A highway active interactive induction system and method based on driver state perception

Abstract

This invention discloses a highway active interactive guidance system and method based on driver state perception. It captures driver facial features, physiological signals, vehicle operation behavior, and real-time road environment data from multiple dimensions through infrared cameras, sensors, vehicle bus interfaces, and roadside interactive devices. This overcomes the limitations of single-perception and provides comprehensive, high-quality basic data support for subsequent analysis. Based on this, the invention effectively solves the problems of single perception dimensions and low accuracy in existing technologies by collecting and fusing multi-dimensional data on facial features, physiological signals, vehicle operation behavior, and road environment. It can comprehensively and accurately capture both explicit and implicit driving risks of drivers.

Description

Technical Field

[0001] This invention relates to the field of intelligent transportation and driver behavior monitoring technology, specifically to a proactive interactive guidance system and method for highways based on driver state perception. Background Technology

[0002] With the continuous improvement of the highway network and the surge in the number of motor vehicles, the driver's driving condition has become one of the core factors affecting road traffic safety. Fatigue driving, distracted driving, and abnormal emotional states are highly likely to cause traffic accidents. Currently, highway guidance systems mainly rely on traditional methods such as fixed traffic signs and passive voice reminders, lacking the ability to dynamically perceive and provide personalized guidance based on the individual driver's condition.

[0003] In existing technologies, although some vehicles are equipped with basic driver assistance functions, the following problems exist in actual operation:

[0004] 1. Driver's state perception has only one dimension and low accuracy.

[0005] Existing systems often rely solely on a single camera to capture facial features to determine fatigue levels, failing to integrate physiological signals and vehicle operation behavior for comprehensive analysis. For example, when a driver is mildly fatigued, their facial expression may not change significantly, but their heart rate may fluctuate abnormally; a single sensing method might miss this risk. Furthermore, if a driver experiences sudden illness causing limb stiffness, the facial recognition system may fail to capture this in time, potentially leading to an accident.

[0006] 2. The methods of induction are passive and homogenous, lacking specificity.

[0007] Current guidance methods mostly consist of standardized voice prompts or road warning lights, failing to adjust strategies based on driver condition and road conditions. For example, using the same "slow down reminder" to address the anxiety of novice drivers and the fatigue of experienced drivers may cause novice drivers to panic further due to excessive reminders, while experienced drivers may ignore repeated reminders. Summary of the Invention

[0008] To address the aforementioned technical problems of limited driver state perception dimensions and passive, homogenized guidance methods, this invention provides the following technical solution:

[0009] A proactive interactive guidance system for highways based on driver state perception, comprising:

[0010] The driver multi-dimensional state perception module is used to capture driver facial features, physiological signals, vehicle operation behavior and real-time road environment data in multiple dimensions through infrared cameras, sensors, vehicle bus interfaces and roadside interaction devices, breaking through the limitations of single perception and providing comprehensive and high-quality basic data support for subsequent analysis.

[0011] The state fusion analysis and risk assessment module is used to preprocess and remove abnormal information, and fuse multimodal data through deep learning algorithms to establish an assessment model to dynamically determine the risk level. It also combines individual driver characteristics to calibrate the warning threshold and accurately output the risk assessment results, providing a reliable basis for the decision-making module.

[0012] The personalized proactive interactive decision-making module is used to generate multi-dimensional guidance combination strategies based on risk assessment results, optimize and adapt to scenarios by combining road conditions and weather, adjust guidance methods by learning driver preferences, and clarify emergency priorities, thereby achieving an upgrade from generalized to personalized proactive decision-making.

[0013] The multi-dimensional interactive guidance execution module is used to transform decision-making strategies into visual, auditory, and tactile multi-modal guidance actions, and to link roadside equipment to achieve vehicle-road cooperative protection. Simultaneously, it collects driver execution feedback data and sends it back to the analysis module to complete the guidance implementation and closed-loop optimization, ensuring the efficient implementation of guidance effects.

[0014] As a preferred embodiment of the driver state perception-based active interactive guidance system for highways described in this invention, the driver multi-dimensional state perception module includes:

[0015] The facial feature acquisition unit is used to acquire the driver's facial feature parameters in real time using an infrared high-definition camera to identify the driver's status.

[0016] The physiological signal monitoring unit is used to collect the driver's physiological signals and capture latent states through flexible sensors integrated in the steering wheel grip and seat pressure sensors;

[0017] The vehicle operation behavior capture unit is used to interface with the vehicle's CAN bus, acquire operation data, and determine whether the driver's operation is standardized and whether there is a tendency to misoperate.

[0018] The environmental collaborative perception unit is used to collect environmental data through interaction with vehicle-mounted radar, high-definition road condition cameras and roadside equipment;

[0019] The driver identity and baseline adaptation unit is used to automatically match the driver's identity through in-vehicle facial recognition or fingerprint recognition, and retrieve the driver's historical physiological baseline data and driving preference profile; and automatically establishes an initial baseline when used for the first time, and then dynamically updates it through continuous learning.

[0020] As a preferred embodiment of the driver state perception-based proactive interactive guidance system for highways described in this invention, the state fusion analysis and risk assessment module includes:

[0021] The data preprocessing unit is used to perform noise reduction and normalization on multi-source data, remove abnormal data, and transform unstructured data into structured data to ensure data quality.

[0022] The multimodal data fusion unit is used to integrate facial features, physiological signals, operational behaviors, and environmental data using deep learning algorithms to establish a comprehensive driver status assessment model.

[0023] The risk level determination unit is used to preset four levels: safe, low risk, medium risk, and high risk, and dynamically determine the current driving risk in combination with other factors.

[0024] The risk trend prediction unit is used to predict future risk trends based on real-time driver status data and historical risk data, using the LSTM time series prediction algorithm, and outputs prediction results of risk increase / decrease / stable, while also labeling key influencing factors.

[0025] The dynamic warning threshold calibration unit is used to adjust the risk judgment threshold in a personalized manner based on the driver's age, driving experience, and historical status data.

[0026] As a preferred embodiment of the driver state perception-based proactive interactive guidance system for highways described in this invention, the personalized proactive interactive decision-making module includes:

[0027] The inducement strategy generation unit is used to generate multi-dimensional inducement combination strategies based on risk level and scenario type;

[0028] The scene adaptation and optimization unit is used to optimize the guidance strategy by combining environmental data;

[0029] The driver preference learning unit is used to learn the preferred guidance methods by analyzing the driver's execution and operational feedback of historical guidance strategies.

[0030] The emergency priority determination unit is used to prioritize triggering the highest level of induction strategy for sudden risks and to coordinate with emergency departments at higher levels to ensure the timeliness of emergency response.

[0031] The status recovery guidance decision unit is used to trigger alert-type strategies and generate personalized status recovery plans for mild to moderate risk statuses.

[0032] As a preferred embodiment of the driver state perception-based proactive interactive guidance system for highways described in this invention, the multi-dimensional interactive guidance execution module includes:

[0033] The visual guidance unit is used to control the in-vehicle central control screen, instrument panel indicator lights, and head-up display to display personalized reminder information, and to link with the roadside variable information board to display vehicle-specific guidance prompts simultaneously;

[0034] The auditory guidance unit is used to output personalized voice prompts through the car audio system, and adjust the volume and speech rate, and set different voice styles for different drivers;

[0035] The haptic feedback unit is used to control steering wheel vibration and seat zone vibration to trigger different vibration modes for different types of risk.

[0036] The emotional relief and guidance unit is used to alleviate the driver's adverse state through emotional means;

[0037] The vehicle-road cooperative guidance unit is used to connect with roadside equipment through vehicle-to-everything (V2X) technology to trigger roadside guidance actions and send early warning information to surrounding vehicles to form a cooperative protection system.

[0038] A proactive interactive guidance method for highways based on driver state perception includes the following specific steps:

[0039] S1 captures driver facial features, physiological signals, vehicle operation behavior, and real-time road environment data from multiple dimensions through infrared cameras, sensors, vehicle bus interfaces, and roadside interaction devices, breaking through the limitations of single perception and providing comprehensive and high-quality basic data support for subsequent analysis.

[0040] S2, after preprocessing to remove abnormal information, and through deep learning algorithms to fuse multimodal data, establish an assessment model to dynamically determine the risk level, and combine driver individual characteristics to calibrate the warning threshold, accurately output the risk assessment results, and provide a reliable basis for the decision-making module;

[0041] S3 generates multi-dimensional guidance combination strategies based on risk assessment results, optimizes and adapts to scenarios by combining road conditions and weather, adjusts guidance methods by learning driver preferences, and clarifies emergency priorities, thus achieving an upgrade from general to personalized proactive decision-making.

[0042] S4 transforms decision-making strategies into multimodal guidance actions involving vision, hearing, and touch, and links with roadside equipment to achieve vehicle-road cooperative protection. Simultaneously, it collects driver execution feedback data and sends it back to the analysis module to complete guidance implementation and closed-loop optimization, ensuring efficient implementation of guidance effects.

[0043] As a preferred embodiment of the proactive interactive guidance method for highways based on driver state perception described in this invention, the specific steps of S1 are as follows:

[0044] The S11 uses an infrared high-definition camera to collect driver facial feature parameters in real time and identify the driver's status.

[0045] S12 collects the driver's physiological signals and captures latent states through flexible sensors integrated in the steering wheel grip and seat pressure sensors.

[0046] S13 connects to the vehicle's CAN bus to acquire operational data and determine whether the driver's operation is standardized and whether there is a tendency to make mistakes.

[0047] S14 collects environmental data through interaction between vehicle-mounted radar, high-definition road condition cameras, and roadside equipment;

[0048] The S15 automatically matches the driver's identity through in-vehicle facial recognition or fingerprint recognition, and retrieves the driver's historical physiological baseline data and driving preference profile; it also automatically establishes an initial baseline upon first use, and then dynamically updates it through continuous learning.

[0049] As a preferred embodiment of the active interactive guidance method for highways based on driver state perception described in this invention, the specific steps of S2 are as follows:

[0050] S21 performs noise reduction and normalization on multi-source data, removes abnormal data, and transforms unstructured data into structured data to ensure data quality.

[0051] S22 uses deep learning algorithms to integrate facial features, physiological signals, operational behavior and environmental data to establish a comprehensive driver status assessment model.

[0052] S23 has four preset levels: safe, low risk, medium risk, and high risk, and dynamically determines the current driving risk based on factors.

[0053] S24, based on real-time driver status data and historical risk data, uses the LSTM time series prediction algorithm to predict future risk change trends and outputs prediction results of risk increase / decrease / stable, and marks key influencing factors;

[0054] S25 adjusts the risk assessment thresholds individually based on the driver's age, driving experience, and historical status data.

[0055] As a preferred embodiment of the active interactive guidance method for highways based on driver state perception described in this invention, the specific steps of S3 are as follows:

[0056] S31, Generate multi-dimensional inducement combination strategies based on risk level and scenario type;

[0057] S32, optimize the guidance strategy by combining environmental data;

[0058] S33, by analyzing drivers' execution of historical guidance strategies and operational feedback, learns their preferred guidance methods;

[0059] S34, in response to sudden risks, prioritize triggering the highest level of induction strategy and coordinate with emergency departments at higher levels to ensure the timeliness of emergency response;

[0060] S35, for mild to moderate risk states, not only triggers alert-type strategies, but also generates personalized state recovery plans.

[0061] As a preferred embodiment of the active interactive guidance method for highways based on driver state perception described in this invention, the specific steps of S4 are as follows:

[0062] S41 controls the vehicle's central control screen, instrument panel indicator lights, and head-up display, showing personalized reminder information, and links with the roadside variable message signs to simultaneously display vehicle-specific guidance prompts;

[0063] The S42 outputs personalized voice prompts through the car audio system, and adjusts the volume and speech rate, and sets different voice styles for different drivers;

[0064] S43 controls steering wheel vibration and seat zone vibration to trigger different vibration modes for different risk types;

[0065] S44, using emotional means to alleviate the driver's poor condition;

[0066] S45 connects to roadside equipment via vehicle-to-everything (V2X) technology to trigger roadside guidance actions and send warning information to surrounding vehicles, forming a collaborative protection system.

[0067] Compared with existing technologies:

[0068] This invention effectively solves the problems of single perception dimension and low accuracy in existing technologies by collecting and fusing multi-dimensional data on facial features, physiological signals, vehicle operation behavior, and road environment. It can comprehensively and accurately capture both explicit and implicit driving risks of drivers. By combining individual driver characteristics, driving experience, and real-time scenarios, it generates personalized guidance strategies, abandoning the passive and homogeneous drawbacks of traditional guidance methods. It adapts to the needs of different drivers and various complex road conditions, significantly improving the targeting of guidance instructions and driver acceptance. Through a complete interactive closed loop of "execution-feedback-calibration," it achieves real-time verification and dynamic optimization of guidance effects, avoiding the problems of guidance instructions failing and being unable to iterate. At the same time, it deeply links driver status with real-time road conditions, weather, and other environmental factors, dynamically adjusting the guidance method, intensity, and timing. This completely solves the defect of insufficient coordination and adaptation between road environment and driving status, ensuring the driving safety of individual vehicles and expanding the protection range through vehicle-road cooperation, further improving the overall safety and traffic efficiency of highways. Attached Figure Description

[0069] Figure 1 This is a schematic diagram of the overall framework of the present invention;

[0070] Figure 2 This is a schematic diagram of the driver multi-dimensional state perception module framework of the present invention;

[0071] Figure 3 This is a schematic diagram of the state fusion analysis and risk assessment module framework of the present invention;

[0072] Figure 4 This is a schematic diagram of the personalized proactive interactive decision-making module framework of the present invention;

[0073] Figure 5 This is a schematic diagram of the framework of the multi-dimensional interactive induced execution module of the present invention. Detailed Implementation

[0074] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be described in further detail below with reference to the accompanying drawings.

[0075] This invention provides a proactive interactive guidance system for highways based on driver state perception. Please refer to [link / reference]. Figure 1 ,include:

[0076] The driver multi-dimensional state perception module is used to capture driver facial features, physiological signals, vehicle operation behavior and real-time road environment data in multiple dimensions through infrared cameras, sensors, vehicle bus interfaces and roadside interaction devices, breaking through the limitations of single perception and providing comprehensive and high-quality basic data support for subsequent analysis.

[0077] The state fusion analysis and risk assessment module is used to preprocess and remove abnormal information, and fuse multimodal data through deep learning algorithms to establish an assessment model to dynamically determine the risk level. It also combines individual driver characteristics to calibrate the warning threshold and accurately output the risk assessment results, providing a reliable basis for the decision-making module.

[0078] The personalized proactive interactive decision-making module is used to generate multi-dimensional guidance combination strategies based on risk assessment results, optimize and adapt to scenarios by combining road conditions and weather, adjust guidance methods by learning driver preferences, and clarify emergency priorities, thereby achieving an upgrade from generalized to personalized proactive decision-making.

[0079] The multi-dimensional interactive guidance execution module is used to transform decision-making strategies into visual, auditory, and tactile multi-modal guidance actions, and to link roadside equipment to achieve vehicle-road cooperative protection. Simultaneously, it collects driver execution feedback data and sends it back to the analysis module to complete the guidance implementation and closed-loop optimization, ensuring the efficient implementation of guidance effects.

[0080] Please see Figure 2The driver multi-dimensional state perception module includes:

[0081] The facial feature acquisition unit uses an infrared high-definition camera to collect facial feature parameters such as the driver's eyelid closure, eye movement frequency, and facial muscle tension in real time, regardless of strong or dark light environments, and to identify states such as fatigue, distraction, and emotional fluctuations.

[0082] The physiological signal monitoring unit is used to collect physiological signals such as driver's heart rate, skin conductivity, and respiratory rate through flexible sensors integrated in the steering wheel grip and seat pressure sensors, and to capture latent states such as sudden physical discomfort, tension and anxiety.

[0083] The vehicle operation behavior capture unit is used to interface with the vehicle's CAN bus to acquire operation data such as steering wheel angle, accelerator / brake pedal travel, shift frequency, and vehicle speed stability, and to determine whether the driver's operation is standardized and whether there is a tendency to make mistakes.

[0084] The environmental collaborative perception unit is used to collect real-time environmental data such as road conditions (congestion, construction, curves), weather (rain, snow, fog), and traffic events through interaction with vehicle-mounted radar, high-definition road condition cameras, and roadside equipment, providing scenario support for guidance strategies.

[0085] The driver identity and baseline adaptation unit is used to automatically match the driver's identity through in-vehicle facial recognition or fingerprint recognition, and retrieve the driver's historical physiological baseline data (such as normal heart rate, standard operating habits, fatigue tolerance) and driving preference profile; and automatically establish an initial baseline upon first use, and then dynamically update it through continuous learning, which solves the shortcomings of the existing system that evaluates different driver states with a "general standard", making the perception data more individualized; for example, when a novice driver is identified, the fault tolerance baseline of the operation behavior is automatically lowered; when an elderly driver is identified, the warning benchmark value of physiological signals is calibrated in advance.

[0086] Please see Figure 3 The state fusion analysis and risk assessment module includes:

[0087] The data preprocessing unit is used to perform noise reduction and normalization on multi-source data, remove abnormal data (such as sensor instantaneous fault data), and transform unstructured data (images, voice) into structured data to ensure data quality.

[0088] The multimodal data fusion unit is used to integrate facial features, physiological signals, operational behaviors, and environmental data using deep learning algorithms to establish a comprehensive driver status assessment model, thus solving the problem of the one-sidedness of judging based on single data.

[0089] The risk level assessment unit is used to preset four levels: safe, mild risk, moderate risk, and high risk. It also dynamically assesses the current driving risk by taking into account factors such as highway speed limits and lane type. For example, if a driver on a highway has an abnormal heart rate and frequent sudden braking, the risk level is assessed as moderate.

[0090] The risk trend prediction unit is used to predict the risk change trend within the next 5-30 minutes based on the driver's real-time status data and historical risk data, using the LSTM time series prediction algorithm. It outputs the prediction results of risk increase / decrease / stable and marks key influencing factors. For example, if it predicts that the driver will enter a moderate fatigue state in 10 minutes, it can trigger a preventive guidance strategy in advance, rather than waiting for the risk to occur before responding.

[0091] The dynamic warning threshold calibration unit is used to adjust the risk judgment threshold in a personalized manner based on the driver's age, driving experience, and historical status data. For example, for elderly drivers, the warning threshold for fatigue status can be appropriately lowered.

[0092] Please see Figure 4 The personalized proactive interactive decision-making module includes:

[0093] The inducement strategy generation unit is used to generate multi-dimensional inducement combination strategies based on risk level and scenario type, such as "voice reminder + seat vibration" for mild fatigue and "forced speed reduction + roadside warning + emergency contact" for high risk.

[0094] The scene adaptation and optimization unit is used to optimize the guidance strategy by combining environmental data, such as reducing the visual guidance brightness on rainy days, increasing the frequency of auditory reminders on construction sections, and extending the guidance time in advance on curves.

[0095] The driver preference learning unit is used to learn the driver's preferred guidance methods by analyzing the driver's execution and operational feedback of historical guidance strategies. For example, some drivers prefer voice prompts to tactile feedback, and the system automatically adjusts the guidance priority.

[0096] The emergency priority determination unit is used to prioritize triggering the highest level of guidance strategy for sudden risks (such as sudden illness of the driver or signs of loss of vehicle control) and to coordinate with emergency departments at higher levels to ensure the timeliness of emergency response.

[0097] The status recovery guidance decision unit is used to trigger warning strategies and generate personalized status recovery plans for mild to moderate risk states. For example, for fatigue, it recommends the nearest service area and plans a temporary rest route; for tension, it adjusts the in-vehicle air conditioning temperature, plays soothing music, and turns on the seat ventilation; for distraction, it temporarily blocks unnecessary notifications from the in-vehicle entertainment system.

[0098] Please see Figure 5The multi-dimensional interactive induced execution module includes:

[0099] The visual guidance unit is used to control the in-vehicle central control screen, instrument panel indicator lights, and head-up display (HUD) to display personalized reminder information (such as "You are tired, it is recommended to rest at the service area"), and to link with the roadside variable information board to display vehicle-specific guidance prompts simultaneously;

[0100] The auditory guidance unit is used to output personalized voice reminders through the car audio system, and adjust the volume and speech rate. It can also set different voice styles (gentle and serious) for different drivers to avoid harsh reminders that may cause resentment.

[0101] The haptic feedback unit is used to control steering wheel vibration and seat zone vibration to trigger different vibration modes for different risk types, such as triggering slight steering wheel vibration when driving while distracted, and triggering strong seat vibration in an emergency.

[0102] The emotional soothing and induction unit is used to alleviate the driver's adverse state through emotional means; for example, it can relieve muscle tension through seat zone massage, release soothing fragrances through the in-vehicle fragrance system to calm anxiety, and provide gentle visual feedback through the dynamic breathing light on the instrument panel (blue-green indicates safety, orange-red indicates risk) to avoid secondary interference to the driver from strong light and rapid voice.

[0103] The vehicle-road cooperative guidance unit is used to connect with roadside equipment through vehicle-to-everything (V2X) technology to trigger roadside guidance actions such as adjusting speed limit signs, illuminating lane guidance lights, and reserving emergency lanes. At the same time, it sends early warning information to surrounding vehicles to form a collaborative protection system.

[0104] A proactive interactive guidance method for highways based on driver state perception includes the following specific steps:

[0105] S1 captures driver facial features, physiological signals, vehicle operation behavior, and real-time road environment data from multiple dimensions through infrared cameras, sensors, vehicle bus interfaces, and roadside interaction devices, breaking through the limitations of single perception and providing comprehensive and high-quality basic data support for subsequent analysis.

[0106] The specific steps of S1 are as follows:

[0107] The S11 uses an infrared high-definition camera, which is unaffected by strong light or low light environments. It collects facial feature parameters such as the driver's eyelid closure, eye movement frequency, and facial muscle tension in real time, and identifies states such as fatigue, distraction, and emotional fluctuations.

[0108] S12 collects physiological signals such as driver's heart rate, skin conductivity, and respiratory rate through flexible sensors integrated in the steering wheel grip and seat pressure sensors, and captures latent states such as sudden physical discomfort, tension and anxiety.

[0109] S13 connects to the vehicle's CAN bus to obtain operational data such as steering wheel angle, accelerator / brake pedal travel, shift frequency, and vehicle speed stability, and to determine whether the driver's operation is standardized and whether there is a tendency to make mistakes.

[0110] S14 uses vehicle-mounted radar, high-definition road condition cameras, and roadside equipment to collect real-time road conditions (congestion, construction, curves), weather (rain, snow, fog), traffic events, and other environmental data to provide scenario support for guidance strategies.

[0111] The S15 automatically matches the driver's identity through in-vehicle facial recognition or fingerprint recognition, and retrieves the driver's historical physiological baseline data (such as normal heart rate, standard operating habits, fatigue tolerance) and driving preference profile. It automatically establishes an initial baseline upon first use and continuously learns and updates it dynamically, overcoming the shortcomings of existing systems that use "general standards" to evaluate different driver states, making the perceived data more individualized. For example, when a novice driver is identified, the error tolerance baseline of the operating behavior is automatically lowered; when an elderly driver is identified, the warning benchmark value of physiological signals is calibrated in advance.

[0112] S2, after preprocessing to remove abnormal information, and through deep learning algorithms to fuse multimodal data, establish an assessment model to dynamically determine the risk level, and combine driver individual characteristics to calibrate the warning threshold, accurately output the risk assessment results, and provide a reliable basis for the decision-making module;

[0113] The specific steps of S2 are as follows:

[0114] S21 performs noise reduction and normalization on multi-source data, removes abnormal data (such as instantaneous sensor fault data), and transforms unstructured data (images, voice) into structured data to ensure data quality;

[0115] S22 employs deep learning algorithms to integrate facial features, physiological signals, operational behavior, and environmental data to establish a comprehensive driver status assessment model, addressing the issue of the one-sidedness of judging based on single data.

[0116] S23 has four preset levels: safe, low risk, medium risk, and high risk. It also dynamically determines the current driving risk by taking into account factors such as highway speed limits and lane type. For example, if a driver on a highway has an abnormal heart rate and frequent sudden braking, it is judged as a medium risk.

[0117] S24, based on real-time driver status data and historical risk data, uses the LSTM time series prediction algorithm to predict the risk change trend within the next 5-30 minutes and outputs the prediction results of risk increase / decrease / stable, and marks key influencing factors; for example, if it is predicted that the driver will enter a moderate fatigue state in 10 minutes, a preventive induction strategy will be triggered in advance, rather than waiting for the risk to occur before responding;

[0118] S25 adjusts risk assessment thresholds individually based on driver age, driving experience, and historical status data. For example, it appropriately lowers the warning threshold for fatigue for elderly drivers.

[0119] S3 generates multi-dimensional guidance combination strategies based on risk assessment results, optimizes and adapts to scenarios by combining road conditions and weather, adjusts guidance methods by learning driver preferences, and clarifies emergency priorities, thus achieving an upgrade from general to personalized proactive decision-making.

[0120] The specific steps of S3 are as follows:

[0121] S31 generates multi-dimensional guidance combination strategies based on risk level and scenario type, such as "voice reminder + seat vibration" for mild fatigue and "forced speed reduction + roadside warning + emergency contact" for high risk.

[0122] S32, combined with environmental data to optimize guidance strategies, such as reducing visual guidance brightness on rainy days, increasing the frequency of auditory reminders on construction sections, and extending guidance time in advance on curves;

[0123] S33 learns drivers' preferred guidance methods by analyzing their execution and operational feedback of historical guidance strategies. For example, if some drivers prefer voice prompts to tactile feedback, the system will automatically adjust the guidance priority.

[0124] S34, in response to sudden risks (such as sudden illness of the driver or signs of loss of vehicle control), prioritizes triggering the highest level of guidance strategy and coordinates with emergency departments at higher levels to ensure the timeliness of emergency response;

[0125] For mild to moderate risk conditions, S35 not only triggers alert-type strategies but also generates personalized recovery plans. For example, for fatigue, it recommends the nearest service area and plans a temporary rest route; for tension, it adjusts the in-car air conditioning temperature, plays soothing music, and turns on the seat ventilation; for distraction, it temporarily blocks unnecessary notifications from the in-car entertainment system.

[0126] S4 transforms decision-making strategies into multimodal guidance actions involving vision, hearing, and touch, and links with roadside equipment to achieve vehicle-road cooperative protection. Simultaneously, it collects driver execution feedback data and sends it back to the analysis module to complete guidance implementation and closed-loop optimization, ensuring efficient implementation of guidance effects.

[0127] The specific steps of S4 are as follows:

[0128] S41 controls the vehicle's central control screen, instrument panel indicator lights, and head-up display (HUD) to display personalized reminder information (such as "You are tired, it is recommended to rest at a service area"), and links with the roadside variable message signs to display vehicle-specific guidance prompts simultaneously;

[0129] The S42 outputs personalized voice reminders through the car audio system, and adjusts the volume and speech speed. It also sets different voice styles (gentle and serious) for different drivers to avoid harsh reminders that may cause resentment.

[0130] S43 controls steering wheel vibration and seat zone vibration to trigger different vibration modes for different risk types, such as triggering slight steering wheel vibration when distracted driving and triggering strong seat vibration in an emergency.

[0131] S44 alleviates driver distress through emotional means; for example, it relieves muscle tension through seat zone massage, releases soothing fragrances to calm anxiety through the in-car fragrance system, and provides gentle visual feedback through the instrument panel's dynamic breathing lights (blue-green for safety and orange-red for risk), avoiding secondary interference from bright lights and urgent voices to the driver.

[0132] S45 connects to roadside equipment through vehicle-to-everything (V2X) technology to trigger roadside guidance actions such as adjusting speed limit signs, illuminating lane guidance lights, and reserving emergency lanes. At the same time, it sends early warning information to surrounding vehicles to form a collaborative protection system.

[0133] Although the present invention has been described above with reference to embodiments, various modifications can be made and components can be replaced with equivalents without departing from the scope of the invention. In particular, as long as there is no structural conflict, the features in the disclosed embodiments can be combined with each other in any manner. The lack of an exhaustive description of these combinations in this specification is merely for the sake of brevity and resource conservation. Therefore, the present invention is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A proactive interactive guidance system for highways based on driver state perception, characterized in that, include: The driver multi-dimensional state perception module is used to capture driver facial features, physiological signals, vehicle operation behavior and real-time road environment data in multiple dimensions through infrared cameras, sensors, vehicle bus interfaces and roadside interaction devices. The State Fusion Analysis and Risk Assessment module is used to preprocess the data collected by the driver's multi-dimensional state perception module to remove abnormal information, and to fuse multi-modal data through deep learning algorithms to establish an assessment model to dynamically determine the risk level. It also combines the driver's individual characteristics to calibrate the warning threshold and accurately output the risk assessment results. The personalized proactive interactive decision-making module is used to generate multi-dimensional guidance combination strategies based on the risk assessment results output by the state fusion analysis and risk assessment module. It optimizes and adapts to the scenario by combining road conditions and weather, adjusts the guidance method by learning driver preferences, and clarifies the priority of emergency situations. The multi-dimensional interactive guidance execution module is used to transform the decision-making strategy generated by the personalized proactive interactive decision-making module into visual, auditory, and tactile multi-modal guidance actions, and link roadside equipment to achieve vehicle-road cooperative protection. The state fusion analysis and risk assessment module includes: The data preprocessing unit is used to perform noise reduction and normalization on multi-source data, remove outlier data, and transform unstructured data into structured data. The multimodal data fusion unit is used to integrate facial features, physiological signals, operational behaviors, and environmental data using deep learning algorithms to establish a comprehensive driver status assessment model. The risk level determination unit is used to preset four levels: safe, low risk, medium risk, and high risk, and dynamically determine the current driving risk in combination with other factors. The risk trend prediction unit is used to predict future risk trends based on real-time driver status data and historical risk data, using the LSTM time series prediction algorithm, and outputs prediction results of risk increase / decrease / stable, while also labeling key influencing factors. The dynamic warning threshold calibration unit is used to adjust the risk judgment threshold in a personalized manner based on the driver's age, driving experience, and historical status data.

2. The active interactive guidance system for highways based on driver state perception according to claim 1, characterized in that, The driver multi-dimensional state perception module includes: The facial feature acquisition unit is used to acquire the driver's facial feature parameters in real time using an infrared high-definition camera to identify the driver's status. The physiological signal monitoring unit is used to collect the driver's physiological signals and capture latent states through flexible sensors integrated in the steering wheel grip and seat pressure sensors; The vehicle operation behavior capture unit is used to interface with the vehicle's CAN bus, acquire operation data, and determine whether the driver's operation is standardized and whether there is a tendency to misoperate. The environmental collaborative perception unit is used to collect environmental data through interaction with vehicle-mounted radar, high-definition road condition cameras and roadside equipment; The driver identity and baseline adaptation unit is used to automatically match the driver's identity through in-vehicle facial recognition or fingerprint recognition, and retrieve the driver's historical physiological baseline data and driving preference profile; and automatically establishes an initial baseline when used for the first time, and then dynamically updates it through continuous learning.

3. The active interactive guidance system for highways based on driver state perception according to claim 1, characterized in that, The personalized proactive interactive decision-making module includes: The inducement strategy generation unit is used to generate multi-dimensional inducement combination strategies based on risk level and scenario type; The scene adaptation and optimization unit is used to optimize the guidance strategy by combining environmental data; The driver preference learning unit is used to learn the preferred guidance methods by analyzing the driver's execution and operational feedback of historical guidance strategies. The emergency priority determination unit is used to prioritize triggering the highest level of induction strategy for sudden risks and to coordinate with emergency departments at higher levels. The status recovery guidance decision unit is used to trigger alert-type strategies and generate personalized status recovery plans for mild to moderate risk statuses.

4. The active interactive guidance system for highways based on driver state perception according to claim 1, characterized in that, The multi-dimensional interactive induced execution module includes: The visual guidance unit is used to control the in-vehicle central control screen, instrument panel indicator lights, and head-up display to display personalized reminder information, and to link with the roadside variable information board to display vehicle-specific guidance prompts simultaneously; The auditory guidance unit is used to output personalized voice prompts through the car audio system, and adjust the volume and speech rate, and set different voice styles for different drivers; The haptic feedback unit is used to control steering wheel vibration and seat zone vibration to trigger different vibration modes for different types of risk. The emotional relief and guidance unit is used to alleviate the driver's adverse state through emotional means; The vehicle-road cooperative guidance unit is used to connect with roadside equipment through vehicle-to-everything (V2X) technology to trigger roadside guidance actions and send warning information to surrounding vehicles.

5. A proactive interactive guidance method for highways based on driver state perception, characterized in that, The specific steps are as follows: S1 captures driver facial features, physiological signals, vehicle operation behavior, and real-time road environment data from multiple dimensions through infrared cameras, sensors, vehicle bus interfaces, and roadside interaction devices. S2 allows the data collected by S1 to be preprocessed to remove abnormal information, and multimodal data to be fused through deep learning algorithms to establish an assessment model to dynamically determine the risk level. It also combines the individual characteristics of the driver to calibrate the warning threshold and accurately output the risk assessment results. S3, based on the risk assessment results output by S2, generates targeted multi-dimensional guidance combination strategies, optimizes and adapts to scenarios by combining road conditions and weather, adjusts guidance methods by learning driver preferences, and clarifies emergency priorities; S4 transforms the decision-making strategy generated by S3 into multimodal induced actions involving vision, hearing, and touch, and links with roadside equipment to achieve vehicle-road cooperative protection.

6. The method for proactive interactive guidance on highways based on driver state perception according to claim 5, characterized in that, The specific steps of S1 are as follows: The S11 uses an infrared high-definition camera to collect driver facial feature parameters in real time and identify the driver's status. S12 collects the driver's physiological signals and captures latent states through flexible sensors integrated in the steering wheel grip and seat pressure sensors. S13 connects to the vehicle's CAN bus to acquire operational data and determine whether the driver's operation is standardized and whether there is a tendency to make mistakes. S14 collects environmental data through interaction between vehicle-mounted radar, high-definition road condition cameras, and roadside equipment; S15 automatically matches the driver's identity through in-vehicle facial recognition or fingerprint recognition, and retrieves the driver's historical physiological baseline data and driving preference profile. Furthermore, it automatically establishes an initial baseline upon first use and dynamically updates it through continuous learning.

7. A method for proactive interactive guidance on highways based on driver state perception according to claim 5, characterized in that, The specific steps of S2 are as follows: S21 performs noise reduction and normalization on multi-source data, removes outlier data, and transforms unstructured data into structured data. S22 uses deep learning algorithms to integrate facial features, physiological signals, operational behavior and environmental data to establish a comprehensive driver status assessment model. S23 has four preset levels: safe, low risk, medium risk, and high risk, and dynamically determines the current driving risk based on factors. S24, based on real-time driver status data and historical risk data, uses the LSTM time series prediction algorithm to predict future risk change trends and outputs prediction results of risk increase / decrease / stable, and marks key influencing factors; S25 adjusts the risk assessment thresholds individually based on the driver's age, driving experience, and historical status data.

8. A method for proactive interactive guidance on highways based on driver state perception according to claim 5, characterized in that, The specific steps of S3 are as follows: S31, Generate multi-dimensional inducement combination strategies based on risk level and scenario type; S32, optimize the guidance strategy by combining environmental data; S33, by analyzing drivers' execution of historical guidance strategies and operational feedback, learns their preferred guidance methods; S34, in response to sudden risks, prioritize triggering the highest level of induction strategy and coordinate with emergency response departments at higher levels; S35, for mild to moderate risk states, not only triggers alert-type strategies, but also generates personalized state recovery plans.

9. A method for proactive interactive guidance on highways based on driver state perception according to claim 5, characterized in that, The specific steps of S4 are as follows: S41 controls the vehicle's central control screen, instrument panel indicator lights, and head-up display, showing personalized reminder information, and links with the roadside variable message signs to simultaneously display vehicle-specific guidance prompts; The S42 outputs personalized voice prompts through the car audio system, and adjusts the volume and speech rate, and sets different voice styles for different drivers; S43 controls steering wheel vibration and seat zone vibration to trigger different vibration modes for different risk types; S44, using emotional means to alleviate the driver's poor condition; S45 connects to roadside equipment via vehicle-to-everything (V2X) technology to trigger roadside guidance actions and send warning information to surrounding vehicles.

Images