A digital processing-based intelligent window breaking control method and system for a passenger vehicle

Abstract

This application relates to the field of vehicle safety technology, specifically to a digitally processed intelligent window-breaking control method and system for passenger vehicles. It includes: acquiring operational and environmental information, preprocessing this information to obtain first operational and environmental information; determining whether multiple information elements conform to preset emergency characteristics based on the first operational and environmental information; identifying the emergency type based on the emergency characteristics when multiple information elements conform to the preset emergency characteristics; determining at least one target window from a preset set of windows based on the emergency type, the physical properties of each window, and the status information of each window; determining a window-breaking scheme based on the emergency type and the physical properties of the target window; and issuing an operation command corresponding to the window-breaking scheme to the window-breaking actuator to form an escape route. This solves the technical problem that existing vehicle emergency window-breaking devices rely on manual operation and have a limited success rate in breaking windows.

Description

Technical Field

[0001] This application relates to the field of vehicle safety technology, and more specifically, to a method and system for intelligent window breaking control of passenger vehicles based on digital processing. Background Technology

[0002] In emergencies such as vehicle submersion, spontaneous combustion, or carbon monoxide leaks, rapid occupant escape is crucial for ensuring safety. Currently, most passenger vehicles are equipped with emergency window-breaking devices that are primarily mechanical safety hammers or built-in window breakers. Their effectiveness depends on the occupant's discovery and manual operation in an emergency, making it a key but also relatively passive link in the escape chain. For occupants who are immobile, confused, or unable to operate effectively due to panic, there is considerable uncertainty regarding whether they can break the window in a timely and successful manner to create an escape route.

[0003] Existing mechanical window breaking solutions suffer from two main technical drawbacks: First, their triggering relies entirely on manual intervention, which can lead to failure in sudden emergencies due to delayed reaction, disability, or improper operation, thus failing to achieve autonomous emergency response. Second, their operation typically involves a single physical impact point, making the window-breaking effect overly dependent on the local strength of the impact point and the instantaneous force exerted by the occupant. The reinforced design and tinted windows of modern automobiles may affect the breaking effect, posing a risk of single-point failure. Therefore, how to reduce the absolute dependence on manual operation in the window-breaking process and improve the success rate and reliability of window breaking under complex window structures is a pressing technical problem to be solved in the field of vehicle passive safety.

[0004] To address the aforementioned issues, existing technologies urgently need improvement. Summary of the Invention

[0005] This application discloses a digital processing-based intelligent window breaking control method and system for passenger vehicles, aiming to solve the technical problem that existing vehicle emergency window breaking devices rely on manual operation and have a limited success rate in breaking windows.

[0006] The technical solution of this application is as follows: In a first aspect, this application discloses a digital processing-based intelligent window-breaking control method for passenger vehicles, comprising: The system acquires operational and environmental information inside and outside the vehicle, and preprocesses the operational and environmental information to obtain first operational and environmental information. The operational and environmental information includes multiple information elements, including external water level information, water pressure information, smoke information inside the vehicle compartment, carbon monoxide concentration information, temperature information, vehicle attitude information, and door and / or window status information. The preprocessing includes filtering and / or calibration. Based on the first operational and environmental information, a correlation analysis is performed on multiple information elements to determine whether the combination pattern of multiple information elements within the first preset time window conforms to the preset emergency situation characteristics. When the combination pattern of multiple information elements within the first preset time window matches the preset emergency characteristics, the emergency type is identified based on the emergency characteristics. Emergency types include vehicle falling into water, spontaneous combustion, and / or carbon monoxide leakage. Based on the type of emergency, the physical properties of each window, and the status information of each window, at least one window to be broken is determined from a preset set of windows as the target window. The physical properties include glass type, thickness, and / or whether it is tinted. Based on the type of emergency and the physical properties of the target vehicle window, determine the window breaking strategy, which includes impact energy parameters, number of impact points, distribution parameters, and / or impact timing parameters. The corresponding window-breaking scheme is issued to the window-breaking mechanism to create an escape route.

[0007] Furthermore, based on the first operational and environmental information, a correlation analysis is performed on multiple information elements to determine whether the combination pattern of the multiple information elements within a first preset time window conforms to preset emergency characteristics, which also includes: When the combination pattern of multiple information elements within the first preset time window does not meet the characteristics of an emergency, and at least two of the multiple information elements are in an abnormal state but have not reached the preset trigger threshold for identifying the type of emergency, the information elements in the abnormal state are identified as fuzzy abnormal signals. Risk assessment is performed based on the combination pattern and duration of fuzzy abnormal signals to obtain risk level assessment results; When the risk level assessment result points to a medium risk level, a voice prompt is issued through the vehicle's loudspeaker, and at least one window in the window assembly is controlled to lower. Within the second preset time window after issuing a voice prompt and performing the descent operation, the system continuously monitors whether there are any signs of active operation by the occupant and obtains the result of the occupant's active operation judgment. After performing the descent operation, the operating and environmental information is reacquired and preprocessed to obtain the second set of operating and environmental information. Compare and evaluate the changes in multiple information elements of the first and second operational and environmental information to obtain the change assessment results; The risk level assessment results are adjusted based on the change assessment results and the occupants' active operation judgment results: when the change assessment results indicate a significant deterioration of the danger and the occupants' active operation judgment results indicate that the occupants are unresponsive, the risk level indicated by the risk level assessment results is upgraded to a high-risk level, and the combination pattern of multiple information elements within the first preset time window is determined to meet the characteristics of an emergency.

[0008] Furthermore, signs of active operation include voice response, door unlocking or door handle operation, window operation, and / or seatbelt unfastening. Continuous monitoring of whether occupants exhibit signs of active operation yields a judgment result on occupant active operation, including: When at least one indication of active action is detected, the result of the occupant active action judgment is determined as an occupant response; When no signs of active operation are detected, micro-motion signals in the occupant area and image information of the occupant's face are acquired, and the micro-motion signals and image information are used as information elements in the operation and environmental information. The micro-motion signals are time-series signals that characterize the chest and abdomen fluctuations or small movements of the body surface caused by the occupant's breathing and / or heartbeat, and the image information is a sequence of occupant facial image frames including the eye area. Periodic features are extracted from micromotion signals to determine whether the occupant has respiratory and / or cardiac activity, thus obtaining micromotion signal analysis results; The image information is analyzed to obtain the eye movement trajectory and / or pupil changes and to determine the occupant's state of consciousness, thus obtaining the image information analysis results; When the micro-motion signal analysis results indicate that the occupant has breathing and / or heartbeat activity, and the image information analysis results indicate that the occupant is conscious, the occupant is determined to be in a state of consciousness but incapacity to act. The voice prompt is issued again, the second preset time window is extended, and after detecting signs of active operation, if no signs of active operation are detected, the occupant's active operation judgment result is determined to be that the occupant is unresponsive. When the micro-motion signal analysis results indicate that the occupant has no breathing and / or heartbeat activity, or the image information analysis results indicate that the occupant is unconscious, the occupant is determined to be in an unconscious or disabled state, and the occupant's active operation judgment result is determined to be occupant unresponsive.

[0009] Furthermore, image information of the occupants' faces is acquired, including: The distance between the occupant and multiple image acquisition units with near-infrared supplementary lighting function are obtained. The intensity of the near-infrared supplementary lighting is adjusted according to the distance so that the image acquisition units can acquire images of the occupant's facial area from different angles to obtain multi-angle occupant facial images. Multi-angle occupant facial images are fused to generate a synthetic image containing complete facial information of the occupants; The occupant's eye region image is identified and extracted from the synthesized image, and the occupant's eye region image is enhanced to highlight the features of the eyeball and pupil; A sequence of occupant facial image frames containing the eye region is generated based on the enhanced eye region image, serving as image information.

[0010] Furthermore, issue another voice prompt, extending the second preset time window, including: Assess the rate of evolution of hazardous situations based on the results of change assessments; Obtain the vehicle's remaining battery power information; Based on the evolution speed and remaining battery power information, determine the extension duration of the second preset time window and the frequency of voice prompts; The system will issue voice prompts again within the extended second preset time window, following the frequency of the voice prompts, and will continue to monitor for any signs of active operation by the occupants.

[0011] Furthermore, the rate of evolution of the hazardous situation is assessed based on the change assessment results, including: Based on the sampling stability and / or signal attenuation degree of at least one information element in the first operation and environment information and the second operation and environment information, determine the reliability parameter corresponding to at least one information element; When the reliability parameter is lower than the preset reliability threshold, the weight of the change rate of the corresponding information element is reduced and / or eliminated. Based on the information elements of the first and second operational and environmental information, the corresponding rate of change is calculated. The corresponding rate of change is then weighted and fused based on the weights after weight reduction and / or elimination to obtain the change assessment result. The evolution rate of the hazardous situation is then determined based on the change assessment result.

[0012] Furthermore, the image information is analyzed to obtain eye movement trajectories and / or pupil changes and to determine the occupant's state of consciousness, including: Acquire vehicle vibration data and occupant head posture changes; The eye movement trajectory is extracted from the image information. The eye movement trajectory is correlated with vehicle vibration data and occupant head posture change information. The eye displacement component caused by vehicle vibration and / or occupant head posture change is identified and filtered out to obtain the filtered eye movement trajectory. Based on the filtered eye movement trajectory, it is determined whether there is voluntary eye rotation and / or saccadic behavior, and the eye movement analysis results are obtained; The image information is analyzed to obtain changes in pupil diameter, and based on the changes in pupil diameter, it is determined whether the pupil contracts and / or expands in response to changes in the intensity of near-infrared supplementary light, thus obtaining the pupil change analysis results. By combining the results of eye movement analysis and pupil change analysis, the occupant's state of consciousness was determined.

[0013] Furthermore, based on the filtered eye movement trajectory, the presence of voluntary eye rotation and / or saccades is determined, yielding eye movement analysis results, including: The filtered eye movement trajectories are subjected to motion pattern recognition to identify whether there are preset autonomous rotation patterns and / or saccade patterns in the filtered eye movement trajectories. The degree of autonomy of the eye movement patterns is evaluated based on the duration, repetition frequency, and stability of the movement direction of the autonomous rotation patterns and / or saccade patterns. Determine the frequency and amplitude information of vehicle vibration based on vehicle vibration data; The degree of autonomy is correlated with the frequency and amplitude information of vehicle vibration. Vibration correlation correction is performed on the degree of autonomy to reduce the impact of residual drift related to vehicle vibration in the filtered eye movement trajectory on the degree of autonomy, and the corrected degree of autonomy is obtained. The intentional judgment of eye movement patterns is made based on the corrected degree of autonomy. The intentional judgment includes: judging whether the eye movement pattern shows tracking of a specific stimulus, regular scanning of the in-vehicle environment and / or directional response to voice prompts. The specific stimulus is a stimulus actively emitted by a pre-set stimulus device. Based on the results of the intention judgment, the eye movement analysis results are obtained.

[0014] Furthermore, intentional judgments are made on eye movement patterns based on the corrected level of autonomy, including: Obtain the spatial location of a specific stimulus in the vehicle coordinate system and its corresponding time series; When the corrected level of autonomy meets the preset autonomy judgment criteria, the correlation between the filtered eye movement trajectory used to characterize the eye movement pattern and the spatial location and time sequence of the specific stimulus is analyzed to determine whether the eye movement pattern exhibits a continuous following behavior towards the specific stimulus, in order to make an intention judgment: When an eye movement pattern exhibits following behavior toward a specific stimulus, the result of the intention judgment is determined as the first intention judgment result. The first intention judgment result indicates that the eye movement pattern has a tracking intention toward the specific stimulus. When the eye movement pattern does not follow a specific stimulus, the result of the intention judgment is determined as the second intention judgment result. The second intention judgment result indicates that the eye movement pattern does not have the intention to track the specific stimulus. When the corrected degree of autonomy does not meet the preset autonomy judgment conditions, the result of the intention judgment is determined as the second intention judgment result. The result of the first intention judgment or the result of the second intention judgment shall be taken as the result of the intention judgment.

[0015] Secondly, this application also discloses a digital processing-based intelligent window-breaking control system for passenger vehicles, comprising: The information acquisition module is used to acquire the internal and external operating and environmental information of the vehicle, and to preprocess the operating and environmental information to obtain the first operating and environmental information. The operating and environmental information includes multiple information elements, including external water level information, water pressure information, smoke information inside the vehicle compartment, carbon monoxide concentration information, temperature information, vehicle attitude information, and door and / or window status information. The preprocessing includes filtering and / or calibration. The emergency situation correlation analysis module is used to perform correlation analysis on multiple information elements based on the first operation and environmental information, and to determine whether the combination pattern of multiple information elements within the first preset time window conforms to the preset emergency situation characteristics. The emergency judgment module is used to identify the emergency type when the combination pattern of multiple information elements within a first preset time window meets the preset emergency characteristics. Emergency types include vehicle falling into water, spontaneous combustion and / or carbon monoxide leakage. The target window determination module is used to determine at least one window to be broken as the target window from a preset set of windows based on the type of emergency, the physical properties of each window, and the status information of each window. The physical properties include glass type, thickness, and / or whether it is covered with film. The window breaking scheme determination module is used to determine the window breaking scheme based on the type of emergency and the physical properties of the target vehicle window. The window breaking scheme includes impact energy parameters, number of impact points, distribution parameters and / or impact timing parameters. The operation instruction issuing module is used to issue operation instructions for the window breaking mechanism according to the corresponding window breaking scheme to form an escape route.

[0016] Beneficial Effects: The intelligent window-breaking control method for passenger vehicles based on digital processing disclosed in this application can intelligently determine whether the vehicle is in an emergency by acquiring the vehicle's internal and external operational and environmental information, preprocessing and correlating this information. When an emergency is detected, the system can intelligently determine the target window to break based on the type of emergency, the physical properties of the window, and its status information, and generate an optimized window-breaking plan. Finally, it issues an operation command to the window-breaking actuator to form an escape route. This method overcomes the absolute dependence of existing mechanical window-breaking schemes on manual operation and solves the problem that occupants may be unable to effectively break windows in emergency situations due to delayed reaction, disability, or improper operation. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating a digital processing-based intelligent window-breaking control method for passenger vehicles provided in this application.

[0018] Figure 2 A flowchart of a digital processing-based intelligent window-breaking control system for passenger vehicles provided in this application.

[0019] In the diagram: 1. Information acquisition module; 2. Emergency situation correlation analysis module; 3. Emergency situation judgment module; 4. Target window judgment module; 5. Window breaking scheme determination module; 6. Operation command issuance module. Detailed Implementation

[0020] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0021] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, the terms "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0022] Reference Figure 1 This application proposes a digital processing-based intelligent window breaking control method for passenger vehicles, including: S1000: Acquires internal and external operating and environmental information of the vehicle, and preprocesses the operating and environmental information to obtain the first operating and environmental information; Specifically, the operation and environmental information includes multiple information elements, such as external water level information, water pressure information, smoke information inside the carriage, carbon monoxide concentration information, temperature information, vehicle attitude information, and door and / or window status information. Preprocessing includes filtering and / or calibration.

[0023] S2000: Based on the first operating and environmental information, perform correlation analysis on multiple information elements to determine whether the combination pattern of multiple information elements within the first preset time window conforms to the preset emergency situation characteristics. S3000: When the combination pattern of multiple information elements within the first preset time window conforms to the preset emergency situation characteristics, the emergency situation type is identified based on the emergency situation characteristics; Specifically, emergency types include vehicles falling into water, spontaneous combustion, and / or carbon monoxide leaks.

[0024] S4000: Based on the type of emergency, the physical properties of each window, and the status information of each window, at least one window to be broken is determined as the target window from a preset set of windows. Specifically, physical properties include glass type, thickness, and / or whether a film is applied.

[0025] S5000: Determine the window breaking method based on the type of emergency and the physical properties of the target vehicle window; The window breaking scheme includes impact energy parameters, number of impact points, distribution parameters, and / or impact timing parameters.

[0026] S6000: Issues operation instructions for the window breaking mechanism corresponding to the window breaking scheme to create an escape route.

[0027] Specifically, the first step is to acquire operational and environmental information both inside and outside the vehicle as the basic input for determining whether the vehicle is in an emergency. This operational and environmental information includes multiple elements, such as external water level, water pressure, smoke inside the vehicle, carbon monoxide concentration, temperature, vehicle attitude, and door and / or window status. These information elements can be collected by various sensors integrated into the vehicle. For example, external water level and water pressure information can be obtained using external water level and pressure sensors; internal smoke and carbon monoxide concentration information can be obtained using internal smoke and carbon monoxide sensors; temperature information can be obtained using internal and external temperature sensors; vehicle attitude information can be obtained using an inertial measurement unit (IMU); and door and / or window status information can be obtained using door lock sensors and window position sensors. The acquired operational and environmental information needs to be preprocessed to obtain the initial operational and environmental information. Preprocessing includes filtering and / or calibration to suppress noise and system errors, improving the accuracy and reliability of the data. For example, low-pass filters can be used to filter analog signals acquired by sensors to remove high-frequency noise; for sensor data with system bias, calibration can be performed to make the output value closer to the true value.

[0028] After obtaining the initial operational and environmental information, it is necessary to perform correlation analysis on multiple information elements to determine whether the combination patterns of these information elements within a first preset time window conform to preset emergency situation characteristics. Correlation analysis can be implemented using various algorithms, such as rule-based expert systems, machine learning models (e.g., support vector machines, neural networks), or fuzzy logic systems. Preset emergency situation characteristics describe the combination patterns of information elements corresponding to different emergency situation types and can be defined based on historical data and expert experience. For example, emergency situation characteristics for a vehicle falling into water might include a continuously rising external water level, increased water pressure, and closed doors and windows; emergency situation characteristics for spontaneous combustion might include increased smoke concentration and a rapid rise in temperature inside the vehicle; emergency situation characteristics for carbon monoxide leakage might include a continuously rising carbon monoxide concentration and abnormal temperature in the passenger area. The first preset time window ensures the temporal continuity of the judgment and avoids misjudgments due to instantaneous data fluctuations.

[0029] When the combination pattern of multiple information elements within a first preset time window matches preset emergency characteristics, the system identifies the emergency type based on these characteristics. Emergency types include vehicle submersion in water, spontaneous combustion, and / or carbon monoxide leakage. The identification process can be implemented using a preset classifier or decision tree. For example, if abnormally high external water level and pressure are detected, it is determined that the vehicle has submerged in water; if abnormally high smoke and temperature are detected inside the vehicle, it is determined that the vehicle has spontaneously combusted; if abnormally high carbon monoxide concentration is detected, it is determined that the vehicle has leaked carbon monoxide.

[0030] After identifying the type of emergency, the system determines at least one target window from a pre-set set of windows based on the emergency type, the physical properties of each window, and the window's status information. The physical properties of the window include at least one or more of the following: glass type (e.g., tempered glass, laminated glass), thickness, and whether it is tinted. The window status information indicates whether the window is closed, partially open, or fully open. For example, in the event of a vehicle falling into water, a side window might be prioritized as the target window because side windows are generally easier to break than the windshield; in the event of spontaneous combustion, a window farther from the fire source and easier to create an escape route might be selected. The system comprehensively considers the emergency type, physical properties, and status information to select the most suitable target window for breaking, and can avoid damaged windows or windows obstructed by obstacles.

[0031] Next, based on the type of emergency and the physical properties of the target window, a window-breaking strategy is determined. This strategy includes impact energy parameters, the number of impact points, distribution parameters, and / or impact timing parameters to minimize the risk of secondary injury while ensuring efficient window breaking. For example, the required impact energy parameters will differ for windows of different thicknesses or glass types; for windows with tinted windows, it may be necessary to increase the number of impact points or adjust the distribution parameters to ensure the glass breaks completely and creates a sufficiently large escape route; impact timing parameters control the order and time interval between impact points. These parameters can be stored in a pre-set database and queried and matched based on the identified emergency type and the physical properties of the target window; in the event of a vehicle falling into water, the impact energy and distribution parameters can also be adjusted based on water pressure information to reduce the difficulty of breaking the window due to water pressure.

[0032] Finally, the system sends an operational command to the window-breaking actuator, specifying the appropriate window-breaking scheme, to create an escape route. The window-breaking actuator can be an electromagnetic, pneumatic, or mechanical window breaker. Upon receiving the operational command, it strikes the target vehicle window according to the impact energy parameters, the number of impact points, the distribution parameters, and / or the impact timing parameters, thereby quickly and effectively creating an escape route.

[0033] In another embodiment of this application, S2000 further includes: S2100: When the combination pattern of multiple information elements within the first preset time window does not meet the characteristics of an emergency, and at least two of the multiple information elements are in an abnormal state but have not reached the preset trigger threshold for identifying the type of emergency, the information elements in the abnormal state are identified as fuzzy abnormal signals. S2200: Risk assessment is performed based on the combination pattern and duration of fuzzy abnormal signals to obtain the risk level assessment result; S2300: When the risk level assessment result points to a medium risk level, a voice prompt is issued through the vehicle speaker, and at least one window in the window assembly is controlled to perform a lowering operation; S2300: Within the second preset time window after issuing a voice prompt and performing the descent operation, continuously monitor whether there are any signs of active operation by the occupant and obtain the result of the occupant's active operation judgment; S2400: After performing the descent operation, the operating and environmental information is reacquired and preprocessed to obtain the second operating and environmental information; S2500: Compare and evaluate changes in multiple information elements of the first and second operating and environmental information to obtain change assessment results; S2600: Adjust the risk level assessment results based on the change assessment results and the occupant's active operation judgment results: When the change assessment results indicate a significant deterioration in the danger and the occupant's active operation judgment results indicate that the occupant is unresponsive, the risk level indicated by the risk level assessment results is upgraded to a high-risk level, and the combination pattern of multiple information elements within the first preset time window is determined to meet the characteristics of an emergency.

[0034] Among them, fuzzy abnormal signals refer to multiple information elements in the vehicle's internal and external operating and environmental information that, although individually in an abnormal state, have not yet reached a preset threshold sufficient to directly trigger emergency identification. For example, external water level information may be slightly higher than normal but not reaching the trigger level for a water-falling emergency, or smoke information inside the vehicle may detect trace amounts of smoke but not reaching the smoke concentration threshold for a spontaneous combustion emergency. These signals, individually or in combination, fail to form clear emergency characteristics within the first preset time window, but their abnormal state indicates potential risk. Identifying these information elements in an abnormal state but not reaching the threshold as fuzzy abnormal signals is a key step in the system's capture of early, unclear signs of danger.

[0035] The purpose of risk assessment is to quantify the degree of danger of the current situation based on the combination pattern and duration of the fuzzy abnormal signals. For example, if external water level and water pressure information show slight anomalies simultaneously and persist for a period of time, the system may assess it as a medium-risk level. This assessment can be implemented using machine learning models, expert system rules, or fuzzy logic reasoning, and its output is a risk level assessment result, such as low risk, medium risk, or high risk.

[0036] When the risk level assessment results indicate a medium risk level, the system issues a voice prompt through the vehicle's loudspeaker, such as "An abnormal situation has been detected, please pay attention to safety," to alert occupants to potential dangers. At the same time, it controls at least one window in the window assembly to perform a lowering operation, such as lowering the driver's side window or the window closest to the occupant to a certain height, to provide occupants with initial ventilation or an escape opportunity, and to observe the occupants' reactions.

[0037] Within a second preset time window following the issuance of the voice prompt and the execution of the lowering operation, the system continuously monitors for signs of active operation by the occupant, and obtains a judgment result on the occupant's active operation. The signs of active operation may include the occupant's verbal response to the voice prompt, attempting to unlock the door or operate the door handle, actively raising or lowering the window, or unfastening the seat belt, etc. By monitoring these signs, it can be determined whether the occupant is awake, conscious, and capable of taking action independently.

[0038] After the window is lowered, the system re-acquires and preprocesses the operational and environmental information to obtain second operational and environmental information. The purpose of this second information is to acquire the latest environmental data after the initial intervention in order to assess whether the situation has improved or worsened. For example, after the window is lowered, information about smoke or carbon monoxide concentration inside the vehicle may change.

[0039] The system compares and evaluates changes in multiple information elements of both primary and secondary operational and environmental information to obtain change assessment results, which are used to quantify the evolution trend of hazardous situations. For example, if the external water level continues to rise rapidly after the windows are lowered, or if the smoke level inside the carriage increases instead of decreasing, the change assessment results will indicate a significant deterioration of the hazard.

[0040] The system adjusts the risk level assessment results based on the change assessment results and the occupant's active operation judgment results: when the change assessment results indicate a significant deterioration of the danger and the occupant's active operation judgment results indicate that the occupant is unresponsive, the risk level indicated by the risk level assessment results is upgraded to a high-risk level, and the combination pattern of multiple information elements within the first preset time window is determined to meet the preset emergency characteristics, thereby triggering the subsequent emergency type identification and window breaking scheme determination process.

[0041] In some preferred embodiments, the following specific example illustrates the situation: Suppose a passenger vehicle accidentally drives into a flooded area. Initially, external water level and pressure information may only show slight anomalies; for example, the water level sensor might detect that the water depth is slightly above the bottom of the wheels, but not yet reaching the preset trigger threshold for an emergency situation. Meanwhile, the temperature and carbon monoxide concentration inside the vehicle remain normal. In this scenario, if the system relies solely on a direct judgment based on emergency characteristics, it may not take immediate action because the combination of multiple information elements does not conform to the preset emergency characteristics.

[0042] However, according to this solution, the system will identify external water level and water pressure information as abnormal but not reaching the preset trigger threshold, and determine them as fuzzy abnormal signals. Subsequently, the system will conduct a risk assessment based on the combination pattern of these fuzzy abnormal signals (e.g., water level and water pressure are abnormal at the same time) and duration (e.g., a continuous rise for 10 seconds). If the risk level assessment result points to a medium risk level, the system will immediately issue a voice prompt through the vehicle's speaker, such as "The vehicle may be entering a flooded area, please pay attention to safety," and control the driver's side window to perform a lowering operation, such as lowering it by 10 centimeters.

[0043] Within a second preset time window (e.g., 30 seconds) after issuing the voice prompt and performing the lowering operation, the system continuously monitors whether the occupants show any signs of active operation: if the occupants immediately attempt to raise the window or unlock the door after hearing the voice prompt, it indicates that the occupants have responded, and the system can maintain a medium risk level and continue monitoring; if the occupants do not show any signs of active operation within 30 seconds, and the operating and environmental information (second operating and environmental information) re-acquired by the system during this period shows that the external water level and water pressure information continue to rise rapidly, and the change assessment result indicates that the danger has significantly worsened (e.g., the water level rise rate exceeds a preset threshold), and at the same time, the occupants' active operation judgment result indicates that the occupants have not responded, the system will raise the risk level assessment result to a high risk level, and determine the combination pattern of multiple information elements within the first preset time window as conforming to the preset emergency situation characteristics (i.e., the vehicle falls into the water), thereby triggering the subsequent emergency situation type identification, target window determination, and window breaking scheme determination process, and finally issuing an operation command to the window breaking execution mechanism to form an escape route.

[0044] In another embodiment of this application, it is further proposed to continuously monitor whether there are signs of active operation by the occupant and obtain a judgment result of active operation by the occupant, including: S2310: When at least one indication of active operation is detected, the result of the occupant active operation judgment is determined as an occupant response; S2320: When no signs of active operation are detected, acquire micro-motion signals of the occupant area and image information of the occupant's face, and use the micro-motion signals and image information as information elements in the operation and environmental information. The micro-motion signals are time-series signals that characterize chest and abdominal fluctuations or small movements of the body surface caused by the occupant's breathing and / or heartbeat, and the image information is a sequence of occupant facial image frames including the eye area. S2330: Extract periodic features from micromotion signals to determine whether the occupant has breathing and / or heartbeat activity, and obtain micromotion signal analysis results; S2340: Analyze the image information to obtain the eye movement trajectory and / or pupil changes and determine the occupant's state of consciousness, and obtain the image information analysis results; S2350: When the micro-motion signal analysis results indicate that the occupant has breathing and / or heartbeat activity, and the image information analysis results indicate that the occupant is conscious, the occupant is determined to be in a state of consciousness but no ability to act, the voice prompt is issued again, the second preset time window is extended, and after detecting signs of active operation, if no signs of active operation are detected, the occupant's active operation judgment result is determined to be that the occupant is unresponsive. S2360: When the micro-motion signal analysis results indicate that the occupant has no breathing and / or heartbeat activity, or the image information analysis results indicate that the occupant is unconscious, the occupant is determined to be in an unconscious or disabled state, and the occupant's active operation judgment result is determined to be occupant unresponsive.

[0045] Specifically, active operational indications can be understood as actions or behaviors that occupants can proactively take in an emergency and that can be recognized by the system. Their purpose is to indicate the occupant's state of consciousness and ability to act to the system. These active operational indications specifically include voice response indications, door unlocking or door handle operation indications, window raising / lowering operation indications, and / or seatbelt unfastening operation indications. Voice response indications refer to the occupant responding to prompts from the vehicle's speakers, such as saying "I'm okay" or "I need help." Door unlocking or door handle operation indications refer to the occupant attempting to unlock the door or operate the door handle to open the door. Window raising / lowering operation indications refer to the occupant attempting to raise or lower the window using the window control buttons. Seatbelt unfastening operation indications refer to the occupant unfastening their seatbelt. When the system detects at least one of these active operational indications, it can determine the occupant's active operation as a response.

[0046] When the system does not detect the aforementioned signs of active operation, in order to more accurately determine the occupant's true state, the system further acquires micro-motion signals of the occupant's area and image information of the occupant's face. The micro-motion signals are time-series signals characterizing chest and abdominal fluctuations or minute body surface movements caused by the occupant's breathing and / or heartbeat, with the aim of monitoring the occupant's vital signs in a non-contact manner; the image information is a sequence of occupant facial image frames including the eye area, with the aim of determining the occupant's state of consciousness through visual analysis.

[0047] After acquiring the micro-motion signal, the system extracts periodic features from the micro-motion signal, such as through Fourier transform or wavelet analysis, to determine whether the occupant has breathing and / or heartbeat activity, and obtains the micro-motion signal analysis results; at the same time, the system analyzes the image information to obtain the eye movement trajectory and / or pupil changes, and judges the occupant's state of consciousness based on the eye movement trajectory and / or pupil changes, and obtains the image information analysis results.

[0048] Furthermore, when the micro-motion signal analysis results indicate that the occupant has respiratory and / or heartbeat activity, and the image information analysis results indicate that the occupant is conscious, the system determines that the occupant is in a conscious but incapacitated state. In this conscious but incapacitated state, to give the occupant more opportunities to respond, the system issues another voice prompt and extends the second preset time window. Within the extended second preset time window, the system continuously detects whether the occupant shows any signs of active operation. If no signs of active operation are detected, the occupant's active operation judgment result is determined as unresponsive. Conversely, when the micro-motion signal analysis results indicate that the occupant has no respiratory and / or heartbeat activity, or the image information analysis results indicate that the occupant is unconscious, the system determines that the occupant is in an unconscious or incapacitated state and determines the occupant's active operation judgment result as unresponsive, in order to accelerate the triggering of subsequent handling procedures.

[0049] By introducing hierarchical analysis of micro-motion signals in the occupant area and image information of the occupant's face, the system can objectively confirm the occupant's status even when no signs of active operation are detected. This distinguishes between a conscious but incapacitated state and an unconscious or disabled state. For a conscious but incapacitated state, the system provides additional response opportunities by issuing a second pre-set voice prompt and extending the second preset time window. For an unconscious or disabled state, the system can quickly determine the occupant's active operation as unresponsive, providing more timely input for subsequent risk level assessment adjustments based on changes in the hazardous situation, and avoiding delays in critical decisions due to misjudgment of the occupant's status.

[0050] In another embodiment of this application, obtaining image information of the occupant's face is further proposed, including: S2321: Obtain the distance between the occupant and multiple image acquisition units with near-infrared supplementary lighting function preset in the vehicle, adjust the intensity of near-infrared supplementary lighting according to the distance, so that the image acquisition units can acquire images of the occupant's facial area from different angles to obtain multi-angle occupant facial images. S2322: Perform fusion processing on multi-angle occupant facial images to generate a synthetic image containing complete information about the occupant's face; S2323: Identify and extract the occupant's eye region image from the synthesized image, and enhance the occupant's eye region image to highlight the features of the eyeball and pupil; S2324: Generate a sequence of occupant facial image frames containing the eye region based on the enhanced eye region image, as image information.

[0051] Specifically, to overcome the impact of complex lighting conditions and changes in occupant posture on image acquisition within the vehicle, multiple image acquisition units can be pre-positioned at different locations inside the vehicle, such as above the dashboard, inside the A-pillar, or on the roof. All image acquisition units are equipped with near-infrared supplementary lighting. Near-infrared supplementary lighting ensures clear capture of occupant facial images even in low-light or no-light environments, without easily causing visual interference to the occupants. When acquiring images, the distance between the occupant and each image acquisition unit is first determined, and the intensity of the near-infrared supplementary lighting is dynamically adjusted based on this distance to ensure proper image exposure and avoid overexposure or underexposure. By having the image acquisition units capture images of the occupant's facial area from different angles, multi-angle images of the occupant's face can be obtained from multiple perspectives, effectively addressing situations where the occupant's head is turned or part of their face is obscured, such as when the occupant is turned to the side, looking down, or wearing a mask.

[0052] Furthermore, the acquired multi-angle occupant facial images are fed into an image processing unit for fusion processing. This fusion processing aims to integrate image information from different perspectives, eliminate redundancy, and fill in blind spots or occluded areas that may exist in a single perspective, ultimately generating a synthetic image containing complete information about the occupant's face. This synthetic image provides a comprehensive view of the occupant's face, laying the foundation for subsequent detailed analysis.

[0053] Building upon this, identifying and extracting the occupant's eye region image from the synthesized image is a crucial step. This occupant's eye region image is a core area for determining the occupant's state of consciousness, containing important physiological information such as eye movements and pupil changes. The extracted occupant's eye region image will undergo further enhancement processing, such as through contrast enhancement, sharpening, or noise reduction algorithms, to highlight the features of the eyeballs and pupils, making them more clearly discernible in the image and facilitating subsequent precise analysis.

[0054] Finally, based on the enhanced images of the occupant's eye region, a sequence of occupant facial image frames containing the eye region is generated. This sequence of occupant facial image frames is a continuous time-series image that can dynamically record changes in the occupant's eye region, such as minute eye movements, saccades, and pupil contraction and dilation, providing a stable data carrier for subsequent extraction of eye movement trajectories and / or pupil changes from the image information.

[0055] This application's solution effectively addresses the problems of poor image quality and incomplete information that traditional single-view or visible light acquisition methods may encounter in complex in-vehicle environments by employing a multi-angle, near-infrared illumination, and distance-adaptive image acquisition strategy. Multi-angle acquisition ensures comprehensive coverage of occupant facial information, acquiring sufficient effective data even if the occupant's posture changes or there is partial occlusion. Near-infrared illumination, combined with dynamic adjustment of near-infrared illumination intensity, ensures that appropriately exposed occupant facial images can be obtained even under low light, fluctuating strong light, or no light conditions. The fusion processing of multi-angle occupant facial images further enhances the integrity and consistency of facial information, providing reliable input for subsequent recognition, extraction, and enhancement processing of occupant eye area images.

[0056] In another embodiment of this application, the step of issuing a voice prompt again to extend the second preset time window includes: S2351: Assess the rate of evolution of a hazardous situation based on the results of a change assessment; S2352: Obtain the vehicle's remaining battery power information; S2353: Based on the evolution speed and remaining battery power information, determine the extension duration of the second preset time window and the frequency of voice prompts; S2354: Issue another voice prompt within the extended second preset time window at the frequency of the voice prompt, and continuously monitor whether the occupant has any signs of active operation.

[0057] Specifically, assessing the rate of evolution of a hazardous situation refers to quantifying the degree and speed of deterioration of the current hazardous situation by analyzing the changing trends and rates of multiple information elements in the first and second operational and environmental information. For example, the rate of increase in temperature, smoke, or water level information can be used to determine whether a hazardous situation is developing rapidly or slowly. Obtaining the vehicle's remaining battery power information means that the system monitors the current battery level in real time. This is crucial for subsequent decision-making because the operation of the window-breaking actuator, continuous monitoring, and voice prompts all consume electrical energy. In practical applications, determining the extension duration of the second preset time window and the frequency of voice prompts based on the rate of evolution and remaining battery power information means that the system will intelligently adjust according to the urgency of the danger and available resources. For example, when the hazardous situation evolves rapidly and the remaining battery power is sufficient, a shorter extension duration and a higher voice prompt frequency can be set to strive for a rapid response; while when the hazardous situation evolves slowly and the remaining battery power is low, a longer extension duration and a lower voice prompt frequency may be set to conserve power and give occupants more time. Therefore, the system will issue voice prompts again within the extended second preset time window according to the dynamically determined voice prompt frequency, and continuously monitor whether the occupant has any signs of active operation, thereby achieving more adaptive and efficient occupant status confirmation and intervention.

[0058] This application's solution, by introducing an assessment of the rate of evolution of the hazardous situation and the acquisition of vehicle remaining battery power information, makes the strategy of issuing voice prompts again and extending the second preset time window more intelligent and adaptive. In some preferred embodiments, assuming a vehicle falls into water, the system, through a comparison of first and second operating and environmental information, assesses that the rate of evolution of the hazardous situation is rapidly deteriorating, such as a rapid increase in external water level and water pressure. Simultaneously, the system acquires information indicating that the vehicle's remaining battery power is moderate. Based on this, the system intelligently determines the extension duration of the second preset time window to be a relatively short 30 seconds and sets the frequency of voice prompts to once every 5 seconds. During the next 30 seconds, the system will issue voice prompts through the vehicle's speakers every 5 seconds, requesting passengers to unlock the doors or operate the windows as soon as possible, and continuously monitor for signs of active operation by the occupants, such as unlocking doors, raising or lowering windows, or unfastening seat belts.

[0059] As another specific implementation, suppose a carbon monoxide leak occurs in the vehicle, but the leak rate is slow. The system assesses that the hazardous situation is deteriorating slowly, and the vehicle's remaining battery power is sufficient. In this case, the system might determine that the second preset time window is extended to a longer duration of 90 seconds, and set the voice prompt frequency to once every 15 seconds. In this way, the system provides occupants with more reaction time while avoiding the discomfort that may be caused by overly frequent prompts, effectively conserving battery power and reserving more energy for possible subsequent window-breaking operations.

[0060] In another embodiment of this application, step S2351 is further proposed to include: S2351-1: Based on the sampling stability and / or signal attenuation degree of at least one information element in the first and second operating and environmental information, determine the reliability parameter corresponding to at least one information element; S2351-2: When the reliability parameter is lower than the preset reliability threshold, the weight of the change rate of the corresponding information element shall be reduced and / or eliminated. S2351-3: Based on the information elements of the first and second operational and environmental information, calculate the corresponding rate of change, and perform weighted fusion on the corresponding rate of change based on the weights after weight reduction and / or elimination processing to obtain the change assessment result, and determine the evolution rate of the hazardous situation based on the change assessment result.

[0061] Specifically, sampling stability refers to the smoothness of numerical fluctuations of information elements over a period of time, which can be measured, for example, by calculating the variance or standard deviation of information elements within a specific time window. Signal attenuation refers to the degree to which the signal strength or clarity of information elements changes with time or environmental conditions, which can be evaluated, for example, by the signal-to-noise ratio or signal strength attenuation curve. Through the analysis of these indicators, a reliability parameter can be determined for each information element, quantifying the trustworthiness of the current data for that information element. A preset reliability threshold is an empirical value or a boundary set according to system design requirements, used to distinguish whether the information element data is sufficiently reliable. When the reliability parameter of an information element is lower than this threshold, it indicates that its data may have a large deviation or be unreliable. In this case, to avoid the negative impact of inaccurate data on the overall evaluation results, the weight of the information element's rate of change can be reduced, i.e., its influence in weighted fusion can be decreased; or it can be eliminated, i.e., the rate of change of that information element can be completely excluded from subsequent weighted fusion. In practical applications, the rate of change refers to the magnitude or trend of change in the values ​​of each information element between the first and second preset time windows. For example, the rate of change of temperature information can be expressed as the increase or decrease in temperature per unit time. By weighting and / or eliminating weights, a more accurate comprehensive change assessment result can be obtained. This change assessment result can more realistically reflect the evolution trend and speed of the hazardous situation; for example, a larger change assessment result usually indicates a faster deterioration of the hazardous situation.

[0062] The solution proposed in this application effectively addresses the problem of uncertainty or bias in information element data under complex or harsh environments by introducing a reliability assessment mechanism for information elements.

[0063] In another embodiment of this application, it is further proposed to analyze image information to obtain eye movement trajectories and / or pupil changes and determine the occupant's state of consciousness, including: S2341: Acquire vehicle vibration data and occupant head posture change information; S2342: Extract eye movement trajectory from image information, perform correlation analysis with vehicle vibration data and occupant head posture change information, identify and filter out eye displacement components caused by vehicle vibration and / or occupant head posture change, and obtain the filtered eye movement trajectory. S2343: Based on the filtered eye movement trajectory, determine whether there is autonomous eye rotation and / or saccadic behavior, and obtain the eye movement analysis results; S2344: Analyze the image information to obtain changes in pupil diameter, and determine whether the pupil contracts and / or expands in response to changes in near-infrared illumination intensity based on the changes in pupil diameter, thereby obtaining the pupil change analysis results; S2345: Combine the results of eye movement analysis and pupil change analysis to determine the occupant's state of consciousness.

[0064] Specifically, acquiring vehicle vibration data and occupant head posture change information means that: vehicle vibration data can be collected in real time by sensors such as accelerometers and gyroscopes integrated inside the vehicle; occupant head posture change information can be obtained by acquiring images of the occupant's head area through a pre-set image acquisition unit inside the vehicle, combined with image processing technology (e.g., deep learning-based head posture estimation algorithms), or by using a dedicated head posture sensor. The aforementioned vehicle vibration data and occupant head posture change information are used to provide a reference benchmark for external disturbances.

[0065] The process involves extracting eye movement trajectories from image information and performing correlation analysis with vehicle vibration data and occupant head posture changes to identify and filter out eye displacement components caused by vehicle vibration and / or occupant head posture changes, resulting in a filtered eye movement trajectory. The extraction of the eye movement trajectory can be achieved through image processing and feature point tracking techniques (e.g., iris center tracking, pupil edge tracking) on ​​image frame sequences of the occupant's eye region. Correlation analysis involves synchronizing the eye movement trajectory, vehicle vibration data, and occupant head posture changes in time, and performing pattern matching and correlation discrimination to identify non-voluntary eye displacement components, such as passive displacement components caused by vehicle bumps or inertial head swaying. Filtering out these passive displacement components from the original eye movement trajectory yields a filtered eye movement trajectory that better reflects the occupant's conscious activity.

[0066] Furthermore, based on the filtered eye movement trajectories, the presence of voluntary eye rotation and / or saccades is determined, yielding eye movement analysis results. Voluntary eye rotation and / or saccades refer to conscious eye movements by the occupant, such as tracking a specific target, reading, or observing the surrounding environment, which differs from unconscious drifting or fixed gaze. By analyzing the speed, acceleration, frequency of directional changes, and duration of the filtered eye movement trajectories, the presence of voluntary eye rotation and / or saccade patterns can be determined, thus obtaining eye movement analysis results.

[0067] Simultaneously, image information is analyzed to obtain pupil diameter changes, and based on these changes, it is determined whether the pupil responds to changes in near-infrared illumination intensity with contraction and / or dilation, thus obtaining pupil change analysis results. Pupil diameter changes can be measured through image segmentation and edge detection of the eye region; near-infrared illumination intensity changes can be actively controlled by the system, for example, by periodically fine-tuning the illumination intensity. The physiological reflex of the occupant's pupils to changes in light typically manifests as contraction when illumination intensity increases and dilation when illumination intensity decreases. By monitoring this contraction and / or dilation response, pupil change analysis results are obtained, which are used to assist in assessing the occupant's state of consciousness.

[0068] Finally, the occupant's state of consciousness is determined by combining the results of eye movement analysis and pupillary change analysis. This determination of consciousness is a comprehensive assessment process. For example, if eye movement analysis shows voluntary rotation and / or saccades, and pupillary change analysis shows that the pupils contract and / or dilate in response to changes in near-infrared illumination intensity, the occupant is considered conscious. If eye movement analysis does not show voluntary rotation and / or saccades, and pupillary change analysis does not show contraction and / or dilation, the occupant is considered unconscious or incapacitated.

[0069] In another embodiment of this application, S2343 specifically includes: S2343-1: Perform motion pattern recognition on the filtered eye movement trajectory, identify whether there are preset autonomous rotation patterns and / or saccade patterns in the filtered eye movement trajectory, and evaluate the degree of autonomy of the eye movement pattern based on the duration, repetition frequency and stability of the movement direction of the autonomous rotation patterns and / or saccade patterns. S2343-2: Determining the frequency and amplitude information of vehicle vibration based on vehicle vibration data; S2343-3: Correlation analysis is performed on the degree of autonomy, vehicle vibration data, and occupant head posture change information. Vibration correlation correction is applied to the degree of autonomy to reduce the influence of residual drift related to vehicle vibration in the filtered eye movement trajectory on the degree of autonomy, and the corrected degree of autonomy is obtained. S2343-4: Based on the corrected degree of autonomy, the intentional judgment of eye movement patterns is performed. The intentional judgment includes: judging whether the eye movement pattern shows tracking of a specific stimulus, regular scanning of the in-vehicle environment and / or directional response to voice prompts. The specific stimulus is a stimulus actively emitted by a pre-set stimulus device. S2343-5: Based on the results of the intention judgment, obtain the eye movement analysis results.

[0070] Specifically, motion pattern recognition refers to the analysis of filtered eye movement trajectories using pattern recognition algorithms, such as those based on machine learning or deep learning models, to identify specific patterns of eye movements that conform to conscious human control (such as smooth tracking and saccades). Predefined autonomous movement patterns can include the trajectory of the eye smoothly following a moving target, while saccades refer to the movement of the eye rapidly jumping from one fixation point to another. By analyzing the duration, repetition frequency, and stability of the movement direction of these patterns, the degree of autonomy of the eye movement patterns can be preliminarily assessed. For example, smooth tracking or saccades with long duration, high repetition frequency, and stable direction usually indicate a higher degree of autonomy.

[0071] Vehicle vibration data can be acquired by inertial measurement units such as accelerometers and gyroscopes inside the vehicle. This data characterizes the vehicle's vibration properties when it is moving or stationary, including its frequency and amplitude information. After determining the frequency and amplitude information of the vehicle vibration based on the data, it can be used as a correction reference to reduce the interference of vehicle vibration on the assessment of autonomy.

[0072] In practical applications, vibration correlation correction aims to further reduce the impact of residual drift related to vehicle vibration in the filtered eye movement trajectory on the level of autonomy. Specifically, it involves correlation analysis between the initially assessed level of autonomy and the frequency and amplitude information of vehicle vibration: when some features of the eye movement pattern show a high correlation with the frequency or amplitude of vehicle vibration, it is considered that these features may be induced by vehicle vibration rather than being entirely autonomous, thus reducing or adjusting the weight of autonomy to obtain the corrected level of autonomy. For example, a vibration-eye movement coupling model can be established, using real-time vehicle vibration data to dynamically adjust the assessment weights of autonomy, making the level of autonomy closer to the actual level of occupant's conscious activity.

[0073] Furthermore, intentionality assessment is a deeper analysis based on the corrected level of autonomy, aiming to determine whether eye movement patterns have a clear purpose or responsiveness. Intentionality assessment includes judging whether eye movement patterns show tracking of specific stimuli, regular scanning of the in-vehicle environment, and / or directional responses to voice prompts. Specific stimuli can be understood as signals with clear spatial or temporal characteristics actively emitted by the system to test occupant consciousness, such as light spots displayed on the in-vehicle screen, flashing LEDs, sound pulses of specific frequencies, directional cues from sound sources, etc. Through the above assessment, it is possible to more accurately distinguish between purposeful autonomous eye behavior and environmental disturbances or unconscious physiological responses, and thereby obtain eye movement analysis results, serving as one of the important bases for judging the occupant's state of consciousness.

[0074] The solution proposed in this application effectively solves the problem that relying solely on preliminary filtering may still leave residual vehicle vibration effects in dynamic vehicle environments, leading to biases in the autonomous judgment of eye movements. This is achieved by introducing motion pattern recognition, vibration correlation correction, and intent judgment.

[0075] In another embodiment of this application, a method for determining the intent of eye movement patterns based on the corrected degree of autonomy is further proposed, specifically including: S2343-41: Obtain the spatial location of a specific stimulus in the vehicle coordinate system and its corresponding time series; S2343-42: When the corrected level of autonomy meets the preset autonomy judgment conditions, analyze the correlation between the filtered eye movement trajectory used to characterize the eye movement pattern and the spatial location and time sequence of the specific stimulus source, and determine whether the eye movement pattern exhibits a continuous following behavior towards the specific stimulus source, in order to make an intention judgment: S2343-43: When an eye movement pattern exhibits following behavior with a specific stimulus, the result of the intention judgment is determined as the first intention judgment result, which indicates that the eye movement pattern has a tracking intention toward the specific stimulus. S2343-44: When the eye movement pattern does not follow a specific stimulus, the result of the intention judgment is determined as the second intention judgment result. The second intention judgment result indicates that the eye movement pattern does not have the intention to track the specific stimulus. S2343-45: When the corrected degree of autonomy does not meet the preset autonomy judgment conditions, the result of the intention judgment shall be determined as the second intention judgment result; S2343-46: The result of the first intention judgment or the result of the second intention judgment shall be taken as the result of the intention judgment.

[0076] Specifically, a specific stimulus source can be understood as a device pre-installed inside the vehicle that can actively emit visual, auditory, or tactile stimuli, such as a movable LED indicator, a speaker emitting a sound at a specific frequency, or a vibrating touchpad. Obtaining the spatial position of the specific stimulus source in the vehicle coordinate system and its corresponding time series refers to monitoring the positional changes of the stimulus source in real time through the vehicle's internal positioning system (such as ultrasonic sensors, visual sensors, or inertial measurement units) and recording the precise time point at which it emits its stimulus signal, thereby providing baseline data for subsequent correlation analysis.

[0077] The corrected level of autonomy refers to the quantitative index used to assess the autonomy of eye movement patterns after vehicle vibration correlation correction. The preset autonomy judgment condition can be a threshold. For example, when the corrected level of autonomy is higher than a certain preset value, the eye movement is considered to have sufficient autonomy, worthy of a deeper intention judgment; when the corrected level of autonomy does not meet the preset autonomy judgment condition, the eye movement autonomy is considered insufficient to support a reliable judgment.

[0078] When the corrected level of autonomy meets the preset autonomy judgment conditions, the correlation between the filtered eye movement trajectory used to characterize the eye movement pattern and the spatial location and time sequence of a specific stimulus source is analyzed. This aims to determine whether the eye movement exhibits synchronicity or predictability with the movement or appearance of the stimulus source. For example, parameters such as the distance error between the eye fixation point and the stimulus source location, the consistency of the movement direction, and the time delay can be calculated. Continuous following behavior towards a specific stimulus source can be understood as the eye stably tracking the movement of the stimulus source over a period of time, with its movement trajectory closely matching the trajectory of the stimulus source, or the eye rapidly and accurately turning towards the stimulus source when it appears.

[0079] When the following behavior is detected, the result of the intention determination is determined as the first intention determination result, which indicates that the eye movement pattern has a tracking intention towards a specific stimulus source. Conversely, when the following behavior is not detected, the result of the intention determination is determined as the second intention determination result, which indicates that the eye movement pattern does not have a tracking intention towards a specific stimulus source.

[0080] Furthermore, when the corrected level of autonomy does not meet the preset autonomy judgment conditions, the result of the intent judgment is directly determined as the second intent judgment result, thus avoiding misjudging unconscious eye drift or random movement as having a tracking intent in the case of insufficient autonomy or insufficient data quality.

[0081] In some preferred embodiments, it is assumed that a small laser pointer is installed inside the vehicle as a specific stimulus source. This pointer can project light spots at different locations inside the vehicle according to a preset pattern and record their precise positions and time sequences in the vehicle coordinate system. After the system acquires images of the occupant's eye area through the image acquisition unit and processes them to obtain a filtered eye movement trajectory, it first assesses the corrected degree of autonomy. If the degree of autonomy meets the preset autonomy judgment criteria (e.g., an autonomy score higher than 0.7), the system analyzes the correlation between the filtered eye movement trajectory and the laser point movement trajectory: for example, when the laser point moves from the left to the right, the occupant's eyes continuously track the point with a similar speed and direction, and the average distance error between the fixation point and the point position is less than a certain threshold, then the eye movement pattern is determined to exhibit continuous following behavior towards a specific stimulus source, and the result of the intent judgment is determined as the first intent judgment result; conversely, if the eye movement and the point movement are not significantly correlated, the result of the intent judgment is determined as the second intent judgment result. Furthermore, if the corrected degree of autonomy does not meet the preset autonomy judgment criteria (e.g., the autonomy score is below 0.7), then there is no need to proceed with the correlation analysis; the result of the intention judgment can be directly determined as the second intention judgment result.

[0082] Reference Figure 2 This application also discloses a digital processing-based intelligent window-breaking control system for passenger vehicles. This system aims to overcome the reliance on manual operation in traditional window-breaking methods, improve the efficiency and success rate of window breaking in emergencies, and thus provide occupants with a more timely and reliable escape route.

[0083] The system includes: Information acquisition module 1 is used to acquire the internal and external operating and environmental information of the vehicle, and to preprocess the operating and environmental information to obtain the first operating and environmental information. The operating and environmental information includes multiple information elements, including external water level information, water pressure information, smoke information inside the vehicle compartment, carbon monoxide concentration information, temperature information, vehicle attitude information, and door and / or window status information. The preprocessing includes filtering and / or calibration. Emergency situation correlation analysis module 2 is used to perform correlation analysis on multiple information elements based on the first operation and environment information, and to determine whether the combination pattern of multiple information elements within the first preset time window conforms to the preset emergency situation characteristics. Emergency situation judgment module 3 is used to identify the emergency type based on the emergency characteristics when the combination pattern of multiple information elements within a first preset time window meets the preset emergency characteristics. Emergency types include vehicle falling into water, spontaneous combustion and / or carbon monoxide leakage. The target window determination module 4 is used to determine at least one window to be broken as the target window from a preset set of windows based on the type of emergency, the physical properties of each window and the status information of each window. The physical properties include glass type, thickness and / or whether it is covered with film. The window breaking scheme determination module 5 is used to determine the window breaking scheme based on the type of emergency and the physical properties of the target window. The window breaking scheme includes impact energy parameters, number of impact points, distribution parameters and / or impact timing parameters. Operation command issuing module 6 is used to issue operation commands corresponding to the window breaking scheme to the window breaking execution mechanism in order to form an escape route.

[0084] This application provides a digital processing-based intelligent window-breaking control system for passenger vehicles. Through a modular design, this system achieves intelligent identification of vehicle emergencies and precise control of window-breaking operations.

[0085] The intelligent window-breaking control system for passenger vehicles based on digital processing described in this application significantly improves the passive safety performance of vehicles in emergency situations through its modular multi-source information perception, intelligent correlation analysis, emergency situation recognition, target window selection, customized window-breaking scheme generation, and automated execution chain. It provides occupants with more timely and reliable escape protection and represents a major breakthrough in existing vehicle passive safety technologies.

[0086] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for intelligent window breaking control of passenger vehicles based on digital processing, characterized in that, include: The system acquires internal and external operating and environmental information of the vehicle and preprocesses the operating and environmental information to obtain first operating and environmental information. The operating and environmental information includes multiple information elements, including external water level information, water pressure information, smoke information inside the vehicle compartment, carbon monoxide concentration information, temperature information, vehicle attitude information, and door and / or window status information. The preprocessing includes filtering and / or calibration. Based on the first operational and environmental information, a correlation analysis is performed on multiple information elements to determine whether the combination pattern of the multiple information elements within a first preset time window conforms to preset emergency situation characteristics. When the combination pattern of the multiple information elements within a first preset time window conforms to the preset emergency characteristics, the emergency type is identified according to the emergency characteristics. The emergency type includes vehicle falling into water, spontaneous combustion and / or carbon monoxide leakage. Based on the emergency type, the physical properties of each window, and the status information of each window, at least one window to be broken is determined from a preset set of windows as the target window. The physical properties include glass type, thickness, and / or whether it is tinted. Based on the type of emergency and the physical properties of the target vehicle window, a window breaking scheme is determined, which includes impact energy parameters, number of impact points, distribution parameters, and / or impact timing parameters. The window-breaking mechanism is given an operational command corresponding to the window-breaking scheme to create an escape route; Based on the first operational and environmental information, a correlation analysis is performed on multiple information elements to determine whether the combination pattern of the multiple information elements within a first preset time window conforms to preset emergency characteristics, further including: When the combination pattern of the multiple information elements within the first preset time window does not meet the emergency situation characteristics, and at least two of the multiple information elements are in an abnormal state but have not reached the preset trigger threshold for identifying the emergency situation type, the information elements in the abnormal state are determined as fuzzy abnormal signals. Risk assessment is performed based on the combination pattern and duration of the fuzzy abnormal signals to obtain the risk level assessment result. When the risk level assessment result points to a medium risk level, a voice prompt is issued through the vehicle speaker, and at least one window in the set of windows is controlled to perform a lowering operation; Within a second preset time window after the voice prompt is issued and the descent operation is performed, the system continuously monitors whether there are any signs of active operation by the occupant and obtains the result of the occupant's active operation judgment. After performing the descent operation, the operating and environmental information is reacquired and the preprocessing is performed to obtain the second operating and environmental information; The changes in the multiple information elements of the first operating and environmental information and the second operating and environmental information are compared and evaluated to obtain the change evaluation results; The risk level assessment result is adjusted based on the change assessment result and the occupant's active operation judgment result: when the change assessment result indicates a significant deterioration of the danger and the occupant's active operation judgment result indicates that the occupant is unresponsive, the risk level indicated by the risk level assessment result is raised to a high-risk level, and the combination pattern of the multiple information elements within the first preset time window is determined to conform to the emergency situation characteristics.

2. The intelligent window-breaking control method for passenger vehicles based on digital processing according to claim 1, characterized in that, The signs of active operation include voice response, door unlocking or door handle operation, window raising / lowering operation, and / or seatbelt unfastening operation. The continuous monitoring of whether occupants exhibit signs of active operation, and the resulting determination of occupant active operation, includes: When at least one of the aforementioned active operation indications is detected, the occupant active operation judgment result is determined to be that the occupant has responded; When no active operation signs are detected, micro-motion signals of the occupant area and image information of the occupant's face are acquired, and the micro-motion signals and image information are used as information elements in the operation and environmental information. The micro-motion signals are time-series signals that characterize chest and abdominal fluctuations or small movements of the body surface caused by the occupant's breathing and / or heartbeat, and the image information is a sequence of occupant facial image frames including the eye area. Periodic features are extracted from the micro-motion signals to determine whether the occupant has breathing and / or heartbeat activity, thus obtaining the micro-motion signal analysis results; The image information is analyzed to obtain the eye movement trajectory and / or pupil changes and to determine the occupant's state of consciousness, thus obtaining the image information analysis results; When the micro-motion signal analysis result indicates that the occupant has breathing and / or heartbeat activity, and the image information analysis result indicates that the occupant is conscious, it is determined that the occupant is in a state of consciousness but incapacity to act, and a voice prompt is issued again, the second preset time window is extended, and after detecting the signs of active operation, if the signs of active operation are still not detected, the occupant's active operation judgment result is determined as the occupant is unresponsive. When the micro-motion signal analysis results indicate that the occupant has no breathing and / or heartbeat activity, or when the image information analysis results indicate that the occupant is unconscious, the occupant is determined to be in an unconscious or disabled state, and the occupant's active operation judgment result is determined to be that the occupant is unresponsive.

3. The intelligent window-breaking control method for passenger vehicles based on digital processing according to claim 2, characterized in that, Acquire image information of the occupants' faces, including: The distance between the occupant and multiple image acquisition units with near-infrared supplementary lighting function are obtained, and the intensity of near-infrared supplementary lighting is adjusted according to the distance so that the image acquisition units can acquire images of the occupant's facial area from different angles to obtain multi-angle occupant facial images. The multi-angle occupant facial images are fused to generate a composite image containing complete facial information of the occupants; The occupant's eye region image is identified and extracted from the synthesized image, and the occupant's eye region image is enhanced to highlight the features of the eyeball and pupil; A sequence of occupant facial image frames containing the eye region is generated based on the enhanced eye region image, and is used as the image information.

4. The intelligent window-breaking control method for passenger vehicles based on digital processing according to claim 2, characterized in that, The voice prompt is issued again, extending the second preset time window, including: The rate of evolution of the hazardous situation is assessed based on the results of the change assessment; Obtain the vehicle's remaining battery power information; Based on the evolution speed and the remaining battery power information, determine the extension duration of the second preset time window and the frequency of the voice prompt; The system will issue another voice prompt within the extended second preset time window according to the frequency of the voice prompt, and continuously monitor whether the occupant shows any signs of active operation.

5. The intelligent window-breaking control method for passenger vehicles based on digital processing according to claim 4, characterized in that, Assess the rate of evolution of the hazardous situation based on the changes assessed, including: Based on the sampling stability and / or signal attenuation degree of at least one information element in the first operation and environment information and the second operation and environment information, the reliability parameter corresponding to the at least one information element is determined; When the reliability parameter is lower than a preset reliability threshold, the weight of the change rate of the corresponding information element is reduced and / or eliminated. Based on the information elements of the first operation and environment information and the information elements of the second operation and environment information, the corresponding rate of change is calculated, and the corresponding rate of change is weighted and fused based on the weights after weight reduction and / or elimination processing to obtain the change assessment result, and the evolution rate of the hazardous situation is determined according to the change assessment result.

6. The intelligent window-breaking control method for passenger vehicles based on digital processing according to claim 3, characterized in that, Analyzing the image information to obtain eye movement trajectories and / or pupil changes and determine the occupant's state of consciousness includes: Acquire vehicle vibration data and occupant head posture changes; The eye movement trajectory is extracted from the image information. The eye movement trajectory is correlated with the vehicle vibration data and the occupant head posture change information. The eye displacement component caused by vehicle vibration and / or occupant head posture change is identified and filtered out to obtain the filtered eye movement trajectory. Based on the filtered eye movement trajectory, it is determined whether there is voluntary eye rotation and / or saccadic behavior, and the eye movement analysis results are obtained. The image information is analyzed to obtain changes in pupil diameter, and based on these changes, it is determined whether the pupil contracts and / or expands in response to changes in near-infrared illumination intensity, thus obtaining pupil change analysis results. By combining the results of the eye movement analysis and the results of the pupil change analysis, the occupant's state of consciousness is determined.

7. The intelligent window-breaking control method for passenger vehicles based on digital processing according to claim 6, characterized in that, Based on the filtered eye movement trajectory, it is determined whether there is voluntary eye rotation and / or saccadic behavior, and the eye movement analysis results are obtained, including: The filtered eye movement trajectory is subjected to motion pattern recognition to identify whether there are preset autonomous rotation patterns and / or saccade patterns in the filtered eye movement trajectory. The degree of autonomy of the eye movement pattern is evaluated based on the duration, repetition frequency and stability of the movement direction of the autonomous rotation patterns and / or saccade patterns. The frequency and amplitude information of vehicle vibration are determined based on the vehicle vibration data; The degree of autonomy is correlated with the frequency and amplitude information of vehicle vibration, and vibration correlation correction is performed on the degree of autonomy to reduce the influence of residual drift related to vehicle vibration in the filtered eye movement trajectory on the degree of autonomy, so as to obtain the corrected degree of autonomy. The intentional determination of the eye movement pattern is performed based on the corrected degree of autonomy. The intentional determination includes: determining whether the eye movement pattern shows tracking of a specific stimulus source, regular scanning of the in-vehicle environment and / or directional response to voice prompts. The specific stimulus source is a stimulus source actively emitted by a pre-set stimulus source device. The eye movement analysis results are obtained based on the results of the intent judgment.

8. The intelligent window-breaking control method for passenger vehicles based on digital processing according to claim 7, characterized in that, Intentional determination of the eye movement pattern based on the corrected degree of autonomy includes: Obtain the spatial location of the specific stimulus source in the vehicle coordinate system and its corresponding time series; When the corrected level of autonomy meets the preset autonomy judgment conditions, the correlation between the filtered eye movement trajectory used to characterize the eye movement pattern and the spatial location and time sequence of the specific stimulus is analyzed to determine whether the eye movement pattern exhibits a continuous following behavior towards the specific stimulus, in order to perform the intention judgment: When the eye movement pattern exhibits the following behavior in relation to the specific stimulus source, the result of the intention determination is determined as the first intention determination result, which indicates that the eye movement pattern has a tracking intention to point to the specific stimulus source. When the eye movement pattern does not follow the specific stimulus source, the result of the intention determination is determined as the second intention determination result, which indicates that the eye movement pattern does not have the intention to track the specific stimulus source. When the corrected degree of autonomy does not meet the preset autonomy determination condition, the result of the intention determination is determined as the second intention determination result; The result of the first intent determination or the result of the second intent determination shall be used as the result of the intent determination.

9. A digital processing-based intelligent window-breaking control system for passenger vehicles, characterized in that: include: The information acquisition module is used to acquire the vehicle's internal and external operating and environmental information, and to preprocess the operating and environmental information to obtain the first operating and environmental information. The operating and environmental information includes multiple information elements, including external water level information, water pressure information, smoke information inside the vehicle compartment, carbon monoxide concentration information, temperature information, vehicle attitude information, and door and / or window status information. The preprocessing includes filtering and / or calibration. The emergency situation correlation analysis module is used to perform correlation analysis on multiple information elements based on the first operation and environment information, and to determine whether the combination pattern of the multiple information elements within a first preset time window conforms to the preset emergency situation characteristics. An emergency judgment module is used to identify the type of emergency when the combination pattern of the multiple information elements within a first preset time window matches the preset emergency characteristics. The emergency types include vehicle falling into water, spontaneous combustion, and / or carbon monoxide leakage. The target window determination module is used to determine at least one window to be broken as the target window from a preset set of windows based on the emergency situation type, the physical properties of each window and the status information of each window. The physical properties include glass type, thickness and / or whether it is covered with film. The window breaking scheme determination module is used to determine the window breaking scheme based on the emergency type and the physical properties of the target window. The window breaking scheme includes impact energy parameters, number of impact points, distribution parameters and / or impact timing parameters. The operation command issuing module is used to issue operation commands corresponding to the window breaking scheme to the window breaking execution mechanism in order to form an escape route; Based on the first operational and environmental information, a correlation analysis is performed on multiple information elements to determine whether the combination pattern of the multiple information elements within a first preset time window conforms to preset emergency characteristics, further including: When the combination pattern of the multiple information elements within the first preset time window does not meet the emergency situation characteristics, and at least two of the multiple information elements are in an abnormal state but have not reached the preset trigger threshold for identifying the emergency situation type, the information elements in the abnormal state are determined as fuzzy abnormal signals. Risk assessment is performed based on the combination pattern and duration of the fuzzy abnormal signals to obtain the risk level assessment result. When the risk level assessment result points to a medium risk level, a voice prompt is issued through the vehicle speaker, and at least one window in the set of windows is controlled to perform a lowering operation; Within a second preset time window after the voice prompt is issued and the descent operation is performed, the system continuously monitors whether there are any signs of active operation by the occupant and obtains the result of the occupant's active operation judgment. After performing the descent operation, the operating and environmental information is reacquired and the preprocessing is performed to obtain the second operating and environmental information; The changes in the multiple information elements of the first operating and environmental information and the second operating and environmental information are compared and evaluated to obtain the change evaluation results; The risk level assessment result is adjusted based on the change assessment result and the occupant's active operation judgment result: when the change assessment result indicates a significant deterioration of the danger and the occupant's active operation judgment result indicates that the occupant is unresponsive, the risk level indicated by the risk level assessment result is raised to a high-risk level, and the combination pattern of the multiple information elements within the first preset time window is determined to conform to the emergency situation characteristics.

Images