dynamic collaborative triggering mechanism based on the depth of cross-disciplinary intersection of the problem, and integrated rules for the differential responses of multiple agents.

[0177] Specifically, a dynamic collaborative triggering mechanism is added to the interdisciplinary intelligent agent joint guidance rules. This mechanism can automatically call up the corresponding number of intelligent agents based on the interdisciplinary depth of the student's question.

[0178] At the same time, the system configures response result fusion rules. When there are differences in the responses from multiple agents, the system automatically integrates the answers according to the subject relevance weights and marks the supplementary viewpoints of each agent to avoid information conflicts.

[0179] Personalized Strategy Unit: This unit configures the difficulty of guidance and the style of feedback based on students' historical data and real-time performance.

[0180] Specifically, based on students' historical data, a real-time dynamic adjustment function has been added. This function fine-tunes the guidance strategy in real time by monitoring students' current interaction data.

[0181] Meanwhile, the dimensions of feedback style adaptation are further refined, including "encouragement intensity" (such as increasing the proportion of affirmative language for introverted students) and "correction method" (such as using "suggestive correction" instead of "direct negation" for sensitive students), and support for adapting the guidance interval according to the student's learning pace to avoid feedback that is too fast or too slow from interfering with the learning state.

[0182] The interactive execution module 103 receives and parses multimodal input from students, including voice, text, operations, facial expressions, and body movements. Based on a scene library and agent configuration, it generates targeted guidance content such as question-and-answer prompts, operation hints, and process progression, and displays real-time feedback in various formats, including text, visualization, and dynamic effects. The interactive execution module 103 includes the following units:

[0183] Multimodal input parsing unit: Used to receive and parse students' multimodal input information, and combine semantic analysis to improve the accuracy of intent recognition.

[0184] Specifically, in addition to basic speech-to-text, keyword parsing, and operation trajectory recording, it can also capture students' facial expressions or body movements through cameras and combine semantic analysis to judge the students' real-time status, thereby improving the accuracy of intent recognition.

[0185] Guidance Logic Generation Unit: Based on the scenario library and agent configuration, it generates guidance content adapted to the scenario and student needs. This includes:

[0186] Question-and-answer guidance: Promptly ask students questions (e.g., if a student asks "What should I do if the unmanned surface vessel veers off course?", reply "Check if the rudder angles are symmetrical? Have you tried calibrating the coordinate system?"); Add error attribution questions to guide students to think more deeply.

[0187] Operation prompts: When students set unreasonable parameters (such as excessive rudder angle of the unmanned surface vessel), real-time warnings and suggestions for adjustment will be provided (e.g., "The current rudder angle of 30° may cause the vessel to capsize; it is recommended to adjust it to within 15°"); alternative solutions will be suggested.

[0188] Process progression: When the experiment stalls, push the next step guidance (such as "Please record the navigation distance at 3 different speeds").

[0189] In addition to the three types of guidance content mentioned above, this unit also strengthens the function of in-depth scenario-based guidance. For complex experimental steps, it generates linked guidance that combines principles and operations; the Q&A guidance adds error attribution-based counter-questions instead of simply open-ended questions; and the operation prompts add suggestions for alternative solutions.

[0190] Real-time feedback display unit: Displays the content guided by the intelligent agent in multiple formats.

[0191] Specifically, this includes text feedback (displaying guiding dialogue in a pop-up window on the interface), visual feedback (such as using arrows to mark the unmanned surface vessel's yaw direction and using waveform graphs to show the speech transcription effect), and dynamic effects (such as triggering a "sprinkling flower effect" when the experiment is successful).

[0192] The data acquisition and analysis module 104 collects and stores student-agent interaction data, experimental operation data, and feedback data locally. It then uses a locally deployed large model to perform real-time learning analysis, agent performance analysis, and anomaly detection on this multimodal data, while simultaneously generating a real-time visualization interface showcasing key classroom data. The data acquisition and analysis module 104 includes the following units:

[0193] Multimodal data acquisition unit: Collects interaction data between students and the intelligent agent, experimental operation data, and student feedback data on guidance, and uses local storage to ensure data security.

[0194] Interaction data: Records the voice / text dialogue content between students and the intelligent agent, including information such as question frequency and keywords;

[0195] Operational data: Records of parameter adjustments during the experiment (such as unmanned surface vessel rotation speed, image recognition threshold), operation time, and success rate;

[0196] Feedback data: Collect students' satisfaction ratings for the intelligent agent guidance, and self-labeled difficulty tags (such as "the model training steps are difficult to understand").

[0197] In this embodiment of the invention, all collected data is stored locally to reduce the risk of privacy leakage and comply with educational data security standards.

[0198] Real-time analysis unit: This unit performs learning analysis, agent performance analysis, and abnormal operation detection on the collected multimodal data based on a locally deployed large model, realizing real-time data processing.

[0199] Learning analysis: Identify high-frequency problems of students through word frequency analysis (e.g., if "neural network" appears too frequently, it is marked as a weak point), and judge students' inquiry ability by combining operation trajectory (e.g., whether they try multi-parameter combination);

[0200] Agent performance analysis: The accuracy of the response to the guiding script is evaluated by calculating the matching degree between the student's question and the agent's response;

[0201] Anomaly detection: Identify dangerous operations by students (such as unmanned surface vessel parameters exceeding safe range) and trigger an alert to ensure experimental safety.

[0202] Data Dashboard Unit: This unit generates a real-time visualization interface to centrally display key classroom data.

[0203] Key classroom data includes the following:

[0204] Student engagement metrics (such as the percentage of students who actively ask questions and the number of times they interact with the system);

[0205] Experiment progress statistics (such as the percentage of students who have completed a specific experimental step);

[0206] A ranking of trending topics (such as sorting high-frequency questions like "how to optimize image recognition accuracy").

[0207] These key data points allow teachers to keep abreast of classroom dynamics.

[0208] The evaluation report module 105 assesses student performance across dimensions such as participation, inquiry ability, and knowledge mastery. Based on interaction data, it provides suggestions for optimizing the agent's language, functions, and logic. By evaluating scenario suitability and interaction efficiency, it summarizes the overall classroom effect and proposes areas for improvement, ultimately integrating these into a multi-dimensional interaction effect evaluation report. Module 105 includes the following units:

[0209] Learning Ability Assessment Unit: This unit comprehensively assesses student performance from three dimensions: participation, inquiry ability, and knowledge mastery, and generates personalized learning suggestions.

[0210] Participation: This is measured by indicators such as the number of times students actively interact, the duration of experimental operations, the number of times they attempt independent exploration, and the percentage of students who actively ask questions.

[0211] Inquiry ability: It is evaluated by analyzing indicators such as the students' innovation in the problem-solving process (e.g., whether they propose unique solutions) and the rationality of the experimental plan (e.g., the logic of parameter setting);

[0212] Knowledge mastery: By assessing students' ability to apply knowledge points related to the scenario (such as whether they can explain experimental phenomena using professional concepts), personalized learning suggestions are generated (such as recommending practice content for weak areas).

[0213] Intelligent Agent Optimization Unit: Based on classroom interaction data, this unit provides directions for improving the intelligent agent in terms of speech optimization, function supplementation, and guidance logic adjustment.

[0214] Script optimization: Make the agent's responses to general questions more accessible and its responses to domain-specific questions more precise;

[0215] Additional features: Add relevant knowledge bases for specific domains to the intelligent agent (such as the principles and operational details of specific experiments).

[0216] Adjusting the guidance logic: Optimize guidance strategies for different types of students (e.g., reduce step prompts and increase open-ended questions to stimulate thinking for innovative and inquiry-based students).

[0217] Classroom Effectiveness Summary Unit: This unit outputs an overall evaluation of the classroom by assessing the suitability of the scenario and the efficiency of interaction, and proposes directions for improvement to optimize the classroom.

[0218] Scenario adaptability: This is evaluated by analyzing the degree of matching between the current teaching scenario and the preset teaching objectives (such as the degree to which the scenario covers the knowledge points);

[0219] Interaction efficiency: Measured by statistical data such as the average response time of the intelligent agent and the student problem-solving rate;

[0220] Areas for improvement: Based on the above analysis, propose suggestions for classroom optimization (such as supplementing specific scenarios to enhance the understanding of abstract knowledge, adjusting the pace of interaction, etc.).

[0221] Figure 3 This is a schematic block diagram of a computer device provided in an embodiment of this application. Figure 3 As shown, the device includes at least one processor 201, a memory 202, at least one network interface 204, and a user interface 206. The various components in the device are coupled together via a bus system 205. It will be understood that the bus system 205 is used to implement communication between these components. In addition to a data bus, the bus system 205 also includes a power bus, a control bus, and a status signal bus. However, for clarity, in… Figure 3 The general will label all buses as bus systems.

[0222] The user interface 206 may include a monitor, keyboard, mouse, trackball, clicker, button, touchpad, or touch screen.

[0223] It is understood that memory 202 can be volatile memory or non-volatile memory, or both. Non-volatile memory can be read-only memory (ROM) or programmable read-only memory (PROM), which serves as an external cache. By way of example, but not limitation, many forms of RAM are available, such as static random access memory (SRAM) and synchronous static random access memory (SSRAM). The memories described in the embodiments of this invention are intended to include, but are not limited to, these and any other suitable categories of memory.

[0224] In this embodiment of the invention, the memory 202 is used to store various types of data to support the operation of the electronic terminal. Examples of this data include: any executable program for operation on the electronic terminal, such as operating system 2021 and application program 2022; operating system 2021 includes various system programs, such as framework layer, core library layer, driver layer, etc., for implementing various basic services and handling hardware-based tasks. Application program 2022 may include various applications, such as media player, browser, etc., for implementing various application services. The AI-driven interactive assistance method for science and technology innovation classrooms provided in this embodiment of the invention can be included in application program 2022.

[0225] The methods provided in the above embodiments of the present invention can be applied to processor 201, or executed by processor 201. Processor 201 may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuit of the hardware in processor 201 or by instructions in software form. The processor 201 may be a general-purpose processor, a digital signal processor (DSP), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Processor 201 can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of the present invention. General-purpose processor 201 may be a microprocessor or any conventional processor, etc. The steps of the accessory optimization method provided in the embodiments of the present invention can be directly manifested as the hardware decoding processor executing the steps, or the hardware and software modules in the decoding processor combining to execute the steps. The software modules may be located in a storage medium, which is located in memory. The processor reads the information in the memory and combines it with its hardware to complete the steps of the aforementioned method.

[0226] In this embodiment of the invention, the electronic terminal may be used by one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), or complex programmable logic devices (CPLDs) to execute the aforementioned method.

[0227] The advantages of this invention are as follows: First, by using a segmented scenario library and personalized strategies for intelligent agents, it adapts the guidance logic to different student abilities, achieving personalized interactive guidance and enhancing students' interest in exploration and learning efficiency. Second, it employs a locally deployed large model to process multimodal data, achieving real-time and secure data processing, avoiding cloud transmission delays, ensuring real-time early warning and problem response for experimental operations, and reducing the risk of privacy leaks through local data storage, complying with educational data security standards. Third, it generates evaluation reports from multiple dimensions, including student abilities, intelligent agent performance, and overall classroom effectiveness, focusing not only on experimental results but also on the innovation and knowledge application abilities during the exploration process, helping teachers to accurately optimize teaching strategies. Fourth, it designs exclusive guidance logic and resource packages for science and technology innovation experiments, which can be deeply integrated with science and technology innovation scenarios, achieving linked teaching of experimental operations and theoretical knowledge, and enhancing students' understanding of complex science and technology innovation concepts. Fifth, through a multi-agent collaborative mechanism, it supports joint guidance of interdisciplinary projects, cultivating students' interdisciplinary thinking and comprehensive application abilities. In summary, the AI-driven interactive classroom technology for science and technology innovation provided by this invention, through the deep integration of large-scale model technology with science and technology innovation teaching scenarios, can achieve personalized, safe, controllable, and multi-dimensional evaluation of classroom interaction effects.

[0228] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. An AI-driven interactive assistance method for science and technology innovation classrooms, characterized in that, The AI-driven interactive support method for science and technology innovation classrooms includes the following steps: Construct an interactive scenario library for science and technology innovation classrooms. The scenario library is divided into scenario levels according to academic stage, and each scenario is associated with experimental resource packages and knowledge point mapping relationships. Classroom interactive intelligent agents are configured based on a large model platform, and these agents support lesson preparation assistance and interdisciplinary joint guidance. The intelligent agent performs real-time interactive guidance, which includes identifying the student's intent and generating guidance content. Collect multimodal interaction data and store it locally, and conduct real-time learning analysis through a large locally deployed model; The student performance was assessed based on participation, inquiry ability, and knowledge mastery. Suggestions for optimizing the intelligent agent's language, functions, and logic were proposed. The classroom effect was summarized by combining scenario adaptability and interaction efficiency, and a visual report was generated. The classroom interactive intelligent agent includes: The AI-powered teaching assistant for science and technology can automatically extract the core elements of an experiment to generate lesson plans that include objectives, steps, and precautions, and link them to relevant scenario resources. Collaborative rules for interdisciplinary intelligent agents include triggering mechanisms based on the depth of interdisciplinary intersection of problems and priority response rules based on scenario matching degree; Personalized guidance strategies and dynamic adjustment mechanisms based on student profile tags.

2. The AI-driven interactive assistance method for science and technology innovation classrooms according to claim 1, characterized in that, The construction of the interactive scenario library for science and technology innovation classrooms also includes further subdividing the scenario levels according to students' inquiry abilities and the types of science and technology innovation projects, and supports the construction of interdisciplinary composite scenarios, including: The scenarios are divided into levels based on the educational stage, students' inquiry abilities, and the type of science and technology innovation projects. The educational stages include primary school, junior high school, and senior high school, and the construction of interdisciplinary composite scenarios is supported. Each level of scenario is associated with an experimental resource package and scenario-specific safety specifications that match its corresponding learning stage and ability level. The experimental resource package includes tools, simulation resources and experimental procedure breakdown documents, and the difficulty and presentation format of the resources can be dynamically adjusted based on classroom interaction data. We construct a dynamic mapping relationship between scenario levels and multidisciplinary knowledge points, link it with the ability development goals such as scientific inquiry and interdisciplinary application, and adjust the mapping weights based on student interaction data to achieve gradient coverage of the same knowledge point in different scenarios.

3. The AI-driven interactive assistance method for science and technology innovation classrooms according to claim 1, characterized in that, The classroom interactive intelligent agent configured based on the large model platform includes: Deploy the basic intelligent agent framework, and configure the natural language processing interface, experimental scenario association interface, and local data interaction interface; Import the science and technology innovation curriculum outline and experimental standard procedures, automatically extract the core links of the experiment to generate lesson plans, support personalized modification of lesson plans and synchronously update related resources and knowledge points; Register multidisciplinary intelligent agents and assign domain labels, set up a dynamic collaborative triggering mechanism based on the depth of cross-disciplinary issues, and set up multi-agent response priority rules based on scenario matching degree; Based on students' historical experiment records and interactive data, profile tags are generated, guidance difficulty and feedback style parameters are set, and a dynamic strategy adjustment mechanism is configured.

4. The AI-driven interactive assistance method for science and technology innovation classrooms according to claim 1, characterized in that, The collaborative rules include agent collaboration triggering conditions and multi-agent collaborative response priorities, and the configuration of personalized guidance strategies includes: Generate student profile tags; Set guidance strategy parameters based on student profile tags; Configuration strategy dynamic adjustment mechanism.

5. The AI-driven interactive assistance method for science and technology innovation classrooms according to claim 1, characterized in that, The real-time interactive guidance performed by the intelligent agent includes: Analyze student multimodal input, which includes voice input, text input, and operation input; Generate targeted guidance content, which includes Q&A guidance, operation prompts, and process advancement; Display real-time feedback, including text feedback, visual feedback, and dynamic effect feedback.

6. The AI-driven interactive assistance method for science and technology innovation classrooms according to claim 1, characterized in that, The collection and analysis of multimodal interaction data includes: Multimodal data is collected, including interaction data, operation data, and feedback data. The interaction data includes the voice / text dialogue content between students and the intelligent agent, the frequency of questions, and keywords. The operation data includes experimental parameter adjustment records, operation duration, step success rate, and operation trajectory. The feedback data includes students' satisfaction ratings of the guidance content and self-labeled difficulty tags. All data is stored locally with encryption to ensure the privacy and security of educational data. The interaction data also includes student facial expression data and body movement data collected through classroom cameras, which are extracted using the OpenCV algorithm and then included in the analysis. The collected multimodal data is analyzed in real time using a local large model. The real-time analysis includes learning analysis, agent performance analysis, and anomaly detection. The learning analysis includes identifying weak knowledge points through word frequency analysis and judging inquiry ability by combining operation trajectory. The agent performance analysis includes calculating the matching degree between student questions and guidance content and evaluating response accuracy. The anomaly detection includes identifying dangerous parameter settings and equipment operation errors in experimental operations and triggering real-time warnings. Generate a data dashboard that displays student participation metrics, experiment progress statistics, hot topic rankings, and abnormal operation alerts in real time, supporting teachers to dynamically intervene in the classroom.

7. The AI-driven interactive assistance method for science and technology innovation classrooms according to any one of claims 1 to 6, characterized in that, The multi-dimensional interaction effect evaluation report generated based on data analysis results includes: The system generates student competency assessment content based on participation, inquiry ability, and knowledge mastery. Participation is quantified based on the number of active interactions, experimental operation time, number of independent inquiry attempts, and the proportion of questions asked. Inquiry ability includes analyzing the rationality of experimental plans, the innovation of problem-solving, and the frequency of interdisciplinary knowledge application. Knowledge mastery includes assessing the depth of understanding of scenario-related knowledge points and generating personalized practice recommendations. The paper proposes suggestions for agent optimization, which include speech optimization, function supplementation, and guidance logic adjustment. Speech optimization includes simplifying responses to general questions and improving the accuracy of responses to domain-specific questions. Function supplementation includes adding a knowledge base for scarce domains and optimizing the accuracy of multimodal input parsing. Guidance logic adjustment includes reducing step prompts and adding open-ended questions for innovative and inquiry-based students, and strengthening guidance on operational details for students with weak foundations. The overall classroom effect is summarized and evaluated by combining scenario adaptability and interaction efficiency. Scenario adaptability includes evaluating the matching degree between scenario resources and teaching objectives and the completeness of cross-disciplinary knowledge coverage. Interaction efficiency includes statistical analysis of the average response time of the intelligent agent, student problem-solving rate, and the percentage of experimental downtime. The student ability assessment content, agent optimization suggestions, and overall classroom effectiveness are integrated into a complete report in the form of text and visual charts.

8. An AI-driven interactive auxiliary system for science and technology innovation classrooms, characterized in that: The AI-driven interactive support system for science and technology innovation classrooms includes: The scenario construction module (101) divides the interactive scenario levels of science and technology innovation classrooms according to the learning stage, associates and adapts experimental resource packages and safety specifications for each level of scenario, establishes a dynamic mapping relationship between scenarios and multi-disciplinary knowledge points, and supports dynamic adjustment of scenario levels, resources and knowledge point mapping based on classroom interaction data. The intelligent agent configuration module (102) deploys the basic framework of classroom interactive intelligent agents based on the large model platform, and configures the lesson preparation assistance and interactive speech customization functions of the science and technology innovation AI teaching assistant, the joint guidance rules of cross-disciplinary intelligent agents, and the personalized guidance strategy based on students' historical data and real-time performance. The interactive execution module (103) receives and parses the student's multimodal input, such as voice, text, operation, and facial expressions or body movements. Based on the scene library and intelligent agent configuration, it generates targeted guidance content such as question and answer guidance, operation prompts, and process advancement, and displays real-time feedback in the form of text, visualization and dynamic effects. The data acquisition and analysis module (104) collects and stores the interaction data, experimental operation data and feedback data between students and intelligent agents in a localized manner. It performs real-time learning analysis, intelligent agent performance analysis and anomaly detection on these multimodal data through a large model deployed locally, and generates a real-time visualization interface to display key classroom data. Evaluation report module (105) evaluates student performance from the dimensions of participation, inquiry ability and knowledge mastery, provides suggestions for optimizing the agent's language, functions and logic based on interactive data, summarizes the overall classroom effect by evaluating the scene adaptability and interaction efficiency, proposes improvement directions and finally integrates them into a multi-dimensional interactive effect evaluation report.

9. The AI-driven interactive auxiliary system for science and technology innovation classrooms according to claim 8, characterized in that, The scene construction module (101) includes: The hierarchical division unit is used to divide the scene hierarchy according to the learning stage, student ability and project type, supports the construction of cross-disciplinary composite scenes, and dynamically adjusts the hierarchical structure and related resources based on classroom data; The resource association unit is used to associate and adapt experimental resource packages and safety specifications for each scenario, and dynamically adjust the difficulty and presentation of resources. The knowledge point mapping unit is used to establish a dynamic mapping relationship between scenarios and multidisciplinary knowledge points, associate it with ability development goals, and adjust the mapping weights based on interactive data to achieve gradient mapping of the same knowledge point in different scenarios. The agent configuration module (102) includes: The teaching assistant configuration unit is used to configure lesson preparation assistance and customized interactive scripts for science and technology innovation AI teaching assistants; The collaborative rule unit is used to set joint guidance rules for interdisciplinary intelligent agents. The joint guidance rules include a dynamic collaborative triggering mechanism based on the depth of cross-disciplinary intersection of the problem and an integration rule for the differential responses of multiple intelligent agents. Personalized strategy unit, used to configure guidance difficulty and feedback style based on students' historical data and real-time performance; The interactive execution module (103) includes: A multimodal input parsing unit is used to receive and parse students' multimodal input information and combine it with semantic analysis to improve the accuracy of intent recognition; The guidance logic generation unit is used to generate guidance content that adapts to the scenario and student needs based on the scenario library and intelligent agent configuration. The real-time feedback display unit is used to display the intelligent agent's guidance content in multiple formats; The data acquisition and analysis module (104) includes: The multimodal data acquisition unit is used to collect interaction data between students and intelligent agents, experimental operation data, and student feedback data on guidance, and uses local storage to ensure data security. The real-time analysis unit is used to perform learning analysis, agent performance analysis, and abnormal operation detection on the collected multimodal data based on a locally deployed large model, thereby realizing real-time data processing. The data dashboard unit is used to generate a real-time visual interface to centrally display key classroom data. The evaluation report module (105) includes: The learning ability assessment unit is used to comprehensively evaluate students' performance from three dimensions: participation, inquiry ability, and knowledge mastery, and generate personalized learning suggestions. The agent optimization unit is used to provide direction for agent improvement based on classroom interaction data, including speech optimization, function supplementation, and guidance logic adjustment. The classroom effectiveness summary unit is used to output an overall evaluation of the classroom by assessing the suitability of the scenario and the efficiency of interaction, and to propose directions for improvement to optimize the classroom.

10. A computer device, characterized in that, The computer device includes: a memory (202), a processor (201), and a computer program stored on the memory (202), characterized in that the processor executes the computer program to implement the AI-driven interactive assistance method for science and technology innovation classrooms as described in any one of claims 1 to 7.

Images