Interactive teaching system for labor education

Abstract

The invention discloses a labor education interactive teaching system, belongs to the technical field of education, and aims to solve the problems of rigid evaluation and non-personalized guidance of traditional labor skill training. The system comprises a multi-modal data acquisition module, a baseline management module, a contradiction analysis module, a strategy generation module and a multi-modal interaction terminal module. According to the method, a personal dynamic ability baseline is established, an operation-state contradiction is diagnosed based on deep coupling of real-time behavior data and physiological data, and a composite guidance strategy containing body relaxation guidance or step-by-step technology guidance can be generated in a targeted manner. According to the system, visual, auditory and tactile feedbacks are output in real time through the multi-mode interaction terminal, and the accuracy of labor skill teaching, the fairness of evaluation and the skill generalization ability of a learner in a high-voltage situation are remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of education, in particular to a labor education interactive teaching system. BACKGROUND

[0002] Labor education and vocational skill training are key links to cultivate technical and skilled talents needed by society, and the core lies in the development of practical operation ability. With the development of information technology, interactive teaching systems, especially those based on virtual reality, augmented reality and sensor technology, have gradually been applied in various skill training. They provide a safe, controllable and repeatable training environment for learners by simulating real work scenarios and operation tasks.

[0003] In the prior art, some interactive teaching systems usually guide and evaluate learners based on a set of preset expert models or standard operation procedures. Such systems capture learners' operation actions through motion capture devices and compare them with standardized action sequences stored in the system. When deviations are detected between learners' operations and standards, the system will give feedback through screen prompts, highlighting error areas or playing demonstration videos. After completing the task, the system will give a quantitative score based on the accuracy of the operation, completion time and other objective indicators, as an evaluation of the learner's training effect.

[0004] However, the above prior art solution has obvious shortcomings. First, the evaluation standard it uses is uniform and fixed, which cannot adapt to individual differences in basic ability, learning progress and psychological tolerance among different learners, leading to a disconnection between teaching content and learners' actual needs. Second, the feedback mechanism of such systems is relatively superficial. They can only point out the deviations in operation behavior, but cannot explore the underlying reasons behind the deviations. For example, the system cannot distinguish between skill inexperience and learner's inner nervousness, resulting in a lack of targeted guidance. Finally, the interaction mode of existing systems is relatively single, mainly relying on visual prompts, and the guidance to learners is one-way indoctrination, lacking dynamic response to learners' real-time state, making it difficult to form effective teaching interaction. SUMMARY

[0005] To solve the above problems, the present application provides a labor education interactive teaching system, which uses multi-modal data collection to construct individual dynamic ability baseline, identifies the internal contradictions between operation and state to generate targeted guidance strategies, and can achieve personalized and deep adaptive teaching of learners' skills and psychology.

[0006] The above objectives can be achieved by the following solutions:

[0007] The application discloses an interactive teaching system for labor education, which comprises a multi-modal data acquisition module, a baseline management module, a contradiction analysis module, a strategy generation module and a multi-modal interactive terminal module.

[0008] Optionally, the multi-modal data acquisition module comprises an action capture unit, a non-contact physiological sensing unit, an embedded sensor unit, a task information retrieval unit and a historical performance acquisition unit.

[0009] Optionally, the baseline management module comprises a skill baseline construction unit, a psychological baseline modeling unit and a dynamic adjustment unit.

[0010] Optionally, the contradiction analysis module comprises: an operation deviation analysis unit configured to analyze deviation of the current action sequence data relative to the skill performance baseline to generate an operation specification deviation feature; a psychological load analysis unit configured to analyze deviation of the current physiological response data relative to the psychophysiological response baseline to generate a psychological load state feature; a contradiction type determination unit configured to determine a dominant causal relationship between the operation specification deviation feature and the psychological load state feature in a time sequence to determine a contradiction type; and a priority calculation unit configured to fuse the operation specification deviation feature, the psychological load state feature and the contradiction type to calculate a contradiction resolution priority index.

[0011] Optionally, the strategy generation module comprises: a first type strategy generation unit configured to generate a composite guidance strategy comprising an embodied relaxation guiding step and a task target simplification step if the contradiction type is determined to be a first type of operation deformation caused by psychological load; a second type strategy generation unit configured to generate a composite guidance strategy comprising a micro-success experience construction step and a step-by-step technical guidance step if the contradiction type is determined to be a second type of psychological load caused by operation unfamiliarity; and an instruction conversion unit configured to convert the generated composite guidance strategy into an instruction sequence executable by a multi-modal interactive terminal according to the task context information.

[0012] Optionally, the multi-modal interactive terminal comprises: an augmented reality display unit configured to present a virtual instructor image, wherein the virtual instructor performs a visual guidance part in the instruction sequence; an audio unit configured to play a guidance voice matching the contradiction type and the instruction sequence; and a somatosensory interaction unit configured to apply a tactile feedback corresponding to the instruction sequence to a relevant part of the learner's body.

[0013] Optionally, the system further comprises: a skill structure disassembly module configured to disassemble a skill structure of the labor task to obtain a core principle feature and a peripheral condition feature; a transfer task generation module configured to generate a skill transfer training task by keeping the core principle feature unchanged and changing the peripheral condition feature; and a skill generalization evaluation module configured to guide the learner to perform the skill transfer training task and reacquire multi-modal perception data to evaluate skill generalization ability.

[0014] Optionally, the transfer task generation module comprises: a contradiction type statistical unit configured to statistically analyze historical contradiction type data of the learner to determine a peripheral condition feature associated with a high-frequency appearing contradiction type; and a gradient difficulty setting unit configured to generate a skill transfer training task with gradient difficulty according to the peripheral condition feature associated with the high-frequency appearing contradiction type.

[0015] Optionally, the system further comprises: a feedback data acquisition unit, configured to acquire new multi-modal perception data obtained in the skill transfer training task and the identified new operation-state contradiction as feedback data; and a model optimization update unit, configured to optimize and update the model used to construct the personal dynamic ability baseline and the model used to identify the operation-state contradiction by using the feedback data.

[0016] Based on the same inventive concept, the application further provides a labor education interactive teaching method, which comprises: acquiring multi-modal perception data, task context information and historical performance data of a learner generated when the learner performs a labor task, wherein the multi-modal perception data comprises action sequence data, physiological response data and physical parameter data; constructing and updating a personal dynamic ability baseline comprising a skill performance baseline and a psychophysiological response baseline based on the historical performance data of the learner and current task context information; comparing and analyzing the current multi-modal perception data with the personal dynamic ability baseline to identify and quantify operation-state contradiction representing the influence of operation on state, and generating a contradiction resolution priority index; generating a contradiction resolution guidance strategy comprising specific guidance steps based on the operation-state contradiction and the contradiction resolution priority index; and executing the contradiction resolution guidance strategy to output multi-modal guidance information comprising visual, auditory or tactile guidance to the learner.

[0017] Compared with the prior art, the application has the following advantages:

[0018] The application realizes real individualized teaching by constructing a personal dynamic ability baseline. Instead of using rigid and unified expert standards, the system generates a dynamically changing personalized evaluation benchmark according to the historical performance of the learner and the current task situation. This enables the difficulty of teaching and the evaluation standard to match the ability boundary and psychological state of the learner in real time, effectively stimulating the potential of the learner and avoiding frustration or complacency caused by inappropriate standards, thereby improving the pertinence and efficiency of teaching.

[0019] The application can perform deep diagnosis on the performance of the learner, breaking through the limitation of traditional teaching that can only judge right or wrong. By introducing the concept of operation-state contradiction, the causal relationship between operation failure and psychophysiological state can be analyzed and identified, and it can be judged whether the failure is due to skill deficiency or psychological stress. This accurate insight into the root cause enables the provision of a root-cause-oriented guidance strategy, realizing the synchronous cultivation and coordinated improvement of the operation skills and psychological quality of the learner.

[0020] The application provides an immersive, multi-dimensional interactive guidance experience. Abstract guidance strategies are translated into visual demonstrations by virtual instructors, synchronized voice explanations, and tactile guidance of key body parts. This integrated visual, auditory, and tactile multi-modal interaction makes guidance information more intuitive, easier to understand, and able to be memorized by the body, reducing the cognitive load of learners, enhancing the participation and reality of the learning process, and thus effectively accelerating the process from theoretical cognition to skill internalization.

[0021] The application establishes a complete closed loop from skill training to skill generalization evaluation, and then to system self-optimization. Not only a single task is taught, but also the ability of the learner to draw inferences from a single case is actively cultivated and evaluated through skill structure disassembly and transfer task generation. Furthermore, the performance data of the learner in the transfer task can be used as feedback to iteratively update the analysis model and evaluation baseline of the system, so that the teaching system itself has the ability to learn and evolve, and can continuously improve its teaching accuracy and effectiveness.

[0022] Other features and advantages of the present application will be set forth in the following description, and in part will become apparent to those skilled in the art from the description, or can be learned by practice of the application. The objects and other advantages of the application will be realized and attained by the structure particularly pointed out in the written description and claims hereof as well as the appended drawings. BRIEF DESCRIPTION OF DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0024] Figure 1 is a structural schematic diagram of a labor education interactive teaching system according to an embodiment of the present application.

[0025] Figure 2 is a real-time resolution curve of operation-state contradiction according to an embodiment of the present application.

[0026] Figure 3 is a comparison diagram of the improvement effect of a composite guidance strategy on skill accuracy according to an embodiment of the present application.

[0027] Figure 4 is a flowchart of a labor education interactive teaching method according to an embodiment of the present application. DETAILED DESCRIPTION

[0028] In order to make the purposes, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some but not all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0029] With reference to Figure 1 One embodiment of the present application proposes a labor education interactive teaching system, which adopts the technical means of collecting multi-modal data to construct a personal dynamic ability baseline, identifying the internal contradiction between operation and state to generate a targeted guidance strategy, and can realize personalized and deep self-adaptive teaching of the skills and psychology of learners.

[0030] The system of the embodiment specifically comprises:

[0031] A multi-modal data collection module, configured to acquire multi-modal perception data and task context information generated by a learner when performing a labor task, and historical performance data of the learner, wherein the multi-modal perception data comprises action sequence data, physiological response data and physical parameter data;

[0032] A baseline management module, configured to construct and update a personal dynamic ability baseline comprising a skill performance baseline and a psychophysiological response baseline based on the historical performance data of the learner and current task context information;

[0033] A contradiction analysis module, configured to compare and analyze the current multi-modal perception data and the personal dynamic ability baseline, identify and quantify operation-state contradiction bodies representing the correlation and influence of operation and state, and generate a contradiction resolution priority index;

[0034] A strategy generation module, configured to generate a contradiction resolution guidance strategy comprising specific guidance steps based on the operation-state contradiction bodies and the contradiction resolution priority index;

[0035] A multi-modal interactive terminal module, configured to execute the contradiction resolution guidance strategy and output multi-modal guidance information comprising visual, auditory or tactile guidance to the learner.

[0036] Specifically, through the multi-modal data acquisition module, the action, physiology and physical interaction data of the learner in performing the task are comprehensively captured, and the historical performance and the current task context are combined. Instead of using a unified expert standard, the baseline management module dynamically constructs a personalized dynamic ability baseline for each learner, which includes both skill and psychological dimensions. The contradiction analysis module compares the current performance of the learner with the personalized baseline in real time. The key innovation lies in identifying and quantifying the "operation-state contradiction" between the operation behavior and the psychological and physiological state, and evaluating the urgency of its solution. Based on the deep diagnosis of the contradiction, the strategy generation module can develop a guidance strategy rooted in the nature of the problem, which contains specific steps. The strategy is converted into guidance information that the learner can intuitively perceive through multi-modal interaction terminals, such as visual, auditory and tactile sensory channels, thereby forming a complete and adaptive teaching closed loop from data acquisition, personalized analysis, intelligent decision-making to multi-modal feedback.

[0037] Optionally, the multi-modal data acquisition module comprises:

[0038] An action capture unit is configured to capture the whole body and hand movements of the learner to obtain action sequence data.

[0039] A non-contact physiological sensing unit is configured to capture facial thermal imaging and visual images, and calculate heart rate variability trends and eye tracking data as physiological response data from the images.

[0040] An embedded sensor unit is configured to capture operation force and angle data as physical parameter data.

[0041] A task information retrieval unit is configured to retrieve preset steps, difficulty levels and tool material attributes corresponding to the labor task from a task library as task context information.

[0042] A historical performance acquisition unit is configured to retrieve and acquire historical multi-modal perception data and operation result data of the learner under different task difficulties and different psychological states from a learner archive database to obtain historical performance data.

[0043] Specifically, the action capture unit is configured to accurately quantify the operation posture and limb trajectory of the learner. The unit uses a wearable device containing at least 17 inertial measurement units (IMUs) fixed on the main joint parts of the learner, supplemented by a data glove integrating bending and inertial sensors. The raw data of each sensor is continuously collected at a sampling frequency of 60 Hz, and is fused and processed by Kalman filtering algorithm to output action sequence data. The action sequence data is a timestamp-aligned data stream, and each frame of data contains the three-dimensional space coordinates of 23 key joint nodes of the whole body and quaternion attitude , so as to accurately reproduce the learner's whole body and hand micro-operation. The motion sequence data is the basis for subsequent baseline comparison of skill performance, to evaluate the fluency, stability and normativity of the operation. The non-contact physiological sensing unit realizes the non-interference monitoring of the learner's psychological and physiological state. The unit deploys a dual-spectrum camera facing the learner's face, integrating visible light RGB and long-wave infrared LWIR sensors. The face key point detection algorithm is used to lock the skin area around the forehead and nasal ala as the region of interest (ROI) in the infrared thermal imaging video stream. The remote photoplethysmography rPPG algorithm is run on the ROI to extract the weak signal of the pixel mean change over time, and then calculate the beat signal BVP. The peak value detection of the BVP signal obtains the beat-to-beat interval RRi, and the standard deviation SDNN of RRi is calculated in a 30-second sliding time window, so as to quantitatively obtain the heart rate variability trend. The heart rate variability trend is an objective indicator to measure the activity of the autonomic nervous system, and its decrease is usually related to the increase of psychological load or tension. The eye tracking algorithm is synchronously run on the visible light video stream, and the real-time gaze point coordinates of the learner on the operation table are calculated by identifying the pupil center position and combining the pre-labeled scene camera. By analyzing the dwell time and saccade path of the gaze point, eye tracking data including gaze duration, saccade frequency and saccade amplitude are obtained. The heart rate variability trend and eye tracking data together constitute the physiological response data, which are used to evaluate the cognitive load and attention allocation state of the learner. Thin film force sensing resistor FSR is integrated at the handle of the key operation tool such as screwdriver to collect operation force data; six-axis inertial measurement unit is integrated inside to collect operation angle data. These sensors convert the measured analog signals into digital signals through analog-to-digital converter ADC and sample at a frequency of 100 Hz. The output physical parameter data is time-synchronized with the aforementioned data stream, including time series of operation force in Newton and tool posture angle in degree, to judge whether there are specific operation problems such as excessive force and inaccurate angle. When the learner selects a labor task, a query request based on task ID is sent to the task library database in the form of background service. Each record in the task library defines the preset steps of the task, i.e. the decomposition items of the standard operation procedure SOP; the difficulty level, a whole number scale of 1 to 5; and the tool material attributes, a JSON object describing the required tool specifications and material properties. The output of this unit is the task context information, which provides a clear reference standard for subsequent baseline dynamic adjustment and strategy generation. The historical performance acquisition unit provides data support for personalized baseline modeling. According to the unique identity ID of the current learner, the learner archive database is retrieved. This database persistently stores the full set of multi-modal perception data of the learner's previous task execution, task context information, and final operation result data, such as completion time and finished product quality score.The query logic of the unit will first retrieve the historical records matching the difficulty level and task type in the current task context information, and package these records into a dataset. This dataset is the historical performance data, which is the input source for building the individual dynamic ability baseline, ensuring the personalization and adaptability of the teaching guidance.

[0044] Optionally, the baseline management module comprises:

[0045] a skill baseline construction unit for extracting the action pattern features of the learner under different task difficulties from the historical performance data to form a skill performance baseline;

[0046] a psychological baseline modeling unit for establishing a physiological response feature model of the learner under different psychological states from the historical performance data to form a psychological and physiological response baseline;

[0047] a dynamic adjustment unit for dynamically adjusting the expected parameter range in the skill performance baseline according to the current task context information and the real-time acquired physiological response data.

[0048] Specifically, the skill baseline building unit processes the action sequence data obtained from the historical performance data. It first groups the historical data according to the task difficulty level in the task context information. For a certain task step under a certain difficulty, the unit uses a time series clustering algorithm, such as K-Means clustering based on dynamic time warping (DTW) distance, to align and group the action sequence data of multiple repeated operations, identifying the mainstream operation mode. For the action sequences belonging to the optimal category, extract their key action pattern features, such as the peak angular velocity of the key joints, the spatial smoothness of the operation path, and the duration of the sub-actions. By calculating the mean and standard deviation of these features, a multi-dimensional Gaussian model is formed. This model is the skill performance baseline for this step under this difficulty, defining the expected operation range of the learner in the "proficient state". The psychological baseline modeling unit establishes the mapping relationship between the learner's physiological indicators and psychological state. It interprets the psychological meaning behind the physiological response data. This unit uses historical performance data to conduct correlation analysis between physiological response data and corresponding operation result data. Using a supervised learning method, it labels the segments with excellent performance in the historical data as "normal load", and labels the segments with poor performance and dramatic fluctuations in physiological indicators as "high load". Based on these labeled data, a classification model is trained, such as a support vector machine (SVM) or a gradient boosting decision tree (GBDT). This model constitutes the psychological and physiological response baseline. In real-time teaching, the current physiological response data is input into this model to obtain a probabilistic assessment of the learner's current psychological load, such as "high load state probability is 85%". The skill performance baseline is adjusted in real time by the dynamic adjustment unit, so that it can adapt to the current task situation and learner state. Avoid misjudgment due to task difficulty or learner tension. The triggering condition of this unit is the change of the current task context information or the significant deviation of the real-time physiological response data from the psychological and physiological response baseline. When triggered, the unit temporarily modifies the parameters in the skill performance baseline according to the adjustment function. The adjustment process is represented by the following formula:

[0049] ;

[0050] wherein, is the expected parameter in the adjusted skill performance baseline. is the static baseline parameter obtained from the skill baseline building unit. is the task complexity factor calculated according to the current task context information, such as the normalized difference obtained by comparing the current task difficulty level with the difficulty level used in the baseline building. is the psychological load factor obtained by comparing the current physiological response data with the psychological and physiological response baseline, such as the "high load state" probability value output by the psychological baseline modeling unit. and are preset weight coefficients respectively used to adjust the influence degree of task complexity and psychological load on baseline adjustment, the value of which is usually between 0.1 and 0.5, and is obtained by expert knowledge or offline optimization. This formula ensures that when the task is more difficult or the learner is more nervous, the evaluation standard is appropriately relaxed, so that the individual dynamic ability baseline is closer to the instantaneous ability boundary of the learner.

[0051] Optionally, the contradiction analysis module comprises:

[0052] An operation deviation analysis unit is configured to analyze the deviation degree of the current action sequence data relative to the skill performance baseline, and generate an operation specification deviation feature;

[0053] A psychological load analysis unit is configured to analyze the deviation degree of the current physiological response data relative to the psychophysiological response baseline, and generate a psychological load state feature;

[0054] A contradiction type determination unit is configured to determine the dominant causal relationship in time sequence between the operation specification deviation feature and the psychological load state feature, and determine the contradiction type;

[0055] A priority calculation unit is configured to fuse the operation specification deviation feature, the psychological load state feature and the contradiction type, and calculate a contradiction resolution priority index.

[0056] Specifically, the operation deviation analysis unit quantifies the gap between the current operation and the individual standard. This unit compares the real-time collected motion sequence data with the skill performance baseline in the individual dynamic ability baseline. Using the dynamic time warping (DTW) algorithm, the distance between the current motion sequence and the optimal operation mode average sequence stored in the skill performance baseline for this step is calculated. This distance value is normalized to generate a value ranging from 0 to 1 as the operation specification deviation feature. The larger the feature value, the more serious the deviation of the current operation from the learner's own best level, and there may be problems such as motion errors, rhythm disorders, or hesitation. The psychological load analysis unit assesses the learner's current psychophysiological state. This unit takes the real-time acquired physiological response data, i.e., heart rate variability trend and eye tracking data, as input, and sends it to the trained psychological state classification model in the psychophysiological response baseline. The model outputs the probability value of being in a "high load" state, which is the psychological load state feature. The closer the feature value is to 1, the higher the learner's current level of tension, anxiety, or cognitive load, and they may be facing challenges beyond their current psychological capacity. The contradiction type determination unit explores the dominant causal relationship between operation deviation and psychological load. This unit performs lead-lag correlation analysis on the operation specification deviation feature and the psychological load state feature, which are two time series, within a sliding time window of about 5 seconds. By calculating the Pearson correlation coefficient of the two sequences at different time offsets, the largest correlation offset is found. If the peak value of the psychological load state feature is on average 0.5 to 1.5 seconds ahead of the peak value of the operation specification deviation feature, it is determined that the contradiction type is the first type, i.e., the operation is distorted due to high psychological load; otherwise, if the peak value of the operation specification deviation feature is ahead, it is determined that the contradiction type is the second type, i.e., the psychological load is caused by unskilled operation. This determination result provides a fundamental basis for subsequent selection of guidance strategies. The priority calculation unit integrates all the analysis results mentioned above to generate a comprehensive intervention index. It decides when and how intensively to intervene in guidance. This unit uses a weighted model to calculate the contradiction resolution priority index, and its calculation formula is as follows:

[0057] ;

[0058] wherein, represents the final contradiction resolution priority index. is the operation specification deviation feature generated by the operation deviation analysis unit, and its value comes from the normalized DTW distance. is the psychological load state feature generated by the psychological load analysis unit, and its value is the probability of being in a high load state. and are the weight coefficients of operation deviation and mental load respectively, usually preset between 0.3 and 0.7, and their sum is 1, which are adjusted according to the task nature, for example, the weight of fine operation task is higher. is the contradiction type adjustment factor determined by the contradiction type determination unit, if the first type is determined, i.e. the mental problem is the root cause, the value of can be 1.2, indicating that more priority is needed for processing; if the second type is determined, the value of can be 1.0. The index is a dimensionless comprehensive score, when it exceeds the preset threshold 0.75, the subsequent strategy generation module is triggered. As shown in Figure 2 , the changes of the two core indicators, "operation norm deviation feature" and "mental load state feature", in the time series of task execution of the learner are shown. After the composite guidance strategy intervention at the moment, both indicators show a rapid decline and convergence trend, verifying the effectiveness of contradiction analysis and real-time guidance.

[0059] Optionally, the strategy generation module comprises:

[0060] a first type strategy generation unit, configured to generate a composite guidance strategy comprising a somatic relaxation guidance step and a task target simplification step if the contradiction type determination is the first type that the operation is deformed due to mental load;

[0061] a second type strategy generation unit, configured to generate a composite guidance strategy comprising a micro-success experience construction step and a step-by-step technical guidance step if the contradiction type determination is the second type that the mental load is caused by operation unfamiliarity;

[0062] an instruction conversion unit, configured to convert the generated composite guidance strategy into an instruction sequence executable by a multi-modal interactive terminal according to the task context information.

[0063] ​Specifically, when the contradiction type determination unit determines the contradiction type as the first type, i.e., the operation is distorted due to psychological load, the first type strategy generation unit is activated. This unit aims to first alleviate the negative emotions of the learner, and then guide him / her to return to the correct operation. A composite guidance strategy containing two parts will be generated. The first part is the embodiment relaxation guidance step, which triggers a series of instructions aimed at reducing the physiological arousal level. For example, an instruction is generated, asking the virtual instructor to say "deep breath, relax your shoulders" in a slow and soft tone, while triggering the virtual instructor to make a demonstrative deep breathing action in the augmented reality display unit. The second part is the task target simplification step. The preset steps of the current task are analyzed, and the most difficult sub-action is located based on the operation specification deviation characteristics. Then, the target of this sub-action is temporarily simplified, for example, if the original target is "tighten the screw to 5 Newton-meters torque at one time", it is simplified to "first tighten the screw halfway". This strategy helps the learner regain confidence by reducing cognitive load, thereby restoring the operation level. If the contradiction type determination unit determines the contradiction type as the second type, i.e., the psychological load is caused by the lack of operation proficiency, the second type strategy generation unit is activated. By improving skill proficiency, the root cause of negative emotions is eliminated. This unit also generates a composite guidance strategy. The first part is the micro-success experience construction step, which identifies the small part of the current operation that is basically correct even with deviations. For example, even if the learner's angle of holding the screwdriver is incorrect, as long as the screw cap position is found, an instruction is generated to give immediate positive feedback through the audio unit, such as "good, you got it". This immediate affirmation helps to break the negative emotional cycle caused by frustration. The second part is the step-by-step technical guidance step, which decomposes the erroneous action into more detailed micro-action units according to the operation specification deviation characteristics. For example, for "incorrect angle of holding the screwdriver", a series of instructions are generated to first guide "adjust the wrist angle" and then "keep the elbow stable", and a correct angle indicator is superimposed in the learner's field of view through the augmented reality display unit as visual guidance. This decomposition and guidance help the learner accurately master the correct technical essentials. Regardless of the composite guidance strategy generated, the instruction conversion unit will convert it into a sequence of instructions that can be directly executed by the multi-modal interaction terminal. Abstract guidance strategies are translated into specific machine instructions. This unit is a protocol converter that receives strategy descriptions from the first or second type strategy generation unit and generates time-synchronized instruction sequences based on the current task context information. This instruction sequence is a JSON format script, where each entry contains a timestamp, target interaction device, instruction type, and specific parameters. For example, an instruction of a step-by-step technical guidance step may be converted into three parallel instructions: one instruction drives the augmented reality display unit to superimpose a rotation arrow on the learner's wrist, one drives the audio unit to play "please rotate your wrist in the direction of the arrow", and one drives the somatosensory interaction unit to apply a slight rotation direction tactile cue to the learner's wrist.This executable instruction sequence is the key to realize multi-modal and immersive guidance. Figure 3 As shown in the table, the average percentage of improvement in labor skill accuracy using the composite guidance strategy of the present application and the traditional standard guidance is compared. The data shows that the present application strategy can be significantly higher than the traditional method, effectively verifying the effectiveness of the personalized guidance based on the "operation-state contradiction".

[0064] Optionally, the multi-modal interactive terminal comprises:

[0065] An augmented reality display unit is configured to present a virtual tutor image, wherein the virtual tutor performs visual guidance in the instruction sequence;

[0066] An audio unit is configured to play guidance voice matched with the contradiction type and the instruction sequence;

[0067] A somatosensory interaction unit is configured to apply tactile feedback corresponding to the instruction sequence to the relevant part of the learner's body.

[0068] Specifically, intuitive visual guidance is provided by the augmented reality display unit. This unit employs a pair of augmented reality (AR) glasses that integrate a semi-transparent display screen and a front-facing camera, with a field of view (FOV) no less than 50 degrees and a refresh rate maintained at 90 Hz to avoid dizziness. When receiving the part of the instruction sequence generated by the instruction conversion unit that pertains to visual guidance, the rendering engine of the AR glasses loads and presents a virtual instructor image in real time. This virtual instructor is a three-dimensional character model whose movements are driven by animation data in the instruction sequence. For example, when executing the embodiment relaxation guidance step, the virtual instructor performs slow and deep breathing movements; when executing the step-by-step technique guidance step, the virtual instructor precisely demonstrates the correct operation gestures. The unit is also responsible for superimposing auxiliary visual elements, such as highlighting the correct operation area, drawing the ideal tool movement trajectory, or displaying whether the physical parameter data is within the expected range in the form of numbers and scales. These visual information is registered and superimposed on the learner's real field of view, forming a guidance scene that seamlessly blends with the real world. Synchronous verbal guidance and feedback are delivered through the audio unit. This unit is usually a bone conduction earphone or a micro speaker integrated into the AR glasses' temples. According to the instructions in the instruction sequence, the guidance speech is called from the local audio library or synthesized in real time by a text-to-speech (TTS) engine. The played audio content closely matches the contradiction type and guidance strategy. For example, when determining that the first type of contradiction is caused by psychological load leading to operation deformation, a guidance speech with a soothing tone and a speed of about 120 words per minute is matched, with the content being "Don't worry, let's slow down and follow me"; when determining that the second type of contradiction is caused by psychological load due to operation unfamiliarity, a clear and decisive instructional speech is matched, such as "Pay attention to the wrist angle, adjust it upward by about 15 degrees." In addition, the audio unit is also responsible for playing immediate feedback sounds, such as in the micro-success experience construction step, once the learner's small movements meet the expectations, a crisp "ding" sound is played to reinforce positive behavior. Direct physical guidance is applied through the somatosensory interaction unit. This unit consists of a group of wearable linear resonant actuators (LRAs) or eccentric rotor motors (ERMs) integrated into wearables on body parts related to common operation errors, such as wrists, arm bands, or gloves. When the instruction sequence contains tactile feedback instructions, the central controller sends pulse width modulation (PWM) signals to the designated actuators through Bluetooth Low Energy (BLE) protocol. The duty cycle and frequency of the PWM signal determine the intensity and mode of the tactile feedback. For example, to guide the learner to adjust the wrist angle, the actuators above or below the learner's wrist can be activated to produce continuous and directional vibrations, providing the learner with intuitive physical cues on which direction to adjust. During the embodiment relaxation guidance step, the somatosensory interaction unit can also apply gentle and rhythmic pulse vibrations to the learner's shoulders to simulate a massage effect, assisting in physiological relaxation.In this way, tactile feedback compensates for the shortcomings of visual and auditory guidance in conveying spatial and force information, creating a more complete and embodied guidance experience.

[0069] Optionally, the system further includes:

[0070] The skill structure decomposition module is used to decompose the labor task into a skill structure to obtain the core principle features and peripheral condition features;

[0071] The transfer task generation module is used to maintain the core principle features unchanged, change the peripheral condition features, and generate skill transfer training tasks.

[0072] The skill generalization assessment module is used to guide the learner to perform the skill transfer training task, reacquire multimodal perception data, and assess skill generalization ability.

[0073] Specifically, a skill structure decomposition module performs in-depth analysis of labor tasks to identify their transferable core. This module performs ontology-based knowledge modeling for each labor task in the task library. A task is decomposed into core principle features and peripheral condition features. Core principle features refer to the fundamental physical laws, operational logic, or safety rules behind the task that do not change with the context. For example, in all screw-tightening tasks, "applying axial pressure to prevent slippage" and "clockwise rotation for tightening" are core principle features. Peripheral condition features refer to the specific implementation methods of the task, including but not limited to the type or size of the tool, the material or location of the workpiece, and the lighting or spatial constraints of the operating environment. These features are structured and stored in XML or JSON files, providing templates for subsequent transfer task generation. Based on the decomposition results, a transfer task generation module dynamically creates new training challenges. This module generates new tasks that are related to but different from the original tasks, forcing learners to recall and apply the core principles. Its core algorithm is "keep the core, change the periphery." This module receives completed labor tasks as input, keeping their core principle features unchanged. It systematically modifies one or more peripheral condition features to generate skill transfer training tasks. For example, if the original task is "to screw an M4 screw into a wooden board using a standard Phillips screwdriver," the transfer task might be generated as "to screw an M3 hex screw into a metal plate using an electric screwdriver." This generation process follows a certain difficulty gradient, potentially starting by changing only one external feature and gradually increasing the number and magnitude of the changed features, thus constructing a training path from near transfer to far transfer. The learner's performance in the new task is examined through a skill generalization assessment module. When a learner begins a skill transfer training task, they are guided through the operation, and the multimodal data acquisition module is restarted to capture their complete performance in the new context. This module inputs the newly acquired multimodal perception data into the established contradiction analysis module. By observing whether the learner encounters new or different types of operation-state contradictions in the new task, and the time and number of system interventions required to resolve these contradictions, their skill generalization ability is comprehensively evaluated. For example, if a learner can quickly adjust the force and angle of operation when faced with a new tool, and does not trigger high-priority contradiction resolution guidance strategies throughout the process, their skill generalization ability is considered strong. The assessment results will be recorded in the learner profile database as a generalization ability score from 0 to 100, serving as another important dimension of their learning progress.

[0074] Optionally, the migration task generation module includes:

[0075] The contradiction type statistics unit is used to statistically analyze the learner's historical contradiction type data and determine the peripheral condition characteristics associated with frequently occurring contradiction types.

[0076] The gradient difficulty setting unit is used to generate a skill transfer training task with gradient difficulty based on the peripheral condition features associated with the frequently occurring contradiction types.

[0077] Specifically, the contradiction type statistics unit deeply analyzes learners' historical performance to accurately pinpoint their skill gaps. This unit is triggered after learners complete a series of basic labor tasks. It initiates an aggregate query to the learner's profile database, retrieving all historical contradiction type data for that learner. This process statistically analyzes the frequency of each operation-state contradiction, particularly the first and second types, and correlates them with the external conditional features that led to the contradiction. For example, it might be found that when the external conditional feature is "using a power tool with a torque greater than 2 Nm," the learner's frequency of the second type of contradiction is 30% higher than average. The unit's output is a ranked list of external conditional features strongly correlated with the learner's frequently occurring contradiction types; these features are then identified as the learner's key weaknesses. Based on the analysis results, a targeted advanced training program is designed using a gradient difficulty setting unit. A smooth learning curve is created to guide learners to systematically overcome their identified weaknesses. This unit receives the key weaknesses output by the contradiction type statistics unit as input. For each weak external conditional feature, a series of skill transfer training tasks with gradient difficulty are automatically generated. For example, if the identified weakness is "operating on reflective metal surfaces," the gradient difficulty setting unit will generate the following task sequence: Task 1, operating on a slightly reflective matte metal surface; Task 2, operating on a moderately reflective brushed metal surface; Task 3, operating on a highly reflective mirror metal surface. Throughout this process, the core principle of the task remains unchanged; only the peripheral condition feature identified as the weakness gradually increases in difficulty. This progressive challenge design ensures that learners can practice effectively at the edge of their comfort zone, avoiding frustration from overly difficult tasks and stagnation from overly easy tasks, thus effectively overcoming specific skill gaps.

[0078] Optionally, the system further includes:

[0079] The feedback data acquisition unit is used to collect new multimodal perception data and newly identified operation-state contradictions from the skill transfer training task as feedback data.

[0080] The model optimization and update unit is used to optimize and update the model used to construct the personal dynamic capability baseline and the model used to identify the operation-state contradiction using the feedback data.

[0081] Specifically, the feedback data acquisition unit specifically captures high-value information generated during the skill generalization phase. This unit is activated when a learner performs a skill transfer training task created by the transfer task generation module. It not only re-invokes the multimodal data acquisition module to obtain new multimodal perception data but also encapsulates this data, along with newly identified operation-state contradictions in the transfer task, as a complete data package. This data package is labeled as feedback data, revealing the learner's adaptability, vulnerabilities, and newly generated coping strategies in the face of unfamiliar external conditions. This data is a crucial input for the system's self-optimization. Using this feedback data, the model optimization and update unit iteratively upgrades the core algorithm model. This unit enables the system's "brain" to develop in sync with the learner's abilities. This process primarily targets two core models. First, it optimizes the model used to construct an individual's dynamic ability baseline. When feedback data indicates that a learner has successfully and efficiently completed a skill transfer training task, the new multimodal perception data from this successful operation is considered a new "excellent example." The system employs online or incremental learning algorithms to add this new data point to the original dataset used to train the skill performance baseline and psychophysiological response baseline, then fine-tunes the model with a low learning rate. This allows the updated individual dynamic ability baseline to cover a wider range of scenarios, and its definition of "proficiency" is also improved. Secondly, the model used to identify operand-state contradictions is optimized. If novel error patterns not previously seen by the system appear in transfer tasks, or if the existing model's judgment of a contradiction is inaccurate, these cases will serve as valuable counterexamples or hard examples. This feedback data will be used to supplement the training of the temporal analysis model or classifier in the contradiction type determination unit, thereby improving its accuracy and sensitivity in identifying complex and novel contradictions in the future. This optimization process is typically performed in the background in batch mode, for example, automatically triggered when the system is idle or after accumulating more than 10 sets of valid feedback data, ensuring continuous iteration and accuracy improvement of the teaching system model.

[0082] Based on the same inventive concept, such as Figure 4 As shown, the present invention also provides an interactive teaching method for labor education, the method comprising:

[0083] The system acquires multimodal perception data and task context information generated by learners when performing labor tasks, as well as learners' historical performance data. The multimodal perception data includes action sequence data, physiological response data, and physical parameter data.

[0084] Based on the learner’s historical performance data and current task context information, a personal dynamic ability baseline, including a skill performance baseline and a psychophysiological response baseline, is constructed and updated.

[0085] By comparing and analyzing the current multimodal perception data with the personal dynamic capability baseline, the operation-state contradictions that characterize the influence of operation and state are identified and quantified, and a contradiction resolution priority index is generated.

[0086] Based on the aforementioned operation-state contradiction and contradiction resolution priority index, a contradiction resolution guidance strategy containing specific guidance steps is generated.

[0087] The aforementioned conflict resolution guidance strategy is implemented, and multimodal guidance information including visual, auditory, or tactile guidance is output to the learner.

[0088] Example 1:

[0089] To verify the feasibility of this invention in practice, it was applied to an electronics assembly training course at a vocational college. In this embodiment, a learner, codenamed "Student X," was tasked with completing a "precision circuit board soldering" task. This task required the learner to solder a TQFP-44 surface-mount integrated circuit onto a designated pad on a PCB board. This process demands high levels of operational stability, precision, and mental fortitude. This invention aims to monitor, analyze, and guide learner "Student X" throughout the entire training process.

[0090] Before training began, data from trainee X's past performance on relatively simple "through-hole component soldering" tasks was retrieved using the historical performance acquisition unit, and a personal dynamic capability baseline was constructed for him through the baseline management module. This baseline shows that trainee X's skill performance under normal load is: an average time of 3.5 seconds to solder a single pin, with a standard deviation of [missing information]. For seconds, the angle deviation between the soldering iron tip and the pad is within ±5°, and the baseline of its psychophysiological response is: the standard deviation of resting heart rate variability (SDNN) is 65ms, and the threshold for judging a high cognitive load state is SDNN below 55ms.

[0091] At the 3-minute mark of the task, student X began soldering the second pin of the TQFP-44 chip. The multimodal data acquisition module captured the following data streams in real time: data from the motion capture unit showed that the hand holding the soldering iron trembled slightly, and the soldering time for the pin reached 5.8 seconds; physical parameter data from the embedded sensor unit indicated that the average operating force applied by the soldering iron tip to the pin was 0.8 N, exceeding the recommended range of 0.5 N; physiological response data collected by the non-contact physiological sensing unit showed that the heart rate variability (SDNN) calculated from thermal imaging of the forehead region of the student's face dropped sharply to 46 ms.

[0092] These data are then processed by the contradiction analysis module. First, the operation deviation analysis unit compares the current welding time of 5.8 seconds and unstable posture with the expected skill performance baseline of 3.5 ± 0.8 seconds. After calculation using a dynamic time warping algorithm, it generates an operation standard deviation feature D of 0.78. The psychological load analysis unit inputs the 46ms SDNN value into the psychophysiological response baseline model, determines that the probability of it being in a high-load state is extremely high, and generates a psychological load state feature L of 0.92.

[0093] Subsequently, the contradiction type determination unit performed a lead-lag correlation analysis on the time series of these two features within the last 5 seconds, and found that the peak decrease of the SDNN value preceded the peak deterioration of the operational normativity deviation feature by about 1.2 seconds. Based on this, the current operational-state contradiction was determined to be of the first type: operational deformation caused by excessive psychological load.

[0094] Based on this judgment, the priority calculation unit is activated, according to the formula... Calculate the conflict resolution priority index. In this precision operation scenario, the weighting coefficient is set to the operation deviation weight. Psychological load weight The moderating factor of type I contradiction The value is 1.2. Substitute the value into the calculation: The index of 1.04 far exceeded the preset intervention threshold of 0.75, and the strategy generation module was immediately triggered.

[0095] Since it was determined to be a Type I contradiction, the Type I strategy generation unit generated a composite guidance strategy that included embodied relaxation guidance and task goal simplification. The instruction conversion unit then converted it into an instruction sequence executable by the multimodal interactive terminal:

[0096] The audio unit played a voice message at a soothing pace of 120 words per minute: "Don't be nervous, let's pause for a moment and take a deep breath with the virtual teacher."

[0097] The virtual instructor displayed in the augmented reality (AR) glasses makes a slow, deep breathing motion, while a simulated heart rate fluctuation is overlaid on the right side of the student's field of vision.

[0098] Approximately 5 seconds later, the motion-sensing interaction unit applied two gentle pulse vibrations to the wearable device on the student's shoulder.

[0099] Once the student's SDNN latency is detected to rise to 57ms, the AR glasses highlight the next pin and corresponding pad on the chip that has not yet been soldered, and play a voice message: "Very good, the condition is much better. Now, we only need to focus on this one point and touch it gently."

[0100] After system intervention, student X successfully soldered the subsequent pins, with the average soldering time reduced to 4.1 seconds, the operating force stabilized at around 0.4 Newtons, and the SDNN maintained at the 60ms level.

[0101] Upon completion of the original task, a skill transfer training task was generated to improve skill generalization ability. The skill structure decomposition module defined the core principle of the original task as "precise heat conduction on a tiny target in a dense array," with the peripheral conditions being "using a constant-temperature soldering iron" and "TQFP packaged chip." The transfer task generation module kept the core principle unchanged but changed the peripheral conditions to "using a hot air gun" and "QFN packaged chip," generating a new task: "hot air gun soldering of QFN chips."

[0102] During the transfer task, the skill generalization assessment module recorded that when trainee X first operated the heat gun, the conflict resolution priority index reached 0.6, but no intervention was triggered. Within 30 seconds, trainee X mastered the control of wind speed and distance through self-adjustment and successfully completed the task. Based on his performance in quickly adapting to the new tool and packaging, his generalization ability score for this skill was assessed as 85 / 100. This successful transfer data was used as feedback data by the model optimization and update unit to fine-tune trainee X's personal dynamic ability baseline model, enabling his skill assessment for "heat conduction control" to cover more diverse tools and scenarios.

[0103] Table 1. Data on the Conflict Analysis and Intervention Process

[0104] Time stamp Task step SDNN (ms) Operational norm deviance feature (D) Psychological load state feature (L) Contradiction type Priority index (I) Intervention strategy 03:08 Welding pin 2 46 0.78 0.92 First type 1.04 Embodied relaxation + goal simplification 03:25 Welding pin 3 59 0.21 0.35 No significant contradiction 0.23 No intervention 03:31 Welding pin 4 61 0.15 0.28 No significant contradiction 0.19 No intervention

[0105] Table 2. Skills Transfer and Generalization Assessment Data Table

[0106] Task type Core principle feature Peripheral condition feature First attempt highest contradiction index (I) Self-correction time (sec) Skill generalization ability score Original task Precise heat conduction Constant temperature iron, TQFP package 1.04 (intervened) - - Transfer task Precise heat conduction Heat gun, QFN package 0.60 (not intervened) 30 85 / 100

[0107] Through the engineering parameters and testing procedures of the above embodiments, it is clear that this invention does not simply judge right or wrong, but rather achieves precise technical effects through a quantitative data derivation process. Table 1 shows how the system uses data such as SDNN and operational deviation features to determine the conflict type as "psychologically dominant" through time-series analysis, and further calculates a priority index as high as 1.04, thereby triggering an "embodied relaxation" strategy targeting the psychological state, rather than simple operational correction. After intervention, all indicators quickly returned to normal, proving the accuracy of the diagnosis and the effectiveness of the strategy.

[0108] The data in Table 2 validates the system's ability to transfer and assess skills. When faced with new tasks that share the same core principles but differ in external conditions, the system can quantitatively assess learners' adaptation process, with a maximum contradiction index of 0.60, a self-correction time of 30 seconds, and a specific generalization ability score (85 / 100). This process not only evaluates learners' ability to apply knowledge to new situations, but the results also serve as feedback data to optimize the system model, forming a complete, personalized, and adaptive teaching loop, thus enhancing the depth and efficiency of labor education.

[0109] It should be noted that the electrical connections between the various units described above do not necessarily represent direct or indirect connections. Any indirect connection method can be applied to the embodiments of the present invention as long as it achieves the purpose of the present invention. The above descriptions are merely exemplary embodiments of the present invention and should not be construed as limiting the scope of the present invention.

[0110] All equivalent changes and modifications made in accordance with the teachings of this invention are still within the scope of this invention. Those skilled in the art will readily conceive of other embodiments of this invention upon considering the specification and the disclosure of practical truth. This application is intended to cover any variations, uses, or adaptations of this invention that follow the general principles of this invention and include common knowledge or conventional techniques in the art not described herein.

Claims

1. An interactive teaching system for labor education, characterized in that, The system includes: The multimodal data acquisition module is used to acquire multimodal perception data and task context information generated by learners when performing labor tasks, as well as learners' historical performance data. The multimodal perception data includes action sequence data, physiological response data, and physical parameter data. The baseline management module is used to construct and update a personal dynamic ability baseline that includes a skill performance baseline and a psychophysiological response baseline based on the learner's historical performance data and current task context information. The contradiction analysis module is used to compare and analyze the current multimodal perception data with the personal dynamic capability baseline, identify and quantify the operation-state contradiction that represents the influence of operation and state association, and generate a contradiction resolution priority index. The strategy generation module is used to generate a conflict resolution guidance strategy containing specific guidance steps based on the operation-state contradiction and the conflict resolution priority index. The multimodal interactive terminal module is used to execute the conflict resolution guidance strategy and output multimodal guidance information containing visual, auditory, or tactile guidance to the learner.

2. The interactive teaching system for labor education according to claim 1, characterized in that, The multimodal data acquisition module includes: A motion capture unit is used to collect the learner's full-body and hand movements to obtain motion sequence data; The non-contact physiological sensing unit is used to acquire facial thermal imaging and visual images, and to deduce heart rate variability trends and eye tracking data as physiological response data. An embedded sensor unit is used to collect operational force and angle data as physical parameter data; The task information retrieval unit is used to retrieve preset steps, difficulty levels, and tool and material attributes corresponding to the labor task from the task library as task context information; The historical performance acquisition unit is used to retrieve and acquire the learner's historical multimodal perception data and operation result data under different task difficulties and different psychological states from the learner's profile database, so as to obtain historical performance data.

3. The interactive teaching system for labor education according to claim 1, characterized in that, The baseline management module includes: The skill baseline construction unit is used to extract the learner's movement pattern features under different task difficulties from the historical performance data to form a skill performance baseline. The psychological baseline modeling unit is used to establish a model of the learner's physiological response characteristics under different psychological states from the historical performance data, forming a psychological and physiological response baseline; The dynamic adjustment unit is used to dynamically adjust the expected parameter range in the skill performance baseline based on the current task context information and the real-time acquired physiological response data.

4. The interactive teaching system for labor education according to claim 1, characterized in that, The contradiction analysis module includes: The operation deviation analysis unit is used to analyze the deviation of the current action sequence data from the skill performance baseline and generate operation standard deviation features; The psychological load analysis unit is used to analyze the deviation of the current physiological response data from the psychological and physiological response baseline and generate psychological load state characteristics. The contradiction type determination unit is used to determine the dominant causal relationship between the operational norm deviation characteristics and the psychological load state characteristics in the time series, and to determine the contradiction type. The priority calculation unit is used to integrate the operational norm deviation characteristics, psychological load state characteristics, and conflict type to calculate the conflict resolution priority index.

5. The interactive teaching system for labor education according to claim 4, characterized in that, The strategy generation module includes: The first type of strategy generation unit is used to generate a composite guidance strategy that includes embodied relaxation guidance steps and task goal simplification steps if the contradiction type is determined to be the first type of operational deformation caused by psychological load. The second type of strategy generation unit is used to generate a composite guidance strategy that includes micro-success experience construction steps and step-by-step technical guidance steps if the contradiction type is determined to be the second type caused by unfamiliarity with operation. The instruction conversion unit is used to convert the generated composite guidance strategy into an instruction sequence that can be executed by a multimodal interactive terminal based on the task context information.

6. The interactive teaching system for labor education according to claim 5, characterized in that, The multimodal interaction terminal includes: An augmented reality display unit is used to present a virtual instructor image, wherein the virtual instructor executes the visual guidance portion of the instruction sequence; An audio unit is used to play guidance voice that matches the contradiction type and instruction sequence; The somatosensory interaction unit is used to apply tactile feedback corresponding to the instruction sequence to relevant parts of the learner's body.

7. The interactive teaching system for labor education according to claim 1, characterized in that, The system also includes: The skill structure decomposition module is used to decompose the labor task into a skill structure to obtain the core principle features and peripheral condition features; The transfer task generation module is used to maintain the core principle features unchanged, change the peripheral condition features, and generate skill transfer training tasks. The skill generalization assessment module is used to guide the learner to perform the skill transfer training task, reacquire multimodal perception data, and assess skill generalization ability.

8. The interactive teaching system for labor education according to claim 7, characterized in that, The migration task generation module includes: The contradiction type statistics unit is used to statistically analyze the learner's historical contradiction type data and determine the peripheral condition characteristics associated with frequently occurring contradiction types. The gradient difficulty setting unit is used to generate a skill transfer training task with gradient difficulty based on the peripheral condition features associated with the frequently occurring contradiction types.

9. The interactive teaching system for labor education according to claim 7, characterized in that, The system also includes: The feedback data acquisition unit is used to collect new multimodal perception data and newly identified operation-state contradictions from the skill transfer training task as feedback data. The model optimization and update unit is used to optimize and update the model used to construct the personal dynamic capability baseline and the model used to identify the operation-state contradiction using the feedback data.

10. An interactive teaching method for labor education, characterized in that, The system is used in an interactive teaching system for labor education as described in any one of claims 1-9, and the method includes: The system acquires multimodal perception data and task context information generated by learners when performing labor tasks, as well as learners' historical performance data. The multimodal perception data includes action sequence data, physiological response data, and physical parameter data. Based on the learner’s historical performance data and current task context information, a personal dynamic ability baseline, including a skill performance baseline and a psychophysiological response baseline, is constructed and updated. By comparing and analyzing the current multimodal perception data with the personal dynamic capability baseline, the operation-state contradictions that characterize the influence of operation and state are identified and quantified, and a contradiction resolution priority index is generated. Based on the aforementioned operation-state contradiction and contradiction resolution priority index, a contradiction resolution guidance strategy containing specific guidance steps is generated. The aforementioned conflict resolution guidance strategy is implemented, and multimodal guidance information including visual, auditory, or tactile guidance is output to the learner.

Images