Intelligent driving assistance method and system based on driver state grading and medium

Abstract

The invention relates to an intelligent driving assistance method and system based on driver state grading and a medium. The method comprises the steps that behavior characteristic data of a driver is collected through a driver monitoring system; based on the behavior characteristic data, calculating and generating a risk score and a confidence coefficient of a driver state through multi-evidence fusion; obtaining a driving scene risk factor, and carrying out coupling calculation on the risk score, the confidence coefficient and the driving scene risk factor to generate a comprehensive risk index; determining a target intervention level based on the comprehensive risk index; generating an auxiliary driving system control parameter set corresponding to the target intervention level according to the target intervention level; and issuing the control parameter set to a corresponding controller of the auxiliary driving system so as to realize self-adaptive adjustment of the auxiliary driving function. According to the method, the response strategy of the auxiliary driving function can be adaptively adjusted according to the real-time state of the driver, so that gradient safety guarantee measures matched with the driver are provided under different risk levels.

Description

Technical Field

[0001] This invention relates to the field of intelligent driving, and in particular to an intelligent driving assistance method, system and medium based on driver state classification. Background Technology

[0002] With the rapid development of the automotive industry, Advanced Driver Assistance Systems (ADAS), such as Adaptive Cruise Control (ACC), Lane Keeping Assist (LKA), and Automatic Emergency Braking (AEB), have been widely applied in modern vehicles, providing drivers with greater driving comfort and safety. Simultaneously, Driver Monitoring Systems (DMS) technology has also made significant progress. These systems use in-vehicle cameras to capture facial images of the driver and employ computer vision technology to identify behavioral characteristics such as eye position, head posture, and yawning, thereby determining whether the driver is fatigued or distracted. When an abnormal driver condition is detected, the DMS system typically issues alerts to the driver through visual, auditory, or tactile means. Furthermore, some existing technologies attempt to classify and assess driver condition based on fatigue levels and take corresponding warning measures or control the vehicle to perform actions such as stopping based on different fatigue levels.

[0003] However, existing technologies still have some shortcomings. On the one hand, traditional ADAS systems typically operate based on preset fixed parameters, such as fixed following distance and fixed warning thresholds. These parameters cannot be dynamically adjusted according to the driver's real-time state, making it difficult to adapt to the safety needs of different driving scenarios. On the other hand, the existing DMS system and ADAS system have weak linkage. They often operate as relatively independent systems. The DMS system is mainly responsible for detecting the driver's state and issuing passive warnings, while the ADAS system executes driver assistance functions according to predetermined logic. There is a lack of an effective collaborative decision-making mechanism between the two. In addition, for driver states at different risk levels, the intervention measures of existing technologies are often relatively simple, lacking a gradient response strategy that matches the risk level. This may lead to over-intervention in low-risk situations, affecting the driving experience, or under-intervention in high-risk situations, failing to effectively ensure driving safety. Summary of the Invention

[0004] In view of the shortcomings of the prior art described above, the purpose of this invention is to provide an intelligent driving assistance method, system and medium based on driver state classification. By combining driver state classification assessment with dynamic adjustment of ADAS system parameters, the response strategy of the driving assistance function can be adaptively adjusted according to the real-time state of the driver, thereby providing graded safety protection measures that match different risk levels.

[0005] To achieve the above objectives, the present invention adopts the following technical solution.

[0006] In a first aspect, the present invention provides an intelligent driving assistance method based on driver state classification, which adopts the following technical solution: An intelligent driving assistance method based on driver state classification includes: The driver monitoring system collects driver behavior data. Based on the behavioral feature data, a risk score and confidence level of the driver's state are calculated and generated through multi-evidence fusion. Obtain driving scenario risk factors, and couple the risk score and confidence level with the driving scenario risk factors to generate a comprehensive risk index; The target intervention level is determined based on the aforementioned comprehensive risk indicators; Generate a set of control parameters for the driver assistance system corresponding to the target intervention level; and The set of control parameters is sent to the corresponding controller of the driver assistance system to achieve adaptive adjustment of the driver assistance function.

[0007] Furthermore, in the above method, the behavioral feature data includes visual behavioral feature data, which includes at least one of the following: Eye closure degree; The direction and duration of visual deviation; Head posture angle; and Frequency of yawning.

[0008] Furthermore, in the above method, the degree of eye closure is determined by calculating the aspect ratio of the eyes, and based on the aspect ratio of the eyes, the PERCLOS value, blinking frequency and duration of a single eye closure are statistically analyzed within a sliding time window.

[0009] Furthermore, the above method also includes: Collect vehicle handling behavior characteristic data, wherein the vehicle handling behavior characteristic data includes at least one of the following: Steering wheel correction range and frequency; Steering wheel torque fluctuation; Lane centering deviation and its rate of change; and Dispersion of accelerator or brake pedal input interval; The multi-evidence fusion calculation includes fusing the visual behavior feature data and the vehicle handling behavior feature data.

[0010] Furthermore, in the above method, the confidence level is determined by at least one of the following factors: Interior lighting conditions; The degree of facial obstruction on the driver's face; Lane line quality; and Sensor health status.

[0011] Furthermore, in the above method, when the confidence level is lower than a preset threshold, entry into the strong intervention strategy is restricted, and only the early warning or light intervention strategy is allowed to be executed.

[0012] Furthermore, in the above method, the driving scenario risk factors include at least one of the following: Vehicle speed; Road curvature; The time of collision with the vehicle in front; Relative velocity; Traffic density; and Environmental factors.

[0013] Furthermore, in the above method, determining the target intervention level based on the comprehensive risk index includes: The risk level is determined using a hysteresis mechanism and a dwell time mechanism. When the comprehensive risk index continuously meets the upgrade conditions for a first preset number of sampling periods, the risk level is upgraded. When the downgrade conditions are met, the risk level must be continuously met for a second preset number of sampling periods before downgrade is allowed. The second preset number of sampling periods is greater than the first preset number of sampling periods.

[0014] Furthermore, in the above method, the set of control parameters includes at least one of the following: Following distance parameters for adaptive cruise control; Sensitivity parameters of automatic emergency braking; Steering intervention gain parameters for lane keeping assist; Speed ​​limit parameters; and Human-computer interaction prompt frequency parameter.

[0015] Furthermore, in the above method, sending the set of control parameters to the corresponding controller of the assisted driving system includes: A rate of change constraint is imposed on the parameters in the set of control parameters to achieve a smooth transition of the parameters.

[0016] Secondly, the present invention provides an intelligent driving assistance system based on driver state classification, which adopts the following technical solution: An intelligent driving assistance system based on driver state classification includes: The driver monitoring module is configured to collect driver behavior characteristic data; The status assessment module is configured to calculate and generate a risk score and confidence level of the driver's status based on the behavioral feature data through multi-evidence fusion. The risk coupling module is configured to acquire driving scenario risk factors and couple the risk score, the confidence level and the driving scenario risk factors to generate a comprehensive risk index. The intervention decision module is configured to determine a target intervention level based on the comprehensive risk index, and generate a corresponding set of control parameters for the driver assistance system according to the target intervention level; and The parameter distribution module is configured to distribute the set of control parameters to the corresponding controller of the driver assistance system in order to achieve adaptive adjustment of the driver assistance function.

[0017] Thirdly, the present invention provides a readable storage medium, which adopts the following technical solution: A readable storage medium storing computer instructions that, when executed by a processor, implement the intelligent driving assistance method based on driver state classification as described in any one of the first aspects above.

[0018] In summary, compared with the prior art, the present invention has at least one of the following beneficial technical effects: This invention provides an intelligent driving assistance method based on driver state grading. By collecting driver behavioral characteristic data and performing multi-evidence fusion calculations, a driver state risk score with confidence assessment can be generated. This score is then coupled with driving scenario risk factors to obtain a comprehensive risk index. Based on this comprehensive risk index, a target intervention level is determined, and a corresponding set of control parameters for the assisted driving system is generated, thereby achieving adaptive adjustment of the assisted driving function. This method enables an effective collaborative decision-making mechanism between the driver monitoring system and the assisted driving system, allowing the control parameters of the assisted driving system to be dynamically adjusted according to the driver's real-time state. Furthermore, by introducing the coupled calculation of confidence assessment and driving scenario risk factors, the accuracy and reliability of risk assessment can be improved. Simultaneously, the graded intervention level determined based on the comprehensive risk index allows the system to adopt response strategies matching different risk levels, thus ensuring both driving safety and a good driving experience. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 A flowchart of an embodiment of the intelligent driving assistance method based on driver state classification of the present invention is shown.

[0021] Figure 2 A schematic diagram of an embodiment of the driver state feature threshold matrix of the present invention is shown.

[0022] Figure 3 A flowchart of an embodiment of the intervention strategy selection method based on confidence threshold judgment of the present invention is shown.

[0023] Figure 4 The flowchart of an embodiment of the method for determining the level of risk based on comprehensive risk indicators of the present invention is shown.

[0024] Figure 5 A flowchart of an embodiment of the control parameter distribution process of the present invention is shown.

[0025] Figure 6 The diagram shows an architecture block diagram of an embodiment of the intelligent driving assistance system based on driver state classification of the present invention. Detailed Implementation

[0026] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application. Furthermore, it should be understood that the specific embodiments described herein are only for illustration and explanation of this application and are not intended to limit this application.

[0027] It should be noted that the order of description of the following embodiments is not intended to limit the preferred order of the embodiments of this application. Furthermore, the descriptions of each embodiment in the following embodiments have their own emphasis; for parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0028] The method steps described in this embodiment of the invention can be executed in the order described in the specific implementation, or the execution order of each step can be adjusted according to actual needs, provided that the technical problem can be solved. These are not listed one by one here.

[0029] The present invention will be further described in detail below with reference to the accompanying drawings.

[0030] Reference Figure 1 , Figure 1 A flowchart of an intelligent driving assistance method 100 based on driver state classification is shown. Method 100 implements a complete processing flow from driver state data acquisition to adaptive adjustment of driving assistance system parameters, and achieves collaborative decision-making between driver state and driving assistance functions through multi-evidence fusion and scenario risk coupling calculation.

[0031] Method 100 begins with step 102, in which the driver's behavioral characteristic data is collected through a driver monitoring system. The driver monitoring system employs an infrared camera or an RGB+infrared dual-mode camera, equipped with infrared illumination to improve recognition stability in nighttime and backlit scenes. In some embodiments, the DMS camera captures driver facial images at 20-60fps. The in-vehicle network can be CAN / CAN FD or in-vehicle Ethernet, used to connect the cooperative control unit with the brake actuator, steering actuator, and human-machine interface (HMI). The cooperative control unit can be integrated into a domain controller or ADAS ECU.

[0032] Continue to refer to Figure 1 The behavioral feature data includes visual behavioral feature data, which includes at least one of the following: eye closure degree, gaze deviation direction and duration, head posture angle, and yawning frequency. Visual behavioral feature extraction includes face detection and keypoint localization. In each frame, a face bounding box is located, and keypoints such as the corners of the eyes, mouth, and nose tip are output, along with their confidence scores. Eye closure degree is determined by calculating the eye aspect ratio, and based on this ratio, the PERCLOS value, blinking frequency, and duration of a single eye closure are calculated within a sliding time window. Gait deviation features estimate the gaze direction based on the iris center or eye posture, calculating the deviation angle between the gaze and the reference direction ahead of the road. Head posture features utilize keypoints and a 3D head model to calculate posture, obtaining pitch, yaw, and roll angles. Yawning features calculate the mouth opening ratio or opening / closing area based on mouth keypoints, and statistically analyze yawning frequency and duration.

[0033] In some embodiments, the behavioral feature data also includes physiological feature data, including heart rate variability (HRV), respiratory rate, or skin conductance signals, to form a multi-source feature vector. In some embodiments, method 100 further includes acquiring vehicle handling behavior feature data, including at least one of steering wheel correction amplitude and frequency, steering wheel torque fluctuation, lane centering deviation and its rate of change, and accelerator or brake pedal input interval dispersion. When sunglasses, masks, or strong backlighting are detected to cause instability at key points, the weight of visual features is reduced and the weight of other evidence sources is increased.

[0034] like Figure 1As shown, method 100 proceeds to step 104 after step 102. In step 104, a risk score and confidence level for the driver's state are generated based on behavioral feature data through multi-evidence fusion calculation. The multi-evidence fusion calculation calculates the visual risk score R_v and the behavioral risk score R_b respectively, and arbitrates them in conjunction with the fusion confidence level C. In some embodiments, the multi-evidence fusion calculation includes fusing visual behavioral feature data and vehicle handling behavioral feature data. When the driver's state is normal and meets the preset lucidity and stability window, the system establishes a personalized baseline parameter set for the driver. The baseline parameter set includes the mean, variance, and acceptable fluctuation range of each feature. Subsequent level determination thresholds are dynamically generated based on the baseline parameters.

[0035] Continue to refer to Figure 1 Method 100 proceeds to step 106 after step 104, where driving scenario risk factors are obtained. Driving scenario risk factors include at least one of the following: vehicle speed, road curvature, collision time with the vehicle ahead, relative speed, traffic density, and environmental factors. Environmental factors include conditions such as nighttime, rainfall, and lane line quality.

[0036] Method 100 proceeds to step 108 after step 106, where the risk score, confidence level, and driving scenario risk factors are coupled and calculated to generate a comprehensive risk index. This comprehensive risk index enables differentiated strategies to be adopted for the same driver in different scenarios.

[0037] like Figure 1 As further shown, method 100 proceeds to step 110 after step 108, where the target intervention level is determined based on a comprehensive risk index. Method 100 then proceeds to step 112, where a set of control parameters for the assisted driving system is generated according to the target intervention level. Method 100 then proceeds to step 114, where the set of control parameters is sent to the corresponding controller of the assisted driving system to achieve adaptive adjustment of the assisted driving function.

[0038] For example, refer to Figure 2 , Figure 2 This diagram illustrates an embodiment of a driver state characteristic threshold matrix. The matrix is ​​presented in tabular form, containing five characteristic items and their corresponding threshold ranges for mild, moderate, and severe levels, as well as related notes.

[0039] Continue to refer to Figure 2The first feature in the feature threshold matrix is ​​the duration of gaze deviation, with a threshold of 1-2 seconds for mild, 2-5 seconds for moderate, and ≥5 seconds for severe. The second feature is the duration of a single eye closure, with a threshold of less than 0.5 seconds for mild, 0.5-1 seconds for moderate, and ≥1 second for severe. The third feature is the PERCLOS sliding window, with a slight increase for mild, a significant increase for moderate, and a high percentage for severe. The fourth feature is the head lateral deviation angle and its duration, with slight deviation for mild, continuous lateral deviation for moderate, and ≥45 degrees and ≥3 seconds for severe. The fifth feature is the yawning frequency sliding window, with occasional yawns for mild, ≥2 yawns for moderate, and frequent and continuous yawns for severe.

[0040] like Figure 2 As further illustrated, the feature threshold matrix is ​​used to compare the driver's real-time feature values ​​with thresholds at various levels to determine the driver's current state level. In some implementations, the thresholds in the feature threshold matrix can be modified by incorporating scenario risk factors. In some implementations, the thresholds are dynamically generated from a personalized baseline to avoid cross-driver misjudgments caused by using fixed thresholds. In some implementations, transition strategies can be triggered earlier based on the upward trend of the comprehensive risk index, achieving earlier and smoother risk intervention.

[0041] Continue to refer to Figure 2 When the driver's condition is determined to be at a mild level, the corresponding intervention strategies include enhancing the clarity and urgency of warning prompts, such as increasing the alarm volume and flashing warning boxes on the screen. When the driver's condition is determined to be at a moderate level, the corresponding intervention strategies include actively enhancing driver assistance functions, such as increasing the following distance of ACC, improving the steering intervention sensitivity of LKA under the rate of change constraint, and pre-tensioning seat belts. When the driver's condition is determined to be at a severe level, the corresponding intervention strategies include activating the strongest intervention measures: hazard lights to warn following vehicles, automatically reducing vehicle speed, finding a safe parking point through navigation, and ultimately performing the least-risk maneuver to achieve a safe stop.

[0042] Reference Figure 3 , Figure 3 A flowchart of an intervention strategy selection method 200 based on confidence threshold judgment is shown. Method 200 demonstrates the decision logic for selecting intervention strategies based on confidence assessment results. By limiting the execution of strong intervention strategies when the confidence level is low, the probability of forced takeover due to false detection is reduced, thereby improving the reliability of the system response.

[0043] Method 200 begins at step 202, where a confidence level is determined. The confidence level is determined by at least one of the following factors: in-vehicle lighting conditions, driver facial occlusion level, lane line quality, and sensor health. In some implementations, the confidence level is obtained by combining key point confidence, lighting quality, occlusion level, lane line quality, and sensor health. The system is configured with a DMS health diagnostic: when the camera is offline, the image is overexposed or underexposed, the key point confidence level remains below a threshold, or the fusion confidence level remains below a threshold, a degradation strategy is initiated.

[0044] Continue to refer to Figure 3 Method 200 proceeds to step 204 after step 202, where it determines whether the confidence level is lower than a preset threshold. If the confidence level is lower than the preset threshold, method 200 proceeds to step 206, where it restricts the implementation of strong intervention strategies and only allows the execution of early warning or mild intervention strategies. The degradation strategy only outputs HMI prompts and mild strategies, and writes abnormal sensor states into fault codes for reporting, to ensure that system behavior is predictable and verifiable.

[0045] like Figure 3 As further shown, if the confidence level is not lower than a preset threshold, method 200 proceeds to step 208, where all intervention strategies, including strong intervention, are allowed to be executed. In some implementations, strong intervention is only allowed when the confidence level is high and the visual risk score and behavioral risk score consistently point to higher risk over N consecutive periods, in order to reduce the probability of forced takeover due to false detection.

[0046] Furthermore, refer to Figure 4 , Figure 4 A flowchart of a risk level assessment method 300 based on comprehensive risk indicators is shown. Method 300 employs a hysteresis mechanism and a dwell time mechanism for risk level assessment to avoid frequent risk level fluctuations and ensure the stability of the system response.

[0047] Method 300 begins at step 302, where a comprehensive risk index is obtained. This comprehensive risk index is generated by coupling risk score, confidence level, and driving scenario risk factors. After step 302, Method 300 proceeds to step 304, where it determines whether the comprehensive risk index meets the upgrade conditions.

[0048] Continue to refer to Figure 4 If the comprehensive risk indicators meet the upgrade conditions, method 300 proceeds to step 306, where it is determined whether the first preset sampling period number is continuously met. If the first preset sampling period number is continuously met in step 306, method 300 proceeds to step 310 to perform a level upgrade. If the first preset sampling period number is not continuously met in step 306, method 300 proceeds to step 312 to maintain the current level.

[0049] like Figure 4 Further, if the comprehensive risk index does not meet the upgrade conditions in step 304, method 300 proceeds to step 308, where it is determined whether the downgrade conditions are met and the second preset sampling period number is continuously satisfied. If the downgrade conditions are met and the second preset sampling period number is continuously satisfied in step 308, method 300 proceeds to step 314 to allow the level downgrade. If the above conditions are not met in step 308, method 300 proceeds to step 316 to maintain the current level.

[0050] Continue to refer to Figure 4 The second preset sampling period number is greater than the first preset sampling period number, thereby achieving asymmetric control of risk level increases and decreases. The design of relatively rapid risk level upgrades and relatively slow risk level downgrades allows the system to respond promptly when risk is detected to be rising, while remaining cautious when risk is falling, avoiding premature removal of intervention measures.

[0051] In some implementations, a rapid escalation pathway can be triggered in the event of a severe critical event. Severe critical events include eyes closed for more than a preset duration or gaze deviating for more than a preset duration. The rapid escalation pathway bypasses the conventional dwell time requirement, enabling the system to immediately escalate to a higher intervention level when a serious risk is detected.

[0052] In some implementations, the system's operating modes are abstracted as a state machine: normal mode, early warning mode, enhanced assistance mode, and MRM minimum risk maneuver mode, switching between modes according to state machine rules. The system receives controller execution feedback and driver status updates; if the risk decreases, it gradually reverts to normal mode according to dwell time rules. If the risk increases or a critical event is triggered, it enters the MRM closed-loop execution.

[0053] Furthermore, refer to Figure 5 , Figure 5 A flowchart illustrating the control parameter distribution process is provided. Method 400 is used to distribute a set of control parameters to the driver assistance system controller. Method 400 demonstrates the sequential execution process from obtaining the target intervention level to completing the control parameter distribution.

[0054] Method 400 begins with step 402, in which the target intervention level is obtained. The target intervention level is determined by a comprehensive risk index, which is generated by coupling risk score, confidence level, and driving scenario risk factors.

[0055] Continue to refer to Figure 5Method 400 proceeds to step 404 after step 402, where a set of control parameters is generated based on the acquired target intervention level. The set of control parameters includes at least one of the following distance parameters for adaptive cruise control, sensitivity parameters for automatic emergency braking, steering intervention gain parameters for lane keeping assist, speed limit parameters, and human-machine interaction prompt frequency parameters. In some implementations, parameter generation uses a lookup table to obtain baseline parameters for each level, and interpolation is performed when the comprehensive risk index is between adjacent levels to achieve a smooth transition.

[0056] like Figure 5 As further shown, method 400 proceeds to step 406 after step 404, where a rate-of-change constraint is applied to the control parameters. The rate-of-change constraint includes the following distance rate of change, the target deceleration rate of change, and the lateral control gain rate of change. Method 400 then proceeds to step 408 after step 406, where a smooth transition of the parameters is achieved. By applying rate-of-change constraints to the parameters in the control parameter set and achieving a smooth transition, comfort degradation or secondary risks caused by sudden parameter changes can be avoided.

[0057] Continue to refer to Figure 5 Method 400 proceeds to step 410 after step 408, in which the set of control parameters is sent to the driver assistance system controller. In some embodiments, the parameters are sent to the corresponding controller via CAN messages or Ethernet service interfaces, and the controller sends back an execution receipt for closed-loop verification.

[0058] In some implementations, the human-machine interface (HMI) automatically simplifies the interface based on the driver's condition level, using clearer and easier-to-understand voice and text for communication, ensuring the highest efficiency of information delivery and the lowest cognitive load when the driver is not in good condition.

[0059] This invention also discloses an intelligent driving assistance system based on driver state classification.

[0060] Reference Figure 6 , Figure 6 The diagram shows the architecture of an intelligent driving assistance system 500 based on driver state classification. The intelligent driving assistance system 500 includes a driver monitoring module 502, a state assessment module 504, a risk coupling module 506, an intervention decision module 508, and a parameter distribution module 510.

[0061] Continue to refer to Figure 6The driver monitoring module 502 is configured to collect driver behavior characteristic data. This behavior characteristic data includes visual behavior characteristic data, which includes at least one of the following: eye closure degree, direction and duration of gaze deviation, head posture angle, and yawning frequency. In some embodiments, the driver monitoring module 502 also collects vehicle handling behavior characteristic data and physiological characteristic data.

[0062] like Figure 6 As further shown, the state assessment module 504 is connected to the driver monitoring module 502. The state assessment module 504 is configured to generate a risk score and confidence level of the driver's state based on behavioral feature data through multi-evidence fusion calculation. The state assessment module 504 performs fusion calculation on visual behavioral feature data and vehicle handling behavioral feature data, and outputs the risk score and confidence level.

[0063] Continue to refer to Figure 6 The risk coupling module 506 is connected to the state assessment module 504. The risk coupling module 506 is configured to acquire driving scenario risk factors and couple the risk score and confidence level with the driving scenario risk factors to generate a comprehensive risk index. The driving scenario risk factors include at least one of the following: vehicle speed, road curvature, collision time with the vehicle in front, relative speed, traffic density, and environmental factors.

[0064] like Figure 6 As further shown, the intervention decision module 508 is connected to the risk coupling module 506. The intervention decision module 508 is configured to determine the target intervention level based on a comprehensive risk index and generate a corresponding set of control parameters for the driver assistance system according to the target intervention level. The set of control parameters includes at least one of the following distance parameters for adaptive cruise control, sensitivity parameters for automatic emergency braking, steering intervention gain parameters for lane keeping assist, speed limit parameters, and human-machine interaction prompt frequency parameters.

[0065] Continue to refer to Figure 6 The parameter distribution module 510 is connected to the intervention decision module 508. The parameter distribution module 510 is configured to distribute a set of control parameters to the driver assistance system controller 512 to achieve adaptive adjustment of the driver assistance function. The driver assistance system controller 512 is located outside the intelligent driving assistance system 500 and receives the set of control parameters from the parameter distribution module 510. The modules of the intelligent driving assistance system 500 are sequentially connected to form an information processing link, realizing a complete data flow from driver status data acquisition, status assessment, risk coupling calculation, intervention decision to parameter distribution.

[0066] In some implementations, when the target intervention level reaches the highest risk, the intelligent driving assistance system 500 enters the Minimum Risk Maneuvering (MRM) mode and executes a safe parking procedure. The MRM safe zone generation step generates candidate parking areas based on navigation and lane models and assesses their feasibility. Candidate parking areas include emergency lanes, rightmost lane parking areas, service areas, or exits.

[0067] In some implementations, the MRM longitudinal speed curve planning generates a deceleration curve while meeting the maximum deceleration threshold and maintaining a TTC (Total Traffic Conversion Rate) no lower than the threshold with respect to the preceding and following vehicles. The MRM lateral lane change or lane-keeping control executes a lane change to the rightmost lane or emergency lane when the lane line quality meets the threshold. If the lane line quality is insufficient, it is downgraded to controlled deceleration and stopping within the lane.

[0068] In some implementations, the MRM completion judgment and post-processing include: maintaining hazard lights and continuous audible and visual warnings when the vehicle speed is 0 and parking is completed, and optionally triggering a remote assistance interface. If external risks such as insufficient lane-changing space are detected during MRM execution, the lane-keeping deceleration strategy is maintained and the safe zone is reassessed, achieving controllable degradation in case of failure.

[0069] This invention also discloses a readable storage medium.

[0070] A computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the intelligent driving assistance method based on driver state classification described in any of the above embodiments. The computer-readable storage medium may include any entity or device capable of carrying a computer program, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), and a software distribution medium, etc. The computer program includes computer program code. The computer program code may be in the form of source code, object code, an executable file, or some intermediate form, etc. The computer-readable storage medium may include any entity or device capable of carrying computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), and a software distribution medium, etc.

[0071] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing a particular logical function or process, and the scope of the preferred embodiments of the invention includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as will be understood by those skilled in the art to which embodiments of the invention pertain.

[0072] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus or device (such as a computer-based system, a system including a processing module or other system that can fetch and execute instructions from, an instruction execution system, apparatus or device).

[0073] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions 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 the present invention.

Claims

1. An intelligent driving assistance method based on driver state classification, characterized in that, include: The driver monitoring system collects driver behavior data. Based on the behavioral feature data, a risk score and confidence level of the driver's state are calculated and generated through multi-evidence fusion. Obtain driving scenario risk factors, and couple the risk score and confidence level with the driving scenario risk factors to generate a comprehensive risk index; The target intervention level is determined based on the aforementioned comprehensive risk indicators; Generate a set of control parameters for the driver assistance system corresponding to the target intervention level; as well as The set of control parameters is sent to the corresponding controller of the driver assistance system to achieve adaptive adjustment of the driver assistance function.

2. The method according to claim 1, characterized in that, The behavioral feature data includes visual behavioral feature data, which includes at least one of the following: Eye closure degree; The direction and duration of visual deviation; Head posture angle; and Frequency of yawning.

3. The method according to claim 2, characterized in that, The degree of eye closure is determined by calculating the aspect ratio of the eyes, and based on the aspect ratio of the eyes, the PERCLOS value, blinking frequency and duration of a single eye closure are statistically analyzed within a sliding time window.

4. The method according to claim 1, characterized in that, Also includes: Collect vehicle handling behavior characteristic data, wherein the vehicle handling behavior characteristic data includes at least one of the following: Steering wheel correction range and frequency; Steering wheel torque fluctuation; Lane centering deviation and its rate of change; as well as Dispersion of accelerator or brake pedal input interval; The multi-evidence fusion calculation includes fusing the visual behavior feature data and the vehicle handling behavior feature data.

5. The method according to claim 1, characterized in that, The confidence level is determined by at least one of the following factors: Interior lighting conditions; The degree of facial obstruction on the driver's face; Lane line quality; and Sensor health status.

6. The method according to claim 5, characterized in that, When the confidence level is lower than a preset threshold, entry into the strong intervention strategy is restricted, and only the early warning or light intervention strategy is allowed.

7. The method according to claim 1, characterized in that, The driving scenario risk factors include at least one of the following: Vehicle speed; Road curvature; The time of collision with the vehicle in front; Relative velocity; Traffic density; and Environmental factors.

8. The method according to claim 1, characterized in that, Determining the target intervention level based on the aforementioned comprehensive risk indicators includes: The risk level is determined using a hysteresis mechanism and a dwell time mechanism. When the comprehensive risk index continuously meets the upgrade conditions for a first preset number of sampling periods, the risk level is upgraded. When the downgrade conditions are met, the risk level must be continuously met for a second preset number of sampling periods before downgrade is allowed. The second preset number of sampling periods is greater than the first preset number of sampling periods.

9. The method according to claim 1, characterized in that, The set of control parameters includes at least one of the following: Following distance parameters for adaptive cruise control; Sensitivity parameters of automatic emergency braking; Steering intervention gain parameters for lane keeping assist; Speed ​​limit parameters; and Human-computer interaction prompt frequency parameter.

10. The method according to claim 9, characterized in that, Sending the set of control parameters to the corresponding controller of the driver assistance system includes: A rate of change constraint is imposed on the parameters in the set of control parameters to achieve a smooth transition of the parameters.

11. An intelligent driving assistance system based on driver state classification, characterized in that, include: The driver monitoring module is configured to collect driver behavior characteristic data; The status assessment module is configured to calculate and generate a risk score and confidence level of the driver's status based on the behavioral feature data through multi-evidence fusion. The risk coupling module is configured to acquire driving scenario risk factors and couple the risk score, the confidence level and the driving scenario risk factors to generate a comprehensive risk index. The intervention decision module is configured to determine the target intervention level based on the comprehensive risk index, and generate a set of control parameters for the assisted driving system corresponding to the target intervention level. as well as The parameter distribution module is configured to distribute the set of control parameters to the corresponding controller of the driver assistance system in order to achieve adaptive adjustment of the driver assistance function.

12. A readable storage medium, characterized in that, The readable storage medium stores computer instructions that, when executed by a processor, implement the intelligent driving assistance method based on driver state classification as described in any one of claims 1-10.

Images