Vehicle early warning method and related equipment

Abstract

The invention discloses a vehicle early warning method and related equipment, and relates to the technical field of vehicle control, and the method comprises the steps: obtaining multi-dimensional perception data of a target vehicle in a current scene; determining an early warning threshold value of the target vehicle in the current scene based on the multi-dimensional perception data; determining a target early warning level of the target vehicle in the current scene based on the early warning threshold; and generating and executing a target early warning interaction strategy based on the target early warning level and a pre-constructed user driving behavior portrait. According to the method, more accurate and adaptive vehicle early warning conforming to driver characteristics is realized through mutual combination of multi-dimensional perception, dynamic threshold adjustment, graded early warning and personalized interaction, and the early warning accuracy, safety and user experience are improved.

Description

Technical Field

[0001] This application relates to the field of vehicle control technology, and more specifically, to a vehicle early warning method and related equipment. Background Technology

[0002] With the rapid development of intelligent connected vehicles and autonomous driving technologies, vehicle active safety warning systems are playing an increasingly important role in improving driving safety and reducing accident risks. These systems perceive and analyze vehicle operating status, surrounding traffic environment, and driver behavior to identify potential hazards in advance and issue warnings to the driver, thereby effectively assisting driving decisions and reducing traffic accidents. However, the complex and ever-changing driving scenarios, differences in driver states, and variations in vehicle dynamic characteristics make the precise setting of warning trigger conditions and alert strategies a challenging task.

[0003] In existing technologies, the decision-making logic of vehicle warning systems is mostly based on fixed thresholds or simple rule-based judgments. For example, the system typically triggers collision warnings by setting fixed distance or time thresholds, or lane departure warnings by the degree of vehicle deviation from the lane line. While this fixed parameterization approach is simple in structure and easy to implement, it fails to reflect the combined effects of factors such as changes in driver attention, variations in road environment complexity, and changes in vehicle dynamics, resulting in high false alarm and false negative rates in practical applications. In other words, related technologies suffer from the technical problem of vehicle warning systems failing to deliver effective risk information to drivers in a timely and accurate manner in complex and ever-changing driving scenarios. Summary of the Invention

[0004] The summary section of this application introduces a series of simplified concepts, which will be further explained in detail in the detailed description section. The summary section of this application is not intended to limit the key features and essential technical features of the claimed technical solution, nor is it intended to determine the scope of protection of the claimed technical solution.

[0005] The vehicle warning method and related equipment provided in this application can achieve more accurate, adaptive vehicle warnings that are consistent with driver characteristics by combining multi-dimensional perception, dynamic threshold adjustment, graded warning and personalized interaction, thereby improving the accuracy, safety and user experience of the warnings.

[0006] In a first aspect, this application provides a vehicle warning method, comprising: acquiring multi-dimensional perception data of a target vehicle in a current scenario, wherein the multi-dimensional perception data includes driver status data, vehicle operating condition data, and external environment perception data; determining a warning threshold for the target vehicle in the current scenario based on the multi-dimensional perception data; determining a target warning level for the target vehicle in the current scenario based on the warning threshold; and generating and executing a target warning interaction strategy based on the target warning level and a pre-constructed user driving behavior profile.

[0007] In some implementations, determining the warning threshold for the target vehicle in the current scenario based on the multidimensional perception data includes: performing feature quantification analysis on the multidimensional perception data to obtain multiple risk feature values, wherein the multiple risk feature values ​​include a first risk coefficient determined based on the driver state data, a second risk coefficient determined based on the external environment perception data, and a third risk coefficient determined based on the vehicle operating condition data; and determining the warning threshold based on the first risk coefficient, the second risk coefficient, and the third risk coefficient using a preset weighting strategy, wherein the warning threshold is negatively correlated with the first risk coefficient and the second risk coefficient, and positively correlated with the third risk coefficient.

[0008] In some implementations, determining the target warning level of the target vehicle in the current scenario based on the warning threshold includes: obtaining the real-time risk value of the target vehicle in the current scenario; and determining the target warning level from a preset warning level mapping table based on the difference between the real-time risk value and the warning threshold, wherein the preset warning level mapping table defines risk difference ranges corresponding to different warning levels.

[0009] In some implementations, obtaining the real-time risk value of the target vehicle in the current scenario includes: determining at least one traffic participant around the target vehicle based on the external environment perception data; predicting the potential movement trajectory of the at least one traffic participant within a preset time window based on the motion state information of the at least one traffic participant; determining the probability of interaction conflict between the target vehicle and the at least one traffic participant based on the real-time motion state of the target vehicle and the potential movement trajectory; and determining the real-time risk value based on the interaction conflict probability.

[0010] In some implementations, generating and executing a target warning interaction strategy based on the target warning level and a pre-constructed user driving behavior profile includes: matching a target interaction pattern based on the user driving behavior profile, wherein the user driving behavior profile is constructed based on the current driver's historical driving behavior data and is used to characterize the current driver's dimensional score on a preset dimension; generating the target warning interaction strategy based on the target warning level and the target interaction pattern, wherein the target warning interaction strategy includes a target interaction device, content complexity, and prompt intensity; and scheduling the target interaction device of the target vehicle to output warning information based on the current driver's real-time gaze focus data and the target warning interaction strategy to execute the target warning interaction strategy.

[0011] In some implementations, the preset dimensions include at least one of a risk propensity dimension, a skill level dimension, and a focus level dimension; the step of matching the target interaction mode based on the user's driving behavior profile includes: determining the target interaction mode as a first interaction mode when the score of the skill level dimension is less than a preset skill threshold, wherein the warning type corresponding to the first interaction mode includes at least one of an audio type and a visual type; determining the target interaction mode as a second interaction mode when the score of the skill level dimension is greater than or equal to the preset skill threshold and the score of the risk propensity dimension is less than a preset risk threshold, wherein the warning type corresponding to the second interaction mode includes at least one of the audio type, the visual type, and the tactile type; and determining the target interaction mode as a third interaction mode when the score of the risk propensity dimension is greater than or equal to the preset risk threshold and the score of the focus level dimension is less than a preset focus threshold, wherein the warning type corresponding to the third interaction mode includes at least one of the audio type, the visual type, the tactile type, and the control intervention type.

[0012] In some implementations, the step of scheduling the target interaction device of the target vehicle to output warning information based on the current driver's real-time gaze focus data and the target warning interaction strategy includes: determining the target interaction device from a plurality of candidate interaction devices based on the real-time gaze focus data and the target warning level, wherein the plurality of candidate interaction devices include a head-up display, an instrument panel, a central control screen, an audio speaker, a haptic actuator, and a vehicle control actuator; and controlling the target interaction device to output the warning information according to the prompt intensity and content complexity defined in the target warning interaction strategy.

[0013] Secondly, this application also provides a vehicle warning device, comprising: a data acquisition unit, configured to acquire multi-dimensional perception data of a target vehicle in the current scenario, wherein the multi-dimensional perception data includes driver status data, vehicle operating condition data, and external environment perception data; a threshold determination unit, configured to determine a warning threshold for the target vehicle in the current scenario based on the multi-dimensional perception data; a level determination unit, configured to determine a target warning level for the target vehicle in the current scenario based on the warning threshold; and a warning execution unit, configured to generate and execute a target warning interaction strategy based on the target warning level and a pre-built user driving behavior profile.

[0014] Thirdly, this application also provides an electronic device, including: a memory and a processor, wherein the processor is configured to execute a computer program stored in the memory to implement the steps of the vehicle warning method described in the first aspect.

[0015] Fourthly, this application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the vehicle warning method described in the first aspect.

[0016] Fifthly, this application also provides a computer program product, including a computer program or computer-executable instructions, which, when executed by a processor, implement the vehicle warning method provided in the embodiments of this application.

[0017] In summary, this application achieves comprehensive perception and analysis of the target vehicle's current driving scenario by acquiring multi-dimensional perception information, including driver status data, vehicle operating condition data, and external environment perception data. Compared to traditional early warning methods that rely on single parameters or fixed rules, it can more comprehensively and accurately reflect the real-time status of the vehicle and its surrounding environment, providing a more reliable data foundation for subsequent risk assessment and early warning decisions, thereby improving the accuracy and stability of early warnings. By dynamically determining the early warning threshold based on multi-dimensional perception data, the early warning triggering conditions can be adaptively adjusted according to changes in factors such as driver attention, vehicle operating status, and road environment, effectively avoiding false alarms that are prone to occur under fixed threshold methods. This approach addresses potential issues like missed warnings, ensuring that early warning assessments better reflect the actual risk level and achieving a balance between sensitivity and stability. By determining the target warning level based on warning thresholds, it can output corresponding warning signals at different risk levels, resulting in a more hierarchical and refined warning response and enhancing vehicle safety in changing scenarios. Furthermore, by combining user driving behavior profiles to generate and execute personalized warning interaction strategies, it can select appropriate prompting methods and intensities based on the driver's driving habits and behavioral characteristics. This not only reduces driver aversion or interference from frequent or inappropriate prompts but also improves the acceptability and response efficiency of warning information, thereby enhancing the driving experience while ensuring safety. In summary, the vehicle warning method provided in this application, through a combination of multi-dimensional perception, dynamic threshold adjustment, hierarchical warning, and personalized interaction, achieves more accurate, adaptive vehicle warnings that conform to driver characteristics, improving warning accuracy, safety, and user experience. Attached Figure Description

[0018] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit this specification. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 A schematic flowchart illustrating a vehicle warning method provided in an embodiment of this application; Figure 2 This is a schematic diagram of the composition structure of a vehicle warning device provided in an embodiment of this application; Figure 3 This is a schematic diagram of the composition structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0019] The terms used in the specification, claims, and drawings of this application, such as "first," "second," "third," "fourth," etc. (if any), are used to distinguish similar objects and not to describe a specific order or sequence. Therefore, it is to be understood that these terms can be used interchangeably where appropriate, allowing the described embodiments to be used in different orders, unless specifically required by the illustrations or description. Furthermore, the terms "is" and "has," and any variations thereof, are intended to cover, non-exclusively, all possible constituent elements. For example, a process, method, system, product, or apparatus comprising several steps or units is not necessarily limited to the steps or units explicitly listed, but may also include other steps or units not explicitly listed, or steps or units inherent to the process, method, product, or apparatus.

[0020] In this application, a "module" or "unit" refers to a computer program or part of a computer program that has a specific function and works in conjunction with other related parts to achieve a predetermined goal. These modules or units can be implemented by software, hardware (e.g., processing circuitry or memory), or a combination of both. One or more processors or memories can implement one or more modules or units. Furthermore, each module or unit can also be part of a larger module or unit.

[0021] The technical solutions of this application will be described in detail below with reference to the accompanying drawings of the embodiments. It should be noted that the described embodiments are only a part of this application, and not all embodiments. In the following description, the "some embodiments" mentioned are only a subset of all possible embodiments, which may be the same or different subsets, and different embodiments can be combined with each other without conflict.

[0022] Figure 1 This is a schematic flowchart illustrating a vehicle warning method provided in an embodiment of this application. For example, see [link to example]. Figure 1 The vehicle warning method provided in this application embodiment may include the following steps 101 to 104: Step 101: Obtain multi-dimensional perception data of the target vehicle in the current scenario. The aforementioned multi-dimensional perception data may include driver status data, vehicle operating condition data, and external environment perception data. In some examples, the target vehicle is a motor vehicle equipped with an intelligent driving warning system, as applied in this application. The target vehicle can achieve real-time monitoring and warning of driving risks to itself and its surroundings through the intelligent driving warning system, such as passenger cars, commercial buses, or freight vehicles. The current scenario is the comprehensive driving environment in which the target vehicle is located at a specific moment. The current scenario includes a combination of elements such as time dimension (e.g., daytime, nighttime, weekday morning rush hour), spatial dimension (e.g., urban main roads, highway ramps, rural intersections), and traffic participation dimension (e.g., traffic density, presence of pedestrians crossing, presence of construction areas). Multidimensional perception data is a collection of multi-source data covering the driver, the vehicle itself, and the external environment, collected to comprehensively depict the current driving scenario. It is collected collaboratively by multiple types of devices, providing comprehensive data support for subsequent warning decisions. Driver status data is a collection of information reflecting a driver's physiological state and driving concentration. Driver status data can be obtained through a driver monitoring system (DMS) installed in the cockpit of the target vehicle. The driver monitoring system may include components such as infrared cameras and facial recognition sensors. For example, it can obtain the driver's gaze direction (such as whether the gaze is focused on the road ahead) through eye-tracking technology, the blinking frequency and eye closure duration obtained through eyelid opening detection (used to determine the fatigue level), and the head turning angle collected by head posture sensors (used to determine whether the driver is distracted). Vehicle operating condition data is a set of parameters characterizing the real-time operating status of a target vehicle. Vehicle operating condition data can be obtained through the Controller Area Network (CAN) bus inside the target vehicle, which is the communication network between various electronic control units inside the vehicle. For example, the current driving speed (e.g., 60 km / h) transmitted by the vehicle speed sensor, the longitudinal and lateral acceleration (e.g., 0.8g during rapid acceleration) collected by the acceleration sensor, the yaw rate (e.g., 5 degrees / second during turning) fed back by the steering system, the turn signal switch status (e.g., left turn signal on) transmitted by the body control module, and the brake pedal travel and pressure (e.g., half-pressed brake state) fed back by the braking system, etc.External environment perception data is a collection of information describing the characteristics of the environment surrounding a target vehicle. This data can be collected through the fusion of multiple types of sensors installed around the target vehicle, including forward-facing cameras, millimeter-wave radar (MMWR), lidar (Light Detection and Ranging, LiDAR), and environmental sensors. Examples include weather conditions (such as light rain or heavy fog) and light intensity (such as 10,000 lux when driving against the light) detected by environmental sensors; lane line positions (such as the left lane line being 0.5 meters from the vehicle) and traffic participant information (such as a small passenger car traveling at 40 km / h 50 meters ahead and two pedestrians crossing the road on the right) identified by camera and radar fusion; and traffic flow density (such as 50 vehicles per kilometer of road) and scene complexity (such as a high complexity level at an intersection with traffic coming from multiple directions).

[0023] For example, when the target vehicle starts and activates the intelligent driving warning system, all data acquisition devices start working simultaneously. The driver monitoring system captures the driver's facial features and body movements in real time, generating driver status data; the controller local area network bus continuously receives operating parameters from various vehicle components and summarizes them into vehicle operating condition data; external sensors such as the forward-facing camera, millimeter-wave radar, and lidar collaboratively scan the vehicle's surrounding environment, and output external environment perception data after data fusion. The above multi-source data is summarized and integrated within the system to form complete multi-dimensional perception data, providing a foundation for subsequent warning threshold calculation and warning strategy generation.

[0024] By implementing step 101, multi-dimensional perception information, including driver status data, vehicle operating condition data, and external environment perception data, is obtained. This enables a comprehensive and integrated perception analysis of the driving scenario in which the vehicle is located, allowing for more accurate judgment of real-time risks. It provides sufficient and reliable data support for subsequent threshold calculation and early warning decisions, thereby improving the accuracy and stability of the overall early warning.

[0025] Step 102: Based on multi-dimensional perception data, determine the warning threshold for the target vehicle in the current scenario; In some examples, the warning threshold is a dynamically generated critical value based on the comprehensive risk characteristics of the current driving scenario. It is used to determine whether the real-time risk faced by the target vehicle reaches a level requiring a warning. Its value is adjusted in real-time according to changes in driver status, vehicle condition, and the external environment, rather than being a fixed value. The core function of this warning threshold is to serve as a benchmark for risk assessment. When the real-time risk value exceeds this threshold, a warning of the corresponding level is triggered; otherwise, it is not triggered, thus achieving accuracy and scenario adaptability in the warning. The warning threshold can be determined through quantitative analysis and weighted calculation of multi-dimensional perception data. For example, when the target vehicle is in heavy rain (external environmental perception data shows visibility of 50 meters and traffic flow...), the warning threshold can be adjusted accordingly. When the density is 80 vehicles per kilometer, and the driver monitoring system detects that the driver blinks 5 times per minute (judged as moderate fatigue, with a first risk coefficient of 0.7), and the vehicle's current speed is 60 km / h with slight slippage (a third risk coefficient of 0.3), the weighted calculation of the warning threshold may be 0.4 (value range 0-1); while when the target vehicle is traveling at a constant speed on a highway in clear weather (speed 100 km / h, acceleration fluctuation ±0.1g, a third risk coefficient of 0.8), the driver's gaze is steadily focused ahead (a first risk coefficient of 0.2), and the surrounding traffic density is 20 vehicles per kilometer (a second risk coefficient of 0.3), the warning threshold may be 0.8.

[0026] By implementing step 102, the warning threshold is dynamically determined based on multi-dimensional perception data. The warning triggering conditions can be adjusted in real time according to factors such as driver attention level, vehicle dynamic status and changes in the external environment. This effectively avoids the common false alarm and missed alarm problems in the fixed threshold mode, and enables the warning sensitivity to adapt to different risk states. This achieves personalized and flexible warning judgment and improves the adaptability to complex scenarios.

[0027] Step 103: Based on the warning threshold, determine the target warning level of the target vehicle in the current scenario; In some examples, the target warning level is a hierarchical identifier used to characterize the severity of the risk faced by the target vehicle, based on a comparison between the current warning threshold and the real-time risk value. The target warning level directly determines the intensity and form of subsequent warning interaction strategies. Its core function is to clarify the level of intensity at which risk information is conveyed, avoiding insufficient warnings (e.g., only a mild warning for high risk) or excessive warnings (e.g., strong intervention for low risk) due to ambiguous risk levels, ensuring that the driver can accurately perceive the urgency of the risk. First, the risk calculation module can be invoked to identify traffic participants around the target vehicle (e.g., vehicles ahead, pedestrians crossing, non-motorized vehicles to the side) based on external environmental perception data, and predict the time of each traffic participant within a preset time window (e.g., 3 seconds). The potential movement trajectory within the target vehicle is analyzed, and combined with the real-time movement status of the target vehicle itself, the probability of interaction and conflict between the target vehicle and each traffic participant is calculated. Finally, the conflict probability is quantified into a real-time risk value. The calculated real-time risk value is then compared with the warning threshold determined in step 102 to obtain the risk difference result. Finally, a preset warning level mapping table is called to match the corresponding level according to the risk difference result. For example, the preset mapping table can be defined as follows: when the risk difference is ≤0, there is no warning level (no need to trigger a warning); when 0 < risk difference is ≤0.3, it is "prompt level" (mild risk); when 0.3 < risk difference is ≤0.6, it is "alarm level" (moderate risk); when the risk difference is >0.6, it is "emergency level" (high risk).

[0028] By implementing step 103, the target warning level is determined based on the calculated warning threshold, and a matching warning response can be output under different risk levels, realizing graded and refined risk warnings. When the risk is low, a mild warning can be issued, while when the risk is high, a strong alarm or active intervention can be executed, thereby making the vehicle's response more reasonable and effective and improving the vehicle's safety protection performance in dynamic driving environments.

[0029] Step 104: Based on the target warning level and the pre-built user driving behavior profile, generate and execute the target warning interaction strategy; In some examples, the user driving behavior profile is a multi-dimensional quantitative model built based on the current driver's historical driving behavior data to characterize their driving features. The user driving behavior profile aims to capture the driver's personalized driving habits and risk response preferences, providing a basis for subsequent warning interactions. The user driving behavior profile can be quantified through scores on preset dimensions (such as risk propensity, skill level, and focus level), with each dimension ranging from 0 to 1 (0 indicating the weakest feature and 1 indicating the strongest). The target warning interaction strategy is a specific execution plan generated based on the target warning level and the user driving behavior profile, used to convey risk information to the driver. This includes the target interaction device (such as the hardware device for outputting the warning), content complexity (such as the level of detail in the warning information), and alert intensity (such as the intensity of the warning signal), aiming to ensure that the warning information is both effectively perceived by the driver and conforms to their acceptance habits. First, based on the dimensional scores of the user driving behavior profile, a target interaction mode can be matched from a preset interaction mode library. Then, combined with the target warning level (such as alert level, warning level, and emergency level), the device, content, and intensity in the interaction mode are refined to form the final target warning interaction strategy.

[0030] For example, firstly, the current driver's three-dimensional score (e.g., risk propensity 0.7, skill level 0.8, focus level 0.4) is retrieved from the user profile database. The interaction mode matching module determines the target interaction mode as the second interaction mode. Next, combining the target warning level (e.g., alarm level) output in step 103, the strategy generation module is invoked to determine the target interaction devices as a head-up display (HUD) and an audio speaker. The content complexity is set to "only key risk parameters (e.g., 'vehicle braking suddenly 30 meters ahead')," and the prompt intensity is set to "medium volume prompt tone + yellow highlighted icon." Subsequently, the driver's real-time gaze focus data (e.g., gaze focused on the road ahead) is obtained through the driver monitoring system. The display dispatcher then controls the HUD to output the highlighted icon according to the strategy, simultaneously triggering the audio speaker to play the prompt tone, thus completing the execution of the target warning interaction strategy. The entire process ensures both the adaptability of the warning information to the driver's characteristics and the effective transmission of information through device scheduling.

[0031] By implementing step 104, a personalized warning interaction strategy is generated and executed by combining the target warning level with the pre-built user driving behavior profile. This allows for the selection of appropriate prompting methods and intensities based on the driver's behavioral characteristics and preferences. For example, strong intervention prompts can be reduced for cautious drivers, while stronger reminders can be given in advance for aggressive drivers. This not only avoids unnecessary interference but also improves the acceptability of warning information and the driver's response efficiency, thereby optimizing the human-computer interaction experience while ensuring safety.

[0032] In summary, this application's embodiments achieve comprehensive perception and analysis of the target vehicle's current driving scenario by acquiring multi-dimensional perception information, including driver status data, vehicle operating condition data, and external environment perception data. Compared to traditional early warning methods that rely on single parameters or fixed rules, this approach can more comprehensively and accurately reflect the real-time status of the vehicle and its surrounding environment, providing a more reliable data foundation for subsequent risk assessment and early warning decisions, thereby improving the accuracy and stability of early warnings. By dynamically determining the early warning threshold based on multi-dimensional perception data, the early warning triggering conditions can be adaptively adjusted according to changes in factors such as driver attention, vehicle operating status, and road environment, effectively avoiding the errors that easily occur with fixed threshold methods. The method addresses the issue of missed or over-reported warnings, making early warning judgments more aligned with the actual risk level, thus achieving a balance between sensitivity and stability. By determining the target warning level based on warning thresholds, it can output corresponding warning signals under different risk levels, making the warning response more hierarchical and refined, and improving the vehicle's safety protection capabilities in changing scenarios. Furthermore, by combining user driving behavior profiles to generate and execute personalized warning interaction strategies, it can select appropriate prompting methods and intensities based on the driver's driving habits and behavioral characteristics. This not only reduces driver aversion or interference caused by frequent or inappropriate prompts but also improves the acceptability and response efficiency of warning information, thereby improving the driving experience while ensuring safety. In summary, the vehicle warning method provided in this application, through the combination of multi-dimensional perception, dynamic threshold adjustment, hierarchical warning, and personalized interaction, achieves more accurate, adaptive, and driver-characteristic-compliant vehicle warnings, improving warning accuracy, safety, and user experience.

[0033] In some embodiments, step 102 may include: performing feature quantification analysis on the multidimensional perception data to obtain multiple risk feature values, wherein the multiple risk feature values ​​include a first risk coefficient determined based on driver state data, a second risk coefficient determined based on external environment perception data, and a third risk coefficient determined based on vehicle operating condition data; and determining a warning threshold based on the first risk coefficient, the second risk coefficient, and the third risk coefficient using a preset weighting strategy, wherein the warning threshold is negatively correlated with the first risk coefficient and the second risk coefficient, and positively correlated with the third risk coefficient.

[0034] In some examples, the process of feature quantification analysis involves transforming unstructured or semi-structured information reflecting driving risks in multidimensional perception data into numerical features that can be directly used for calculation. The purpose of feature quantification analysis is to eliminate differences in data formats, enabling risk information of different dimensions to have a unified quantitative standard, thus providing a basis for subsequent risk coefficient calculation. Feature quantification analysis can include three steps: data cleaning (removing outliers, such as extreme data from instantaneous false alarms of sensors), feature extraction (extracting key risk indicators from raw data, such as extracting "percentage of time with eyes closed" from eyelid opening data), and normalization processing (mapping feature values ​​of different dimensions to the 0-1 range to ensure data comparability).

[0035] Multiple risk feature values ​​are the output of feature quantification analysis, representing a set of quantified risk values ​​for different dimensions of the target vehicle in the current scenario. These can include a first risk coefficient, a second risk coefficient, and a third risk coefficient, corresponding to the risk levels of the driver, the external environment, and the vehicle itself, respectively, collectively constituting the core indicators of risk assessment. The first risk coefficient is a numerical value representing the driver's own risk level, quantified based on driver state data. The higher the risk reflected in the driver state data (e.g., fatigue, distraction), the larger the first risk coefficient. The driver risk assessment module can be invoked to analyze eye-tracking data (e.g., the percentage of time the gaze deviates from the road ahead), eyelid opening data (e.g., the number of times the eyes close per minute), and head posture data (e.g., the frequency of head deflection exceeding 30 degrees) collected by the driver monitoring system, using a preset algorithm (e.g., a classification model based on support vector machines). The fatigue level (0-1, 1 being extreme fatigue) and distraction level (0-1, 1 being severe distraction) are calculated separately. Then, combined with the "percentage of high-risk operations" in the driver's historical driving behavior (such as the frequency of sudden braking and continuous lane changes), a weighted average is calculated to obtain the first risk coefficient. For example, if the driver's fatigue level is 0.6, the distraction level is 0.5, and the percentage of historical high-risk operations is 0.4, the first risk coefficient is calculated by weighting (with weights of 0.4, 0.4, and 0.2 respectively) as 0.6×0.4+0.5×0.4+0.4×0.2=0.52. The second risk coefficient is a numerical value representing the level of risk in the external environment, quantified based on external environmental perception data. The more complex and dangerous the external environment, the larger the second risk coefficient. The environmental risk assessment module can be called to analyze the environmental data after sensor fusion and extract indicators such as traffic density (e.g., the number of motor vehicles per kilometer of road, mapped to 0-1), weather visibility (e.g., 50 meters of visibility corresponds to 1, 500 meters corresponds to 0.2), and traffic scene complexity (e.g., a crossroads with vehicles coming from multiple directions corresponds to 0.9, and an open straight road corresponds to 0.1). The second risk coefficient is obtained by weighting the indicators with preset weights (e.g., traffic density 0.4, visibility 0.3, scene complexity 0.3). For example, if traffic density corresponds to 0.7, visibility corresponds to 0.6 (foggy weather), and scene complexity corresponds to 0.8 (construction area), then the second risk coefficient is 0.7×0.4+0.6×0.3+0.8×0.3=0.7.The third risk coefficient is a numerical value representing the stability of vehicle operation, obtained by quantifying vehicle operating data. The more stable and controllable the vehicle operation, the higher the third risk coefficient. The vehicle status assessment module can be called to analyze the vehicle speed data (such as the speed fluctuation amplitude, the smaller the fluctuation, the higher the value), acceleration data (such as the standard deviation of longitudinal / lateral acceleration, the smaller the standard deviation, the higher the value), and yaw rate (such as steering smoothness, the smaller the fluctuation, the higher the value) transmitted by the controller area network bus. After normalization of the indicators, the weighted average (such as speed stability 0.4, acceleration stability 0.3, steering stability 0.3) is used to obtain the third risk coefficient. For example, if the speed fluctuation corresponds to 0.8, the acceleration fluctuation corresponds to 0.7, and the steering fluctuation corresponds to 0.6, then the third risk coefficient is 0.8×0.4+0.7×0.3+0.6×0.3=0.73.

[0036] The preset weighted strategy is a pre-defined mathematical rule used to calculate the warning threshold by combining the first risk coefficient, the second risk coefficient, and the third risk coefficient. Its core is to reflect the degree of influence of different risk coefficients on the warning threshold through weight allocation. The weight parameters of the preset weighted strategy are obtained by training with a large amount of real vehicle test data (covering different scenarios and driver types) and stored in the vehicle's local algorithm library. In this embodiment, the mathematical expression of the preset weighted strategy is: Warning threshold = α × (1 - first risk coefficient) + β × (1 - second risk coefficient) + γ × third risk coefficient, where α, β, and γ are preset weights (α + β + γ = 1, such as α = 0.3, β = 0.4, γ = 0.3). The processing of "1 - first risk coefficient" and "1 - second risk coefficient" reflects the negative correlation between the warning threshold and the first two (the higher the coefficient, the lower the corresponding value, and the smaller the threshold). The third risk coefficient is directly introduced to reflect its positive correlation with the threshold (the higher the coefficient, the higher the corresponding value, and the larger the threshold). The warning threshold is a critical value calculated using a preset weighted strategy to determine whether a warning is triggered and the level of triggering. As mentioned earlier, it is negatively correlated with the first risk coefficient and the second risk coefficient (the higher the risk to the driver or the environment, the lower the threshold, and the easier it is for the system to trigger a warning), and positively correlated with the third risk coefficient (the more stable the vehicle, the higher the threshold, and the less likely the system is to trigger a warning to reduce interference).

[0037] Through the implementation of the above embodiments, feature quantification analysis of multidimensional perception data and the introduction of a risk coefficient weighting mechanism are performed, realizing the quantitative modeling of the relationship between driver state, environmental complexity and vehicle stability. This enables the warning threshold to change dynamically according to different risk sources. The threshold can be automatically lowered and early warnings issued when driver attention decreases or environmental risks increase, while the threshold can be raised and interference reduced when the vehicle is running smoothly or the environment is simple. This enhances the adaptability and rationality of the warning triggering mechanism and improves the robustness and effectiveness of the method in complex road environments.

[0038] In some embodiments, step 103 may include: obtaining the real-time risk value of the target vehicle in the current scenario; determining the target warning level from a preset warning level mapping table based on the difference between the real-time risk value and the warning threshold, wherein the preset warning level mapping table defines the risk difference range corresponding to different warning levels.

[0039] In some examples, the real-time risk value is a numerical value that is quantified based on the interaction status between the target vehicle and surrounding traffic participants in the current scenario. It represents the probability of the vehicle facing a collision or dangerous event. Its core function is to directly reflect the immediate risk level of the vehicle in the current scenario and provide a real-time risk benchmark for subsequent comparison with the warning threshold. The difference result is obtained by subtracting the real-time risk value of the target vehicle in the current scenario from the warning threshold determined in step 102. That is, difference result = real-time risk value - warning threshold. The purpose of the difference result is to quantify the degree to which the real-time risk exceeds (or falls below) the warning benchmark by its numerical value, providing a direct basis for matching the warning level. When the difference result is positive, it means that the real-time risk exceeds the warning threshold and the corresponding level of warning needs to be triggered. When the difference result is negative, it means that the real-time risk does not reach the warning threshold and no warning needs to be triggered. For example, if the real-time risk value is 0.66 and the warning threshold is 0.483, then the difference result = 0.66 - 0.483 = 0.177 (positive value, warning needs to be triggered); if the real-time risk value is 0.3 and the warning threshold is 0.483, then the difference result = 0.3 - 0.483 = -0.183 (negative value, warning does not need to be triggered). The preset warning level mapping table is a table that stores the relationship between risk difference ranges and warning levels in the vehicle's local database, based on extensive real-vehicle scenario testing, user behavior feedback, and safety standard calibration. This provides a clear basis for matching the difference results to the warning levels, avoiding subjectivity in warning level determination and ensuring consistency and rationality in level classification across different scenarios. For example, the specific correspondence of a preset warning level mapping table is as follows: a risk difference range of "difference result ≤ 0" corresponds to a warning level of "no warning"; a risk difference range of "0 < difference result ≤ 0.3" corresponds to a warning level of "notice level"; a risk difference range of "0.3 < difference result ≤ 0.6" corresponds to a warning level of "alarm level"; and a risk difference range of "difference result > 0.6" corresponds to a warning level of "emergency level".

[0040] Through the implementation of the above embodiments, the target warning level is determined based on the difference between the real-time risk value and the dynamic threshold, enabling the vehicle to output graded warning information according to the degree of risk. Mild prompts can be used in low-risk scenarios, while strong intervention measures can be used in high-risk scenarios, realizing hierarchical and refined warning response. This avoids driving interference caused by excessive prompts, while ensuring that effective alarms can be triggered in a timely manner when danger approaches, thereby improving the vehicle's safety protection capabilities in dynamic traffic scenarios.

[0041] In some embodiments, the aforementioned acquisition of the real-time risk value of the target vehicle in the current scenario may include: identifying at least one traffic participant around the target vehicle based on external environment perception data; predicting the potential movement trajectory of at least one traffic participant within a preset time window based on the motion state information of at least one traffic participant; determining the probability of interaction conflict between the target vehicle and at least one traffic participant based on the real-time motion state and potential movement trajectory of the target vehicle; and determining the real-time risk value based on the probability of interaction conflict.

[0042] In some examples, at least one traffic participant is a moving object or road user within the target vehicle's driving range that may affect its driving safety. Its core function is to identify the source of real-time risk, forming the basis for subsequent risk calculations. Traffic participants include, but are not limited to, other motor vehicles (such as passenger cars ahead, trucks to the side, and vehicles following behind), non-motorized vehicles (such as electric bicycles and bicycles), pedestrians (such as pedestrians crossing the road or walking on the roadside), and obstacles in special scenarios (such as construction equipment temporarily occupying the road). Motion state information is a set of parameters used to describe the real-time motion characteristics of traffic participants. Its core function is to provide a dynamic basis for predicting the potential trajectories of traffic participants, ensuring the accuracy of trajectory prediction. Motion state information can include the absolute speed of traffic participants (such as 5 km / h, 40 km / h), relative speed (the speed difference relative to the target vehicle, such as -10 km / h indicating that the traffic participant is 10 km / h slower than the target vehicle), and acceleration (such as 1.2 m / s² during rapid acceleration). 2 -0.8m / s during rapid deceleration 2), turning trends (such as vehicle turn signal status, pedestrian walking direction angle), and real-time location coordinates (three-dimensional coordinates calculated based on vehicle positioning system and sensor data), etc. A preset time window is a fixed time range set for predicting the potential movement trajectories of traffic participants. Its core function is to balance the foresight and accuracy of prediction. If the time is too short, it will be impossible to detect risks in advance (e.g., predicting a trajectory within 0.5 seconds makes it difficult to respond to sudden lane changes in time). If the time is too long, it will be prone to prediction deviations due to the uncertainty of traffic participant behavior (e.g., predicting a trajectory within 5 seconds may lead to pedestrians temporarily changing their walking direction). In practical applications, the current speed of the target vehicle can be obtained from the controller area network bus, and the window length can be dynamically adjusted according to preset rules. The higher the speed, the longer the window length (e.g., a 3-second window length at 100 km / h to ensure sufficient distance to deal with risks); the lower the speed, the shorter the window length (e.g., a 1-second window length at 10 km / h to reduce invalid predictions). For example, when the target vehicle is traveling at 110 km / h on a highway, the preset time window is adjusted to 3 seconds based on the speed data obtained from the CAN bus. When the target vehicle enters a residential area and the speed drops to 15 km / h, the preset time window is automatically shortened to 1.2 seconds. The potential motion trajectory is based on the current motion state information of traffic participants and predicts the path curve that they may travel within a preset time window. Its core function is to intuitively present the future position changes of traffic participants and provide spatial basis for judging whether a conflict with the target vehicle will occur. The potential motion trajectory is presented in the form of a time-position coordinate sequence (e.g., one position point is recorded every 0.1 seconds, and 30 position points correspond to a 3-second window), covering the displacement changes of traffic participants in the lateral (perpendicular to the direction of travel of the target vehicle) and longitudinal (parallel to the direction of travel of the target vehicle) directions. Interaction conflict probability is the likelihood that the real-time trajectory of a target vehicle will spatially overlap with the potential trajectory of a traffic participant within a preset time window. Its core function is to quantify the risk level of a single traffic participant to the target vehicle. The value ranges from 0 to 1, where 0 indicates no possibility of conflict and 1 indicates inevitable conflict. The system spatially compares the expected trajectory of the target vehicle with the potential trajectory of the traffic participant, calculating the overlapping area at each time point (e.g., the proportion of the overlapping area to the target vehicle's body area). Then, combining the time difference of the trajectory overlap (e.g., the interval between the earliest overlap time point and the current time; the shorter the interval, the higher the probability of conflict), the system outputs the interaction conflict probability through a preset probability calculation model (e.g., a risk assessment model based on Bayes' theorem). Finally, the interaction conflict probabilities of the target vehicle with all surrounding traffic participants are weighted, summarized, and normalized to obtain a single numerical value reflecting the overall risk level of the target vehicle (range 0 to 1, with a higher value indicating higher overall risk). Its core function is to transform multi-source risks (multiple traffic participants) into a unified risk benchmark, facilitating subsequent comparison with warning thresholds.

[0043] For example, the external environment perception module can be invoked first to fuse environmental data collected by millimeter-wave radar, lidar, and forward-facing camera, identifying the types of traffic participants around the target vehicle (e.g., passenger cars ahead, non-motorized vehicles on the side, pedestrians crossing), their positions (e.g., lateral and longitudinal distances relative to the target vehicle), and their real-time motion states (e.g., speed, acceleration, and steering trend). Then, based on the real-time motion states of the traffic participants, a trajectory prediction algorithm (e.g., a prediction model based on a long short-term memory network) is used to predict the potential trajectory of each traffic participant within a preset time window (e.g., 1-3 seconds, which can be dynamically adjusted according to vehicle speed). The system first considers whether the vehicle will slow down and whether pedestrians will continue to cross the road. Then, it combines the real-time motion status of the target vehicle (current speed, steering angle, and braking status obtained from the controller area network bus) with spatial trajectory overlap analysis and time arrival prediction to calculate the probability of interaction and conflict between the target vehicle and each traffic participant within a preset time window (e.g., the higher the trajectory overlap and the smaller the time difference, the higher the probability of conflict). Finally, it weights and summarizes the conflict probabilities of all traffic participants (e.g., assigning higher weight to the conflict probability of vehicles close to the front and lower weight to non-motorized vehicles far away), and normalizes the summary result to the 0-1 interval to obtain the final real-time risk value.

[0044] Through the implementation of the above embodiments, the movement trajectories of surrounding traffic participants are predicted based on external environment perception data, and the probability of interaction conflict is calculated in combination with the dynamic state of the target vehicle, thereby obtaining a real-time risk value that is more consistent with real traffic behavior. This enables the early identification of potential collision or mutual interference risks in vehicle control scenarios, and achieves a forward-looking judgment of dangerous situations in the near future. Compared with traditional algorithms that are based only on current distance or speed, it has stronger predictive power and scene adaptability, which helps vehicles achieve more timely and reliable early warning output in high-density traffic or complex road conditions.

[0045] In some embodiments, step 104 may include: matching a target interaction pattern based on a user driving behavior profile, wherein the user driving behavior profile is constructed based on the current driver's historical driving behavior data and is used to characterize the current driver's dimensional score on a preset dimension; generating a target warning interaction strategy based on the target warning level and the target interaction pattern, wherein the target warning interaction strategy may include the target interaction device, content complexity, and prompt intensity; and scheduling the target interaction device of the target vehicle to output warning information based on the current driver's real-time gaze focus data and the target warning interaction strategy, so as to execute the target warning interaction strategy.

[0046] In some examples, the target interaction pattern is a standardized interaction framework matched from a pre-defined interaction pattern library based on the dimensional scoring of the user's driving behavior profile. This framework is adapted to the driver's characteristics and standardizes the basic form of the warning interaction, ensuring that the interaction logic is consistent with the driver's acceptance habits. The pre-defined interaction pattern library contains multiple sets of patterns, each defining the adapted driver characteristics (dimensional scoring range) and basic interaction rules (such as preferred device types and information presentation styles). The user's driving behavior profile's dimensional scores can be compared with the feature threshold ranges of the patterns in the pre-defined interaction pattern library to match a uniquely suitable pattern. The process of generating a target warning interaction strategy based on the target warning level and target interaction pattern involves integrating the severity of the warning risk (level) with the driver-adapted interaction framework (pattern) to output a specific and executable warning solution. This ensures that the warning matches both the risk intensity (level determines the strength) and the driver's habits (pattern determines the form), avoiding excessive or insufficient information delivery. First, a target interaction mode can be used as the basic framework (such as multi-device collaboration in the first interaction mode). Second, the elements in the framework can be refined by combining the target warning level (such as prompt level and alarm level). The higher the level, the stronger the prompt intensity (such as upgrading the sound from "single prompt" to "continuous beeping"), the more concise the content complexity (such as simplifying from "with explanation" to "risk location only"), and the more the target interaction device tends to be a strong stimulus type (such as adding "steering wheel vibration" from "head-up display"). The target interaction device is the hardware device designated in the target warning interaction strategy to output warning information. Its core role is to act as a carrier of warning information, ensuring that the information can be effectively perceived by the driver. It can include visual devices (head-up display, instrument panel, central control screen), audio devices (audio speakers), tactile devices (steering wheel vibration module, seat vibration module, and other tactile actuators), and control intervention devices (seat belt pretensioners, brake assist actuators, and other vehicle control actuators). Content complexity refers to the amount and level of detail of the warning information. Its core function is to balance information completeness and cognitive load, ensuring that drivers can quickly understand the risks while avoiding distraction caused by excessive information. Content complexity can be divided into three levels: low (containing only core risk indicators, such as "risk ahead"), medium (containing risk type and key parameters, such as "vehicle braking suddenly 20 meters ahead"), and high (containing risk reasons and recommended actions, such as "vehicle braking suddenly 20 meters ahead, it is recommended to reduce speed to 30 km / h").The intensity of the warning information refers to the degree of stimulation the driver's senses receives. Its core function is to match the intensity of the stimulation to the risk level, ensuring that high-risk information quickly attracts attention while low-risk information does not cause interference. Specifically, this is manifested in the brightness / flicker frequency of visual devices (e.g., low intensity is a solid green light, high intensity is a high-frequency flashing red light), the volume / frequency of audio devices (e.g., low intensity is a single low-frequency warning, high intensity is a continuous high-frequency beep), and the vibration amplitude / frequency of tactile devices (e.g., low intensity is a slight vibration, high intensity is a strong, continuous vibration). The current driver is the person currently operating the target vehicle. Its core function is to determine the matching object for the user's driving behavior profile, ensuring that the historical data of the current driver is used. The current driver's identity can be confirmed through the facial recognition function of the driver monitoring system, vehicle account login information, or mobile phone Bluetooth pairing records. Real-time gaze focus data is information obtained through the driver monitoring system that reflects the current driver's gaze location. Its core function is to ensure that the warning information is output in an area directly visible to the driver, improving information transmission efficiency. Real-time gaze focus data can include the gaze focus area (e.g., the road ahead, the instrument panel, the central control screen, the left side of the vehicle) and the focus duration (e.g., the gaze has lingered on the central control screen for 3 seconds). Based on the driver's real-time gaze focus data and the target warning interaction strategy, the process of dispatching the target vehicle's target interaction device to output warning information involves controlling the designated device to output warning information according to the driver's current gaze position (ensuring information visibility) and the preset warning scheme (ensuring information compatibility), ultimately achieving precise warning execution. The display scheduler can be invoked to compare the real-time gaze focus data with the target interaction device list in the target interaction strategy, prioritizing the device within the driver's gaze focus area to output information. If the gaze is not in any target device area, the device most likely to attract attention is selected (e.g., when the gaze is distracted, the dashboard's strong light effect is activated first). For example, if the target interaction strategy specifies "head-up display + audio speaker," and the real-time gaze focus is "the road ahead" (where the head-up display is located), then the head-up display is dispatched to display an icon according to the strategy, and the audio speaker is simultaneously triggered to play a warning sound. If the real-time gaze focus is "the center console screen" (off-center), then while activating the head-up display, the dashboard flashes (a backup device in the target interaction device list) to ensure the driver notices the warning.

[0047] Through the implementation of the above embodiments, personalized interaction strategies are generated and executed based on the target warning level and driver behavior profile. This enables the vehicle to automatically select appropriate interaction modes, prompt intensity, and information presentation methods according to the driver's behavioral characteristics, operating habits, and real-time attention status. In the field of vehicle control, this effectively avoids the problem of one-size-fits-all prompts in traditional warning processes, improves the pertinence and acceptability of human-computer interaction, and enables drivers to identify and respond to warning information more efficiently, thereby optimizing the driving experience while ensuring safety.

[0048] In some embodiments, the aforementioned preset dimensions may include at least one of a risk propensity dimension, a skill level dimension, and a focus level dimension; the aforementioned matching of the target interaction mode based on the user's driving behavior profile may include: determining the target interaction mode as a first interaction mode when the score of the skill level dimension is less than a preset skill threshold, wherein the warning type corresponding to the first interaction mode may include at least one of an audio type and a visual type; determining the target interaction mode as a second interaction mode when the score of the skill level dimension is greater than or equal to the preset skill threshold and the score of the risk propensity dimension is less than a preset risk threshold, wherein the warning type corresponding to the second interaction mode may include at least one of an audio type, a visual type, and a tactile type; and determining the target interaction mode as a third interaction mode when the score of the risk propensity dimension is greater than or equal to the preset risk threshold and the score of the focus level dimension is less than a preset focus threshold, wherein the warning type corresponding to the third interaction mode may include at least one of an audio type, a visual type, a tactile type, and a control intervention type.

[0049] In some examples, the risk propensity dimension is a quantitative dimension in the user driving behavior profile used to characterize the aggressiveness of a driver's driving style. Its core function is to distinguish between a driver's tolerance for risk and their willingness to actively avoid it. The risk propensity dimension has a score range of 0-1, with a higher score indicating a more aggressive driving style (e.g., a higher proportion of high-risk operations such as frequent rapid acceleration, continuous lane changes, and following too closely), and a lower score indicating a more conservative driving style (e.g., a higher proportion of low-risk operations such as constant speed driving, maintaining a safe distance, and cautious lane changes). The skill level dimension is a quantitative dimension in the user driving behavior profile used to characterize a driver's driving proficiency and handling stability. Its core function is to distinguish between a driver's ability to control the vehicle and their emergency response capabilities. The skill level dimension has a score range of 0-1, with a higher score indicating a more proficient driver (e.g., smooth handling, rapid emergency response, and high accuracy in responding to warnings), and a lower score indicating a less skilled driver (e.g., large fluctuations in handling, delayed emergency response, and low accuracy in responding to warnings). The focus level dimension is a quantitative dimension in the user driving behavior profile used to characterize the driver's level of attention while driving. Its core function is to distinguish whether the driver is easily distracted (e.g., frequently deviating from the road ahead, operating a mobile phone, etc.). The focus level dimension score ranges from 0 to 1, with a higher score indicating a higher level of driver focus (e.g., long-term focus on the road ahead, low percentage of distraction time) and a lower score indicating a lower level of focus (e.g., frequent deviation of gaze, high percentage of distraction time). The preset skill threshold is a critical value used to classify the driver's skill level. Its core function is to serve as a benchmark for judging the skill level dimension score and to determine whether to match the interaction mode suitable for low-skilled drivers. The first interaction mode is an interaction mode designed specifically for low-skilled drivers whose skill level dimension score is below the preset skill threshold. Its core feature is to help drivers understand risks through multi-sensory basic prompts and avoid misjudging warning information due to insufficient skills. This mode emphasizes the comprehensibility of information, and the corresponding warning types are mainly based on basic sensory stimuli. Audio-based warnings convey information through sound signals. Their core function is to attract the driver's attention through auditory stimulation, making them particularly suitable for scenarios where visual focus is limited. These can include beeps (such as a single "beep"), voice announcements (such as "Vehicle ahead slows down"), and buzzers (such as continuous high-frequency beeps). The volume and frequency of the sound are dynamically adjusted according to the warning level. Visual warnings convey information through images or light effects. Their core function is to visually present the location and type of risk, making them suitable for conveying specific risk parameters. These can include icon displays (such as vehicle icons on a head-up display), text prompts (such as "50 meters ahead" on the dashboard), and flashing light effects (such as flashing yellow / red lights on the dashboard). The brightness and flashing frequency of the display are dynamically adjusted according to the warning level.The preset risk threshold is a critical value used to classify drivers' risk propensity levels. Its core function is to serve as a benchmark for risk propensity scoring, helping to differentiate between conservative and aggressive drivers and thus matching corresponding interaction modes. The second interaction mode is designed specifically for highly skilled and conservative drivers whose skill level score is greater than or equal to the preset skill threshold and whose risk propensity score is less than the preset risk threshold. Its core feature is the addition of tactile stimulation to basic audiovisual cues, conveying risk in a concise and efficient way while avoiding information redundancy. Tactile types are forms of warning that convey warning information through physical vibration or pressure stimulation. Their core function is to enhance warning perception through tactile senses, especially suitable for scenarios requiring rapid attention. These can include steering wheel vibration (via a built-in vibration module in the steering wheel), seat vibration (via a vibration actuator under the seat), and slight tightening of the seatbelt (via a low-intensity action of the seatbelt pretensioner), etc. The amplitude and frequency of the vibration are dynamically adjusted according to the warning level. The preset focus threshold is a critical value used to classify drivers' level of focus. Its core function is to serve as a benchmark for focus level scoring, helping to identify easily distracted drivers and thus matching stronger intervention interaction modes. The third interaction mode is designed specifically for high-risk, easily distracted drivers whose risk propensity score is greater than or equal to a preset risk threshold and whose focus score is less than a preset focus threshold. Its core feature is the addition of vehicle control intervention in addition to visual, auditory, and tactile stimuli. This stronger intervention ensures the driver perceives the risk and takes action. Control intervention types are forms of warnings that transmit warning information through the vehicle's active execution of control actions. Their core function is to force the driver to perceive the risk through physical intervention, and even assist in avoiding it. They are suitable for high-risk scenarios where the driver may ignore the warning. These interventions can include seatbelt pretensioning (tightening the seatbelt via a seatbelt pretensioner), slight vehicle braking (applying a small amount of braking force via a brake actuator to create a sense of deceleration), and slight steering wheel correction (fine-tuning the direction via a steering actuator to generate steering feedback). The intensity of the intervention is dynamically adjusted according to the warning level.

[0050] For example, the current driver's three-dimensional score is retrieved from the user driving behavior profile database, with a skill level score of 0.3 (< preset skill threshold of 0.4), a risk tendency score of 0.7, and a focus score of 0.4. First, based on a skill level score < 0.4, the system directly matches the target interaction mode as the first interaction mode, without needing to judge other dimensions. If another driver's three-dimensional score is skill level 0.6 (≥ 0.4), risk tendency 0.5 (< preset risk threshold 0.6), and focus level 0.7, the system determines that they are a highly skilled and stable driver, matching the second interaction mode. The warning types include audio (medium volume alert tone), visual (dashboard icon), and tactile (slight vibration of steering wheel). If a driver's three-dimensional score is risk tendency 0.8 (≥ 0.6), focus level 0.3 (< preset focus threshold 0.5), and skill level 0.6 (≥ 0.4), the system determines that they are a high-risk and easily distracted driver, matching the third interaction mode. The warning types include audio (high-frequency beep), visual (red flashing HUD), tactile (strong seat vibration), and control intervention (seatbelt pretensioning), ensuring that high-risk information can be effectively perceived.

[0051] By implementing the above embodiments, three dimensions—risk tendency, skill level, and concentration level—are introduced into the driver profile, enabling multi-dimensional quantitative modeling of driver characteristics. Based on the scores of different dimensions, corresponding interaction modes are matched, allowing for differentiated early warning strategies for different groups such as novice, cautious, aggressive, or distracted drivers. For example, visual and voice prompts are added for novice drivers, and tactile and intervention feedback is enhanced for aggressive drivers. This improves the personalization and effectiveness of early warning output, reduces information redundancy and cognitive burden, and further enhances overall safety assurance capabilities.

[0052] In some embodiments, the aforementioned method of scheduling the target interaction device of the target vehicle to output warning information based on the current driver's real-time gaze focus data and the target warning interaction strategy may include: determining the target interaction device from multiple candidate interaction devices based on the real-time gaze focus data and the target warning level, wherein the multiple candidate interaction devices may include a head-up display, instrument panel, central control screen, audio speaker, haptic actuator, and vehicle control actuator; and controlling the target interaction device to output warning information according to the prompt intensity and content complexity defined in the target warning interaction strategy.

[0053] In some examples, the most suitable device for outputting the warning information can be selected from a preset set of devices by combining the driver's current gaze direction (ensuring information visibility) and the severity of the warning risk (determining the intensity of device stimulation). First, based on real-time gaze focus data (e.g., the driver's gaze is focused on the road ahead, the dashboard, or the central control screen), candidate visible devices located within the gaze focus area can be selected from multiple candidate interactive devices (e.g., when the gaze is focused on the road ahead, the head-up display is a visible device). Second, based on the target warning level (e.g., alert level, warning level, emergency level), the candidate visible devices can be supplemented or adjusted. The higher the level, the more devices are supplemented (e.g., for the emergency level, strong stimulation devices are added on top of the visible devices) to enhance the warning effect. For example, if the target warning level is alert level and the real-time gaze focus is on the dashboard, then the target interactive device is determined to be "dashboard + audio speaker" from the candidate interactive devices; if the target warning level is emergency level and the real-time gaze focus is on the central control screen (off-center), then the target interactive device is determined to be "dashboard (strong stimulation) + tactile actuator + vehicle control actuator" to ensure that high-risk information is forcibly perceived.Multiple candidate interactive devices are a collection of all hardware devices pre-installed in the target vehicle that can output warning information. Their core function is to provide a basic range for selecting target interactive devices, ensuring the system can flexibly allocate devices according to the scenario. These multiple candidate interactive devices cover various types, including visual, audio, tactile, and control intervention devices, and can be combined according to warning needs. The instrument panel is a visual device that projects warning information (such as icons and text) onto the windshield or a dedicated display screen in front of the driver. Its core function is to allow the driver to see the information without looking down, reducing eye movement. The instrument panel is a vehicle status display panel located directly in front of the driver, also serving as a visual warning function. Its core function is to convey risk information through light effects, icons, or text, and it has high visibility because it is within the driver's normal line of sight. The central control screen is a touch display screen located on the vehicle's center console, which can serve as an auxiliary visual device to output warning information. Its core function is to provide supplementary prompts when the driver's gaze turns to the central control area. Audio... A loudspeaker is a built-in sound output device in a vehicle that transmits warning information through sound signals. Its core function is to attract attention through auditory stimulation, making it particularly suitable for visually distracted scenarios. A tactile actuator is a device that transmits warning information through physical vibration or pressure. Its core function is to enhance perception through tactile stimulation, making it suitable for scenarios that require quick attention. Examples include steering wheel vibration modules (installed inside the steering wheel to alert drivers of directional risks through vibration) and seat vibration modules (installed under the seat to alert drivers of directional risks through vibration on one or both sides). For example, vibration on the left side of the steering wheel alerts drivers of vehicles approaching from the left. A vehicle control actuator is a device that actively controls vehicle movements to transmit warning information. Its core function is to force the driver to perceive high risks through physical intervention, and even assist in avoiding risks. Examples include seat belt pretensioners (which generate a pulling force alert by tightening the seat belt) and brake assist actuators (which generate a deceleration alert by applying slight braking). For example, seat belts may be instantly pre-tensioned during an emergency warning. The process of controlling the target interactive device to output warning information according to the prompt intensity and content complexity defined in the target warning interaction strategy involves sending control commands to the target interactive device based on the preset prompt intensity (stimulus level) and content complexity (information detail) in the target warning interaction strategy, driving the device to output warning information in a coordinated manner, so as to ensure that the presentation form of the warning information is consistent with the strategy definition and achieve accurate information transmission.

[0054] Through the implementation of the above embodiments, the interactive device is dynamically scheduled based on the driver's real-time gaze focus and warning level. When the driver is focused on the road ahead, the warning can be displayed using low-interference methods such as HUD. When the driver is distracted or the gaze shifts, the device can switch to stronger visual, audio, or tactile prompts. This allows the warning information to be delivered to the driver at the most appropriate time and in the most suitable way, thereby shortening the driver's perception and reaction time and enhancing the response efficiency and safety performance of the vehicle warning method in dynamic driving scenarios.

[0055] Furthermore, as an implementation of the aforementioned method embodiments, this application also provides a vehicle warning device for implementing the aforementioned method embodiments. This device embodiment corresponds to the aforementioned method embodiments. For ease of reading, this vehicle warning device embodiment will not repeat the details of the aforementioned method embodiments one by one, but it should be understood that the device in this application embodiment can correspondingly implement all the contents of the aforementioned method embodiments. For example... Figure 2 As shown, the vehicle warning device 20 includes: a data acquisition unit 201, a threshold determination unit 202, a level determination unit 203, and a warning execution unit 204. The data acquisition unit 201 acquires multi-dimensional perception data of the target vehicle in the current scenario, including driver status data, vehicle operating condition data, and external environment perception data. The threshold determination unit 202 determines the warning threshold for the target vehicle in the current scenario based on the multi-dimensional perception data. The level determination unit 203 determines the target warning level for the target vehicle in the current scenario based on the warning threshold. The warning execution unit 204 generates and executes a target warning interaction strategy based on the target warning level and a pre-built user driving behavior profile.

[0056] In some embodiments, the threshold determination unit 202 is further configured to perform feature quantification analysis on the multidimensional perception data to obtain multiple risk feature values, wherein the multiple risk feature values ​​include a first risk coefficient determined based on driver state data, a second risk coefficient determined based on external environment perception data, and a third risk coefficient determined based on vehicle operating condition data; and based on the first risk coefficient, the second risk coefficient, and the third risk coefficient, a warning threshold is determined by a preset weighting strategy, wherein the warning threshold is negatively correlated with the first risk coefficient and the second risk coefficient, and positively correlated with the third risk coefficient.

[0057] In some embodiments, the level determination unit 203 is further configured to obtain the real-time risk value of the target vehicle in the current scenario; and determine the target warning level from a preset warning level mapping table based on the difference result of comparing the real-time risk value with the warning threshold, wherein the preset warning level mapping table defines the risk difference range corresponding to different warning levels.

[0058] In some embodiments, the rating determination unit 203 is further configured to: determine at least one traffic participant around the target vehicle based on external environment perception data; predict the potential movement trajectory of at least one traffic participant within a preset time window based on the motion state information of at least one traffic participant; determine the probability of interaction conflict between the target vehicle and at least one traffic participant based on the real-time motion state and potential movement trajectory of the target vehicle; and determine the real-time risk value based on the probability of interaction conflict.

[0059] In some embodiments, the warning execution unit 204 is further configured to match a target interaction pattern based on a user driving behavior profile, wherein the user driving behavior profile is constructed based on the current driver's historical driving behavior data and is used to characterize the current driver's dimensional score on a preset dimension; generate a target warning interaction strategy based on the target warning level and the target interaction pattern, wherein the target warning interaction strategy includes the target interaction device, content complexity, and prompt intensity; and schedule the target interaction device of the target vehicle to output warning information based on the current driver's real-time gaze focus data and the target warning interaction strategy, so as to execute the target warning interaction strategy.

[0060] In some embodiments, the preset dimensions include at least one of a risk propensity dimension, a skill level dimension, and a focus level dimension; the warning execution unit 204 is further configured to determine the target interaction mode as a first interaction mode when the score of the skill level dimension is less than a preset skill threshold, wherein the warning type corresponding to the first interaction mode includes at least one of an audio type and a visual type; when the score of the skill level dimension is greater than or equal to the preset skill threshold and the score of the risk propensity dimension is less than a preset risk threshold, the target interaction mode is determined as a second interaction mode when the score of the second interaction mode is greater than or equal to the preset risk threshold and the score of the focus level dimension is less than a preset focus threshold, wherein the warning type corresponding to the third interaction mode includes at least one of an audio type, a visual type, a tactile type, and a control intervention type.

[0061] In some embodiments, the warning execution unit 204 is further configured to determine a target interactive device from a plurality of candidate interactive devices based on real-time gaze focus data and target warning level, wherein the plurality of candidate interactive devices include a head-up display, an instrument panel, a central control screen, an audio speaker, a haptic actuator, and a vehicle control actuator; and control the target interactive device to output warning information according to the prompt intensity and content complexity defined in the target warning interaction strategy.

[0062] This application also provides a computer-readable storage medium storing computer-executable instructions or computer programs, which, when executed by a processor, will cause the processor to perform any step of the vehicle warning method provided in this application.

[0063] In some embodiments, the computer-readable storage medium may be a random access memory (RAM), a read-only memory (ROM), flash memory, a magnetic surface memory, an optical disc, or a compact disc read-only memory (CD-ROM); or it may be a variety of devices that include one or any combination of the above-mentioned memories.

[0064] In some embodiments, computer-executable instructions may take the form of programs, software, software modules, scripts, or code, written in any form of programming language (including compiled or interpreted languages, or declarative or procedural languages), and may be deployed in any form, including as stand-alone programs or as modules, components, subroutines, or other units suitable for use in a computing environment.

[0065] In some embodiments, computer-executable instructions may, but do not necessarily, correspond to files in a file system, and may be stored as part of a file that holds other programs or data, for example, in one or more scripts in a HyperText Markup Language (HTML) document, in a single file dedicated to the program in question, or in multiple co-located files (e.g., files that store one or more modules, subroutines, or code sections).

[0066] In some embodiments, computer-executable instructions may be deployed to execute on an electronic device, or on multiple electronic devices located at one location, or on multiple electronic devices distributed across multiple locations and interconnected via a communication network.

[0067] like Figure 3 As shown, this application also provides an electronic device 30, including a memory 310, a processor 320, and a computer program 311 stored in the memory 310 and executable on the processor. When the processor 320 executes the computer program 311, it implements any step of the above-described vehicle warning method.

[0068] This application also provides a computer program product comprising a computer program or computer-executable instructions stored in a computer-readable storage medium. A processor of an electronic device reads the computer program or computer-executable instructions from the computer-readable storage medium and executes the computer program or computer-executable instructions, causing the electronic device to perform any step of the vehicle warning method described above.

[0069] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A vehicle early warning method, characterized in that, include: Acquire multi-dimensional perception data of the target vehicle in the current scenario, wherein the multi-dimensional perception data includes driver status data, vehicle operating condition data, and external environment perception data; Based on the multi-dimensional perception data, the warning threshold for the target vehicle in the current scenario is determined; Based on the warning threshold, the target warning level of the target vehicle in the current scenario is determined; Based on the target warning level and the pre-built user driving behavior profile, a target warning interaction strategy is generated and executed.

2. The vehicle early warning method according to claim 1, characterized in that, Determining the warning threshold for the target vehicle in the current scenario based on the multi-dimensional perception data includes: The multidimensional perception data is subjected to feature quantification analysis to obtain multiple risk feature values, wherein the multiple risk feature values ​​include a first risk coefficient determined based on the driver state data, a second risk coefficient determined based on the external environment perception data, and a third risk coefficient determined based on the vehicle operating condition data. Based on the first risk coefficient, the second risk coefficient, and the third risk coefficient, the early warning threshold is determined by a preset weighting strategy, wherein the early warning threshold is negatively correlated with the first risk coefficient and the second risk coefficient, and positively correlated with the third risk coefficient.

3. The vehicle early warning method according to claim 1, characterized in that, Determining the target warning level of the target vehicle in the current scenario based on the warning threshold includes: Obtain the real-time risk value of the target vehicle in the current scenario; Based on the difference between the real-time risk value and the warning threshold, the target warning level is determined from a preset warning level mapping table, wherein the preset warning level mapping table defines the risk difference range corresponding to different warning levels.

4. The vehicle early warning method according to claim 3, characterized in that, The step of obtaining the real-time risk value of the target vehicle in the current scenario includes: Based on the external environment perception data, at least one traffic participant is identified around the target vehicle; Based on the motion state information of the at least one traffic participant, predict the potential motion trajectory of the at least one traffic participant within a preset time window; Based on the real-time motion state of the target vehicle and the potential motion trajectory, the probability of interaction conflict between the target vehicle and the at least one traffic participant is determined. The real-time risk value is determined based on the probability of the interaction conflict.

5. The vehicle early warning method according to claim 1, characterized in that, The step of generating and executing a target warning interaction strategy based on the target warning level and a pre-built user driving behavior profile includes: Based on the user driving behavior profile, a target interaction pattern is matched. The user driving behavior profile is constructed based on the current driver's historical driving behavior data and is used to characterize the current driver's dimensional score on a preset dimension. Based on the target warning level and the target interaction mode, a target warning interaction strategy is generated, wherein the target warning interaction strategy includes the target interaction device, content complexity, and prompt intensity; Based on the current driver's real-time gaze focus data and the target warning interaction strategy, the target vehicle's target interaction device is scheduled to output warning information in order to execute the target warning interaction strategy.

6. The vehicle early warning method according to claim 5, characterized in that, The preset dimensions include at least one of the following: risk propensity dimension, skill level dimension, and focus level dimension; The process of matching the target interaction pattern based on the user's driving behavior profile includes: If the score of the skill level dimension is less than a preset skill threshold, the target interaction mode is determined to be the first interaction mode, wherein the warning type corresponding to the first interaction mode includes at least one of audio type and visual type; If the score of the skill level dimension is greater than or equal to the preset skill threshold, and the score of the risk tendency dimension is less than the preset risk threshold, the target interaction mode is determined to be the second interaction mode. The warning type corresponding to the second interaction mode includes at least one of the audio type, the visual type, and the tactile type. If the score of the risk propensity dimension is greater than or equal to the preset risk threshold, and the score of the focus dimension is less than the preset focus threshold, the target interaction mode is determined to be the third interaction mode. The warning type corresponding to the third interaction mode includes at least one of the audio type, the visual type, the tactile type, and the control intervention type.

7. The vehicle warning method according to claim 5, characterized in that, The step of scheduling the target vehicle's target interaction device to output warning information based on the current driver's real-time gaze focus data and the target warning interaction strategy includes: Based on the real-time gaze focus data and the target warning level, the target interaction device is determined from multiple candidate interaction devices, wherein the multiple candidate interaction devices include a head-up display, an instrument panel, a central control screen, an audio speaker, a haptic actuator, and a vehicle control actuator; The target interactive device is controlled to output the warning information according to the prompt intensity and content complexity defined in the target warning interaction strategy.

8. A vehicle warning device, characterized in that, include: The data acquisition unit is used to acquire multi-dimensional perception data of the target vehicle in the current scenario, wherein the multi-dimensional perception data includes driver status data, vehicle operating condition data and external environment perception data. A threshold determination unit is used to determine the warning threshold of the target vehicle in the current scenario based on the multi-dimensional perception data. A level determination unit is used to determine the target warning level of the target vehicle in the current scenario based on the warning threshold. The warning execution unit is used to generate and execute the target warning interaction strategy based on the target warning level and the pre-built user driving behavior profile.

9. An electronic device, comprising: The memory and processor are characterized in that the processor is used to implement the steps of the vehicle warning method as described in any one of claims 1 to 7 when executing a computer program stored in the memory.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the vehicle warning method as described in any one of claims 1 to 7.

Images