AES (Advanced Encryption Standard) trigger threshold adjustment method and system based on driver portrait and medium

Abstract

The invention relates to an AES trigger threshold adjustment method and system based on a driver portrait and a medium. The method comprises the steps that vehicle motion data, environment perception data and driver operation data are collected; calculating driving behavior characteristics based on the collected data; performing clustering analysis on the driving behavior characteristics by using an unsupervised learning algorithm, constructing driver portraits, and classifying drivers into an agile type, a conventional type or a conservative type by the driver portraits; determining a personalized adjustment coefficient according to the driver portrait, and calculating a user personalized collision time threshold value based on the personalized adjustment coefficient and a preset basic threshold value; and comparing the personalized collision time threshold value of the user with the minimum safe collision time threshold value, and taking the larger value as the final triggering threshold value of the AES system. According to the method, the system acceptance and the obstacle avoidance success rate of users with different driving styles can be remarkably improved by constructing the personalized driver portrait and dynamically adjusting the AES trigger threshold, and meanwhile, false triggering and interference feeling are reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of automobile driving assistance, in particular to an AES trigger threshold adjustment method and system based on driver portrait and a medium. BACKGROUND

[0002] Automatic Emergency Steering (AES) system is one of the core functions of Advanced Driver Assistance System (ADAS) and is an active safety technology for vehicles. It relies on sensors (such as cameras, millimeter wave radars, laser radars, etc.) to monitor the driving environment in front, side and side-rear of the vehicle in real time. When the vehicle faces an emergency situation, the system controls the electric power steering system (EPS) to control the vehicle steering to avoid collision or reduce the consequences of collision through autonomous judgment and control. The existing AES system adopts a fixed threshold strategy based on parameters such as Time to Collision (TTC) and Distance to Collision (DTC) to trigger the emergency steering function. These thresholds are usually based on general scene calibration and the uniform trigger conditions are determined through a large amount of test data.

[0003] The fixed threshold strategy of the existing AES system is difficult to adapt to the operation characteristics of users with different driving habits, resulting in user experience problems. Aggressive drivers are used to late steering and large angle operation, and the fixed threshold may cause them to feel "disturbed" due to the system's early intervention. Conservative drivers tend to steer early and adjust with small angles, and the fixed threshold may reduce the obstacle avoidance success rate due to the late intervention. In addition, the existing system lacks the ability to learn and adapt to individual differences of drivers and cannot dynamically adjust the trigger parameters according to the driving behavior characteristics of users, which limits the applicability and acceptance of the AES system under different driving styles. SUMMARY

[0004] In view of the above-mentioned shortcomings of the prior art, the purpose of the present application is to provide an AES trigger threshold adjustment method and system based on driver portrait and a medium. By constructing a personalized driver portrait and dynamically adjusting the AES trigger threshold, the system acceptance and obstacle avoidance success rate of users with different driving styles can be significantly improved, and the false triggering and disturbance can be reduced.

[0005] To achieve the above-mentioned purpose, the present application adopts the following technical solutions.

[0006] In a first aspect, the present application provides an AES trigger threshold adjustment method based on driver portrait, which adopts the following technical solutions: Collecting vehicle motion data, environmental perception data and driver operation data; Calculating driving behavior characteristics based on the collected data, the driving behavior characteristics including collision time, steering angle rate and following time distance when a steering event is triggered; performing clustering analysis on the driving behavior features by using an unsupervised learning algorithm to construct a driver portrait, the driver portrait classifying the driver as agile, regular, or conservative; determining a personalized adjustment coefficient according to the driver portrait, and calculating a user personalized collision time threshold based on the personalized adjustment coefficient and a preset basic threshold; and comparing the user personalized collision time threshold with a minimum safe collision time threshold, and taking the larger value as the final triggering threshold of the AES system.

[0007] Further, in the above-described AES triggering threshold adjustment method, the collection of vehicle motion data, environmental perception data, and driver operation data includes: collecting vehicle longitudinal speed, lateral acceleration, yaw rate, steering angle, and steering angle rate as vehicle motion data; collecting front, side, and rear millimeter wave radar data and forward-looking camera data to obtain the relative distance, relative speed, and cross-sectional width of the target object as environmental perception data; collecting steering torque, release button signal, and steering intervention flag as driver operation data; and performing time synchronization on all collected data to ensure that the jitter of the data timestamp is less than a preset threshold.

[0008] Further, in the above-described AES triggering threshold adjustment method, it further includes collecting lane geometry data and road adhesion coefficient data, the lane geometry data including lane curvature, lane width, and available lateral margin, and the road adhesion coefficient being estimated by tire slip data of an electronic stability program system or a machine learning algorithm.

[0009] Further, in the above-described AES triggering threshold adjustment method, the calculation of driving behavior features based on the collected data includes: statistically calculating the collision time mean value at the triggering moment of the driver's active steering event within a preset time length of a sliding window; calculating the peak angle rate and the average angle rate of the steering event; statistically calculating the following time distance based on the forward nearest target; calculating the leading collision time before lane changing; statistically calculating the average torque threshold and the takeover frequency of the driver when the lateral assistance system takes over; and performing three standard deviation truncation and minimum-maximum normalization processing on all features.

[0010] Further, the above-mentioned AES trigger threshold adjustment method further comprises calculating a risk exposure feature, the risk exposure feature comprising a cut-in event frequency, a city and highway driving proportion, a day and night driving label, and a rain and snow weather label, for correcting a confidence level of the driver portrait.

[0011] Further, in the above-mentioned AES trigger threshold adjustment method, the clustering analysis of the driving behavior feature by using an unsupervised learning algorithm comprises: adopting a hierarchical density-based clustering algorithm or a Gaussian mixture model to cluster the driving behavior feature; outputting a soft assignment probability to obtain a probability distribution of the driver belonging to the agile type, the regular type, and the conservative type; selecting a type corresponding to a maximum probability as a current driver portrait gear; and setting a hysteresis threshold, and only when a difference between a new gear probability and a current gear probability exceeds a preset threshold, the portrait switching is performed.

[0012] Further, the above-mentioned AES trigger threshold adjustment method further comprises calculating a driver portrait confidence level, the confidence level being calculated based on a driving event count and an outlier rate, and when the confidence level is lower than a preset threshold, a weight of the portrait on threshold adjustment is reduced.

[0013] Further, in the above-mentioned AES trigger threshold adjustment method, the determination of the individualized adjustment coefficient according to the driver portrait comprises: calculating an original adjustment coefficient based on a soft portrait probability distribution, the original adjustment coefficient being proportional to a difference between a conservative type coefficient and an agile type coefficient; correcting the original adjustment coefficient according to a driver acceptance of the system, the acceptance being calculated based on a cancellation rate after the AES trigger; dynamically modulating the adjustment coefficient based on a real-time risk score, the adjustment coefficient being increased when the risk score is higher than a preset upper limit, and the adjustment coefficient being decreased when the risk score is lower than a preset lower limit and a takeover rate is low; and limiting the adjustment coefficient in a preset range.

[0014] Further, the above-mentioned AES trigger threshold adjustment method further comprises periodically updating the individualized adjustment coefficient, and when a driving mileage reaches a preset mileage threshold or a cumulative number of AES triggers reaches a preset number threshold, the adjustment coefficient is recalculated based on driving behavior data in a current evaluation period.

[0015] Further, in the above-mentioned AES trigger threshold adjustment method, the minimum safe collision time threshold is calculated based on a minimum displacement required for lateral obstacle avoidance and a maximum lateral acceleration allowed, the maximum lateral acceleration being the smaller of a product of a road adhesion coefficient and a gravitational acceleration and an upper limit of a comfortable lateral acceleration, and the minimum safe collision time threshold further includes a sensor sensing time delay, a calculation processing time delay and an actuator response time delay.

[0016] In a second aspect, the present application provides an AES trigger threshold adjustment system based on driver portrait, which adopts the following technical solution: a data acquisition module configured to acquire vehicle motion data, environment perception data and driver operation data; a feature calculation module configured to calculate driving behavior features based on the acquired data, the driving behavior features including collision time when a steering event is triggered, steering angular velocity and following distance; a portrait construction module configured to perform cluster analysis on the driving behavior features by using an unsupervised learning algorithm to construct a driver portrait, the driver portrait classifying drivers into agile, regular or conservative types; a threshold calculation module configured to determine a personalized adjustment coefficient according to the driver portrait and calculate a user personalized collision time threshold based on the personalized adjustment coefficient and a preset basic threshold; and a threshold determination module configured to compare the user personalized collision time threshold with a minimum safe collision time threshold and take the larger value as the final trigger threshold of the AES system.

[0017] In a third aspect, the present application provides a readable storage medium, which adopts the following technical solution: A readable storage medium, the readable storage medium storing computer instructions, the computer instructions being executed by a processor to implement the AES trigger threshold adjustment method according to any one of the above-mentioned first aspect.

[0018] In summary, compared with the prior art, the present application has at least one of the following beneficial technical effects: By acquiring multi-dimensional driving data and constructing a personalized driver portrait by using an unsupervised learning algorithm, the present application can dynamically adjust the AES trigger threshold according to different driving styles, thereby significantly improving the adaptability and user acceptance of the system while ensuring safety. For agile drivers, the trigger timing can be appropriately delayed to reduce unnecessary intervention and avoid the feeling of being disturbed; for conservative drivers, the intervention can be advanced to improve the success rate of obstacle avoidance. At the same time, through the comparison mechanism with the minimum safe collision time threshold, it is ensured that the basic safety guarantee level will not be reduced in any case, achieving the balance between personalization and safety. BRIEF DESCRIPTION OF DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the description of the embodiments will be briefly introduced. Obviously, the drawings in the following description only represent some of the embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative effort on the basis of these drawings.

[0020] Figure 1 A flow chart of the method for adaptive adjustment of the AES triggering threshold based on the driver portrait is shown.

[0021] Figure 2 A flow chart of the data acquisition and synchronization method in the AES triggering threshold adjustment system is shown.

[0022] Figure 3 A flow chart of the driving behavior feature calculation method is shown.

[0023] Figure 4 A flow chart of the driver portrait clustering and classification method is shown.

[0024] Figure 5 A flow chart of the personalized adjustment coefficient determination method is shown.

[0025] Figure 6 A flow chart of the AES system minimum safe collision time threshold calculation method is shown.

[0026] Figure 7 A module architecture diagram of the AES triggering threshold adjustment system is shown. DETAILED DESCRIPTION

[0027] The technical solutions in the embodiments of the present application will be described clearly and completely in the following description of the embodiments of the present application in conjunction with the drawings. Obviously, the described embodiments only represent some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative effort fall within the scope of the present application. In addition, it should be understood that the specific embodiments described herein are only used to illustrate and explain the present application, and are not intended to limit the present application.

[0028] It should be noted that the description order of the following embodiments is not intended to limit the preferred order of the embodiments of the present application. Moreover, the description of each embodiment in the following embodiments has its own focus, and the parts not described in detail in a certain embodiment can be referred to the related description of other embodiments.

[0029] The execution sequence of the method steps in the embodiments of the present application can be executed according to the sequence described in the specific implementation, or the execution sequence of each step can be adjusted on the premise of being able to solve the technical problems according to actual needs, and here is not listed one by one.

[0030] The present application is further described in detail below with reference to the accompanying drawings.

[0031] Referring to Figure 1 The embodiments of the present application provide an AES trigger threshold adjustment method 100 based on driver portrait, which includes data collection, portrait clustering, threshold decision, execution and feedback, and forms a closed-loop feedback mechanism to realize online adaptive adjustment.

[0032] The method 100 starts from step 102, and collects vehicle motion data, environmental perception data and driver operation data. The vehicle motion data includes longitudinal speed, lateral acceleration, yaw rate, steering angle and steering angle rate. The environmental perception data is obtained through front, side and rear millimeter wave radars and front-view cameras, and includes relative distance, relative speed and cross-section width information of target objects. The driver operation data covers steering torque, release button signal and steering intervention flag.

[0033] In step 104, the method 100 calculates the driving behavior features based on the collected data. The driving behavior features include the collision time when the steering event is triggered, the steering angle rate and the following time distance. In a preset time length sliding window, the average value of the collision time at the moment of the driver's active steering event trigger is counted, the peak angle rate and the average angle rate of the steering event are calculated, and the following time distance based on the nearest forward target is counted.

[0034] Step 106 uses unsupervised learning algorithm to perform clustering analysis on the driving behavior features. This step uses hierarchical density-based clustering algorithm or Gaussian mixture model to perform clustering processing on the driving behavior features, and outputs soft assignment probability to obtain the probability distribution of the driver belonging to different types.

[0035] In step 108, the method 100 constructs the driver portrait and classifies the driver into agile, regular or conservative. The agile driver has higher steering angle rate and shorter following time distance, the conservative driver has lower steering angle rate and longer following time distance, and the characteristic parameters of the regular driver are between the two. This step selects the type corresponding to the maximum probability as the current driver portrait gear.

[0036] Step 110 determines the individualized adjustment coefficient according to the driver profile. This step calculates the original adjustment coefficient based on the soft profile probability distribution, which is proportional to the difference between the conservative coefficient and the agile coefficient. The original adjustment coefficient is modified according to the driver's acceptance of the AES, which is calculated based on the cancellation rate after the AES is triggered.

[0037] In step 112, the method 100 calculates the user individualized collision time threshold based on the individualized adjustment coefficient and the preset basic threshold. The preset basic threshold is set to different values according to different driver profile types, with different basic collision time thresholds corresponding to agile, regular, and conservative types. The user individualized collision time threshold is obtained by multiplying the individualized adjustment coefficient by the corresponding preset basic threshold.

[0038] Step 114 compares the user individualized collision time threshold with the minimum safe collision time threshold. The minimum safe collision time threshold is calculated based on the minimum displacement required for lateral obstacle avoidance and the maximum allowed lateral acceleration, ensuring that the AES can maintain basic safety performance under any individualized adjustment.

[0039] The method 100 takes the larger value as the final trigger threshold of the AES in step 116. By comparing the user individualized collision time threshold with the minimum safe collision time threshold, the threshold with the larger value is selected as the final trigger threshold of the AES, ensuring that the individualized adjustment does not exceed the safety boundary. This closed-loop feedback mechanism records the user's cancellation and takeover behavior through the execution and feedback module, which is used for subsequent online adaptive adjustment.

[0040] Referring to Figure 2 The embodiment of the present application provides a data acquisition and synchronization method 200 for realizing high-precision time synchronization acquisition of multi-source data. The method 200 acquires different types of driving data through a frequency division acquisition mechanism, and ensures the time consistency of the data through a unified time base.

[0041] The method 200 first performs step 202 to acquire vehicle longitudinal speed, lateral acceleration, yaw rate, steering angle, and steering angle rate as vehicle motion data. The vehicle motion data is acquired at a frequency of 100 Hz, ensuring high-precision monitoring of the vehicle's dynamic behavior. The longitudinal speed reflects the change in the vehicle's forward speed, the lateral acceleration represents the lateral motion state of the vehicle, and the yaw rate describes the rotational motion of the vehicle around the vertical axis.

[0042] In step 204, the method 200 collects front, side and rear millimeter-wave radar data and front-facing camera data. The environmental perception data is collected at a frequency of 20-25 Hz, obtaining the relative distance, relative speed and cross-sectional width of the target object as the environmental perception data. The millimeter-wave radar provides accurate distance and speed measurements, and the front-facing camera supplements the geometric feature information of the target object.

[0043] Step 206 collects steering wheel torque, release button signal and steering intervention flag as driver operation data. The driver operation data is collected at a frequency of 100 Hz, capturing the real-time operation intention and intervention behavior of the driver. The steering wheel torque reflects the steering force of the driver, and the release button signal indicates the behavior of the driver actively turning off the assistance function.

[0044] In step 208, the method 200 collects lane geometry data and road surface adhesion coefficient data. The lane geometry data includes lane curvature, lane width and available lateral clearance, collected at a frequency of 20 Hz. The road surface adhesion coefficient is estimated by the tire slip data of the electronic stability program or a machine learning algorithm, updated at a frequency of 10 Hz. The lane curvature describes the degree of road curvature, and the available lateral clearance represents the space range for lateral maneuvering of the vehicle.

[0045] Step 210 synchronizes all collected data in time. This step synchronizes the unified time base of sensors and ECUs through PTP or GNSS PPS, ensuring that data collected at different frequencies have accurate time correspondence. The unified time base synchronization mechanism eliminates the time deviation between different data sources.

[0046] As shown in Figure 2 Step 212 of the method 200 determines whether the data timestamp jitter is less than a preset threshold. The preset threshold is set to 5 milliseconds, and when the data timestamp jitter exceeds the threshold, it indicates that the data synchronization quality does not meet the requirements. This decision step ensures that the data used in subsequent processing has sufficient time accuracy.

[0047] When the timestamp jitter meets the requirements, the method 200 enters step 214 to complete data collection and output synchronized multi-source data for subsequent processing. When the timestamp jitter exceeds the preset threshold, the method 200 performs step 216 to discard out-of-range data and re-collect the affected data. This quality control mechanism ensures the reliability and accuracy of data collection.

[0048] The multi-frequency data collection mechanism sets the corresponding collection frequency according to the characteristics of different data types. Vehicle motion data and driver operation data use a higher frequency of 100 Hz to capture rapidly changing dynamic information. Environment perception data and lane geometry data use medium frequencies of 20-25 Hz and 20 Hz, balancing data accuracy and processing load. Road surface adhesion coefficient data uses a lower frequency of 10 Hz because road conditions change relatively slowly.

[0049] Referring to Figure 3 The embodiment of the present application also provides a driving behavior feature calculation method 300 for feature extraction and processing to realize quantitative analysis of driving behavior. The method 300 uses a 300-second sliding window and a 1-second step for feature statistics, ensuring continuous monitoring and analysis of driving behavior.

[0050] The method 300 starts at step 302, and calculates driving behavior events in a sliding window of a preset time length. The sliding window is set to a time length of 300 seconds, and is updated with a step of 1 second to capture the behavior pattern changes of the driver in different time periods. The sliding window mechanism ensures the time continuity and statistical effectiveness of feature calculation.

[0051] In step 304, the method 300 calculates the average collision time at the triggering moment of the driver's active steering event. This step identifies the occurrence time of the driver's active steering event, records the time-to-collision value between the steering action start moment and the potential collision target, and calculates the statistical mean of these values in the sliding window. The average collision time at the steering event triggering moment reflects the driver's risk perception and reaction timing characteristics.

[0052] Step 306 calculates the peak angular rate and the average angular rate of the steering event. This step monitors the change of the steering wheel angular rate during the driver's steering operation, extracts the maximum angular rate value in a single steering event as the peak angular rate, and calculates the average angular rate of the entire steering process. The peak angular rate represents the intensity of the driver's steering operation, and the average angular rate reflects the overall strength of the steering operation.

[0053] As Figure 3 shown, the method 300 calculates the following time distance based on the forward nearest target in step 308. This step continuously monitors the relative distance and relative speed between the vehicle and the forward nearest target object, and calculates the time-to-collision value under the current motion state as the following time distance. The following time distance statistics reflect the driver's safe distance preference and risk tolerance in the following scene.

[0054] Step 310 calculates the leading collision time before the start of lane change. This step identifies the starting moment of the driver's lane change behavior, records the time-to-collision value between the target lane and the leading vehicle before the start of the lane change action. The leading collision time before the start of lane change reflects the driver's judgment standard for safety gap when making a lane change decision.

[0055] In step 312, the method 300 calculates the average torque threshold and the takeover frequency of the driver when the lateral assistance function is taken over. This step monitors the driver's intervention behavior to the lateral assistance function, records the steering torque value applied by the driver and the frequency of takeover events. The average torque threshold reflects the driver's acceptance of the assistance function, and the takeover frequency characterizes the driver's intervention tendency.

[0056] Step 314 calculates the risk exposure features. The risk exposure features include the cut-in event frequency, the urban and highway driving proportion, the day and night driving label, and the rain and snow weather label. The cut-in event frequency counts the frequency of other vehicles cutting into the lane, the urban and highway driving proportion reflects the complexity of the driving environment, and the day and night driving label and the rain and snow weather label characterize the changes in driving conditions. These risk exposure features are used to correct the confidence of the driver portrait.

[0057] The method 300 performs three times standard deviation truncation and minimum-maximum normalization processing on all features in step 316. The three times standard deviation truncation processing eliminates the influence of abnormal values on feature statistics, and the values exceeding the range of three times standard deviation are truncated. The minimum-maximum normalization processing maps feature values of different dimensions into a unified numerical range, ensuring that each feature has equal weight in subsequent cluster analysis.

[0058] The feature calculation method 300 realizes dynamic monitoring of driving behavior through a sliding window mechanism, and a window length of 300 seconds provides sufficient statistical sample size, and a step size of 1 second ensures the timeliness of feature updating. The introduction of risk exposure features enhances the environmental adaptability of the driver portrait, and improves the accuracy and stability of portrait classification through the correction of confidence mechanism.

[0059] Referring to Figure 4 , the embodiment of the present application also provides a driver portrait clustering and classification method 400, which is used for cluster analysis and probability evaluation, and realizes accurate identification of driver behavior types. The method 400 adopts a soft assignment probability mechanism and a hysteresis threshold control, avoids frequent switching of portrait classification, and provides confidence evaluation.

[0060] The method 400 begins at step 402 by clustering driving behavior features using a hierarchical density-based clustering algorithm or Gaussian mixture model. The hierarchical density-based clustering algorithm can automatically identify clustering regions of different densities and is suitable for processing non-spherical distributed driving behavior data. The Gaussian mixture model models the driving behavior feature space through a linear combination of multiple Gaussian distributions, providing probabilistic clustering results.

[0061] In step 404, the method 400 performs clustering analysis on the driving behavior features. This step takes the average collision time when the steering event is triggered, the steering angle rate, the following time distance, the leading collision time before lane changing, and the takeover frequency as input, and identifies the internal structure of the driving behavior pattern through the clustering algorithm. The clustering analysis process considers the correlation and distribution characteristics between the features, and classifies similar driving behaviors into the same cluster.

[0062] Step 406 outputs soft assignment probabilities to obtain the probability distribution of the driver belonging to the agile, regular, and conservative types. The soft assignment probability mechanism calculates the probability value of each driver sample belonging to each type to form a probability vector (p agg, p reg, p cons). The agile coefficient p agg reflects the possibility of the driver exhibiting aggressive driving behavior, the regular probability p reg represents the matching degree of standard driving behavior, and the conservative coefficient p cons embodies the compliance of cautious driving characteristics.

[0063] As shown in Figure 4 Step 408 of the method 400 selects the type corresponding to the maximum probability as the current driver portrait gear. This step determines the maximum value among the three probability values through the argmax operation and sets the corresponding driver type as the current portrait gear. The probability maximum value principle ensures that the portrait classification is based on the strongest behavior feature matching.

[0064] Step 410 determines whether the difference between the new gear probability and the current gear probability exceeds a preset threshold. This decision step sets the hysteresis threshold to a probability difference of 0.15, and only when the difference between the new gear probability and the current gear probability exceeds 0.15 does it trigger the portrait switching. The hysteresis threshold mechanism prevents frequent portrait switching caused by small fluctuations in probability values and improves the stability of portrait classification.

[0065] When the probability difference exceeds the preset threshold, the method 400 performs step 412 to switch the portrait and calculate the driver portrait confidence. The portrait switching process updates the current driver type label and recalculates the portrait confidence to reflect the reliability of the new classification. When the probability difference does not exceed the preset threshold, the method 400 performs step 414 to maintain the current portrait gear and calculate the driver portrait confidence.

[0066] The driver profile confidence is calculated based on the driving event count and the outlier rate. The confidence calculation formula is Conf = min(1, event count / 200) x (1-outlier rate), where the event count reflects the sufficiency of the statistical sample, and the outlier rate represents the consistency of the data. When the driving event count reaches 200 times, the sample sufficiency factor reaches a maximum value of 1. The outlier rate is calculated by identifying the proportion of data points that deviate from the normal behavior pattern.

[0067] When the confidence is lower than a preset threshold of 0.5, the influence weight of the profile on the threshold adjustment is reduced. This mechanism is realized through the adjustment of the scaling factor a, when the confidence Conf < 0.5, the adjustment coefficient a is corrected according to the formula a = 0.5 x a_raw + 0.5, where a_raw is the original adjustment coefficient. The confidence weight adjustment ensures that the low reliability profile classification does not have an excessive impact on the AES trigger threshold.

[0068] The HDBSCAN algorithm, as a specific implementation based on hierarchical density-based clustering algorithm, shows stronger robustness to multi-modal and outlier data. HDBSCAN algorithm can identify clustering regions of different densities and automatically process noise points by constructing minimum spanning tree and hierarchical clustering structure. This algorithm has an advantage in handling outliers and multi-peak distribution in driving behavior data, improving the accuracy and robustness of profile classification.

[0069] Referring to Figure 5 The embodiment of the present application also provides a personalized adjustment coefficient determination method 500 for calculating and dynamically modulating the personalized adjustment of the AES trigger threshold. This method 500 uses the comprehensive consideration of soft profile probability distribution, driver acceptance and real-time risk score to ensure that the adjustment coefficient reflects the personalized characteristics of the driver and meets the safety requirements.

[0070] The method 500 starts from step 502, and calculates the original adjustment coefficient based on the soft profile probability distribution. The original adjustment coefficient is calculated according to the formula a_raw = k · [1 + β · (p_cons - p_agg)], where k is the basic adjustment coefficient, β is the weight parameter, the recommended value is 0.3, p_cons is the conservative coefficient, and p_agg is the agile coefficient. The difference between the original adjustment coefficient and the conservative coefficient and the agile coefficient is proportional, when the driver shows more conservative characteristics, the adjustment coefficient increases, and when the driver shows more agile characteristics, the adjustment coefficient decreases.

[0071] In step 504, the method 500 calculates the driver's acceptance of AES. Driver acceptance is calculated based on the cancellation rate after AES triggering, by statistically assessing the proportion of AES interventions canceled by the driver within N seconds of triggering. The cancellation rate reflects the driver's satisfaction with the timing and intensity of AES intervention; a higher cancellation rate indicates that the driver perceives AES intervention as too frequent or inappropriate.

[0072] Step 506 adjusts the original adjustment coefficient based on driver acceptance. This step is calculated using the acceptance rate A = 1 - FalseCancelRate, where FalseCancelRate is the error cancellation rate. The adjusted coefficient takes into account the driver's subjective feelings and user experience; when acceptance is high, the original adjustment coefficient is maintained, and when acceptance is low, the coefficient is adjusted appropriately to reduce unnecessary intervention.

[0073] like Figure 5 As shown, in step 508, the method 500 dynamically modulates the adjustment coefficient based on the real-time risk score. The real-time risk score R comprehensively considers factors such as relative speed, cut-in prediction, available lateral margin, road curvature, and pavement adhesion coefficient, and the score range is 0 to 1. When the risk score is higher than the preset upper limit of 0.7, the adjustment coefficient is increased to no less than 1.0 to ensure that AES can intervene in a timely manner under high-risk conditions. When the risk score is lower than the preset lower limit of 0.3 and the takeover rate is low, the adjustment coefficient is appropriately reduced to avoid excessive intervention in low-risk environments.

[0074] Step 510 limits the adjustment factor to a preset range. The adjustment factor is limited to the range of 0.8 to 1.2 to ensure that personalized adjustments do not deviate too far from the safety boundary. This range limitation mechanism prevents potential safety risks from extreme personalized adjustments while reserving sufficient adjustment space to reflect the driver's personalized needs.

[0075] In step 512, method 500 determines whether the mileage has reached a preset mileage threshold or whether the cumulative number of AES triggers has reached a preset number threshold. The preset mileage threshold is set to 1000 kilometers, and the preset number threshold is set to 5 cumulative AES triggers. This decision step determines whether the adjustment coefficient needs to be updated based on new driving behavior data, and maintains consistency between the adjustment coefficient and the driver's current behavioral characteristics through a periodic update mechanism.

[0076] When the driving mileage reaches the preset mileage threshold or the number of accumulated AES trigger times reaches the preset number threshold, the method 500 performs step 514 to recalculate the adjustment coefficient based on the driving behavior data in the current evaluation period. The recalculation process adopts a comprehensive strategy of exponential moving average and user acceptance, and the update formula is k_new = clip(0.8, 1.2, k_old· (1 + γ·d) · (1 + η·(A-0.5))), where d=(TTC_steer_avg – TTC_base) / TTC_base, γ is recommended to be 0.4, and η is recommended to be 0.2. When the conditions are not met, the method 500 performs step 516 to keep the current adjustment coefficient unchanged.

[0077] The personalized adjustment coefficient determination method 500 balances the personalized needs of the driver and the safety requirements through a multi-level modulation mechanism. The soft portrait probability distribution provides basic personalized features, the driver acceptance correction optimizes the user experience, the real-time risk score dynamic modulation guarantees the maintenance of safety, and the range limitation mechanism prevents the risk of extreme adjustment.

[0078] In some embodiments, trajectory generation adopts a three-segment quintic polynomial or a minimum curvature variational trajectory method. The three-segment quintic polynomial trajectory constitutes a complete obstacle avoidance trajectory by three consecutive quintic polynomial segments, each of which maintains the continuity of position, velocity and acceleration at the connection point. The minimum curvature variational trajectory generates a smooth obstacle avoidance path by optimizing the curvature rate to minimize the objective function, reducing the lateral dynamic load of the vehicle.

[0079] The lateral acceleration and lateral jerk are constrained and controlled during the trajectory generation process. The lateral acceleration constraint ensures that the vehicle does not exceed the tire adhesion limit and the comfort requirement of the occupants during obstacle avoidance, and the lateral jerk constraint is limited to below 3 m / s³ to avoid excessive acceleration rate that may adversely affect the stability of the vehicle and the comfort of the occupants.

[0080] In some embodiments, a smooth takeover mechanism is adopted when the AES exits to return control to the driver. The steering torque is smoothly transitioned within a time window of 200-300 milliseconds, and a T_blend torque blending strategy is used to achieve seamless switching from AES control to driver control. This smooth takeover mechanism avoids the discomfort of the driver and the dynamic disturbance of the vehicle that may be caused by sudden torque changes, improving the user acceptance and safety of the AES function.

[0081] Reference Figure 6This invention also provides a method 600 for calculating the minimum safe collision time threshold, which ensures that the AES maintains basic safety performance under any personalized adjustments through multi-step calculations. This method 600 calculates the minimum safe threshold that cannot be exceeded through personalization based on vehicle kinematics principles and safety constraints.

[0082] The method 600 begins at step 602, which calculates the minimum displacement required for lateral obstacle avoidance. The minimum lateral obstacle avoidance displacement s_lat is determined based on vehicle geometry and safety margins, taking into account vehicle width, target object width, and the safety clearance required during obstacle avoidance. The minimum displacement calculation ensures that the vehicle can completely avoid the obstacle without collision, providing a spatial basis for subsequent time threshold calculations.

[0083] In step 604, the method 600 determines the maximum permissible lateral acceleration. Determining the maximum lateral acceleration requires comprehensive consideration of the vehicle's physical limits and occupant comfort requirements, ensuring that obstacle avoidance maneuvers effectively prevent collisions without adversely affecting vehicle stability and occupant safety.

[0084] Step 606 takes the smaller value between the product of the road surface adhesion coefficient and gravitational acceleration, and the upper limit of the comfort lateral acceleration. This step calculates the maximum lateral acceleration using the formula a_lat,max = min(μ·g, a_lat, upper limit of comfort), where μ is the road surface adhesion coefficient and g is the gravitational acceleration. In high-speed scenarios, the upper limit of the comfort lateral acceleration is taken in the range of 2.0-2.5 m / s², prioritizing comfort limits to ensure the occupant's riding experience. In low-speed scenarios, road surface adhesion limits typically become the constraint condition.

[0085] like Figure 6 As shown, in step 608, method 600 calculates sensor perception delay, computation processing delay, and actuator response delay. Sensor perception delay includes the data acquisition and preprocessing time of millimeter-wave radar and cameras; computation processing delay covers the calculation time for collision risk assessment and trajectory planning; and actuator response delay includes the response time of electric power steering. The cumulative effect of these delays directly affects the actual response capability of AES.

[0086] Step 610 calculates the minimum safe collision time threshold based on the minimum displacement, maximum lateral acceleration, and time delay. The calculation of the minimum safe TTC_min requires a time Δt that satisfies the kinematic constraint s_lat ≤ 0.5·a_lat,max·Δt², where the initial lateral velocity v_lat is assumed to be approximately 0. The minimum safe collision time threshold TTC_min = Δt + τ_compute + τ_actuation, which includes all delays from sensing, computation, and execution.

[0087] The method 600 compares the user personalized time-to-collision threshold with the minimum safe time-to-collision threshold in step 612. This comparison step ensures that the personalized adjustment does not break the safety boundary, when the user personalized threshold is lower than the minimum safe threshold, the minimum safe threshold is adopted as a protection mechanism.

[0088] Step 614 selects the larger value as the final triggering threshold of the AES. With the selection mechanism of max(TTC_user, TTC_min), it is ensured that the final triggering threshold embodies both the personalized characteristics of the driver and meets the basic safety requirements. This safety clamping mechanism is a core component of the functional safety design of the AES.

[0089] In some embodiments, the boundary scenario processing strategy formulates special enabling and inhibiting conditions for special driving environments. In the close-cut-in scenario, when the other vehicle suddenly cuts into the lane and the distance is extremely close, the AES adopts a more stringent minimum safe threshold and increases the lateral acceleration limit. In the S-type continuous lane-changing scenario, when the front vehicle is detected to perform a continuous lane-changing maneuver, the AES prolongs the observation time and reduces the triggering sensitivity to avoid misjudgment.

[0090] In the cone barrel work area scenario, the AES identifies the cone barrel sign of the construction area and adjusts the obstacle avoidance strategy, adopts more conservative lateral maneuvering parameters to ensure safe passing through the construction area. In the low road adhesion scenario, when the road adhesion coefficient μ is lower than 0.35, the AES reduces the maximum lateral acceleration limit and prolongs the minimum safe time-to-collision threshold, and in rainy and snowy weather, the adhesion coefficient is further reduced to 0.25 and the AES is limited to low-speed scenarios.

[0091] The on-ramp merging scenario involves complex traffic flow convergence, and the AES enhances the monitoring of the side rear vehicle in this scenario and adjusts the curvature constraint of the obstacle avoidance trajectory. In the intersection pedestrian crossing scenario, the AES prioritizes pedestrian protection and adopts more aggressive obstacle avoidance parameters, while coordinating the arbitration relationship with the AEB braking function.

[0092] In the stationary large truck scenario, the AES identifies large stationary targets and evaluates the available lateral maneuvering space, and when the lateral margin is insufficient, the AES is inhibited and relies on the AEB braking function. In the tunnel entrance strong light scenario, the camera sensor may be affected by light, and the AES reduces the dependence on the visual sensor and increases the weight of the millimeter wave radar.

[0093] The night rain and fog scenario combines the dual challenges of low visibility and low adhesion, and the AES adopts the most conservative triggering strategy and prolongs the system response time in this scenario to adapt to the poor sensing conditions. The driving door target scenario involves sudden obstacles, and the AES responds to such sudden situations through fast target recognition and emergency obstacle avoidance trajectory generation.

[0094] The processing strategy of these boundary scenarios is realized by special enabling and inhibiting switches, and each type of scenario corresponds to a specific safety minimum threshold table. In the mass production stage, these parameters are calibrated and optimized in the platform parameter table to ensure that the AES can provide reliable safety protection in various complex driving environments. The identification and classification of boundary scenarios are based on multi-sensor fusion and machine learning algorithms, and the accuracy of scenario identification and the appropriateness of response are improved through continuous data accumulation and algorithm optimization.

[0095] The embodiment of the application also discloses an AES trigger threshold adjustment system based on driver portrait.

[0096] Reference Figure 7 An AES trigger threshold adjustment system 700 based on driver portrait adopts a modular architecture to realize personalized threshold adjustment function. The AES trigger threshold adjustment system 700 realizes the complete processing flow from data acquisition to final threshold determination through the cooperative work of five core modules, and integrates a function manager, a man-machine interface and a functional safety mechanism.

[0097] The AES trigger threshold adjustment system 700 comprises a data acquisition module 702 for acquiring vehicle motion data, environmental perception data and driver operation data. The data acquisition module 702 obtains vehicle motion information such as vehicle longitudinal speed, lateral acceleration, yaw angular velocity, steering angle and steering angular rate through multi-sensor fusion, simultaneously acquires millimeter wave radar data in front, side and rear and front-view camera data as environmental perception information, and monitors driver operation information such as steering torque, release button signal and steering intervention flag.

[0098] The AES trigger threshold adjustment system 700 further comprises a feature calculation module 704 for calculating driving behavior features based on the acquired data. The feature calculation module 704 receives the original data from the data acquisition module 702, and calculates driving behavior features such as collision time, steering angular rate and following distance at the time of steering event triggering. The feature calculation module 704 performs feature statistics in a preset time length sliding window, and performs three standard deviation truncation and minimum-maximum normalization processing on all features.

[0099] As Figure 7 The AES trigger threshold adjustment system 700 further comprises a portrait construction module 706 for performing clustering analysis on the driving behavior features by using an unsupervised learning algorithm to construct a driver portrait. The portrait construction module 706 receives the processed feature data from the feature calculation module 704, and performs clustering analysis by using a hierarchical density-based clustering algorithm or a Gaussian mixture model. The portrait construction module 706 classifies drivers into agile type, regular type or conservative type, and outputs soft assignment probability to obtain the probability distribution of drivers belonging to different types.

[0100] The AES trigger threshold adjustment system 700 comprises a threshold calculation module 708 for determining a personalized adjustment coefficient based on the driver profile and calculating a user personalized collision time threshold based on the personalized adjustment coefficient and a preset base threshold. The threshold calculation module 708 receives the driver profile information from the profile construction module 706, calculates an original adjustment coefficient based on the soft profile probability distribution, and modifies the adjustment coefficient according to the driver's acceptance of AES. The threshold calculation module 708 also dynamically modulates the adjustment coefficient based on the real-time risk score and limits the adjustment coefficient within a preset range.

[0101] The AES trigger threshold adjustment system 700 comprises a threshold determination module 710 for comparing the user personalized collision time threshold with a minimum safety collision time threshold and taking the larger value as the final trigger threshold of AES. The threshold determination module 710 receives the user personalized collision time threshold from the threshold calculation module 708, while calculating the minimum safety collision time threshold based on the minimum displacement required for lateral obstacle avoidance and the maximum lateral acceleration. The threshold determination module 710 ensures that the final trigger threshold reflects the personalized characteristics of the driver and meets the basic safety requirements through the comparison mechanism.

[0102] The data flow between the modules forms a sequential processing architecture, the output of the data acquisition module 702 serves as the input of the feature calculation module 704, the processing result of the feature calculation module 704 is transmitted to the profile construction module 706 for cluster analysis, the driver profile information of the profile construction module 706 flows to the threshold calculation module 708 for personalized adjustment coefficient calculation, and the output of the threshold calculation module 708 is finally transmitted to the threshold determination module 710 for final threshold determination. The sequential data flow ensures that each module can perform corresponding processing and calculation based on the output of the previous module.

[0103] The AES trigger threshold adjustment system 700 integrates a function manager for arbitration with AEB, LKA / ELK, EPS / ESP. The function manager implements an arbitration mechanism that AES intervention priority is lower than AEB braking, when AEB detects emergency braking demand, AES function is suppressed to avoid conflict between lateral maneuvering and longitudinal braking. The function manager also implements a mutual exclusion mechanism for AES and LKA to share lateral control, ensuring that only one lateral control function is active at the same time, avoiding control conflicts between multiple lateral auxiliary functions.

[0104] In some embodiments, the AES trigger threshold adjustment system 700 further includes an HMI display function that displays the current profile level and an AES sensitivity bar. The HMI interface displays the current profile type of the driver, including agile, regular, or conservative, through icons or colors. The AES sensitivity bar visually shows the current trigger threshold setting, allowing the driver to intuitively understand the intervention sensitivity of the AES. The HMI display function also provides a personalized reset and temporary shutdown option, allowing the driver to restore the profile and adjustment coefficient to the initial state through the personalized reset function, and temporarily disable the AES function in specific situations through the temporary shutdown option.

[0105] The AES trigger threshold adjustment system 700 described in the embodiments of the present application realizes ASIL-B / C functional safety decomposition, establishes perception-planning-control link monitoring and sensor self-checking capability. Perception link monitoring includes real-time state detection of millimeter wave radar and camera sensors, planning link monitoring covers collision risk assessment and running state of trajectory planning algorithm, and control link monitoring ensures normal execution response of electric power steering. The sensor self-checking capability is realized through the heartbeat signal mechanism. When the sensor self-checking fails, the system enters a degraded mode, only retaining the AEB or LKA function and prompting the driver to take over.

[0106] At the same time, the AES trigger threshold adjustment system 700 records each AES trigger event and writes detailed event logs in CSV or Protobuf format for regression analysis. The event log contains time stamp, driver profile category, personalized adjustment coefficient, user personalized collision time threshold, minimum safe collision time threshold, real-time risk score, and exit reason, etc. Key information. This log recording mechanism provides data support for subsequent algorithm optimization and performance evaluation, and identifies the improvement direction and parameter tuning requirements of the AES function through regression analysis.

[0107] In some embodiments, when the AES is triggered and canceled by the driver, the HMI interface pops up a light prompt and records the cancellation reason option. The cancellation reason option includes "intervention too early", "intervention too frequent", "no need to intervene", etc. The feedback information of the driver is used for the estimation of acceptance A and the optimization of the adjustment coefficient. This user feedback mechanism enhances the self-learning ability of the system, and improves the accuracy of personalized adjustment by accumulating the subjective evaluation of the driver.

[0108] The modular architecture design makes the AES trigger threshold adjustment system 700 have good scalability and maintainability. The independence of each module allows individual algorithm optimization and function upgrade, and the standardized interface between modules ensures the stability of system integration. The architecture also supports OTA parameter update, which realizes online optimization of adjustment coefficient and threshold parameters through remote parameter delivery, and adapts to the driving feature differences of different regions and user groups.

[0109] The application further discloses a readable storage medium.

[0110] A readable storage medium stores a computer program, and the computer program is executed by a processor to implement steps of the AES trigger threshold adjustment method in any one of the above embodiments. The computer readable storage medium can include any entity or device capable of carrying the computer program, a recording medium, a U disk, a mobile hard disk, 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 can be in a source code form, an object code form, an executable file, or some intermediate form, etc. The computer readable storage medium can include any entity or device capable of carrying the computer program code, a recording medium, a U disk, a mobile hard disk, 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.

[0111] Any process or method descriptions or any other descriptions herein can be understood as representing any manner of executing the functionality, unless specific limitations are explicitly contained therein. The functions of the various elements shown in the figures can be provided through the use of dedicated hardware as well as hardware capable of executing software in association with appropriate software. When provided by a processor, the functions can be provided by a single dedicated processor, by a single shared processor, or by a plurality of individual processors, some of which can be shared.

[0112] The logic and / or steps represented in the flowcharts and / or otherwise described herein, for example, can be considered as a list of instructions to implement the logic function, and the scope of the preferred embodiments of the present application includes additional implementations in which the functions are performed in an order different from that shown or discussed, including substantially concurrently or in reverse order, as will be understood by those having ordinary skill in the art to which embodiments of the present application pertain.

[0113] The above embodiments are only used to illustrate the technical solutions of the present application, not limit it; although the above-mentioned embodiments of the present application are described in detail, those skilled in the art should understand that the technical solutions recorded in the above-mentioned embodiments can be modified, or some technical features can be replaced by equivalent; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for adjusting the AES trigger threshold based on driver profile, characterized in that, include: Collect vehicle motion data, environmental perception data, and driver operation data; Based on the collected data, driving behavior characteristics are calculated, including the collision time, steering angle rate, and following distance when a steering event is triggered. Unsupervised learning algorithms are used to perform cluster analysis on the driving behavior features to construct driver profiles, which classify drivers into agile, conventional, or conservative types. Based on the driver profile, a personalized adjustment coefficient is determined, and a personalized collision time threshold for the user is calculated based on the personalized adjustment coefficient and a preset base threshold; and The user-personalized collision time threshold is compared with the minimum safe collision time threshold, and the larger value is taken as the final trigger threshold of the AES system.

2. The AES trigger threshold adjustment method according to claim 1, characterized in that, The collected vehicle motion data, environmental perception data, and driver operation data include: Collect vehicle longitudinal velocity, lateral acceleration, yaw rate, steering angle, and steering angular rate as vehicle motion data; Collect millimeter-wave radar data from the front, sides, and rear, as well as data from the forward-looking camera, to obtain the relative distance, relative velocity, and cross-sectional width of the target object as environmental perception data; Collect steering wheel torque, release button signal, and steering intervention indicator as driver operation data; and All collected data are synchronized in time to ensure that the jitter of the data timestamps is less than a preset threshold.

3. The AES trigger threshold adjustment method according to claim 2, characterized in that, It also includes collecting lane geometry data and road surface adhesion coefficient data. The lane geometry data includes lane curvature, lane width, and available lateral margin. The road surface adhesion coefficient is estimated using tire slip data from the electronic stability program system or machine learning algorithms.

4. The AES trigger threshold adjustment method according to claim 1, characterized in that, The calculation of driving behavior characteristics based on the collected data includes: Within a sliding window of preset time length, the average collision time at the moment the driver's active steering event is triggered is statistically analyzed; Calculate the peak angular rate and average angular rate of the steering event; The statistics are based on the following distance to the nearest forward target; Calculate the lead-out collision time before the lane change begins; Statistical analysis of the average torque threshold and frequency of driver intervention during lateral assist system takeover; and All features were truncated to three standard deviations and subjected to min-max normalization.

5. The AES trigger threshold adjustment method according to claim 4, characterized in that, It also includes calculating risk exposure features, which include the frequency of entry events, the proportion of driving on urban and highway roads, day and night driving labels, and rain and snow weather labels, to adjust the confidence level of the driver profile.

6. The AES trigger threshold adjustment method according to claim 1, characterized in that, The clustering analysis of the driving behavior features using an unsupervised learning algorithm includes: Hierarchical density-based clustering algorithms or Gaussian mixture models are used to cluster driving behavior features; Output the soft assignment probability to obtain the probability distribution of whether the driver belongs to the agile, conventional, or conservative type; Select the type corresponding to the highest probability as the current driver profile level; and Set a hysteresis threshold so that profile switching only occurs when the difference between the probability of the new gear and the probability of the current gear exceeds a preset threshold.

7. The AES trigger threshold adjustment method according to claim 6, characterized in that, It also includes calculating the confidence level of the driver profile, which is calculated based on the number of driving events and the outlier rate. When the confidence level is lower than a preset threshold, the influence weight of the profile on the threshold adjustment is reduced.

8. The AES trigger threshold adjustment method according to claim 1, characterized in that, The process of determining the personalized adjustment coefficient based on the driver profile includes: The original adjustment coefficient is calculated based on the probability distribution of the soft profile, and the original adjustment coefficient is proportional to the difference between the conservative coefficient and the agile coefficient; The original adjustment coefficient is corrected based on the driver's acceptance of the system, and the acceptance is calculated based on the cancellation rate after AES is triggered; The adjustment coefficient is dynamically modulated based on real-time risk scoring. The adjustment coefficient increases when the risk score exceeds a preset upper limit and decreases when the risk score falls below a preset lower limit and the takeover rate is low. The adjustment coefficient is limited to a preset range.

9. The AES trigger threshold adjustment method according to claim 8, characterized in that, It also includes periodically updating the personalized adjustment coefficient. When the driving mileage reaches a preset mileage threshold or the cumulative number of AES triggers reaches a preset number threshold, the adjustment coefficient is recalculated based on the driving behavior data within the current evaluation period.

10. The AES trigger threshold adjustment method according to claim 1, characterized in that, The minimum safe collision time threshold is calculated based on the minimum displacement required for lateral obstacle avoidance and the maximum permissible lateral acceleration. The maximum lateral acceleration is the smaller of the product of the road surface adhesion coefficient and the gravitational acceleration and the upper limit of the comfort lateral acceleration. The minimum safe collision time threshold also includes sensor perception delay, calculation and processing delay and actuator response delay.

11. An AES trigger threshold adjustment system based on driver profile, characterized in that, include: The data acquisition module is used to collect vehicle motion data, environmental perception data, and driver operation data. The feature calculation module is used to calculate driving behavior features based on the collected data. The driving behavior features include the collision time when the steering event is triggered, the steering angle rate, and the following distance. The driver profile building module is used to perform cluster analysis on the driving behavior features using an unsupervised learning algorithm to build a driver profile, which classifies drivers into agile, conventional, or conservative types. The threshold calculation module is used to determine the personalized adjustment coefficient based on the driver profile, and to calculate the user's personalized collision time threshold based on the personalized adjustment coefficient and the preset basic threshold. as well as The threshold determination module is used to compare the user-personalized collision time threshold with the minimum safe collision time threshold, and take the larger value as the final trigger threshold of the AES system.

12. A readable storage medium, characterized in that, The readable storage medium stores computer instructions that, when executed by a processor, implement the AES trigger threshold adjustment method as described in any one of claims 1-10.

Images