Asynchronous Decoupling Method and System for Implicit Intents and Explicit Commands in Immersive Interaction

Abstract

This invention discloses an asynchronous continuous decoupling method and system for implicit intentions and explicit commands in immersive interaction. Addressing the problem of asynchronous time differences between explicit commands (such as gestures and voice) and implicit physiological intentions (such as regret or hesitation) issued by users in immersive interaction, which easily leads to erroneous operations in traditional systems, this method first detects the explicit command action in real time and records the trigger timestamp, simultaneously acquiring the implicit physiological data stream; then, it temporarily suspends the execution of system operations and constructs an asynchronous physiological feature buffer time window along the time axis starting from the timestamp; it analyzes the absolute value of the rate of change of physiological features within the time window; if none exceed a threshold, the intention is determined to be consistent and the operation is executed; if it exceeds the threshold, it is determined to be a sudden physiological feature, confirming an intention conflict, and then the original operation is canceled and a matching secondary confirmation command is invoked. This invention effectively captures the "physiological trail," achieving precise decoupling between explicit commands and implicit intentions, and improving system security and fault tolerance.

Description

Technical Field

[0001] This invention relates to the field of information technology, and in particular to an asynchronous and continuous decoupling method and system for implicit intentions and explicit instructions in immersive interaction. Background Technology

[0002] With the rapid development of virtual reality (VR), augmented reality (AR), mixed reality (MR) and smart cockpit technologies, immersive interactive systems have gradually penetrated into people's daily lives.

[0003] In these high-frequency interaction scenarios, the system often needs to process both explicit user commands (such as voice commands and specific gestures) and implicit physiological intentions (such as gaze deviation, breathing rhythm, micro-expressions, or skin conductance responses) to achieve highly intelligent and humanized interactive feedback.

[0004] However, existing interactive systems generally employ a "time-aligned" synchronization fusion algorithm when processing these two types of heterogeneous signals.

[0005] The algorithm assumes that the user's explicit command triggering time is completely synchronized and statically corresponding with its internal physiological state in physical time.

[0006] However, in actual complex interactions, there is often a subtle technical time lag between conscious decision-making and subconscious physiological stress response.

[0007] When a user issues a key explicit command such as "confirm send" or "execute delete", their intention may change in a very short instant after the command is issued due to psychological stress or decision reversal.

[0008] Such instantaneous changes in intent are usually accompanied by distinctive physiological fluctuations (such as momentary breath-holding, sudden pupillary changes, or micro-expression stiffness), and these physiological characteristics often lag behind the explicit command by several milliseconds to hundreds of milliseconds, forming the so-called "physiological trail".

[0009] Traditional synchronous fusion algorithms lack data monitoring on the timeline after command triggering, completely severing the capture of "physiological trails." This causes the system to mechanically execute preset commands even when implicit intentions have conflicted, resulting in irreversible misoperations.

[0010] Most existing improvement solutions attempt to introduce complex time-series prediction models to anticipate users' hesitation in advance. However, this not only incurs huge computational costs, but is also prone to bias in predicting such sudden, instinctive stress responses, leading to more erroneous interventions.

[0011] Therefore, how to capture the asynchronous time difference between explicit instructions and implicit intentions in a more concise and reliable way without relying on ultra-large-scale computing resources, and how to achieve continuous decoupling and timely intervention when intention conflicts are detected, is a key technical problem in improving the fault tolerance and interaction security of the current immersive interaction field. Summary of the Invention

[0012] This invention provides an asynchronous and continuous decoupling method for implicit intentions and explicit instructions in immersive interaction, mainly including: S1. Real-time detection of preset explicit command actions issued by the user in an immersive interactive scenario. When a valid explicit command action is identified, the trigger timestamp when the explicit command action is completed is recorded, and the implicit physiological data stream continuously generated by the user on the timeline is simultaneously acquired through the sensor. S2. Temporarily suspend the system control operation corresponding to the explicit instruction action, and construct an asynchronous physiological feature buffer time window of preset duration by extending along the time axis from the trigger timestamp as the starting point; S3. Extract the latent physiological data stream within the asynchronous physiological feature buffer time window, extract the values ​​of preset physiological features at adjacent time nodes from the latent physiological data stream, and calculate the absolute value of the rate of change of the values ​​relative to time; S4. If, within the asynchronous physiological characteristic buffer time window, the absolute value of the rate of change at all adjacent time nodes is less than or equal to a preset rate of change threshold, then it is determined that the latent physiological intention is consistent with the issued explicit instruction, the suspension state is lifted, and the system control operation is executed. S5. If, within the asynchronous physiological feature buffer time window, the absolute value of any of the change rates is detected to be greater than the change rate threshold, then the corresponding preset physiological feature is determined to be a mutation physiological feature, and it is determined that the latent physiological intention and the explicit instruction conflict at the time boundary. S6. Cancel the execution of the system control operation and call a secondary confirmation interaction instruction that matches the mutated physiological characteristics to replace the original system control operation.

[0013] Furthermore, S1 includes: The limb movement trajectory is acquired, and the limb movement trajectory is matched with a preset action template to obtain the action recognition result; If the action recognition result is a successful match, the trigger time when the explicit instruction is completed is recorded; Based on the trigger time, a sliding time window algorithm is used to extract continuous data segments from the sensor acquisition signal, and the continuous data segments are denoised to obtain a latent physiological sequence. Physiological features are extracted based on the implicit physiological sequence to complete real-time detection of preset explicit commands issued by users in immersive interactive scenarios, and when the explicit command is detected, the trigger timestamp is recorded and the implicit physiological data stream is acquired synchronously.

[0014] Furthermore, S2 includes: Extract the trigger timestamp corresponding to the explicit instruction; the trigger timestamp is used to generate an operation interception signal to obtain a system control lock state. An asynchronous physiological feature buffer time window is constructed based on the system control lock state, and the asynchronous physiological feature buffer time window is used to acquire the first physiological feature sequence. The first physiological feature sequence is smoothed and filtered to obtain the second physiological feature sequence; The system control lock state is released according to the second physiological feature sequence, thereby temporarily suspending the execution of the system control operation corresponding to the explicit instruction action and constructing an asynchronous physiological feature buffer time window of preset duration along the time axis starting from the trigger timestamp.

[0015] Furthermore, S3 includes: The raw physiological signal sequence is acquired and smoothed to obtain the implicit physiological data stream; Extract the feature values ​​of preset physiological features at adjacent time nodes from the implicit physiological data stream, and obtain the feature fluctuation difference based on the feature values; The rate of change is calculated based on the characteristic fluctuation difference and the time difference. Determine whether the rate of change is less than zero. If the rate of change is less than zero, invert the rate of change to obtain the absolute value. Complete the extraction of the latent physiological data stream within the asynchronous physiological feature buffer time window. Extract the values ​​of preset physiological features at adjacent time nodes from the latent physiological data stream and calculate the absolute value of the rate of change of the values ​​relative to time.

[0016] Furthermore, S4 includes: Obtain the data sequence of asynchronous physiological characteristics within a buffer time window; Calculate the numerical differences between adjacent time nodes based on the data sequence and perform absolute value conversion to obtain a set of absolute values ​​of the rate of change. Determine whether all values ​​in the set of absolute values ​​of the rate of change are less than or equal to a preset rate of change threshold. If all values ​​in the set of absolute values ​​of the rate of change are less than or equal to a preset rate of change threshold, then it is determined that the latent physiological intention and the explicit instruction possess intent. Figure 1 To the point of being compatible; According to the meaning Figure 1 The activation instruction queue is obtained by releasing the suspended state. An execution signal is issued to the activation instruction queue to determine whether the latent physiological intention is consistent with the issued explicit instruction and to release the suspended state to execute the system control operation.

[0017] Furthermore, S5 includes: Obtain the sequence of physiological characteristic change rates within a time window; If the absolute value of the physiological characteristic change rate sequence is greater than the preset change rate threshold, the corresponding physiological characteristic data is determined to be mutated physiological characteristic data. The timestamps of the extracted mutation physiological characteristic data; Calculate the timing deviation between the timestamp and the explicit instruction reception timestamp; If the timing deviation value is within a preset conflict time window, then based on the mutation physiological characteristic data, it is determined that the latent physiological intention and the explicit instruction conflict at the time boundary.

[0018] Furthermore, S6 includes: Obtain current physiological characteristic data; If the current physiological characteristic data exceeds the mutation threshold range, it is determined that there is an abnormal triggering condition in the current system control operation; A secondary confirmation instruction is extracted based on the abnormal triggering conditions, and the secondary confirmation instruction is matched with the current physiological characteristic data; Generate a secondary confirmation interaction instruction to be executed based on the aforementioned secondary confirmation instruction; Cancel the execution of the current system control operation and call the pending secondary confirmation interaction instruction to replace the original system control operation.

[0019] This invention also provides an asynchronous continuous decoupling system for implicit intentions and explicit instructions in immersive interaction, mainly comprising: The explicit command action detection module is used to detect the preset explicit command actions issued by the user in the immersive interaction scenario in real time. When a valid explicit command action is identified, the trigger timestamp when the explicit command action is completed is recorded, and the implicit physiological data stream continuously generated by the user on the timeline is acquired through the sensor. The trigger timestamp recording and physiological data stream synchronization acquisition module is used to temporarily suspend the system control operation corresponding to the explicit instruction action, and to construct an asynchronous physiological feature buffer time window of preset duration by extending along the time axis from the trigger timestamp as the starting point. The system control operation suspension and asynchronous physiological feature buffer time window construction module is used to extract the latent physiological data stream within the asynchronous physiological feature buffer time window, extract the values ​​of preset physiological features at adjacent time nodes from the latent physiological data stream, and calculate the absolute value of the rate of change of the values ​​relative to time. The latent physiological data stream extraction and absolute value calculation module is used to determine that the latent physiological intention is consistent with the issued explicit instruction if the absolute value of the rate of change at all adjacent time nodes is less than or equal to a preset rate of change threshold within the asynchronous physiological feature buffer time window, thereby releasing the suspending state and executing the system control operation. Latent physiological intention Figure 1 The consistency determination and system control operation execution module is used to determine the corresponding preset physiological feature as a mutation physiological feature if, within the asynchronous physiological feature buffer time window, the absolute value of any of the change rates is detected to be greater than the change rate threshold, and to determine that the latent physiological intention and the explicit instruction conflict at the time boundary. The latent physiological intent conflict determination and mutation physiological characteristic identification module is used to cancel the execution of the system control operation and call a secondary confirmation interaction instruction that matches the mutation physiological characteristic to replace the original system control operation.

[0020] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects: This invention discloses an asynchronous and continuous decoupling method for implicit intentions and explicit commands in immersive interactions. It proposes an innovative solution to the unique business problem of potential conflicts between explicit user commands and implicit physiological intentions in immersive interaction scenarios. The problem lies in the fact that when a user issues an explicit command, implicit physiological data may reflect an inconsistent true intention, leading to system misjudgment of user needs. This invention detects explicit command actions in real time and records the trigger timestamp, while simultaneously collecting implicit physiological data streams. An asynchronous physiological feature buffer time window is constructed starting from the trigger timestamp, and the rate of change of physiological features is analyzed to determine the consistency between implicit intentions and explicit commands. If the rate of change exceeds a threshold, a conflict is identified, the original operation is canceled, and a secondary confirmation interaction command is invoked to ensure decision accuracy. This method effectively avoids operational errors caused by intention conflicts, improves the intelligence and user experience of the interaction system, and achieves precise decoupling and dynamic coordination between explicit commands and implicit intentions. Attached Figure Description

[0021] Figure 1 This is a flowchart of an asynchronous and continuous decoupling method for implicit intentions and explicit instructions in immersive interaction according to the present invention.

[0022] Figure 2 This is a schematic diagram of the structure of an asynchronous and continuous decoupling system for implicit intentions and explicit instructions in immersive interaction according to the present invention. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0024] Terminology Explanation Immersive interactive scenarios refer to complex interactive spatial environments, whether digital or virtual-real fusion, constructed for users through advanced hardware and software technologies to provide a highly immersive multi-sensory experience. In this invention, such scenarios not only require the system to process explicit user input commands in real time, but also demand a high degree of realism and high-frequency interactive attributes. Due to the immediacy of the interaction and the severity of its consequences, it easily induces subconscious psychological and physiological fluctuations in users. In specific embodiments, this overarching concept can be further subdivided into: a virtual reality (VR) metaverse high-value virtual asset trading space environment; an L4-level autonomous driving car intelligent cockpit equipped with infrared eye tracking and an ECG steering wheel; a three-dimensional lesion visualization digital operating room with AR surgical microscope glasses; and a digital twin remote operation room for high-risk explosive demining while wearing a mixed reality (MR) headset, among other specific scenarios.

[0025] Explicit command actions refer to physical interactive behaviors that users, in immersive interactive scenarios, issue to the system through body language, voice, or other active input methods based on subjective awareness and clear execution intentions. These behaviors can be captured and parsed into standardized operation requests by the system's sensors and recognition algorithms. In this invention, it is a prerequisite for triggering subsequent system control operations and initiating asynchronous continuous decoupling mechanisms. In specific embodiments, this higher-level concept can be further subdivided into specific actions such as: a user's "wave right hand downwards" gesture to confirm the sale of assets in three-dimensional space; a driver pressing the "Auto-Pilot full control" physical button on the steering wheel; a doctor issuing a "start laser ablation" voice command via microphone combined with gaze persistence (GazeDwell); and a mine clearance engineer issuing specific spatial gestures through "pinch-drag-release," etc.

[0026] Implicit intention refers to the subconscious psychological activity and true desire that arises in the brain momentarily, without being explicitly expressed through language or large-scale actions, due to the user's perception of the potential risks of the interaction result, the correctness of their own decision, or sudden changes in the external environment after issuing an explicit command. In this invention, implicit intention often manifests as "regret," "hesitation," or a deviation from the rational goal of the just-issued command, and is a core monitoring object that the system needs to deeply explore to prevent misoperation. In specific embodiments, this higher-level concept can be further subdivided into: the regret arising momentarily due to loss aversion in high-value asset transactions; the extreme panic and distrust arising when facing complex road conditions at the moment of handing over control of autonomous driving; the uncertainty and hesitation arising momentarily when a critical blood vessel is caught in the peripheral vision during medical surgery; and the non-subjective muscle stiffness and spasms caused by extreme tension in high-risk explosive environments, etc.

[0027] Asynchronous continuous decoupling: This refers to breaking the default assumption in traditional human-computer interaction systems that "the triggering time of explicit actions and the user's internal psychological state are completely synchronized and statically corresponded in physical time." At the system software's underlying logic, it involves separating, independently collecting, and performing delayed comparative analysis of the instantaneously captured explicit instruction / action signals and the delayed, continuous physiological characteristic change signals representing implicit intentions, in the time dimension. In this invention, this mechanism enables the system to intercept erroneous instructions caused by the conflict between subjective consciousness and subconscious physiological instincts within the time difference between the physical instruction being issued and not yet taking effect. In specific embodiments, this higher-level concept can be further broken down into: an asynchronous separation mechanism between the instruction capture thread and the operation execution thread at the software level; a mechanism for instantiating system control tasks and pushing them into a delayed waiting queue to suspend the main thread; and a mechanism for independently extracting and calculating the characteristic fluctuation derivative of delayed "physiological traces" within a preset buffer time window.

[0028] Trigger timestamp: This refers to the precise absolute time stamp recorded by the system clock at the instant the action is completed when the system's multimodal interaction engine or action recognition module successfully matches and confirms the validity of the explicit command action issued by the user. In this invention, it is the temporal anchor point connecting explicit commands and implicit physiological data. It is not only used to generate system control interception signals, but also serves as the absolute starting point (zero point) for constructing an asynchronous physiological feature buffer time window along the time axis in the forward direction. In specific embodiments, this higher-level concept can be expressed as: absolute UTC time accurate to the millisecond level (e.g., 2026-05-02 10:00:00.000), nanosecond-level integer variables returned by the system's underlying clock function, and long integer system time cycle count values ​​containing time zone information, among other specific time data formats.

[0029] Implicit physiological data stream: refers to the set of raw biological signs that are continuously and uninterruptedly collected over time by a high-frequency sensor array worn on the user's body surface or integrated into an immersive interactive device. This data objectively reflects the activity of the user's autonomic nervous system and is difficult to conceal by conscious will. In this invention, this data stream is acquired synchronously with the occurrence of explicit commands and serves as the underlying data source for the system to understand the user's true implicit intentions. In specific embodiments, this higher-level concept can be further subdivided into: heart rate variability (HRV) data streams collected by photoplethysmography (PPG) sensors; continuous one-dimensional floating-point array signals of skin conductance collected by a 1000Hz sampling rate using a gel skin response (GSR) sensor; continuous pupil diameter pixel value sequences collected by an infrared eye-tracking camera; multi-channel brainwave microvolt voltage signal streams collected by a portable electroencephalography (EEG) sensor; and continuous respiratory rhythm waveform data collected by a piezoelectric breathing band, among other specific data streams.

[0030] System control operations refer to the underlying business logic or tasks that an interactive system is scheduled to invoke after receiving and confirming explicit instructions, used to change the system state or drive external hardware devices. In this invention, these operations typically involve high-value, high-risk, or irreversible consequences, and therefore need to be forcibly intercepted and put into a suspended state under a decoupling mechanism, awaiting final verification of the implicit physiological intent. In specific embodiments, this higher-level concept can be further subdivided into: calling the backend server's asset transfer and account balance deduction API interface program; the complete software transfer program of the vehicle chassis drive-by-wire system control from manual driving to the autonomous driving domain controller; sending an electrical signal command to the high-frequency medical laser emitter to start power output; and requesting control messages to the remote industrial robotic arm via network communication to close the main valve or detonate target explosives, etc., as specific execution-level operations.

[0031] Forward along the time axis: This refers to the direction of data processing and observation that extends progressively towards future time nodes (i.e., the direction t>t0) according to the natural law of the passage of physical time, with a specific historical time node (usually a trigger timestamp) as the absolute coordinate origin. In this invention, this term defines the construction direction of the time window for the system to capture "physiological trails," emphasizing the lagging characteristic of physiological stress response relative to physical action. In specific embodiments, this higher-level concept can be expressed as: a data flow control mechanism that instantiates a circular buffer in memory and continuously writes sensor data to the future time node t0+1.5 seconds; an event-driven logic that starts a forward timer from time t0 and continuously listens to the callback function; and a cursor movement rule that traverses the data index in the continuous sampling array in a monotonically increasing direction according to the timestamp, etc., as specific program execution logic.

[0032] Asynchronous physiological feature buffer time window: This refers to an extremely short continuous listening and data caching period artificially set along the time axis, starting from the trigger timestamp of the explicit instruction action after the system suspends the original system control operation. In this invention, the construction of this time window aims to compensate for the objective technical time difference of tens to hundreds of milliseconds between the human cognitive vocal / motor mechanism and the autonomic nervous physiological stress response, and is specifically used to frame and extract "physiological trails" data that may contain conflicting intentions. In specific embodiments, this higher-level concept can be further subdivided into: a circular memory data buffer with a length of 1.5 seconds, a time window for capturing abnormal pupil dilation image sequences with a length of 2.0 seconds, a time window for capturing minute changes in EEG stress with an extremely short duration (e.g., 800 milliseconds), and a dynamic buffer queue interception period that dynamically and adaptively extends from a default 0.5 seconds to 2 seconds according to the operational risk level, etc., as specific data interception window mechanisms.

[0033] Preset physiological characteristics: These refer to the core physiological indicators that are most sensitive to and quantified by algorithms, selected manually or optimized from massive, smoothed and denoised implicit physiological data streams based on prior knowledge of specific application scenarios or domain expert experience. These indicators are capable of most sensitively representing specific user emotions (such as panic, regret, or high concentration). In this invention, extracting the values ​​of this feature at adjacent time points is the basis for calculating feature fluctuation differences and judging abrupt changes. In specific embodiments, this higher-level concept can be further subdivided into: local peak values ​​of skin conductivity in a time series, the percentage of pupil dilation area pixels obtained from eye-tracking image analysis per unit time, the relative power spectral density of EEG signals in the Beta and Gamma wave bands, and specific quantitative feature indicators such as the muscle contraction intensity potential value extracted every 50 milliseconds based on root mean square (RMS) envelope filtering.

[0034] Suspended state: This refers to a state in which, after recognizing an explicit instruction, the system forcibly intercepts the system control operation corresponding to that instruction through software locks or hardware isolation mechanisms. This suspends the operation in memory, awaiting further condition triggering, creating a non-execution standstill. In this invention, this is a defensive intermediate system state, providing a valuable computational time window for asynchronous continuous decoupling and analysis of latent physiological data. In specific embodiments, this higher-level concept can be further categorized into specific software or hardware suspension techniques, such as instantiating network communication requests and pushing them into the system's delay queue (DelayQueue), generating operation interception signals and setting the underlying control module's running flag to a locked state using Boolean variable control logic, and a normally open electromagnetic relay maintenance mechanism that disconnects the mechanical steering connection in physical hardware-level middleware.

[0035] The first physiological characteristic sequence refers to the array of raw, continuous data segments directly from the hardware sensors, acquired by the system within the asynchronous physiological characteristic buffer time window, without any digital signal processing or noise reduction optimization. In this invention, this sequence, due to the presence of environmental noise, power frequency interference, or artifacts from the user's normal limb movements, cannot be directly used for accurate rate of change calculation and must be used as the input source for a smoothing filtering algorithm. In a specific embodiment, this higher-level concept can be further defined as: a one-dimensional floating-point array signal of raw skin conductivity (GSR) without eliminating high-frequency hardware noise from the workshop environment. The underlying raw data structures include a continuous sequence of original pixel values ​​of pupil diameter containing turbulence and vibration noise, an original multi-channel brainwave potential array doped with power frequency interference from operating room equipment, and an original high-frequency electromyographic potential oscillation sequence without envelope extraction.

[0036] Smoothing filtering: This refers to the technical process of using mathematical algorithms in the field of digital signal processing to clean the data, remove high-frequency spikes, and smooth the trend of the first physiological feature sequence mixed with noise, in order to restore the true waveform of the physiological signal. In this invention, it is the core preprocessing step to ensure the accuracy of feature fluctuation difference calculation and prevent noise-induced system misjudgment (i.e., false positive conflict). In specific embodiments, this higher-level concept can be further broken down into specific signal denoising and reconstruction techniques such as the Moving Average filtering algorithm with a window size of k, the low-pass Gaussian filtering algorithm based on the Gaussian kernel function, the optimal Kalman filter algorithm for dynamic state estimation in bumpy driving environments, the Discrete Wavelet Transform (DWT) denoising and reconstruction algorithm for eliminating strong electrical interference in brain waves, and the Root Mean Square (RMS) envelope filtering algorithm for extracting muscle force intensity.

[0037] The second physiological characteristic sequence refers to the pure data sequence array output after the first physiological characteristic sequence has been smoothed and filtered, removing environmental background noise and high-frequency interference spikes, and clearly reflecting the true gradual or abrupt trend of physiological indicators. In this invention, it is the direct and reliable basis for the system to extract values ​​from adjacent time nodes and then calculate the characteristic fluctuation difference and the absolute value of the rate of change. In a specific embodiment, this higher-level concept can be further defined as: the smoothed skin conductivity sequence after removing high-frequency noise. The system includes high-quality data structures such as the pupil diameter state sequence output by optimal estimation through Kalman filtering, the extremely pure Beta / Gamma wave frequency band energy sequence after wavelet decomposition and reconstruction, and the RMS electromyographic envelope curve array representing the actual force intensity of muscles.

[0038] Characteristic fluctuation difference: refers to the difference between two closely adjacent sampling time points on the time axis in a pure second physiological feature sequence (e.g., and The mathematical subtraction result between the extracted preset physiological characteristic quantification values ​​is used in this invention. In this invention, this difference reflects the absolute increase or decrease of the physiological characteristic within a very short time slice, and is the fundamental numerator for deriving the rate of physiological change (i.e., the rate of change) per unit time. In a specific embodiment, this higher-level concept can be reduced to a lower-level concept: the algebraic difference between skin conductivity values ​​at adjacent nodes. Specific micro-calculations include the percentage difference in blood oxygen saturation between adjacent 1-second intervals (e.g., 94%-95%=-1%), and the voltage floating-point difference in muscle contraction intensity between adjacent 50-millisecond sampling periods.

[0039] Physiological baseline variance: This refers to the statistical variance of a specific user's baseline physiological fluctuations and metabolic levels, calculated by the system continuously collecting physiological characteristic data during a user's non-task state or historically stable interaction periods. In this invention, it serves as a highly individualized baseline, used for dynamic calibration and generation of preset rate-of-change thresholds, thereby eliminating threshold judgment errors caused by differences in physical constitution among different users. In specific embodiments, this higher-level concept can be further defined as: the statistical variance of data within historically stable interaction periods. The specific baseline calibration parameters include the mean fluctuation range of the calm baseline calculated based on the skin conductivity during the 30-second rest period in the game lobby, the background variance of pupil diameter fluctuation continuously recorded under normal driving conditions, and the normal muscle exertion potential baseline envelope variance extracted after the user completes three standard and stable gripping actions.

[0040] The preset rate of change threshold refers to a safe red line limit for the rate of change of physiological characteristics, dynamically calculated or pre-set by the system based on factors such as the user's physiological baseline variance and operational risk level. In this invention, it is the sole mathematical criterion for judging whether abnormal "abrupt changes" occur in the data within the asynchronous physiological characteristic buffer time window. Once the absolute value of the rate of change exceeds this limit, an intent conflict determination is triggered. In a specific embodiment, this higher-level concept can be reduced to the lower-level concept: the dynamic rate of change threshold Vth = α*, dynamically generated by multiplying the baseline standard deviation by a sensitivity coefficient and combining it with a time reference value. The high-risk sensitive alarm threshold is narrowed down based on operational risk level and physiological benchmarks; the dynamic safety redundancy limit is calibrated by combining the mean of the first derivative of the instantaneous baseline; and the normal muscle exertion threshold limit is derived through basic exertion testing. ) and other specific critical judgment values.

[0041] Mutant physiological characteristics: These refer to abnormal physiological data segments within the asynchronous physiological characteristic buffer time window where the absolute value of the rate of change relative to time exceeds a preset rate of change threshold, exhibiting unusually sharp increases or decreases. In this invention, such sudden and drastic physiological fluctuations are characterized by the system as an objective mapping of the user's subconscious and genuine stress response, serving as key data evidence directly proving a contradiction between implicit intentions and explicit commands. In specific embodiments, this higher-level concept can be further subdivided into: abnormal skin conductivity data sequences that surge instantaneously due to sympathetic nerve excitation caused by extreme regret; abnormal pupil diameter dilation area data that suddenly amplifies due to stress-induced panic while driving; the sharp peak fluctuation of Beta brain waves accompanied by instantaneous breath-holding caused by extreme tension; and specific abnormal physiological signal clusters such as the involuntary electromyographic potential surge envelope far exceeding the normal exertion threshold caused by extreme panic or convulsions.

[0042] The temporal boundary refers to the extremely short transitional microscopic time interval immediately following the completion of an explicit instruction action (i.e., the establishment of the trigger timestamp). In this invention, this concept emphasizes the temporal overlap and misalignment between the human consciousness issuing an action instruction and the body's physiological stress response due to perceived errors. It is precisely because the system captures the abrupt physiological characteristics within this crucial temporal gap that it can determine that the apparent nature of the instruction deviates from the user's subconscious. In specific embodiments, this higher-level concept can be further subdivided into: the preset conflict time window (such as an extremely short interval of 0 to 1 second) into which the temporal deviation value obtained by subtracting the timestamp of the abrupt physiological characteristic data from the timestamp of the explicit instruction reception falls; the physiological trail fermentation period of tens to hundreds of milliseconds after the trigger timestamp; and the high-frequency temporal domain intersection region during the microsecond-level comparison between the action capture thread and the physiological monitoring thread, among other specific time judgment boundaries.

[0043] Secondary confirmation interaction command: This refers to a high-security alternative interaction request that is reissued to the user after the system determines, through an asynchronous continuous decoupling algorithm, that a conflict exists between implicit physiological intent and explicit command, and successfully intercepts (cancels) the original system control operation. This request is characterized by strong reminders, prevention of accidental touches, and a high level of security. In this invention, it serves as the system's final safety fallback mechanism when it detects potential user misoperation, aiming to force the user to return from subconscious reactions to rational decision-making by changing the interaction modality. In specific embodiments, this higher-level concept can be further subdivided into: a physical button secondary confirmation logic that pops up a bright red 3D floating anti-accidental touch panel in the center of the VR field of view and triggers high-frequency tactile vibration of the gloves; a safety intervention mechanism that tightens the seatbelt pretensioner for tactile warning and broadcasts the highest-priority directional voice prompt; a strong intervention command that renders the AR marker as a flashing yellow warning color and requires the voice to read the safety code to unlock; and a highest-level lock-off protocol that turns off the ambient backlight and requires both hands to simultaneously press the physical safety valve on the console, among other specific interaction degradation and forced verification strategies.

[0044] The sliding time window algorithm refers to a streaming data processing algorithm that, when processing infinitely long or uninterrupted sensor time series data, sets a fixed-length data block (window) and moves it continuously forward along the time axis with a specific step size, sequentially capturing and calculating local data features within the window. In this invention, this algorithm is particularly suitable for eliminating random burst interference introduced by immersive environments (such as cabin turbulence or vibration of wearable devices), ensuring real-time and smooth extraction of continuous data segments within the buffer time window. In specific embodiments, this higher-level concept can be further categorized into: a step-by-step extraction algorithm that continuously captures the first physiological feature sequence within the buffer time window with a sliding window step size of 100 milliseconds and a window width of 300 milliseconds; and specific time series framing and batch processing techniques such as applying a local smooth moving average extraction mechanism with a window size of 3 or 5 points to the heart rate sequence.

[0045] Action recognition algorithm: refers to a computer vision or machine learning model used by a system to process limb trajectory or posture data captured by input sensors (such as cameras, spatial locators) in real time and determine whether it matches a certain meaningful operation command predefined in the system. In this invention, successful matching by the action recognition algorithm is a necessary prerequisite core step for generating "valid explicit command actions" and recording trigger timestamps. In specific embodiments, this higher-level concept can be further subdivided into: dynamic time warping (DTW) pattern matching algorithm based on the coordinate sequence matrix of three-dimensional hand skeleton keypoints from a depth camera, long short-term memory (LSTM) trajectory classification algorithm for processing spatiotemporal action classification, and gesture pinch-and-release spatial coordinate recognition algorithm based on multi-dimensional spatial bounding box collision detection, etc., as well as specific computer vision and temporal pattern recognition technologies.

[0046] Resampling algorithm: In multimodal data acquisition, this refers to a data conversion mechanism that, given the inherently different original sampling rates of different sensors, uses digital signal processing to uniformly convert various sensor-acquired signals to the same target sampling rate, and uses the trigger timestamp of the completion of the explicit action as a unified time zero point for strict alignment. In this invention, the algorithm breaks down the clock and sampling barriers between heterogeneous hardware, smoothly mapping the complex implicit physiological data stream to the same time reference dimension. This process not only eliminates cumbersome time-frequency domain conversion calculations but is also a core data preprocessing step that ensures accurate extraction of continuous data segments within a sliding time window, elimination of temporal misalignment deviations, and accurate extraction of physiological features.

[0047] Example 1 In Embodiment 1 of this invention, an asynchronous continuous decoupling method is provided for immersive interaction in virtual reality (VR) high-value virtual asset trading scenarios, specifically for decoupling implicit intentions from explicit commands. The immersive interaction scenario in this embodiment is a three-dimensional spatial environment where a user wears a VR headset and data-sensitive haptic gloves with physiological monitoring functions to engage in virtual reality games or metaverse asset trading. The explicit command action is, in this context, a "wave your right hand downwards" gesture performed by the user in three-dimensional space to confirm the sale of the asset. The implicit intention refers to the subconscious psychological activity of "regret" or "hesitation" that arises instantly in the brain after issuing a sell command due to aversion to losing high-value assets. Asynchronous continuous decoupling refers to breaking the traditional assumption that actions and psychological states are always aligned, and using technical means to separate and independently analyze instantaneous hand gestures and delayed physiological changes in the time dimension.

[0048] During actual transactions, the system first detects in real time the preset explicit commands issued by the user in the immersive interactive scenario. When a valid "asset confirmation sale" gesture is recognized by the built-in three-dimensional spatial coordinate recognition algorithm (such as a trajectory classification algorithm based on Long Short-Term Memory Network LSTM), the system immediately records the trigger timestamp (usually expressed as absolute time) when the explicit command is completed. For example, 2026-05-02 10:00:00.000), and simultaneously acquire the implicit physiological data stream continuously generated by the user on the time axis through sensors. The implicit physiological data stream here is specifically reduced to the heart rate variability (HRV) data stream acquired by the photoplethysmography (PPG) sensor embedded in the glove and the skin conductivity data stream acquired by the skin conductance (GSR) sensor.

[0049] To capture the "physiological trail" lagging behind the explicit command, the system temporarily suspends the execution of the system control operation corresponding to the explicit command action (in this embodiment, this is reduced to calling the asset transfer deduction API interface of the backend server), causing the system to enter a suspended waiting state. Simultaneously, the system uses the trigger timestamp... Starting from the beginning, along the positive direction of the timeline (i.e., towards the future time) (Direction) Extend the construction of an asynchronous physiological feature buffer time window of preset duration (e.g., setting the window length) Seconds, that is, the time interval is ).

[0050] After the buffer time window is constructed, the system extracts the latent physiological data stream within the asynchronous physiological feature buffer time window. Because the original data contains environmental noise, the system first obtains the first physiological feature sequence within this time window. The noise is then reduced using a lower-level smoothing filter (such as a Gaussian filter or a moving average filter). Taking a moving average filter with a window size of k as an example, the formula is: After smoothing filtering, the system obtains a second physiological feature sequence with high-frequency noise removed. Based on this second physiological feature sequence, the system extracts preset physiological features (specifically, the peak value of skin conductivity in this embodiment) from the latent physiological data stream at adjacent time nodes (set as follows). and The characteristic fluctuation difference is calculated from the values ​​on the graph. Subsequently, the system calculates the absolute value of the rate of change of the stated value with respect to time, defined mathematically as follows: To assess a user's implicit intent, the system needs to consider the user's baseline state. The system pre-extracts the user's physiological baseline variance during historical stable interaction periods (such as rest periods in the game lobby). As a personalized baseline, its calculation formula is: ,in This is the historical benchmark average. Based on this, the system dynamically generates a preset rate of change threshold Vth=α* (Where σ is the baseline standard deviation,) The reference time step is set for the system to unify the dimensions, and α is the set sensitivity coefficient, which is 1.5 in this embodiment.

[0051] To ensure accuracy, the system dynamically generates a judgment benchmark. Specifically, during the non-interactive preparation period (e.g., the first 30 seconds) of a user's immersive login scenario, the system continuously collects and calculates the benchmark variance of the user's baseline skin conductivity. Subsequently, the system retrieves the operational risk level weight of the current explicit instruction (e.g., the weight coefficient for a virtual asset sale instruction is set to 1.5). The system multiplies the benchmark variance by the operational risk level weight to dynamically calculate a personalized preset rate of change threshold applicable to the current user. This dynamic calculation method eliminates the interference of differences in user constitution, ensuring that the judgment of "physiological trails" is based entirely on the user's relative physiological abnormalities, rather than absolute values.

[0052] If, within the asynchronous physiological characteristic buffer time window, the absolute value of the rate of change at all adjacent time points is less than or equal to a preset rate of change threshold... If the implicit physiological intention (calmness, certainty) is determined to be consistent with the issued explicit instruction, the suspension state is lifted and the system control operation for the asset transfer is immediately executed. If the system detects the absolute value of any of the aforementioned rates of change... (This indicates that regret leads to sympathetic nerve excitation, causing a sudden spike in skin conductivity), thus the corresponding preset physiological characteristic is identified as a mutation physiological characteristic. This indicates that the user's implicit physiological intention (regret, revocation) and explicit instruction (sell) conflict at the time boundary of the instruction issuance (i.e., within a very short period after the trigger timestamp).

[0053] Based on this conflict determination, the system completely cancels the system control operation to execute the asset transfer and replaces the original system control operation with a secondary confirmation interaction command that matches the mutated physiological characteristics. In this scenario, the secondary confirmation interaction command is reduced to "a 3D floating panel marked with a bright red anti-accidental touch indicator pops up in the center of the VR field of view, and at the same time triggers the high-frequency haptic vibration motor of the glove, requiring the user to perform a secondary irreversible confirmation via a physical button." Through this embodiment, the system successfully identifies the user's instantaneous regret after the operation, avoiding the loss of high-value virtual assets due to accidental operation.

[0054] Example 2 In a second embodiment of the present invention, an asynchronous and continuous decoupling method for implicit intentions and explicit commands in immersive interaction scenarios applied to intelligent cockpits and advanced autonomous driving takeover scenarios is provided. The immersive interaction scenario here is specifically defined as an L4-level autonomous driving vehicle cockpit equipped with an infrared eye-tracking camera, steering wheel center electrode, and intelligent voice assistant. In this scenario, the explicit command action specifically refers to the driver pressing the "Auto-Pilot Full Takeover" physical button on the steering wheel; the corresponding system control operation is defined as the complete transfer of control of the vehicle chassis drive-by-wire system from manual driving to the autonomous driving domain controller. The implicit intention here is manifested as the driver's distrust and extreme panic at the moment of handing over control due to complex road conditions (such as heavy rain or construction ahead).

[0055] The system monitors the driver's preset explicit commands in real time via the cockpit controller. When the "Auto-Pilot full control" button is detected to be validly triggered, the system records the trigger timestamp when the action is completed. Simultaneously, it acquires implicit physiological data streams collected by an infrared eye-tracking camera and a steering wheel ECG sensor. This data stream is specifically downdivided into continuous pupil diameter sequences (representing the degree of panic) and transient heart rate interval (RR interval) sequences.

[0056] To ensure driving safety, the system does not immediately disconnect the driver's mechanical steering connection, but instead suspends the system control operation of transferring control. Starting from this point, an asynchronous physiological characteristic buffer time window with a length of 2.0 seconds is opened along the positive time axis.

[0057] When extracting data within the buffer time window, due to the bumpy and vibration-prone driving environment, the system needs to apply a sliding time window algorithm to truncate and extract features from the original continuous data segments. Specifically, the system sets the sliding window step size to 100 milliseconds and the window width to 300 milliseconds, sliding continuously within the buffer time window. For the first physiological feature sequence (original pixel value of pupil diameter) within each sliding window, the system uses a down-level Kalman filter algorithm for state estimation and smoothing to calculate the optimal estimated second physiological feature sequence. The core state update formula of the Kalman filter includes: State prediction: Covariance prediction: Subsequently, the system extracts the values ​​of preset physiological features (here, pupil dilation area per unit time) at adjacent time points from the smoothed latent physiological data stream, calculates the feature fluctuation difference, and then obtains the absolute value of the rate of change. .

[0058] The system retrieves the driver's physiological baseline variance under normal driving conditions and uses this to set a threshold for the rate of change of abnormal pupil dilation. If the absolute value of all rates of change does not exceed the threshold within the time window, it indicates that the driver's intention to hand over the responsibility is firm and their mindset is calm. The system then lifts the suspension state, and the autonomous driving domain controller officially takes over the vehicle. Conversely, if the threshold is not exceeded within the window... The system flagged this as a mutation in physiological characteristics (i.e., a sudden dilation of the pupils due to panic). This clearly demonstrates that the driver's latent physiological intentions (extreme insecurity and a desire to grip the steering wheel) and the recently issued explicit takeover command clashed severely at the time boundary.

[0059] Due to the conflicting judgment, the system immediately canceled the system control operation to transfer control. To ensure driving safety, the system invoked a secondary confirmation interaction command (which, in this case, translates to: tightening the driver's seatbelt pretensioner as a tactile warning, while simultaneously broadcasting a high-priority directional voice prompt through the cabin audio: "Your state of tension has been detected; the transfer of control has been terminated. Please continue manual driving"). This utilizes an asynchronous continuous decoupling mechanism to intercept a potentially fatal danger caused by driver panic within a very short time difference.

[0060] Example 3 In Embodiment 3 of this invention, an asynchronous and continuous decoupling method for implicit intentions and explicit commands in immersive interaction applied to a medical augmented reality (AR) surgical navigation scenario is described. Here, the immersive interaction scenario is categorized as a digital operating room where a doctor wears AR surgical microscope glasses to visualize lesions in three dimensions during neurosurgery. The explicit command action is categorized as the doctor's voice command "Start laser ablation" delivered via microphone, combined with continuous gaze dwell on the lesion model (GazeDwell). The system control operation is categorized as sending a power output start signal to a high-frequency medical laser transmitter. In this ultra-high precision scenario, even a fraction of a second of hesitation by the doctor (i.e., implicit intention) could lead to irreversible damage to healthy neurons.

[0061] During the surgery, the system's multimodal interaction engine monitors the explicit commands and actions issued by the doctor in real time. Once the Natural Language Processing (NLP) module successfully parses "start laser ablation" and the gaze tracking module confirms that the gaze has remained in place for the preset time, the system records the trigger timestamp at this point. Meanwhile, the system continuously and synchronously acquires implicit physiological data streams collected through a portable electroencephalogram (EEG) sensor attached to the doctor's head and a piezoelectric breathing band.

[0062] Given the irreversible nature of the surgical procedure, the system temporarily suspends the laser activation process, preventing immediate laser emission. The system uses a trigger timestamp... Using the origin as the starting point, extending positively along the time axis, an asynchronous physiological feature buffer time window with extremely short latency (set to 800 milliseconds) is constructed to capture instantaneous, minute physiological changes.

[0063] The system extracts the EEG brainwave signal within the 800-millisecond buffer time window as the first physiological feature sequence. To eliminate power frequency interference from the operating room's high-voltage equipment, the system employs a lower-level smoothing filter (specifically, the Discrete Wavelet Transform (DWT) noise reduction algorithm). This is achieved through wavelet decomposition functions... The system separates high-frequency noise and reconstructs an extremely pure second physiological feature sequence (focusing on extracting the Beta and Gamma wave frequency bands reflecting focus and stress). Based on this, the system extracts the values ​​of preset physiological features (reduced to Beta wave relative power spectral density) at adjacent time nodes, obtains the feature fluctuation difference, and then calculates the absolute value of its rate of change relative to time. .

[0064] The system retrieves the physiological baseline variance recorded by the doctor during the pre-operative preparation phase when the doctor is resting with their eyes closed, and calculates the threshold for abrupt brain wave fluctuations. If all during the buffer period This indicates that the doctor is certain that the extent of the resection was correct (implicit physiological intention). Figure 1 (To), the system releases the pause, and the laser fires precisely. However, in this scenario, if the system detects a momentary breath-holding in the breathing rhythm, and This is identified as a mutation in physiological characteristics. It reveals that after a doctor issues an instruction, their peripheral vision may catch a glimpse of a key blood vessel, instantly generating an uncertain, implicit intention. This time lag between cognition and vocalization leads to a conflict of intentions at the temporal boundary.

[0065] Accordingly, the system intercepts and completely cancels the system control operation for laser emission. Instead, it invokes a secondary confirmation interaction command, which in this embodiment is rendered as: "In the surgical field of view of the AR glasses, the previously marked red ablation target area is rendered as a flashing yellow warning color, and a pop-up window appears in the upper right corner of the doctor's field of vision: 'Brainwave stress fluctuation detection: Laser is locked. Please read the security code aloud again to unlock.'" Through asynchronous continuous decoupling, the system grants the doctor a "final right of withdrawal" in life-or-death operations, allowing the body's instincts to precede conscious thought.

[0066] Example 4 In Embodiment 4 of this invention, an asynchronous continuous decoupling method for implicit intentions and explicit commands in immersive interaction in mixed reality (MR) industrial remote control and high-risk explosive mine clearance scenarios is provided. Specifically, the immersive interaction scenario here involves a mine clearance engineer wearing an MR headset and a multi-channel electromyography (EMG) armband, performing spatial remote operations on a remote robotic arm and mine clearance explosives. In this extremely high-risk environment, explicit commands are reduced to "detonate the target detonation point" commands issued by the engineer through specific "pinch-drag-release" spatial gestures. Due to the extremely high pressure of the operating environment, the human limbs may misuse the gestures due to muscle tension; the implicit intention here is the deviation between the non-subjective instinctive reaction caused by limb stiffness or spasm and the actual rational target.

[0067] When the MR spatial tracking camera detects a valid "detonate release" gesture in real time, the system immediately records a timestamp indicating the completion of the explicit command action. Simultaneously, the system acquires a stream of latent physiological data continuously generated by an electromyography (EMG) sensor worn on the engineer's forearm at a high sampling rate of 1000 Hz.

[0068] Detonation is a destructive system control operation. Once triggered, the system's hardware-level middleware immediately and forcibly intercepts the operation, placing it in a suspended state. Starting from zero, an asynchronous physiological characteristic buffer time window with a length of 1.0 second is opened along the positive time axis.

[0069] Within the buffer time window, the system extracts continuous raw electromyographic potentials as the first physiological feature sequence. To accurately quantify muscle tension, the system employs a lower-order root mean square (RMS) envelope filtering algorithm for smoothing, obtaining the second physiological feature sequence characterizing muscle exertion intensity. The RMS feature extraction formula is as follows: Based on this smooth envelope, the system extracts the muscle contraction intensity of adjacent sampling nodes (e.g., every 50 milliseconds) as a preset physiological feature, calculates the feature fluctuation difference, and then obtains the absolute value of the electromyographic rate of change. .

[0070] The system pre-requires engineers to complete three standard, stable gripping actions while wearing the device, extracts the user's physiological baseline variance, and uses this to derive the normal muscle exertion threshold. If all within the time window If the absolute values ​​are all below the threshold, the system confirms that the engineer's muscle control is stable and the implicit physiological intention is firm. The system then releases the pause state and sends a real physical detonation signal to the remote end.

[0071] However, if the system detects during the monitoring process... A sharp rise, far exceeding Furthermore, the duration of the spasm was extremely short, leading the system to classify it as a sudden physiological characteristic caused by convulsions or extreme tension. This involuntary muscle spasm indicates that at the point where the command was triggered, the user was not actually in a stable psychological state ready to detonate; a fatal conflict occurred between physiological instincts and the control gestures.

[0072] At this point, the system, through its core decoupling algorithm, disconnects the physical gesture from the final control intention (the essence of asynchronous continuous decoupling), canceling the system control operation to send a detonation signal to the remote end. To ensure absolute safety, the system invokes the highest-level secondary confirmation interaction command, which here is translated as: "Disable all virtual interactive buttons in the MR space, dim the overall ambient background light (enter safe mode), and require the engineer to simultaneously press the physical safety valves on both sides of the control panel with both hands to unlock the system's highest-level lock." This method greatly enhances the fault tolerance of human-machine interaction in high-risk industrial scenarios.

[0073] Example 5 Embodiment 5 of this invention provides a low-level algorithm and code-level implementation scheme for an asynchronous continuous decoupling method of implicit intentions and explicit commands in immersive interaction for remote control scenarios of high-precision industrial collaborative robots. In this embodiment, the "immersive interaction scenario" is represented by a remote hazardous materials handling digital twin control room where the operator wears a mixed reality head-mounted display and is connected to ECG and electrodermal sensors. The so-called "explicit command action" specifically refers to the operator's hand-clamping action to the system to shut off the main valve through spatial gestures. "Implicit intention" refers to the subconscious psychological state of stopping or hesitating that occurs when the operator observes an abnormal reading of pipeline pressure at the moment of clamping. "Asynchronous continuous decoupling" is manifested in the system's asynchronous separation of the command capture thread and the operation execution thread at the software level, and the independent calculation and analysis of the "physiological trail" that occurs after the delay. To achieve accurate recognition of the above-mentioned gesture actions, the system pre-deploys a trained action recognition model. Specifically, the system collects a large number of three-dimensional hand skeleton key point coordinates of operators under different lighting and field of view angles as positive samples, and non-instructional daily hand waving actions as negative samples to construct a training dataset. During model training, the training dataset is input into a Long Short-Term Memory (LSTM) network. The error between the output confidence probability and the actual action label is calculated using the Cross-Entropy Loss function. The Adam optimizer iteratively updates the weight parameters of the forget gate, input gate, and output gate within the network using backpropagation until the loss function converges, completing the training of the action recognition model. During inference, the model input is a 3D coordinate matrix of 21 hand keypoints from T consecutive frames. Temporal action features are extracted using the LSTM network, and the fully connected layer outputs the confidence probability of a specific explicit instruction action. A valid action is determined and a trigger timestamp is generated only if the confidence exceeds a preset recognition threshold (e.g., 0.95), thus filtering out invalid action afterimages caused by environmental interference.

[0074] The system first executes S1, which detects the user's preset explicit commands and actions in the immersive interactive scenario in real time. When a valid explicit command or action is identified, the system records the trigger timestamp of the action's completion and simultaneously acquires the implicit physiological data stream continuously generated by the user on the timeline through sensors. In the code implementation, the system's main thread runs the action recognition module, whose input parameter is the three-dimensional hand skeleton key point coordinate sequence matrix acquired in real time by the depth camera. The system calls a lower-level Dynamic Time Warping (DTW) algorithm function to perform pattern matching, calculating the alignment distance between the input trajectory sequence and the standard action template. Where Q represents an input trajectory sequence of length N. This represents a standard action template sequence of length M. To normalize the path length, For the first on the path Alignment point coordinates, Indicates the alignment point The Euclidean distance between two sequence elements is calculated. When the return value is less than a set distance threshold, the action is deemed valid, and the trigger timestamp is obtained via the system clock function. For example, it returns a precise nanosecond-level integer variable. At the same time, the sensor monitoring thread in the system background uses a non-blocking callback mechanism to synchronously acquire the implicit physiological data stream, which is then converted into a one-dimensional floating-point array signal of continuous skin conductance (GSR) with a sampling rate of 1000Hz.

[0075] Next, the system executes S2, temporarily suspending the system control operation corresponding to the explicit command action, and constructing an asynchronous physiological feature buffer time window of a preset duration, starting from the trigger timestamp and extending forward along the time axis. In the program logic, the system control operation is quantified as a network communication request to close a valve sent to a remote industrial robot. The system does not immediately execute this request, but instead instantiates the control task and pushes it into the system's delayed waiting queue, causing the system control to enter a locked, suspended state. Simultaneously, the system calls the buffer construction function, with the trigger timestamp as the input parameter. And a preset duration variable, such as 1.5 seconds. This function instantiates a circular buffer in memory, along the positive time axis, i.e., towards the future time. The system continuously writes the implicit physiological data stream collected by the sensor within a given time period, starting from the second. The raw data obtained is defined in the code as an array of first physiological feature sequences. .

[0076] The system then executes S3, extracting multiple heterogeneous latent physiological data streams within the asynchronous physiological feature buffer time window. Using the trigger timestamp as a unified time zero point, a resampling algorithm is employed to align the latent physiological data streams with different sampling rates to the same time reference dimension. Subsequently, the system extracts the values ​​of preset physiological features at adjacent time nodes from the aligned latent physiological data streams and calculates the absolute value of the rate of change of these values ​​relative to time. To eliminate high-frequency hardware noise from the workshop environment, the system first processes the first physiological feature sequence using a lower-order low-pass smoothing filter function, outputting a denoised array of the second physiological feature sequence. Subsequently, the system extracts adjacent time nodes from the smoothed sequence within the loop. and The preset physiological characteristics (i.e., specific skin conductivity values) are used to calculate the characteristic fluctuation difference. Next, the system calls the data change rate calculation function, whose input is the characteristic fluctuation difference and the time difference between adjacent sampling points. The internal mathematical calculation formula is as follows: This process then iterates through the data to generate an array containing all the absolute values ​​of the rates of change. .

[0077] To make comparisons, the system backend extracts the physiological baseline variance recorded by the current user during historical stable interaction periods. The specific formula for calculating this variance is as follows: ,in Using historical averages and considering the risk coefficient of current high-risk operations, a preset rate of change threshold Th is dynamically calculated. At this point, the system proceeds to step S4. If, within the asynchronous physiological characteristic buffer time window, the absolute value of the rate of change at all adjacent time points is less than or equal to the preset rate of change threshold (i.e., the maximum value in the program's judgment array V does not exceed Th), then it is determined that the latent physiological intention is consistent with the issued explicit instruction, indicating that the operator is mentally stable and has confirmed the correctness. The system then executes a dequeue operation, releases the pause state, and issues an execution signal, allowing the remote robotic arm to officially shut off the valve.

[0078] Conversely, in stage S5, if the absolute value of any of the aforementioned rates of change is detected to be greater than the rate of change threshold within the asynchronous physiological characteristic buffer time window, the system identifies the corresponding set of skin conductance data exceeding the threshold as a mutated physiological characteristic. This mutated characteristic triggers an anomaly flag in the code, indicating a serious conflict between the operator's latent physiological intention (extreme panic arising instantaneously from the discovery of a potential hazard) and the recently issued explicit cutoff command within a very short time after the trigger timestamp (i.e., at the time boundary).

[0079] Upon triggering of the anomaly flag, the system enters the anomaly handling branch (S6), cancels the system control operation, and replaces the original system control operation with a secondary confirmation interaction command matching the mutated physiological characteristics. At the underlying implementation level, the system calls the operation termination function to destroy the valve shut-off task in the delay queue, blocking physical execution. Simultaneously, based on the matching algorithm mapping to the high-intervention level secondary confirmation interaction command, a full-screen semi-transparent red visual warning layer is rendered in the center of the operator's head-mounted display field of vision. A high-frequency pulse drive signal for the vibration motor is sent to the operator's hand controller, requiring the operator to click the physical safety confirmation button on the control panel twice consecutively to reactivate subsequent system control operations. Through this series of detailed function calls and mathematical calculations, precise asynchronous decoupling of commands and intentions is achieved.

[0080] like Figure 1 As shown, an embodiment of the present invention provides a method and system for asynchronous and continuous decoupling of implicit intentions and explicit instructions in immersive interaction, which may specifically include: S1. Real-time detection of preset explicit command actions issued by the user in an immersive interactive scenario. When a valid explicit command action is identified, the trigger timestamp of the explicit command action is recorded, and the implicit physiological data stream continuously generated by the user on the timeline is simultaneously acquired through sensors.

[0081] The process involves acquiring limb movement trajectories and matching them with preset action templates to obtain action recognition results. In this embodiment, the implicit physiological data stream is collected in real time by a sensor array worn by the user or integrated into immersive interactive devices (such as VR headsets or smart cockpit seats). These sensors include, but are not limited to: high-frame-rate eye-tracking cameras for capturing micro-expressions and pupil diameter changes; photoplethysmography (PPG) sensors for collecting skin conductivity (GSR) to reflect the intensity of sympathetic nerve activity; and piezoelectric breathing bands for monitoring respiratory rhythm. The data stream continuously generated by these sensors can objectively and without being deliberately obscured by the user's subjective consciousness, reflecting their true subconscious emotional fluctuations. If the action recognition result is successfully matched, the trigger time when the explicit command is completed is recorded; based on the trigger time, a sliding time window algorithm is used to extract continuous data segments from the sensor acquisition signal, and the continuous data segments are denoised to obtain a latent physiological sequence; physiological features are extracted based on the latent physiological sequence to complete the real-time detection of the preset explicit command actions issued by the user in the immersive interactive scenario, and the trigger timestamp is recorded when the explicit command action is recognized to synchronously acquire the latent physiological data stream.

[0082] In immersive interactive scenarios, users issue commands through body movements to achieve real-time synchronous acquisition of physiological data.

[0083] For example.

[0084] For example, in virtual reality games, when a user waves their arm as an explicit command, the system first captures the trajectory of the limb movement through a camera.

[0085] For example, limb movement trajectories can be collected in real time by optical sensors to collect the user's joint position data, forming a spatiotemporal sequence in which each data point includes three-dimensional coordinates and a timestamp, thereby describing the dynamic changes of the movement.

[0086] In one embodiment, when matching the limb movement trajectory with a preset action template, a dynamic time warping algorithm can be used to align the collected trajectory with the template sequence and calculate the similarity score. If the score exceeds a threshold, it is considered a successful match. For example, the template is a standardized trajectory of the "raising hand" action. The user trajectory is normalized and then matched to obtain the action recognition result.

[0087] For example, if a match is successful, the system records the trigger time when the explicit command is completed, such as using the system clock to record a timestamp accurate to milliseconds. Based on the trigger time, continuous data segments are obtained by extracting the sensor acquisition signal.

[0088] For example, from continuous signals acquired by a heart rate sensor, data within 5 seconds before and after the trigger time is extracted to form a segment, which is used to capture the physiological response at the moment of the command. Noise is then removed from the continuous data segment to obtain a latent physiological sequence.

[0089] For example, wavelet transform can be applied to remove high-frequency noise and retain low-frequency physiological signals, such as heart rate variability sequences. Physiological features can then be extracted based on these latent physiological sequences.

[0090] For example, the mean, standard deviation, and frequency domain metrics such as the low-frequency to high-frequency ratio of a sequence can be calculated to assess a user's stress level.

[0091] In one embodiment, the system performs real-time detection of preset explicit commands issued by the user in an immersive interactive scenario and records the trigger timestamp when the explicit command is detected, while simultaneously acquiring implicit physiological data streams.

[0092] For example, in a medical rehabilitation VR system, the timing of a user's leg-raising action triggers the simultaneous collection of EEG data to monitor emotions, thereby enhancing the personalization of the interactive experience.

[0093] S2. Temporarily suspend the system control operation corresponding to the explicit instruction action, and construct an asynchronous physiological feature buffer time window of preset duration by extending along the time axis from the trigger timestamp.

[0094] The trigger timestamp corresponding to the explicit instruction is extracted, and this timestamp is used to generate an operation interception signal to obtain a system control lock state. An asynchronous physiological feature buffer time window is constructed based on the system control lock state, and this window is used to acquire a first physiological feature sequence. The first physiological feature sequence is smoothed and filtered to obtain a second physiological feature sequence. The system control lock state is released based on the second physiological feature sequence, thereby temporarily suspending the system control operation corresponding to the explicit instruction action. An asynchronous physiological feature buffer time window of a preset duration is constructed, extending forward along the time axis from the trigger timestamp.

[0095] In one embodiment, the system first extracts the trigger timestamps from the explicit command actions performed by the user in the virtual reality environment.

[0096] For example, when a user initiates device control via a gesture command, the system records the precise moment the gesture is completed as a timestamp, thus providing a reference for subsequent signal generation.

[0097] In one embodiment, the system generates an operation interception signal based on the trigger timestamp.

[0098] For example, in a smart home interaction scenario, if a user issues a command to close the door lock, the system uses a timestamp to generate a temporary intercept signal. This signal is transmitted to the control module through an internal communication protocol. Upon receiving the signal, the module immediately switches to the lock mode to prevent the command from being executed in real time.

[0099] Specifically, the process of generating the intercept signal involves comparing a timestamp with a preset threshold. If the timestamp falls within a specific range, the signal encoder is activated to convert the timestamp into a binary pulse sequence and add a check bit to ensure transmission reliability. This encoding method is similar to the error correction and verification algorithm in digital communication, thereby obtaining a system control lock state. In the lock state, all related operations are suspended, awaiting further verification.

[0100] In one embodiment, an asynchronous physiological characteristic buffer time window is constructed based on the system control lock state.

[0101] For example, in augmented reality games, after the lockout is triggered, the system creates a buffer window by extending a preset duration, such as 5 seconds, along the time axis starting from the timestamp. This window asynchronously collects the user's physiological data, such as heart rate and skin conductance, and samples it in real time through a sensor array to form a first physiological feature sequence. The sequence includes continuous numerical points, such as the record of heart rate fluctuations from 80 beats / min to 90 beats / min.

[0102] In one embodiment, the first physiological feature sequence is smoothed and filtered to obtain the second physiological feature sequence.

[0103] For example, a moving average filter can be used to take the average of several points before and after each data point in the sequence. For instance, a filter with a window size of 3 can be applied to a heart rate sequence. The calculation method is to take the average of the current point and the two adjacent points, thereby smoothing out noise interference and obtaining a more stable second sequence.

[0104] In one embodiment, the system control lock state is released based on the second physiological characteristic sequence.

[0105] For example, if the second sequence shows normal physiological indicators, the system sends an unlock signal to resume operation execution, thereby temporarily suspending the system control operation corresponding to the explicit instruction action.

[0106] S3. Extract the latent physiological data stream within the asynchronous physiological feature buffer time window, extract the values ​​of preset physiological features at adjacent time nodes from the latent physiological data stream, and calculate the absolute value of the rate of change of the values ​​relative to time.

[0107] The process involves: acquiring the original physiological signal sequence and simultaneously acquiring the environmental stimulus event log of the current moment in the immersive interactive scene; extracting preset high-intensity sensory stimulus time points from the environmental stimulus event log; removing the environmental stress physiological data segments corresponding to the high-intensity sensory stimulus time points from the original physiological signal sequence to obtain a denoised physiological signal sequence; smoothing the denoised physiological signal sequence to obtain a latent physiological data stream; extracting the feature values ​​of preset physiological features at adjacent time points from the latent physiological data stream, and obtaining the feature fluctuation difference based on the feature values; calculating the rate of change based on the feature fluctuation difference and the time difference; determining whether the rate of change is less than zero; if the rate of change is less than zero, inverting the rate of change to obtain an absolute value, thus completing the extraction of the latent physiological data stream within the asynchronous physiological feature buffer time window; extracting the values ​​of preset physiological features at adjacent time points from the latent physiological data stream and calculating the absolute value of the rate of change of the values ​​relative to time.

[0108] In one possible implementation, the user's raw physiological signal sequence is first acquired via sensors, such as continuous pulse waveform data recorded by a heart rate monitoring device. This data may contain noise interference. Subsequently, a smoothing process is applied to the sequence, such as using a moving average filter, setting the window size to 5 sampling points, and calculating the average of each point with its neighbors to eliminate high-frequency noise, thereby obtaining a latent physiological data stream that more smoothly reflects the underlying physiological changes.

[0109] For example, during processing, predefined physiological features, such as blood oxygen saturation values ​​at adjacent time points, are extracted from the implicit physiological data stream. Assuming time points t1 and t2 are 10:00:00 and 10:00:01 respectively, with feature values ​​of 95% and 94%, the feature fluctuation difference is calculated based on these values: 94% minus 95%, resulting in a difference of -1%. This difference captures short-term fluctuations in physiological features, providing fundamental data for subsequent analysis.

[0110] In one possible implementation, the rate of change is then calculated based on the difference in characteristic fluctuations and the time difference. For example, if the time difference is 1 second, the rate of change is -1% divided by 1 second, which gives -1% / s. This rate represents the trend of physiological characteristics over time and helps to identify declining or rising patterns.

[0111] For example, if the rate of change is less than zero, such as -1% / s as mentioned above, then it is inverted to obtain the absolute value of 1% / s, thereby uniformly processing the fluctuation intensity in the positive and negative directions, and completing the extraction and calculation process of the implicit physiological data stream within the asynchronous physiological characteristic buffer time window.

[0112] In one possible implementation, this absolute value calculation ensures a non-negative representation of the rate of change, facilitating subsequent comparisons and threshold determinations.

[0113] It should be further explained that the preset rate of change threshold is not a fixed empirical constant, but rather dynamically generated through the baseline calibration procedure during the system initialization phase. Specifically, when a user first enters the immersive interactive scenario, the system guides the user through a non-task browsing session of a preset duration (e.g., 30 seconds), during which the collected physiological characteristic data is used as a calm baseline. The system calculates the average fluctuation range of this calm baseline data sequence and multiplies it by a preset safety redundancy coefficient (e.g., between 1.5 and 2.0) to obtain a personalized preset rate of change threshold. This dynamic calibration method effectively eliminates individual differences in basal metabolic rate and physiological characteristic baselines among different users, ensuring subsequent... Figure 1 The objectivity and accuracy of consistency determination.

[0114] S4. If, within the asynchronous physiological characteristic buffer time window, the absolute value of the rate of change at all adjacent time nodes is less than or equal to a preset rate of change threshold, then it is determined that the latent physiological intention is consistent with the issued explicit instruction, the suspension state is lifted, and the system control operation is executed.

[0115] Acquire a data sequence of asynchronous physiological characteristics within a buffer time window; calculate the numerical differences between adjacent time nodes based on the data sequence and perform absolute value conversion to obtain a set of absolute values ​​of the rate of change; determine whether all values ​​in the set of absolute values ​​of the rate of change are less than or equal to a preset rate of change threshold; if all values ​​in the set of absolute values ​​of the rate of change are less than or equal to the preset rate of change threshold, then determine that the latent physiological intention and the explicit instruction are intentional. Figure 1 Consistency; according to the stated meaning Figure 1 The system releases the delayed state to obtain an activation command queue; an execution signal is issued to the activation command queue to determine whether the latent physiological intention is consistent with the issued explicit command and to release the delayed state to execute the system control operation.

[0116] In one embodiment, the system first acquires asynchronous physiological characteristics, such as heart rate or skin conductance data, from physiological sensors to form a sequence within a buffer time window, thereby providing a basis for subsequent analysis.

[0117] For example, in medical assistance systems, acquiring the data sequence of asynchronous physiological characteristics within a buffer time window can be achieved through real-time monitoring devices.

[0118] Specifically, the user wears a wearable sensor, and the system sets a buffer time window of 5 seconds to collect heart rate sequences, such as values ​​per second: 80, 82, 81, 79. By calculating the differences between values ​​at adjacent time points (e.g., the difference between 80 and 82 is 2, the difference between 82 and 81 is -1, and the difference between 81 and 79 is -2), absolute value conversion is performed to obtain a set of absolute values ​​of the rate of change {2, 1, 2}. Then, it is determined whether all values ​​in this set are less than or equal to a preset rate of change threshold, such as a threshold of 3. If all values ​​meet this threshold, it is determined that implicit physiological intentions, such as a calm state, and explicit instructions, such as "stay relaxed," are met. Figure 1 Consistency. Based on this consistency, the system releases the paused state and generates an activation command queue, including "start relaxation music" and "adjust the lights." Execution signals are issued to this queue, thereby ensuring that the latent physiological intention is consistent with the issued explicit instructions, and the paused state is released to execute system control operations, such as automatically playing music to assist the user in relaxation. In this approach, a threshold determination algorithm is introduced to ensure consistency between the latent physiological intention and the issued explicit instructions, thus releasing the paused state and executing system control operations, such as automatically playing music to assist the user in relaxation. Figure 1 Consistency assessment is more reliable, thereby optimizing system response efficiency.

[0119] For example, in extended scenarios, this method can be applied to smart home control to enhance the user experience.

[0120] Furthermore, to address the varying fault tolerance rates of different operations in immersive interactive scenarios, this system introduces a dynamic buffering mechanism based on operational risk levels. The system pre-establishes a mapping table between explicit commands and operational risk levels. For example, "view switching" or "interface swiping" is defined as low-risk; "asset transaction confirmation," "irreversible file deletion," or "advanced intelligent driving takeover" is defined as high-risk. Upon obtaining the operational risk level corresponding to an explicit command action, if it is determined to be a high-risk trigger condition, the system will invoke the physiological baseline variance calculated for the current user during historical stable interaction periods. The system employs a proportional mapping strategy, meaning that the higher the operational risk level, the longer the dynamically set preset duration (i.e., the length of the buffer time window). Specifically, the system uses the following linear mapping rule: Preset duration = Basic buffer duration + (Operational risk coefficient * Extension step size). For example, when the risk level is defined as a regular viewpoint switch (risk coefficient of 1), the preset duration remains at 0.5 seconds; when the risk level is defined as an irreversible asset transaction (risk coefficient of 5), the preset duration is dynamically extended to 2.0 seconds to provide sufficient display period for physiological stress responses. Simultaneously, by combining physiological baseline variance with a reduced dynamic change rate threshold, the system becomes more sensitive to subtle physiological abnormalities (such as slight breath-holding) after high-risk operations. Through this dynamic adaptive adjustment based on risk classification, an optimal balance between interaction efficiency and interaction safety is achieved.

[0121] S5. If, within the asynchronous physiological feature buffer time window, the absolute value of any of the change rates is detected to be greater than the change rate threshold, the corresponding preset physiological feature is determined to be a mutation physiological feature, and it is determined that the latent physiological intention and the explicit instruction conflict at the time boundary.

[0122] Obtain the physiological characteristic change rate sequence within the physiological characteristic time window; if the absolute value of the physiological characteristic change rate sequence is greater than a preset change rate threshold, then the corresponding physiological characteristic data is determined to be mutated physiological characteristic data; extract the timestamp of the mutated physiological characteristic data; calculate the timing deviation value between the timestamp and the timestamp of the explicit instruction reception; if the timing deviation value is within a preset conflict time window, then determine that the latent physiological intention and the explicit instruction conflict at the time boundary based on the mutated physiological characteristic data.

[0123] It should be further explained that users' physiological characteristics can be affected by the immersive scene itself (such as intense game visuals). To accurately isolate abrupt physiological changes purely caused by "intent conflict," the dynamic rate of change threshold in this embodiment is not a fixed constant. The system extracts the smoothed curve of physiological characteristics within a first preset time period (e.g., the first 10 seconds) before the explicit command is issued as an immediate baseline and calculates the mean of the first derivative of this immediate baseline. Only when the absolute value of the physiological rate of change within the buffer time window is simultaneously greater than the system's preset basic safety threshold and deviates from the mean of the first derivative by a set multiple is it considered a valid mutation. This dynamic calibration mechanism effectively eliminates physiological artifacts caused by normal user limb movements or startling background noise, greatly improving the robustness of decoupling judgment.

[0124] In one embodiment, the system first collects physiological signals of the user within a specific time window, such as heart rate or brainwave data, to form a sequence of physiological characteristic change rates, thereby capturing potential intention fluctuations.

[0125] For example, in driver assistance systems, acquiring the rate of change sequence of physiological characteristics within a time window involves real-time monitoring of the driver's skin conductivity, sampling once per second, calculating the difference between adjacent sampling points and taking the absolute value to form a sequence. If a value in the sequence is greater than a preset threshold, such as 0.5, the corresponding data is marked as a sudden physiological characteristic data. This step, by introducing a first-order difference calculation algorithm, enables the system to identify physiological spikes such as those caused by sudden stress, thus providing a basis for subsequent conflict detection.

[0126] It should be noted that the preset rate of change threshold can be dynamically adjusted based on the user's historical data. For example, the threshold is 0.3 in a calm state, but increases to 0.7 in a high-intensity task to adapt to different scenarios and ensure the accuracy of detection.

[0127] Specifically, when extracting the timestamps of the mutated physiological characteristic data, the system records the UTC time accurate to milliseconds, such as 2023-10-01 14:30:45.678. Then, it calculates the deviation from the timestamp of the explicit command received. For example, if the command time is 14:30:45.000, the deviation is 0.678 seconds. If this deviation is within a preset conflict time window, such as 0-1 seconds, it is determined that there is a conflict between the latent physiological intention and the explicit command. For example, if the driver issues an "accelerate" command but the physiological mutation displays a panic response, a safety intervention mechanism is triggered. This method optimizes the intention through time-series deviation analysis. Figure 1 The consistency assessment process enables timely responses to potential risks in business operations.

[0128] In one embodiment, this determination can be extended to medical monitoring scenarios to improve the robustness of the system.

[0129] S6. Cancel the execution of the system control operation and call a secondary confirmation interaction instruction that matches the mutated physiological characteristics to replace the original system control operation.

[0130] The system acquires current physiological characteristic data; if the current physiological characteristic data exceeds the mutation threshold range, it determines that the current system control operation has an abnormal triggering condition; it extracts a secondary confirmation instruction based on the abnormal triggering condition, and the secondary confirmation instruction matches the current physiological characteristic data; it generates a secondary confirmation interaction instruction to be executed based on the secondary confirmation instruction; it cancels the execution of the current system control operation and calls the secondary confirmation interaction instruction to be executed to replace the original system control operation. Specifically, the system pre-stores a mapping matrix between mutation characteristics and interaction intervention strategies. When the detected mutation physiological characteristic is a sudden surge in skin conductivity (indicating high-intensity stress or shock), the system extracts a secondary confirmation instruction at a strong intervention level, that is, suspends the current screen and provides a high-frequency tactile vibration prompt, forcing the user to confirm with a physical button; when the detected mutation physiological characteristic is micro-expression stiffness or high-frequency shifting of gaze outside the target area (indicating hesitation or uncertainty), the system extracts a secondary confirmation instruction at a weak intervention level, that is, displays the operation consequences prompt in the form of a semi-transparent floating window on the original interaction interface, and extends the waiting time for the user to cancel the instruction. By employing this differentiated degradation and interception strategy based on physiological mutation types, the system's security is ensured while maximizing the preservation of a seamless immersive interactive experience.

[0131] In one possible implementation, the system first collects the user's physiological signals, such as heart rate and skin conductivity, in real time using sensors as the foundational source of current physiological characteristic data. This data originates from wearable devices, such as smart bracelets, which sample once per second to ensure real-time accuracy, thus providing accurate input for subsequent judgments.

[0132] For example, in a driver assistance system, when a user's heart rate suddenly jumps from a normal 80 beats per minute to 120 beats per minute, the system compares this value with a preset threshold range for sudden changes, such as ±20% of the normal value. If the threshold is exceeded, it is determined that the current system control operation, such as automatic acceleration, has an abnormal trigger condition. This determination is based on a threshold comparison algorithm, where the abnormal trigger condition is confirmed by calculating the deviation between the current data and the historical average.

[0133] Specifically, the system first extracts the average physiological characteristics of the user over the past hour from the database, then calculates the percentage deviation. If the deviation exceeds a threshold, such as 25%, it is marked as abnormal, triggering the next step of processing. This method ensures timely identification of anomalies and avoids misjudgments.

[0134] In one possible implementation, based on the aforementioned anomaly triggering conditions, the system extracts a secondary confirmation command from a preset command library. For example, if the anomaly is a sudden change in heart rate, the corresponding voice confirmation command, such as "Please confirm whether to continue accelerating," is matched. This extraction process involves a rule-based mapping matching algorithm, where the command library is a categorized database grouped by physiological characteristic types such as heart rate and blood pressure. Each group of commands is associated with a specific anomaly threshold. By querying the anomaly type, the system quickly locates and extracts matching commands, ensuring the relevance of the commands to the current physiological characteristic data.

[0135] For example, when generating a secondary confirmation interaction command to be executed, the system will generate a specific form based on the extracted command template and the current context, such as converting the voice command into an interactive method of screen pop-up and voice broadcast. This generation process includes template filling. For example, if the command template is "[physiological characteristics] abnormality detected, [operate]?", after filling it in, it becomes "[heart rate abnormality detected, continue to accelerate]?" thus preparing for execution.

[0136] In one possible implementation, the original operation, such as acceleration, is ultimately canceled, and a generated interactive instruction is invoked, such as a confirmation dialog box that waits for the user to input "yes" or "no," thereby replacing the original operation and maintaining system security.

[0137] See Figure 2 The present invention also provides an asynchronous continuous decoupling system for implicit intentions and explicit instructions in immersive interaction, mainly comprising: The explicit command action detection module is used to detect the preset explicit command actions issued by the user in the immersive interaction scenario in real time. When a valid explicit command action is identified, the trigger timestamp when the explicit command action is completed is recorded, and the implicit physiological data stream continuously generated by the user on the timeline is acquired through the sensor. The trigger timestamp recording and physiological data stream synchronization acquisition module is used to temporarily suspend the system control operation corresponding to the explicit instruction action, and to construct an asynchronous physiological feature buffer time window of preset duration by extending along the time axis from the trigger timestamp as the starting point. The system control operation suspension and asynchronous physiological feature buffer time window construction module is used to extract the latent physiological data stream within the asynchronous physiological feature buffer time window, extract the values ​​of preset physiological features at adjacent time nodes from the latent physiological data stream, and calculate the absolute value of the rate of change of the values ​​relative to time. The latent physiological data stream extraction and absolute value calculation module is used to determine that the latent physiological intention is consistent with the issued explicit instruction if the absolute value of the rate of change at all adjacent time nodes is less than or equal to a preset rate of change threshold within the asynchronous physiological feature buffer time window, thereby releasing the suspending state and executing the system control operation. Latent physiological intention Figure 1 The consistency determination and system control operation execution module is used to determine the corresponding preset physiological feature as a mutation physiological feature if, within the asynchronous physiological feature buffer time window, the absolute value of any of the change rates is detected to be greater than the change rate threshold, and to determine that the latent physiological intention and the explicit instruction conflict at the time boundary. The latent physiological intent conflict determination and mutation physiological characteristic identification module is used to cancel the execution of the system control operation and call a secondary confirmation interaction instruction that matches the mutation physiological characteristic to replace the original system control operation.

[0138] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application 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 do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for asynchronous and continuous decoupling of implicit intentions and explicit instructions in immersive interaction, characterized in that, The method includes: S1. Real-time detection of preset explicit command actions issued by the user in an immersive interactive scenario. When a valid explicit command action is identified, the trigger timestamp when the explicit command action is completed is recorded, and the implicit physiological data stream continuously generated by the user on the timeline is simultaneously acquired through the sensor. S2. Temporarily suspend the system control operation corresponding to the explicit instruction action, and construct an asynchronous physiological feature buffer time window of preset duration by extending along the time axis from the trigger timestamp as the starting point; S3. Extract the latent physiological data stream within the asynchronous physiological feature buffer time window, extract the values ​​of preset physiological features at adjacent time nodes from the latent physiological data stream, and calculate the absolute value of the rate of change of the values ​​relative to time; S4. If, within the asynchronous physiological characteristic buffer time window, the absolute value of the rate of change at all adjacent time nodes is less than or equal to a preset rate of change threshold, then it is determined that the latent physiological intention is consistent with the issued explicit instruction, the suspension state is lifted, and the system control operation is executed.

2. The method according to claim 1, characterized in that, S4 is followed by: S5. If, within the asynchronous physiological feature buffer time window, the absolute value of any of the change rates is detected to be greater than the change rate threshold, then the corresponding preset physiological feature is determined to be a mutation physiological feature, and it is determined that the latent physiological intention and the explicit instruction conflict at the time boundary. S6. Cancel the execution of the system control operation and call a secondary confirmation interaction instruction that matches the mutated physiological characteristics to replace the original system control operation.

3. The method according to claim 1, characterized in that, S1 includes: The limb movement trajectory is acquired, and the limb movement trajectory is matched with a preset action template to obtain the action recognition result; If the action recognition result is a successful match, the trigger time when the explicit instruction is completed is recorded; The resampling algorithm is invoked to align sensor acquisition signals with different original sampling rates collected by different sensors to the same time reference dimension, with the trigger time as a unified time zero point. Based on the trigger time, a sliding time window algorithm is used to extract continuous data segments from the sensor acquisition signal, and the continuous data segments are denoised to obtain a latent physiological sequence. Physiological features are extracted based on the recessive physiological sequences.

4. The method according to claim 1, characterized in that, S2 includes: Extract the trigger timestamp corresponding to the explicit instruction; the trigger timestamp is used to generate an operation interception signal to obtain a system control lock state. An asynchronous physiological feature buffer time window is constructed based on the system control lock state, and the asynchronous physiological feature buffer time window is used to acquire the first physiological feature sequence. The first physiological feature sequence is smoothed and filtered to obtain the second physiological feature sequence; The system control lock state is released based on the second physiological characteristic sequence.

5. The method according to claim 1, characterized in that, S3 includes: The raw physiological signal sequence is acquired and smoothed to obtain the implicit physiological data stream; Extract the feature values ​​of preset physiological features at adjacent time nodes from the implicit physiological data stream, and obtain the feature fluctuation difference based on the feature values; The rate of change is calculated based on the characteristic fluctuation difference and the time difference. Determine whether the rate of change is less than zero. If the rate of change is less than zero, invert the rate of change to obtain the absolute value.

6. The method according to claim 1, characterized in that, S4 includes: Obtain the data sequence of asynchronous physiological characteristics within a buffer time window; Calculate the numerical differences between adjacent time nodes based on the data sequence and perform absolute value conversion to obtain a set of absolute values ​​of the rate of change. Determine whether all values ​​in the set of absolute values ​​of the rate of change are less than or equal to a preset rate of change threshold. If all values ​​in the set of absolute values ​​of the rate of change are less than or equal to a preset rate of change threshold, then the implicit physiological intention and the explicit instruction are determined to be consistent. Based on the consistency of intent, the deferred state is lifted to obtain the activation instruction queue; An execution signal is issued to the activation instruction queue to determine whether the latent physiological intention is consistent with the issued explicit instruction and to release the suspended state to execute the system control operation.

7. The method according to claim 2, characterized in that, S5 includes: Obtain the sequence of physiological characteristic change rates within a time window; If the absolute value of the physiological characteristic change rate sequence is greater than the preset change rate threshold, the corresponding physiological characteristic data is determined to be mutated physiological characteristic data. The timestamps of the extracted mutation physiological characteristic data; Calculate the timing deviation between the timestamp and the explicit instruction reception timestamp; If the timing deviation value is within a preset conflict time window, then based on the mutation physiological characteristic data, it is determined that the latent physiological intention and the explicit instruction conflict at the time boundary.

8. The method according to claim 2, characterized in that, S6 includes: Obtain current physiological characteristic data; If the current physiological characteristic data exceeds the mutation threshold range, it is determined that there is an abnormal triggering condition in the current system control operation; A secondary confirmation instruction is extracted based on the abnormal triggering conditions, and the secondary confirmation instruction is matched with the current physiological characteristic data; Generate a secondary confirmation interaction instruction to be executed based on the aforementioned secondary confirmation instruction; Cancel the execution of the current system control operation and call the pending secondary confirmation interaction instruction to replace the original system control operation.

9. The method according to claim 1, characterized in that, S1 includes: real-time detection of preset explicit command actions issued by the user in an immersive interactive scenario, obtaining the operation risk level corresponding to the explicit command action; when the operation risk level reaches a preset high-risk triggering condition and a valid explicit command action is identified, recording the trigger timestamp when the explicit command action is completed, and simultaneously acquiring the implicit physiological data stream continuously generated by the user on the timeline through sensors. S2 includes: temporarily suspending the system control operation corresponding to the explicit instruction action, calling the physiological baseline variance of the current user during the historical stable interaction period, dynamically setting the corresponding preset duration and dynamic change rate threshold according to the operation risk level and the physiological baseline variance; and constructing an asynchronous physiological feature buffer time window of the preset duration by extending it forward along the time axis from the trigger timestamp.

10. An asynchronous and continuous decoupling system for implicit intentions and explicit instructions in immersive interaction, characterized in that, The system includes: The explicit command action detection module is used to detect the preset explicit command actions issued by the user in the immersive interaction scenario in real time. When a valid explicit command action is identified, the trigger timestamp when the explicit command action is completed is recorded, and the implicit physiological data stream continuously generated by the user on the timeline is acquired through the sensor. The trigger timestamp recording and physiological data stream synchronization acquisition module is used to temporarily suspend the system control operation corresponding to the explicit instruction action, and to construct an asynchronous physiological feature buffer time window of preset duration by extending along the time axis from the trigger timestamp as the starting point. The system control operation suspension and asynchronous physiological feature buffer time window construction module is used to extract the latent physiological data stream within the asynchronous physiological feature buffer time window, extract the values ​​of preset physiological features at adjacent time nodes from the latent physiological data stream, and calculate the absolute value of the rate of change of the values ​​relative to time. The latent physiological data stream extraction and absolute value calculation module is used to determine that the latent physiological intention is consistent with the issued explicit instruction if the absolute value of the rate of change at all adjacent time nodes is less than or equal to a preset rate of change threshold within the asynchronous physiological feature buffer time window, thereby releasing the suspending state and executing the system control operation. The latent physiological intention consistency determination and system control operation execution module is used to determine the corresponding preset physiological feature as a mutation physiological feature if, within the asynchronous physiological feature buffer time window, the absolute value of any of the change rates is detected to be greater than the change rate threshold, and to determine that the latent physiological intention and the explicit instruction conflict at the time boundary. The latent physiological intent conflict determination and mutation physiological characteristic identification module is used to cancel the execution of the system control operation and call a secondary confirmation interaction instruction that matches the mutation physiological characteristic to replace the original system control operation.

Images