Holographic interactive learning system based on ai motion perception and somatosensory recognition
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHANGHAI HAOPAI DIGITAL TECH
- Filing Date
- 2026-04-27
- Publication Date
- 2026-07-21
AI Technical Summary
Existing technologies in fields such as sports training, rehabilitation medicine, special operations, and dance lack multimodal data fusion, resulting in one-sided judgments of movement completion and physical condition, making it difficult to predict and provide feedback on risks, and lacking immersive interactive guidance, which affects learning efficiency and safety.
The system employs a holographic interactive learning system based on AI motion perception and body sensing recognition. Through modules such as motion indicator preset, data acquisition, state perception, progress evaluation, recognition and analysis, risk warning, impact analysis, and holographic interaction, it achieves real-time monitoring and three-dimensional holographic feedback of user actions and physiological states. Combined with deep learning algorithms, it performs risk prediction and dynamic adjustment.
It enables accurate assessment of user actions and physiological states, identifies risks in advance, provides personalized learning feedback, improves learning safety and immersion, and enhances the interactivity and fun of learning.
Smart Images
Figure CN122435680A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of analog interactive recognition technology, specifically a holographic interactive learning system based on AI motion perception and body sensing recognition. Background Technology
[0002] In fields such as sports training, rehabilitation medicine, special operations, and dance, the demands for standardized movements, learning efficiency, and training safety are constantly increasing. With the maturity of computer vision, inertial sensing, and biosignal detection technologies, precise capture of human movements and physiological states has become possible. Simultaneously, augmented reality and holographic projection technologies provide new mediums for creating highly immersive interactive environments. Against this backdrop, intelligent learning systems that integrate multimodal perception, artificial intelligence analysis, and immersive interaction have become an important development direction for improving the scientific rigor, personalization, and enjoyment of training.
[0003] Existing technologies rely solely on visual or inertial data, lacking deep fusion of multimodal data. This leads to biased judgments on action completion and body state, making it difficult to distinguish complex states. Traditional methods often perform retrospective analysis after an action is completed or simply set simple threshold alarms, failing to predict risks based on continuous behavioral patterns and lacking precise identification of the root causes of risks. The limited forms of feedback and interaction, and the lack of immersive guidance that naturally integrates with the three-dimensional action space, result in low efficiency in understanding and correction, and also weaken the interactivity and contextuality of learning. Summary of the Invention
[0004] (a) Technical problems to be solved In view of the above-mentioned shortcomings of the existing technology, the present invention provides a holographic interactive learning system based on AI motion perception and body sensing recognition, which can effectively solve the problems of the existing technology.
[0005] (II) Technical Solution To achieve the above objectives, the present invention provides the following technical solution: This invention discloses a holographic interactive learning system based on AI action perception and motion recognition, comprising: The action indicator preset module is used to preset several action indicators as simulation learning targets, and to define activity thresholds corresponding to different intensity levels for each action indicator. The data acquisition module is used to acquire the user's motion video data through an image acquisition device and acquire the user's physiological and motion sensor data through a motion sensor at the beginning of learning. The state perception module is used to process the collected data in real time, trigger and perceive the position of preset action indicators, and calibrate and detect the body sensation data based on the user's basic vital signs data to judge the changes in the position of the user's actions and the state changes of the body sensation data. The progress assessment module is used to comprehensively calculate the user's completion rate of the overall learning objectives based on changes in the state of action placement and changes in the state of body sensation data. The identification and analysis module is used to train and build a change recognition model through deep learning algorithms. The model takes action position change data and haptic data state change data as input. By analyzing the user's interaction behavior patterns in the current behavior cycle, it determines whether the behavior has a risk coefficient that exceeds a preset threshold. The risk warning module is used to trigger the holographic interaction module to send an alarm prompt to the user that is integrated into the holographic scene when the risk coefficient is detected to exceed the preset threshold. The impact analysis module is used to perform impact analysis based on the risk coefficient performance trend of the current and historical behavior cycles, and to locate and mark specific actions that contribute highly to the risk trend. The holographic interaction module is used to convert learning progress, action guidance, and risk alarm information into three-dimensional holographic images and interactive commands. The optimization and update module is used to generate suggestions on the improvement range of the activity threshold of the preset action index corresponding to the marked action based on the output of the impact analysis module, update the corresponding action index settings according to the suggestions, and execute the action guidance in the holographic interaction module.
[0006] Furthermore, the working logic of the holographic interaction module is as follows: by communicating with the holographic projection device, it receives instruction information, generates corresponding three-dimensional stereoscopic visual elements, integrates a spatial audio system and a natural user interface, and performs real-time fusion and rendering of the generated element information; The integrated spatial audio system provides voice guidance or sound effect feedback from the corresponding direction based on the location and result of the user's actions; the natural user interface allows users to interact with holographic images through specific gestures, postures, or gaze movements to call up function menus, view detailed data, or adjust learning parameters.
[0007] Furthermore, the visual elements include: dynamic guide lines for indicating deviations between the user's skeletal joints and standard movement trajectories; warning halos for highlighting incorrect or high-risk body parts; ambient color temperature changes for representing the overall risk level; and a three-dimensional progress model for displaying learning progress and skill mastery.
[0008] Furthermore, the motion indicator preset module stores a set of standard motion indicators applicable to different learning scenarios and skill levels. Each motion indicator includes one or more parameterized definitions of motion trajectory, key posture angle, strength standard, speed range, and duration. This allows system administrators or instructors to configure multiple intensity levels for each parameter of each motion indicator and set specific activity thresholds for each intensity level.
[0009] Furthermore, the data acquisition module includes a visual acquisition unit, a motion sensing unit, and a preprocessing unit, wherein: The visual acquisition unit is used to deploy several image acquisition components, such as depth cameras or RGB-D cameras, to capture the user's skeletal key points, joint point three-dimensional coordinates and motion sequences from preset angles, forming motion video data; The somatosensory sensing unit is used to collect the user's acceleration, angular velocity, electromuscular signals, heart rate variability, and pressure distribution data of the soles of the feet or hands in real time, forming physiological and motion somatosensory data. The preprocessing unit performs preprocessing operations on the raw data acquired by the vision acquisition unit and the motion sensing unit, including timestamp synchronization, noise filtering, coordinate system unification, and feature point extraction, to form a spatiotemporally aligned data stream.
[0010] Furthermore, the operating logic of the state-aware module is as follows: Real-time comparison of motion video data with preset motion indicators to determine whether the action is triggered, whether the motion trajectory is consistent, and whether the key postures are in place, and output a quantified sequence of motion status changes. It receives pre-processed somatosensory data, combines it with the user's pre-entered basic vital signs data, normalizes and analyzes physiological signals with context awareness, monitors changes in the user's cardiopulmonary load, muscle activation level, balance stability indicators, and identifies abnormal physiological response patterns. The system receives the output data of the above-mentioned sequence of changes in the state of action and abnormal physiological reaction patterns, comprehensively judges the user's combined action and somatosensory state at each moment, and marks the composite state event.
[0011] Furthermore, the composite state event types include: the type where the action meets the standard but the physiological load exceeds the limit, the type where the action compensates but the physiological load is normal, the type where the action is deformed and accompanied by early physiological warning, the type where fatigue accumulates based on time sequence, and the type where psychological stress leads to stiffness.
[0012] Furthermore, the construction process of the change recognition model in the recognition and analysis module is as follows: Collect multi-user data during the historical interactive learning process. The data includes the temporal action state change sequence and the somatosensory data state change sequence output by the state awareness module. Label each complete behavior cycle data with a comprehensive risk level label, and optionally label high-risk behavior type labels. A multi-task temporal deep learning network is constructed, including an input layer, a temporal feature extraction layer, and a multi-task output layer. The input layer is used to receive the concatenated vector of aligned temporal action state feature vector and somatosensory state feature vector. The temporal feature extraction layer adopts a long short-term memory network to extract dynamic behavior pattern features from the input temporal concatenated vector. The multi-task output layer is used to output a continuous risk coefficient score to quantify the overall risk level of the current behavior cycle, and output a probability distribution vector to identify the high-risk type of the current behavior cycle, which includes one or more of the following: over-fatigue, compensatory error, joint over-limit risk, and cardiopulmonary overload risk. Supervised training of a multi-task temporal deep learning network was performed using a labeled dataset. During training, the risk coefficient score was optimized using the mean squared error loss function, the risk type was optimized using the cross-entropy loss function, and the total loss function of the model was the weighted sum of the two branch loss functions. The performance of the trained model is evaluated using an independent validation set, and the model that meets the performance criteria is deployed to the recognition and analysis module.
[0013] Furthermore, during the impact analysis process, the impact analysis module performs time series analysis on the risk coefficients of multiple consecutive behavioral cycles to obtain the risk change rate and fluctuation amplitude, so as to statistically analyze the future evolution trend of risk. When an upward trend of risk or a high-risk event is identified, the module traces back the detailed action and somatosensory data within the associated cycle. By obtaining the feature contribution of the associated action and somatosensory data, it identifies one or more action indicator parameters that lead to the increase in risk, and highlights the high-impact actions in the three-dimensional holographic image of the holographic interaction module.
[0014] Furthermore, the action indicator preset module and the data acquisition module are interconnected via a wireless network; the state perception module, the data acquisition module, and the progress evaluation module are interconnected via a wireless network; the identification and analysis module, the progress evaluation module, the risk warning module, and the impact analysis module are interconnected via a wireless network; the impact analysis module and the holographic interaction module are interconnected via a wireless network; and the holographic interaction module and the optimization and update module are interconnected via a wireless network.
[0015] (III) Beneficial Effects Compared with the known prior art, the technical solution provided by this invention has the following beneficial effects: 1. By integrating visual-motor data and somatosensory physiological data and performing fusion analysis, the system comprehensively assesses the combined state of whether the user's movements are in place and whether the body can tolerate them. It analyzes continuous behavioral patterns and identifies risk trends such as over-fatigue and compensatory errors in advance, which greatly improves the safety of training and prevents cumulative injuries.
[0016] 2. By identifying the key actions or parameters that lead to risks, precise source tracing can be achieved, and personalized threshold adjustment suggestions can be generated. This allows the system's difficulty and safety boundaries to dynamically adapt to changes in user capabilities, ensuring that the learning process is both safe and continuously challenges the user's potential, thus achieving personalized skill advancement.
[0017] 3. By setting up a holographic interactive module, the abstract data analysis results are transformed into three-dimensional visual guidance, warnings and enhancement information that conforms to the user's action space, which greatly reduces the cognitive load, allows users to focus more on the action itself, and enhances the immersion and fun of learning. Attached Figure Description
[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.
[0019] Figure 1 This is a schematic diagram of the overall framework of the present invention; Figure 2 This is a schematic diagram of the data acquisition module in this invention.
[0020] The numbers in the diagram represent: 1. Motion indicator preset module; 2. Data acquisition module; 21. Visual acquisition unit; 22. Motion sensing unit; 23. Preprocessing unit; 3. State perception module; 4. Progress evaluation module; 5. Recognition and analysis module; 6. Risk warning module; 7. Impact analysis module; 8. Holographic interaction module; 9. Optimization and update module. Detailed Implementation
[0021] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.
[0022] The present invention will be further described below with reference to embodiments.
[0023] This embodiment presents a holographic interactive learning system based on AI motion perception and body sensing recognition, such as... Figures 1-2 As shown, it includes: The motion indicator preset module 1 is used to preset several motion indicators as simulated learning objectives and define activity thresholds corresponding to different intensity levels for each motion indicator. The motion indicator preset module 1 stores a set of standard motion indicators applicable to different learning scenarios and skill levels. Each motion indicator includes one or more parameterized definitions of motion trajectory, key posture angle, strength standard, speed range, and duration. It allows system administrators or instructors to configure multiple intensity levels for each parameter of each motion indicator and set specific activity thresholds for each intensity level. The activity thresholds include achievement thresholds, warning thresholds, and risk thresholds. The activity thresholds can be associated with the user's initial physical test data or historical performance data to achieve preliminary personalized setting of the thresholds.
[0024] Data acquisition module 2 is used to acquire user motion video data through an image acquisition device and acquire user physiological and motion sensor data through a motion sensor at the start of learning. Data acquisition module 2 includes a visual acquisition unit 21, a motion sensing unit 22, and a preprocessing unit 23, wherein: The visual acquisition unit 21 is used to deploy several image acquisition components, such as depth cameras or RGB-D cameras, to capture the user's skeletal key points, joint point three-dimensional coordinates and motion sequences from preset angles, forming motion video data; The somatosensory sensing unit 22 is used to collect the user's acceleration, angular velocity, electromyographic signals, heart rate variability and pressure distribution data of the soles of the feet or hands in real time, forming physiological and motion somatosensory data. The preprocessing unit 23 is used to perform preprocessing operations such as timestamp synchronization, noise filtering, coordinate system unification, and feature point extraction on the raw data collected by the visual acquisition unit 21 and the motion sensing unit 22, forming a spatiotemporally aligned data stream.
[0025] The state perception module 3 is used to process the collected data in real time, trigger and sense the position of preset action indicators, and calibrate and detect the body sensation data based on the user's basic vital sign data to determine changes in the user's action position and the state of the body sensation data. The operating logic of the state perception module 3 is as follows: Real-time comparison of motion video data with preset motion indicators to determine whether the action is triggered, whether the motion trajectory is consistent, and whether the key postures are in place, and output a quantified sequence of motion status changes. It receives pre-processed somatosensory data, combines it with the user's pre-entered basic vital signs data, normalizes and analyzes physiological signals with context awareness, monitors changes in the user's cardiopulmonary load, muscle activation level, balance stability indicators, and identifies abnormal physiological response patterns. Receive the output data of the above-mentioned action position change sequence and abnormal physiological reaction pattern, comprehensively judge the user's action and somatosensory state at each moment, and mark the composite state event. The types of complex state events include: the type where the movement meets the standard but the physiological load exceeds the limit, the type where the movement is compensated but the physiological load is normal, the type where the movement is deformed and accompanied by early physiological warning, the type where fatigue is accumulated based on time sequence, and the type where psychological stress leads to movement stiffness. Among them, the "action meets the standard but physiological load exceeds the limit" type is triggered when the user's action reaches or exceeds the preset threshold in terms of trajectory, posture and speed, but the somatosensory detection detects one or more physiological indicators that continuously exceed the safety threshold. The "Motion Compensation but Normal Physiological Load" type is triggered when it is determined that the user's physiological indicators are within the normal load range, but when it is detected that the user has adopted an unconventional and non-standard compensatory action mode in order to achieve the goal. The motion distortion is accompanied by an early physiological warning type. It is triggered when the motion recognition engine detects that the user's motion trajectory or key angle begins to deviate from the standard, and the somatosensory detection detects that the activation level of related muscle groups is insufficient or that the antagonistic muscle groups are abnormally tense. The time-based fatigue accumulation type is triggered when the accuracy, speed or force output of the user's action shows a downward trend, and the user's heart rate recovery rate slows down and the electromyography integral value increases under the same action, through time-series analysis of multiple consecutive behavioral cycles. Stress-induced stiffness type: Triggered when heart rate variability analysis detects increased sympathetic nerve activity in the user's autonomic nervous system, and motion detection shows reduced fluidity of user movements, decreased range of motion of joints, or unnatural tremors.
[0026] The progress assessment module 4 is used to comprehensively calculate the user's completion rate of the overall learning objective based on changes in the state of action placement and changes in the state of somatosensory data. By deconstructing the overall learning objective into multiple sub-objectives, and based on changes in the state of action placement and changes in the state of somatosensory data, it quantifies and calculates four core progress indicators: action accuracy, action stability, physiological load adaptation, and risk control level. Based on the current learning stage, the user's individual ability baseline, and the risk contribution analysis results, it dynamically adjusts the weight of each progress indicator in the overall completion calculation. Based on the quantification results of each progress indicator and its dynamic weight, it calculates the overall completion percentage through a weighted algorithm, and presents the results and detailed analysis through the holographic interactive module 8 or other display devices.
[0027] The identification and analysis module 5 is used to train and construct a change recognition model through deep learning algorithms. The model takes action position change data and haptic data change data as input, and determines whether the behavior exceeds the preset threshold risk coefficient by analyzing the user's interaction behavior pattern in the current behavior cycle.
[0028] The risk warning module 6 is used to trigger the holographic interaction module 8 to issue an alarm prompt to the user that is integrated into the holographic scene when the risk coefficient is detected to exceed the preset threshold.
[0029] The impact analysis module 7 is used to perform impact analysis based on the risk coefficient performance trend of current and historical behavioral cycles, locating and marking specific actions that contribute highly to the risk trend or have great potential harm. During the impact analysis process, the impact analysis module 7 performs time series analysis on the risk coefficients of multiple consecutive behavioral cycles to obtain the risk change rate and fluctuation amplitude in order to statistically analyze the future evolution trend of risk. When an upward trend in risk or a high-risk event is identified, detailed action and sensory data within the related cycle are traced back. By obtaining the characteristic contribution of related action and sensory data, one or more action indicator parameters that lead to increased risk are identified. High-impact actions are highlighted in the three-dimensional holographic image of the holographic interaction module 8, and an analysis report containing risk factors, degree of impact, and context of occurrence is generated.
[0030] Holographic interaction module 8 is used to convert learning progress, action guidance, and risk alarm information into three-dimensional holographic images and interactive commands; the working logic of holographic interaction module 8 is as follows: By communicating with a holographic projection device to receive instruction information and generate corresponding three-dimensional stereoscopic visual elements, and integrating a spatial audio system and a natural user interface, the generated element information is fused and rendered in real time; the visual elements include: dynamic guide lines to indicate the deviation of the user's skeletal joints from the standard movement trajectory; warning halos to highlight errors or high-risk body parts; environmental color temperature changes to characterize the overall risk level; and a three-dimensional progress model to display learning progress and skill mastery. The integrated spatial audio system provides voice guidance or sound effect feedback from the corresponding direction based on the location and result of the user's actions; the natural user interface allows users to interact with holographic images through specific gestures, postures, or gaze movements to access function menus, view detailed data, or adjust learning parameters.
[0031] The optimization and update module 9 is used to generate improvement suggestions for the activity threshold of the preset action indicators corresponding to the marked actions based on the results output by the impact analysis module 7, update the corresponding action indicator settings according to the suggestions, and execute action guidance in the holographic interaction module 8. The attributes of the improvement suggestions are: for action indicators that lead to high risks, it is recommended to increase their warning threshold and risk threshold by a preset scale, or decrease their compliance threshold to reduce the intensity; for action indicators that are low risk, it is recommended to increase the compliance threshold by an appropriate preset scale.
[0032] The action indicator preset module 1 and the data acquisition module 2 are connected via a wireless network. The status perception module 3, the data acquisition module 2, and the progress evaluation module 4 are connected via a wireless network. The identification and analysis module 5, the progress evaluation module 4, the risk warning module 6, and the impact analysis module 7 are connected via a wireless network. The impact analysis module 7 and the holographic interaction module 8 are connected via a wireless network. The holographic interaction module 8 and the optimization and update module 9 are connected via a wireless network.
[0033] Compared with existing technologies, this technology achieves quantitative analysis of user action accuracy and physiological risks through multi-source data fusion and AI models. The analysis results are transformed into three-dimensional holographic images and natural interactive commands, providing users with immersive, spatial real-time visual guidance and risk warnings. Based on long-term behavioral pattern analysis, it automatically locates the causes of high-risk actions and dynamically optimizes personalized learning thresholds. This enhances the learning immersion and efficiency while building a continuously self-optimizing safety protection mechanism.
[0034] At other levels, this embodiment also provides a process for constructing a change recognition model, specifically as follows: Collect multi-user data during the historical interactive learning process. The data includes the temporal action state change sequence and the somatosensory data state change sequence output by the state perception module 3. Label each complete behavior cycle data with a comprehensive risk level label, and optionally label high-risk behavior type labels. A multi-task temporal deep learning network is constructed, including an input layer, a temporal feature extraction layer, and a multi-task output layer. The input layer is used to receive the concatenated vector of aligned temporal action state feature vector and somatosensory state feature vector. The temporal feature extraction layer adopts a long short-term memory network to extract dynamic behavior pattern features from the input temporal concatenated vector. The multi-task output layer is used to output a continuous risk coefficient score to quantify the overall risk level of the current behavior cycle, and output a probability distribution vector to identify the high-risk type of the current behavior cycle, which includes one or more of the following: over-fatigue, compensatory error, joint over-limit risk, and cardiopulmonary overload risk. Supervised training of a multi-task temporal deep learning network was performed using a labeled dataset. During training, the risk coefficient score was optimized using the mean squared error loss function, and the risk type was optimized using the cross-entropy loss function. The total loss function of the model was the weighted sum of the two branch loss functions. The performance of the trained model is evaluated using an independent validation set to ensure the accuracy and generalization ability of its risk prediction. The model that meets the performance standard is deployed to the identification and analysis module 5 for online risk identification and analysis of real-time input user behavior cycle data. Compared with existing technologies, it can simultaneously process the fusion time-series data of motion and body sensation, not only outputting continuous risk coefficient scores to achieve accurate quantitative early warning, but also identifying specific high-risk behavior types in parallel, thereby realizing intelligent diagnosis of risk status.
[0035] In summary, this invention constructs an immersive learning environment by deeply integrating AI motion perception, multimodal body sensing recognition, and holographic interaction technologies. It achieves precise and real-time learning guidance and risk prevention. By synchronously monitoring and quantitatively analyzing user actions and physiological indicators, it ensures the objectivity and comprehensiveness of the assessment. By predicting risks such as sports injuries or learning bottlenecks, it transforms abstract data and analysis conclusions into spatial three-dimensional visual guidance and natural interaction, greatly improving the intuitiveness, immersion, and operational efficiency of learning. Based on continuous performance and risk analysis, the system can adjust the training difficulty and safety boundaries to achieve adaptive and personalized customization of the learning path, thereby improving the efficiency of skill mastery while maximizing the safety and health of users.
[0036] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A holographic interactive learning system based on AI motion perception and body sensing recognition, characterized in that: include: The action indicator preset module is used to preset several action indicators as simulation learning targets, and to define activity thresholds corresponding to different intensity levels for each action indicator. The data acquisition module is used to acquire the user's motion video data through an image acquisition device and acquire the user's physiological and motion sensor data through a motion sensor at the beginning of learning. The state perception module is used to process the collected data in real time, trigger and perceive the position of preset action indicators, and calibrate and detect the body sensation data based on the user's basic vital signs data to judge the changes in the position of the user's actions and the state changes of the body sensation data. The progress assessment module is used to comprehensively calculate the user's completion rate of the overall learning objectives based on changes in the state of action placement and changes in the state of body sensation data. The identification and analysis module is used to train and build a change recognition model through deep learning algorithms. The model takes action position change data and haptic data state change data as input. By analyzing the user's interaction behavior patterns in the current behavior cycle, it determines whether the behavior has a risk coefficient that exceeds a preset threshold. The risk warning module is used to trigger the holographic interaction module to send an alarm prompt to the user that is integrated into the holographic scene when the risk coefficient is detected to exceed the preset threshold. The impact analysis module is used to perform impact analysis based on the risk coefficient performance trend of the current and historical behavior cycles, and to locate and mark specific actions that contribute highly to the risk trend. The holographic interaction module is used to convert learning progress, action guidance, and risk alarm information into three-dimensional holographic images and interactive commands. The optimization and update module is used to generate suggestions on the improvement range of the activity threshold of the preset action index corresponding to the marked action based on the output of the impact analysis module, update the corresponding action index settings according to the suggestions, and execute the action guidance in the holographic interaction module.
2. The holographic interactive learning system based on AI motion perception and body sensing recognition according to claim 1, characterized in that, The working logic of the holographic interaction module is as follows: By communicating with a holographic projection device to receive instruction information and generate corresponding three-dimensional stereoscopic visual elements, and integrating a spatial audio system and a natural user interface, the generated element information is fused and rendered in real time. The integrated spatial audio system provides voice guidance or sound effect feedback from the corresponding direction based on the location and result of the user's actions; the natural user interface allows users to interact with holographic images through specific gestures, postures, or gaze movements to call up function menus, view detailed data, or adjust learning parameters.
3. The holographic interactive learning system based on AI motion perception and body sensing recognition according to claim 2, characterized in that, The visual elements include: dynamic guide lines for indicating deviations between the user's skeletal joints and standard movement trajectories; warning halos for highlighting incorrect or high-risk body parts; ambient color temperature changes for representing the overall risk level; and a three-dimensional progress model for displaying learning progress and skill mastery.
4. The holographic interactive learning system based on AI motion perception and body sensing recognition according to claim 1, characterized in that, The motion indicator preset module stores a set of standard motion indicators applicable to different learning scenarios and skill levels. Each motion indicator includes one or more parameterized definitions of motion trajectory, key posture angle, strength standard, speed range, and duration. System administrators or instructors can configure multiple intensity levels for each parameter of each motion indicator and set specific activity thresholds for each intensity level.
5. The holographic interactive learning system based on AI motion perception and body sensing recognition according to claim 1, characterized in that, The data acquisition module includes a visual acquisition unit, a motion sensing unit, and a preprocessing unit, wherein: The visual acquisition unit is used to deploy several image acquisition components to capture the user's skeletal key points, joint point three-dimensional coordinates and motion sequences from preset angles, forming motion video data. The somatosensory sensing unit is used to collect the user's acceleration, angular velocity, electromuscular signals, heart rate variability, and pressure distribution data of the soles of the feet or hands in real time, forming physiological and motion somatosensory data. The preprocessing unit performs preprocessing operations on the raw data acquired by the vision acquisition unit and the motion sensing unit, including timestamp synchronization, noise filtering, coordinate system unification, and feature point extraction, to form a spatiotemporally aligned data stream.
6. The holographic interactive learning system based on AI motion perception and body sensing recognition according to claim 1, characterized in that, The operating logic of the state awareness module is as follows: Real-time comparison of motion video data with preset motion indicators to determine whether the action is triggered, whether the motion trajectory is consistent, and whether the key postures are in place, and output a quantified sequence of motion status changes. It receives pre-processed somatosensory data, combines it with the user's pre-entered basic vital signs data, normalizes and analyzes physiological signals with context awareness, monitors changes in the user's cardiopulmonary load, muscle activation level, balance stability indicators, and identifies abnormal physiological response patterns. The system receives the output data of the above-mentioned sequence of changes in the state of action and abnormal physiological reaction patterns, comprehensively judges the user's combined action and somatosensory state at each moment, and marks the composite state event.
7. The holographic interactive learning system based on AI motion perception and body sensing recognition according to claim 6, characterized in that, The composite state event types include: the type where the movement meets the standard but the physiological load exceeds the limit, the type where the movement is compensated but the physiological load is normal, the type where the movement is deformed and accompanied by early physiological warning, the type where fatigue accumulation is based on time sequence, and the type where psychological stress leads to movement stiffness.
8. The holographic interactive learning system based on AI motion perception and body sensing recognition according to claim 1, characterized in that, The construction process of the change recognition model in the recognition and analysis module is as follows: Collect multi-user data from historical interactive learning processes, label each complete behavioral cycle with a comprehensive risk level tag, and optionally label high-risk behavior types; A multi-task temporal deep learning network is constructed, including an input layer, a temporal feature extraction layer, and a multi-task output layer. The input layer is used to receive the concatenated vector of aligned temporal action state feature vector and somatosensory state feature vector. The temporal feature extraction layer adopts a long short-term memory network to extract dynamic behavior pattern features from the input temporal concatenated vector. The multi-task output layer is used to output a continuous risk coefficient score to quantify the overall risk level of the current behavior cycle, and output a probability distribution vector to identify the high-risk type of the current behavior cycle. Supervised training of multi-task temporal deep learning networks using labeled datasets; The performance of the trained model is evaluated using an independent validation set, and the model that meets the performance criteria is deployed to the recognition and analysis module.
9. The holographic interactive learning system based on AI motion perception and body sensing recognition according to claim 1, characterized in that, During the impact analysis process, the impact analysis module performs time series analysis on the risk coefficients of multiple consecutive behavioral cycles to obtain the risk change rate and fluctuation amplitude, so as to statistically analyze the future evolution trend of the risk. When an upward trend in risk or a high-risk event is identified, detailed action and sensory data within the relevant period are traced back. By obtaining the feature contribution of the relevant action and sensory data, one or more action indicator parameters that lead to the increased risk are identified, and high-impact actions are highlighted in the 3D holographic image of the holographic interaction module.
10. The holographic interactive learning system based on AI motion perception and body sensing recognition according to claim 1, characterized in that, The action indicator preset module and the data acquisition module are interconnected via a wireless network. The status perception module, the data acquisition module, and the progress evaluation module are interconnected via a wireless network. The identification and analysis module, the progress evaluation module, the risk warning module, and the impact analysis module are interconnected via a wireless network. The impact analysis module and the holographic interaction module are interconnected via a wireless network. The holographic interaction module and the optimization and update module are interconnected via a wireless network.