A risk assessment method, system and vehicle

Abstract

This application relates to a risk assessment method, system, and vehicle, belonging to the field of vehicle control technology. It includes: real-time acquisition of data on multiple road rage triggering factors; inputting the acquired triggering factor data into a pre-trained scoring model, which outputs the factor-induced score for each triggering factor; calculating the comprehensive triggering score and corresponding dynamic weight for each type of triggering factor; and performing a weighted summation operation on the comprehensive triggering score and corresponding dynamic weight for each type of triggering factor to obtain the current road rage risk index. This invention significantly improves the accuracy of road rage risk assessment.

Description

Technical Field

[0001] This application relates to the technical field of vehicle control, and more particularly to a risk assessment method, system, and vehicle. Background Technology

[0002] With the continuous growth of motor vehicle ownership, the pressure on urban roads is constantly increasing, and road rage has become an important risk factor affecting road traffic safety.

[0003] In existing technologies, the risk assessment of road rage in drivers is usually based on scoring each road rage triggering factor with fixed weights. This calculation method can lead to a deviation between the assessment results and the actual emotional fluctuations of drivers, making it difficult to accurately reflect the true level of road rage risk. Summary of the Invention

[0004] To address the discrepancy between road rage assessment results and actual driver emotional fluctuations in existing technologies, this application provides a risk assessment method, system, and vehicle. It employs a technical approach that involves real-time data collection of multiple triggering factors, determining triggering factors and outputting factor scores through a pre-trained model, dynamically calculating the comprehensive triggering score and weights, and weighted summation to obtain a risk index. This solves the technical problem of accurately predicting road rage risk in advance.

[0005] Firstly, a risk assessment method is provided for vehicles, and this assessment method includes: Real-time data collection of multiple road rage triggering factors; each triggering factor includes at least one inducing agent; The collected triggering factor data is input into the pre-trained scoring model, which determines the existing triggering factors and outputs the factor triggering score corresponding to each triggering factor. Based on the existing inducing factors and their corresponding inducing scores, the comprehensive inducing scores and corresponding dynamic weights of each inducing factor are calculated. Based on the comprehensive induced scores of various triggering factors and their corresponding dynamic weights, a weighted summation is performed to obtain the current road rage risk index.

[0006] In this embodiment, a pre-trained scoring model is used to identify triggering factors and output their triggering scores. Based on this, the comprehensive triggering score of each type of triggering factor and its weight in the overall road rage risk composition are calculated in real time according to the triggering factors with different influence scores at the current moment. Then, the road rage risk index is obtained by weighted summation, so that the final output risk index can more accurately reflect the emotional risks faced by drivers in complex and dynamic real driving environments. This significantly improves the accuracy and scenario adaptability of road rage risk prediction, thereby providing more reliable data support for subsequent early warning and intervention measures.

[0007] In conjunction with the first aspect, pre-training in certain implementations of the first aspect includes: The scoring model learns the mapping relationship between triggering factors and road rage behavior; Output the factor evoked scores for each evoked factor, including: The scoring model is based on mapping relationships to predict the road rage behavior that each existing triggering factor may induce. Based on the road rage behaviors that may be induced by triggering factors, and the preset behavior scores for each type of road rage behavior, the factor induction score corresponding to each triggering factor is determined.

[0008] In this embodiment, during the pre-training phase, the scoring model learns the inherent correlation between different combinations of triggering factors and specific road rage behaviors. During actual operation, based on the established mapping relationship, the scoring model predicts the most likely type of road rage behavior induced by each triggering factor, and then calculates the factor-induced score corresponding to each triggering factor based on pre-set road rage behavior scores. This process ensures that the factor-induced scores output by the model objectively reflect the actual tendency of that factor to trigger road rage in similar scenarios, effectively improving the rationality and accuracy of road rage risk assessment results.

[0009] In conjunction with the first aspect, in some implementations of the first aspect, when the number of road rage behaviors that each triggering factor may induce is greater than the first number; Determine the factor evoked score for each evoked factor, including: Calculate the probability of each type of rage disorder behavior that may be induced by each triggering factor; The second most frequent road rage behavior is selected, and the average of its corresponding preset scores is calculated as the factor induction score for that inducing factor; the second number is less than the first number.

[0010] In this embodiment, when a single triggering factor can induce a large number of road rage behaviors, the probability of occurrence of each road rage behavior is first calculated, and a specified number of behaviors with the highest probability are selected. Then, the average value of the preset scores of these behaviors is calculated and used as the score of the corresponding triggering factor. This can effectively eliminate interference from low-probability and atypical road rage behaviors, making the factor-induced scores output by the scoring model more representative and reliable, and providing more accurate basic data for subsequent comprehensive score calculation and dynamic weight allocation.

[0011] In conjunction with the first aspect, some implementations of the first aspect enable the rating model to learn the mapping relationship between triggering factors and road rage behavior, including: Acquire historical driving behavior data; historical driving behavior data includes the driver's road rage behavior and the sequence of triggering factors present at the time of the behavior; The scoring model is iteratively trained using historical driving behavior data. During the training process, the triggering factor sequence is used as the input feature of the scoring model, and the road rage behavior corresponding to the triggering factor sequence is used as the output target of the scoring model, so that the scoring model learns the mapping relationship from triggering factors to road rage behavior.

[0012] In this embodiment, by acquiring historical driving behavior data containing the driver's road rage behavior and the corresponding triggering factor sequence, the scoring model is iteratively trained. This allows the model to fully learn the intrinsic relationship between triggering factors and road rage behavior in real driving scenarios, thereby establishing a mapping relationship that fits the actual driving pattern. This avoids the one-sidedness and bias caused by artificially set rules, making the scoring model's judgment basis for triggering behavior more objective and reliable.

[0013] In conjunction with the first aspect, in certain implementations of the first aspect, the currently existing inducing factors are identified, including: Compare each input trigger factor data with the preset anomaly detection threshold for that trigger factor; If the data of a certain triggering factor exceeds its corresponding anomaly detection threshold, then the triggering factor is determined to be a currently existing triggering factor.

[0014] In this embodiment, by comparing the input data of various triggering factors with the corresponding preset anomaly judgment thresholds, and determining the triggering factors whose data exceeds the thresholds as the currently existing triggering factors, it is possible to accurately screen out the truly effective triggering factors with a unified standard, exclude invalid data that has not reached the triggering level, and ensure that the triggering factors participating in subsequent score calculation and weight allocation are all triggering factors that can cause road rage risk, thereby further improving the accuracy of the road rage risk index calculation results.

[0015] In conjunction with the first aspect, in certain implementations of the first aspect, the comprehensive induced score of various induced factors is calculated, including: Calculate the average factor induced score corresponding to the currently existing induced factors in each category of induced factors, and use it as the comprehensive induced score for that category of induced factors.

[0016] In this embodiment, the average value of the factor induced score corresponding to the current induced factor under each type of induced factor is calculated, and this average value is used as the comprehensive induced score of that type of induced factor. This makes the comprehensive induced score of each type of induced factor no longer affected by the number of induced factors within it, effectively eliminating the calculation bias caused by the difference in the number of induced factors under different categories, and providing an accurate data basis for the subsequent weighted calculation of the road rage risk index.

[0017] In conjunction with the first aspect, in certain implementations of the first aspect, the dynamic weights corresponding to various inducing factors are calculated, including: Calculate the sum of the factor induction scores corresponding to the currently existing induction factors in each category, and use it as the total score for that category of induction factors; Calculate the sum of the factor induction scores corresponding to the currently existing inducing factors among various inducing factors, and use it as the total score of various inducing factors; The ratio of the total score of each type of triggering factor to the total score of all types of triggering factors is calculated and used as the dynamic weight corresponding to that type of triggering factor.

[0018] In this embodiment, the total score for each category is obtained by summing the scores of the current triggering factors under each category. The ratio of the total score of a single category to the total score of all categories is used as the dynamic weight of that category. This can automatically adjust the proportion of each category according to the risk intensity of the triggering factors triggered in real time, overcoming the defect of fixed weights in the prior art that cannot adapt to changes in the scenario. This allows the weight allocation to be adaptively adjusted in real time according to changes in the current driving scenario and the driver's state, thereby making the final road rage risk index closer to the actual driving situation and further improving the accuracy of the assessment results.

[0019] In conjunction with the first aspect, in some implementations of the first aspect, the evaluation method further includes...

[0020] The road rage risk index is compared with a preset threshold range to determine the current road rage risk level. Based on the risk level of road rage, corresponding intervention actions are triggered.

[0021] In this embodiment, after calculating the road rage risk index, the index is further compared with multiple preset threshold ranges to determine the specific level of the current risk. Based on the level, a matching intervention operation is triggered, which achieves a precise match between the risk level and the intensity of intervention. This avoids excessive intervention that could lead to a decline in user experience, while ensuring the timeliness and effectiveness of intervention measures in high-risk scenarios.

[0022] Secondly, a risk assessment system is provided for use in vehicles, the assessment system comprising: The data acquisition module is used to collect data on multiple road rage triggering factors in real time; each triggering factor contains at least one triggering agent. The scoring module is used to receive triggering factor data, determine the existing triggering factors through a pre-trained scoring model, and output the factor triggering score corresponding to each triggering factor. The weight and score calculation module is used to calculate the comprehensive induction score and corresponding dynamic weight of each induction factor based on the existing induction factors and their corresponding factor induction scores. The risk index calculation module is used to perform a weighted summation calculation based on the comprehensive induced scores of various inducing factors and their corresponding dynamic weights to obtain the current road rage risk index.

[0023] Thirdly, a vehicle is provided, the vehicle including a memory, a processor, and a computer program stored in the memory and executable on the processor; when the processor executes the computer program, it causes the processor to implement the risk assessment method of the first aspect described above, or the vehicle includes the risk assessment system of the second aspect described above.

[0024] Fourthly, a computer program product is provided, comprising: computer program code, which, when run on a computer, causes the computer to perform the risk assessment method described in the first aspect.

[0025] Fifthly, a computer-readable storage medium is provided, which stores computer program code that is executed by one or more processors, such that when the computer program code is run on the processor, a device including one or more processors performs the risk assessment method of the first aspect described above.

[0026] The beneficial effects of the technical solutions provided in this application include at least the following: This application provides a risk assessment method, system, and vehicle. A scoring model is trained to score each triggering factor in real time, ensuring that the output factor triggering score reflects the intensity of each triggering factor in the current driving scenario. Then, based on the model's output factor triggering score, the comprehensive score and dynamic weight of each triggering factor are dynamically calculated, allowing the weight allocation of each triggering factor to adaptively adjust with changes in the current driving scenario, driver state, and real-time road conditions. Finally, the comprehensive score of each factor is weighted based on this dynamic weight, enabling the calculated risk index to more accurately reflect the driver's immediate road rage tendency. This solves the problem of existing technologies using fixed-weight scoring, which cannot adapt to changes in driving scenarios, driver state, and real-time road conditions, leading to inaccurate road rage risk prediction and a large deviation from the driver's actual emotions.

[0027] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application, it can be implemented according to the contents of the specification. In order to make the above and other objects, features and advantages of this application more apparent, specific embodiments of this application are given below. Attached Figure Description

[0028] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the 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.

[0029] Figure 1 This is a flowchart illustrating the risk assessment method of an embodiment of this application.

[0030] Figure 2 This is a flowchart illustrating the method for outputting factor-induced scores corresponding to each inducing factor in an embodiment of this application.

[0031] Figure 3 This is a flowchart illustrating the method for calculating the comprehensive induced score and dynamic weight in an embodiment of this application.

[0032] Figure 4 This is a schematic diagram of the architecture of the risk assessment system in an embodiment of this application.

[0033] In the diagram, 100 is the data acquisition module, 200 is the scoring module, 300 is the weight and score calculation module, and 400 is the risk index calculation module. Detailed Implementation

[0034] To make the technical problems to be solved, the technical solutions, and the beneficial effects of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present invention and are not intended to limit the present invention.

[0035] The prefixes such as "first" and "second" used in this application embodiment are merely for distinguishing different descriptive objects and do not limit the position, order, priority, quantity, or content of the described objects. The use of ordinal numbers and other prefixes used to distinguish descriptive objects in this application embodiment does not constitute a limitation on the described objects. The description of the described objects is given in the claims or the context of the embodiments, and should not constitute unnecessary restrictions due to the use of such prefixes. Furthermore, in the description of this embodiment, unless otherwise stated, "multiple" means two or more.

[0036] The technical solutions of the embodiments of this application will be described below with reference to the accompanying drawings. In the description of the embodiments of this application, unless otherwise stated, " / " means "or," for example, A / B can mean A or B; the term "and / or" in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone.

[0037] In the embodiments provided in this application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0038] With the continuous growth of motor vehicle ownership and the increasing pressure on urban roads, road rage caused by drivers losing control of their emotions has become an important risk factor affecting road traffic safety.

[0039] In existing technologies, to reduce the risk of traffic accidents caused by road rage, some vehicles are equipped with Driver Fatigue Monitor Systems (DMS). These systems, along with onboard sensors and dashcams, detect road rage behaviors in real time, such as frequent honking, aggressive cutting in, and heated arguments. Upon detecting such behavior, intervention is provided through voice prompts, alarms, or the intervention of driver assistance systems. However, this approach is a passive response mechanism, intervening only after dangerous road rage behavior has occurred. It misses the optimal opportunity for risk prevention and fails to reduce the risk of traffic accidents caused by road rage at its source.

[0040] To prevent risks earlier, existing technologies typically employ a static weight-based scoring and warning method. This involves pre-setting several categories of triggering factors that influence driver emotions, such as the driver's own state, traffic environment, and vehicle condition, and assigning a fixed weight and scoring rule to each category. During driving, the vehicle control system collects relevant data in real time. When a triggering factor is detected, a risk index is calculated according to the fixed weight and scoring rule to issue a warning before dangerous behavior occurs.

[0041] The drawback of this statically weighted scoring and warning method is that, in real-world driving scenarios, the impact of the same triggering factor changes dynamically in different situations. For example, the emotional response evoked by another vehicle cutting in front of you differs significantly between a smooth expressway and a congested urban main road. This fixed-weight road rage risk index calculation method cannot adaptively adjust to different driving scenarios, individual driver differences, and dynamic changes in real-time road conditions. Consequently, the calculated risk index often does not match the driver's actual emotional fluctuations, making it difficult to serve as a basis for effective warning and intervention decisions.

[0042] Based on the above application scenarios, this application proposes a risk assessment method.

[0043] Figure 1 This is a schematic flowchart illustrating a risk assessment method provided in this embodiment. The method is applicable to vehicles equipped with a risk assessment system. The method includes the following steps.

[0044] S1. Real-time collection of data on multiple road rage triggering factors. Each triggering factor includes at least one inducing agent.

[0045] In this embodiment of the application, road rage triggering factors data refers to the macro-environment or state categories that cause drivers to experience anger.

[0046] Among them, triggering factors are specific, quantifiable micro-indicators or events that constitute the category of triggering factors, such as abnormal driver heart rate, malicious cutting in by other vehicles, or excessively high interior temperature of the vehicle. These triggering factors are the smallest unit of data collection, and their role is to provide structured input variables for subsequent model analysis.

[0047] In some specific embodiments of this application, the data on factors inducing road rage are shown in Table 1, which are divided into four categories: the driver's own physiological and internal state factors, the illegal behavior factors of other traffic participants, external road traffic environment factors, and vehicle and in-vehicle environment factors.

[0048] Table 1 Classification of Inducing Factors

[0049] In this embodiment, by comprehensively collecting data on triggering factors from four dimensions—physiology, other vehicles, environment, and cabin—the source of abnormal emotions in drivers can be fully captured. This is the physical basis for achieving accurate emotion prediction and solves the problems of high misjudgment rate and delayed early warning caused by a single data source.

[0050] In this embodiment, step S1 is executed by an in-vehicle risk assessment system, which periodically reads data from various sensor interfaces. For example, it acquires driver facial images through a DMS camera and extracts features such as frowning frequency and eye-closing duration using image processing algorithms; it acquires in-vehicle audio through a microphone array and extracts volume and speech rate through voice emotion analysis; it detects the cutting speed and lateral distance of surrounding vehicles by fusing forward-looking and side-looking cameras with radar; it obtains the average vehicle speed and congestion index of the current road segment through a navigation module; and it obtains the cabin temperature through an in-vehicle temperature sensor.

[0051] Specifically, regarding the driver's own physiological and internal state factors, biosensors integrated into the steering wheel or seat can be used to collect data such as the driver's heart rate and blood pressure variability in real time to determine whether there is a risk of driver emotional fluctuations due to illness or fatigue. Alternatively, the driver monitoring system can use in-vehicle cameras to observe the driver's facial micro-expressions, such as frowning or clenching teeth, or analyze the volume, speed, and sharpness of speech through a microphone array to identify negative emotions while driving.

[0052] For other traffic violations, surround-view cameras and millimeter-wave radar can be used to detect the target cutting speed, lateral distance, and trajectory changes of surrounding vehicles to identify behaviors such as cutting in line or changing lanes without signaling. Alternatively, vehicle-to-everything (V2X) technology can be used to obtain information on illegal crossings by non-motorized vehicles or pedestrians, or image recognition technology can be used to analyze whether drivers of neighboring vehicles are engaging in provocative behavior.

[0053] Regarding external road traffic environmental factors, information on severe weather and strong light can be obtained through rain sensors and light sensors, while road congestion duration and traffic conditions can be obtained by connecting to navigation map data.

[0054] Regarding vehicle and in-vehicle environmental factors, data such as current temperature comfort level, noise level, and system lag are obtained through in-vehicle temperature and humidity sensors, decibel meters, and vehicle system logs.

[0055] For example, in a specific urban evening rush hour scenario, the vehicle control system simultaneously collects the following data: navigation data shows that the current road congestion duration has exceeded 30 minutes; radar detects a vehicle on the left rapidly entering the lane; and the driver monitoring system's camera captures the driver's facial muscles tense and heart rate continuously increasing. Here, the 30-minute congestion duration is attributed to external road traffic environment factors; the vehicle on the left rapidly entering the lane is attributed to violations by other road users; and the driver's tense facial muscles and continuously increasing heart rate are attributed to the driver's own actions.

[0056] S2. Input the collected triggering factor data into the pre-trained scoring model. The scoring model determines the current triggering factors and outputs the factor triggering score corresponding to each triggering factor.

[0057] In this embodiment of the application, the pre-trained scoring model is specifically a machine learning model built on a deep neural network, which has been trained under supervision using historical driving behavior data.

[0058] Determining the current triggering factors means that the model determines which factors are in an abnormal or active state at the current moment based on the input data, rather than simply including all collected factors in the calculation.

[0059] The factor-induced score is a scalar value that represents the average severity of road rage behavior triggered by the factor in similar historical scenarios.

[0060] Understandably, in step S2, the nonlinear mapping capability of the scoring model is used to automatically identify truly risk-inducing factors from the raw driving data and assign them quantified scores. Compared to manually setting fixed thresholds, the scoring model can learn the coupling relationships between factors, avoiding the limitations of isolated judgments. Simultaneously, the scores output by the scoring model directly reflect the correlation between the factor and the severity of road rage behavior in historical statistics, providing objective input for subsequent weight calculations.

[0061] Specifically, the risk assessment system inputs multi-dimensional triggering factor data into the scoring model. After forward computation, the model outputs a vector, where each element corresponds to a triggering factor, and the element value is the triggering factor score. Internally, the model performs anomaly detection while calculating the triggering factor scores. For triggering factors below the model's internal activation threshold, their output scores are set to zero or ignored.

[0062] For example, inputting fatigue feature values, vehicle trajectory parameters, congestion index, and noise decibel value into the scoring model, the model outputs a fatigue factor score of 3.2, a malicious cutting-in factor score of 5.1, a road congestion factor score of 2.8, and an in-vehicle noise factor score of 3.8. The vector output by the scoring model can then be [3.2, 5.1, 2.8, 3.8]. Among these, the model determines malicious cutting-in as the most significant contributing factor.

[0063] S3. Based on the existing inducing factors and their corresponding inducing scores, calculate the comprehensive inducing score and corresponding dynamic weight of each inducing factor.

[0064] In this embodiment of the application, the comprehensive induced score is a quantification of the overall risk level of a certain type of induced factor, while the corresponding dynamic weight is a coefficient reflecting the relative importance of this type of factor among all abnormal factors at the current moment.

[0065] S4. Based on the comprehensive induced scores of various inducing factors and their corresponding dynamic weights, a weighted summation calculation is performed to obtain the current road rage risk index.

[0066] In this embodiment of the application, the weighted summation operation refers to multiplying the comprehensive induced score of each type of induced factor by its corresponding dynamic weight, and then adding the products of all induced factor categories to obtain a single value, namely the road rage risk index.

[0067] In this embodiment of the application, the formula for calculating the Road Rage Index is as follows: RoadRageIndex=T1(t)×W1(t)+T2(t)×W2(t)+T3(t)×W3(t)+T4(t)×W4(t).

[0068] Where Wi(t) is the dynamic weight of the i-th factor at time t, which is output in real time by the scoring model. Ti(t) is the comprehensive score of the i-th factor at time t, specifically the average score of all factors in that category.

[0069] In this embodiment of the application, the road rage risk index ranges from 0 to 10 points. In practical applications, the update frequency can be set to 1 time / second.

[0070] It should be understood that the road rage risk index is a dimensionless scalar, and its theoretical value range is consistent with the set range of the triggering factor scores. For example, the value range can be set to 0 to 10 points.

[0071] In this embodiment, a complete dynamic assessment loop is formed by real-time collection of data on multiple triggering factors, automatic identification of existing triggering factors and output of triggering factor scores using a pre-trained scoring model, calculation of the comprehensive triggering score and dynamic weight of various triggering factors based on the triggering factor scores, and finally weighted summation to obtain the road rage risk index. Compared with existing road rage assessment methods with fixed weights and fixed scoring rules, this application enables the risk index to adaptively adjust according to real-time changes in driving scenarios and driver states. For example, in congested road sections, the dynamic weight of traffic environment factors automatically increases; when comfort decreases, the weight of vehicle environment factors increases accordingly. The final output risk assessment index can accurately reflect the driver's actual emotional fluctuations, providing reliable data support for subsequent graded interventions.

[0072] In some embodiments, pre-training includes the following steps.

[0073] S21; Use historical driving behavior data to train the scoring model, enabling the scoring model to learn the mapping relationship between triggering factors and road rage behavior.

[0074] In this embodiment of the application, the mapping relationship is the association rule from the triggering factor to the possible road rage behavior obtained by the model through learning.

[0075] It should be understood that in step S21, by constructing the association between triggering factors and road rage behavior, the scoring model is able to automatically learn complex nonlinear relationships from real driving data instead of relying on manual rules.

[0076] For example, 10,000 samples were extracted from real historical driving data. Of these, 3,000 samples showed drivers honking excessively after the occurrence of the malicious lane-cutting factor; 1,500 samples showed drivers forcibly changing lanes; and 500 samples showed drivers getting out of the car to argue their way out. After training, the scoring model, when inputting the malicious lane-cutting factor features, outputs the following probability distributions for each behavior: excessive honking 0.6, forced lane changing 0.3, and getting out of the car to argue their way out 0.1. This probability mapping provides a basis for the subsequent quantitative calculation of the triggering factor scores.

[0077] Outputting the factor-induced scores for each inducing factor includes the following steps.

[0078] S22; The scoring model, based on this mapping relationship, predicts the road rage behavior that each currently existing triggering factor may induce.

[0079] In this embodiment, prediction refers to the process during the forward inference phase where the scoring model receives the triggering factor data collected at the current moment and calculates the probability of each possible road rage behavior using a pre-trained mapping relationship. Possible road rage behaviors refer to road rage behavior types whose probability is greater than zero, as output by the scoring model. In practical applications, the scoring model typically outputs the probability distribution of all road rage behaviors.

[0080] For example, if the malicious lane-cutting factor feature is input into the model, the model output results are: 0.55 probability of loud cursing inside the car, 0.65 probability of honking the horn excessively, 0.28 probability of forcibly changing lanes, and the probabilities of other behaviors are all below 0.05. Therefore, the behaviors that the malicious lane-cutting factor feature may induce include: loud cursing inside the car, honking the horn excessively, and forcibly changing lanes.

[0081] S23; The scoring model determines the factor-induced score for each triggering factor based on the road rage behaviors that may be induced by triggering factors and the preset behavior score for each road rage behavior.

[0082] In this embodiment, the preset behavior score is a standardized value predefined according to the severity of the behavior, used to uniformly measure the risk level of different behaviors.

[0083] The preset behavior scores can be obtained by querying a pre-built road rage behavior scoring table. The road rage behavior scoring table includes the driver's road rage behaviors and the preset behavior scores for each behavior.

[0084] The preset behavior scores can be set based on the severity of the driver's emotions and the level of danger. For example, cursing in the car is set to 2 points, honking the horn repeatedly is set to 3 points, and forcibly changing lanes is set to 6 points.

[0085] In this embodiment, by establishing a mapping relationship between triggering factors and road rage behavior in the scoring model, the scoring model can predict the specific road rage behavior that each triggering factor may induce based on historical statistical data, and then derive the factor's triggering score based on the preset score of the behavior. Compared with the existing technology that directly scores triggering factors using artificial rules, this application enables the triggering factor score to truly reflect the real tendency of the triggering factor to cause road rage behavior in similar scenarios, significantly improving the accuracy of road rage risk assessment.

[0086] In real-world driving scenarios, the same triggering factor may be associated with multiple potential behaviors, but many of these behaviors have an extremely low probability of occurrence. If these low-probability behaviors are included in the average score without screening, the representativeness of high-probability behaviors will be diluted, resulting in a lower triggering factor score.

[0087] In some embodiments, when the number of road rage behaviors that each triggering factor may induce is greater than a first number, the method for determining the factor induced score corresponding to each triggering factor includes the following steps.

[0088] Calculate the probability of each type of rage disorder behavior that may be induced by each triggering factor.

[0089] The second most frequent road rage behavior is selected, and the average of its corresponding preset scores is calculated as the factor induction score for that inducing factor; the second number is less than the first number.

[0090] In this embodiment, the first quantity is a preset integer threshold, indicating that when the number of road rage behavior types predicted by the scoring model exceeds this quantity, a screening strategy needs to be adopted. The occurrence probability refers to the probability value of each road rage behavior category output by the scoring model in step S22, which is usually normalized by softmax and summed to 1.

[0091] In this embodiment of the application, it is first determined whether the number of predicted road rage behavior types is greater than a first number. If it is greater, the probability sorting process is entered; if it is not greater, the average score of all road rage behaviors is used directly.

[0092] For example, if the first quantity is preset to 3, and the in-car temperature factor is input into the scoring model, the scoring model predicts 6 possible road rage behaviors: complaining (probability 0.4), slamming the steering wheel (0.25), opening the window (0.2), accelerating (0.08), swearing (0.05), and others (0.02). Since the quantity is greater than 3, it enters the screening process.

[0093] In this embodiment, the second quantity is an integer less than the first quantity, indicating that only the top M behaviors with the highest probabilities are retained. The average value is the sum of the preset behavior scores corresponding to these behaviors, divided by M. This average value serves as the final factor induction score for the induction factor.

[0094] Specifically, the system sorts the possible road rage behaviors from highest to lowest probability, selects the top M behaviors, retrieves the scores for these M behaviors from a pre-set road rage behavior scoring table, and calculates the average: FactorScore=(Score1+Score2+...+ScoreM) / M.

[0095] Where FactorScore is the final factor evoked score for the evoking factor, Score1 to ScoreM are the preset behavioral scores for each possible road rage behavior, and M is the number of possible road rage behaviors. It should be understood that if M is 1, the score of the behavior with the highest probability is directly taken. This average value is output as the factor-induced score.

[0096] For example, if the second quantity is preset to 2, the scoring model predicts the following order of six possible road rage behaviors: complaining (probability 0.4), slamming the steering wheel (0.25), opening the window (0.2), accelerating (0.08), swearing (0.05), and others (0.02). We then select complaining (preset score 1 point) and slamming the steering wheel (preset score 2 points), and calculate their average score as (1+2) / 2 = 1.5 points. Therefore, the triggering factor of excessively high interior temperature has a triggering score of 1.5.

[0097] In this embodiment, by setting a first quantity threshold and a second quantity screening, the probability ranking and optimal averaging mechanism is activated only when there are too many predicted behavior types. This effectively avoids the dilution of the inducing factor score by low-probability behaviors, making the final inducing factor score more focused on the typical road rage behavior most likely to be triggered by that inducing factor. Compared with the full averaging or single extreme value method, this method retains the stability of the multi-behavior synthesis and enhances the representativeness of the factor inducing score for high-frequency scenarios, thereby improving the accuracy of the factor inducing score.

[0098] In some embodiments, enabling the rating model to learn the mapping relationship between triggering factors and road rage behavior includes the following steps.

[0099] S21.1: Obtain historical driving behavior data. This data includes the driver's road rage behavior and the sequence of triggering factors present at the time of the behavior.

[0100] In this embodiment of the application, historical driving behavior data refers to real data with timestamps recorded during the long-term actual driving process of the vehicle.

[0101] The sequence of triggering factors present at the moment the behavior occurs refers to a vector sequence of measurements of all detected triggering factors within a short time window, such as 30 seconds, prior to the occurrence of the road rage behavior. It should be understood that this sequence is time-aligned.

[0102] Specifically, in a real vehicle test fleet or mass-produced vehicle, whenever a driver exhibits road rage behavior, the system triggers a recording event: saving the time-series data of all triggering factors within 30 seconds prior to the event as an input sequence, and simultaneously saving the road rage behavior as a tag.

[0103] For example, during a drive, the driver suddenly starts honking the horn excessively. Data from the preceding 30 seconds is reviewed: malicious cutting in, in-car noise level of 65 dB, and driver's heart rate of 110 beats / min are detected. These factor values ​​are serialized and saved as input, labeled "excessive honking."

[0104] S21.2: Use historical driving behavior data to iteratively train the scoring model. During the training process, the triggering factor sequence is used as the input feature of the scoring model, and the road rage behavior corresponding to the triggering factor sequence is used as the output target of the scoring model, so that the scoring model learns the mapping relationship from triggering factors to road rage behavior.

[0105] In this embodiment of the application, iterative training refers to using optimization algorithms such as gradient descent to traverse the training dataset multiple times and continuously update the model parameters to minimize the loss function between the predicted output of the scoring model and the true label.

[0106] In this embodiment of the application, the scoring model can adopt a deep learning-based sequence modeling architecture, specifically including: time series classification network, Transformer encoder, multimodal fusion network, or graph neural network.

[0107] In some specific embodiments of this application, the scoring model employs a Transformer encoder, which can effectively capture the long-distance dependencies and temporal dynamic changes between different triggering factors, and outputs the triggering score corresponding to each triggering factor or the probability distribution of each behavioral category. It should be understood that the Transformer encoder, rather than a traditional RNN, is used primarily because there are complex nonlinear interactions among the factors that trigger road rage. For example, when fatigue and malicious lane-cutting occur simultaneously, their effect on triggering road rage is far greater than the linear sum of their individual effects. The Transformer encoder's self-attention mechanism can dynamically calculate the correlation weights between any two factors, thereby capturing this interaction effect. Furthermore, the triggering factor data is often multimodal, including images, audio, and numerical data. The Transformer encoder can align and fuse different modalities through multi-head attention, significantly improving the accuracy of factor score prediction.

[0108] Specifically, in this embodiment, the scoring model employs batch training, dividing historical driving data into training, validation, and test sets in an 8:1:1 ratio. The loss function is defined as cross-entropy loss. The optimizer can be Adam, with a learning rate decay setting. After each iteration, the accuracy is evaluated on the validation set. Training stops when the validation loss no longer decreases for several consecutive iterations. Finally, the model parameters are saved. During training, the model learns to map the input factor sequence to the probability distribution of the output behavior category.

[0109] In this embodiment, by acquiring real historical driving behavior data and using the sequence of inducing factors as input and road rage behavior as output target, the scoring model is trained under supervision. This enables the model to autonomously summarize the mapping relationship between factors and behaviors from a large number of real-world scenarios, thereby establishing a mapping relationship that fits the actual driving patterns. This avoids the bias and deviation caused by manually set rules, making the scoring model's judgment of inducing behaviors more objective and reliable.

[0110] In some embodiments, determining the currently present triggering factors includes the following steps.

[0111] Each input trigger factor data is compared with the preset anomaly detection threshold for that trigger factor.

[0112] If the data of a certain triggering factor exceeds its corresponding anomaly detection threshold, then the triggering factor is determined to be a currently existing triggering factor.

[0113] In this embodiment, the anomaly detection threshold is a preset critical value based on statistical distribution or domain knowledge for each quantifiable triggering factor. For example, the fatigue factor can be set with a threshold based on the percentage of time the eyes are closed, and malicious cutting in can be set with a threshold based on the speed of entry.

[0114] The meaning of "exceeding" depends on the type of factor. For factors with larger values, such as noise decibels, "exceeding" means exceeding the threshold. For factors with smaller values, such as vehicle speed, "exceeding" means falling below the threshold.

[0115] For example, the fatigue factor has a preset threshold of blinking frequency > 25 times / minute. A real-time blinking frequency of 28 times / minute is detected, exceeding the threshold, thus indicating the presence of the fatigue factor. The malicious cutting-in factor has a preset threshold of cutting speed of 3 m / s. A cutting-in speed of 5 m / s is detected, exceeding the threshold, thus indicating the presence of the malicious cutting-in factor. The in-vehicle noise threshold is 50 dB. A noise level of 65 dB is detected, exceeding the threshold, thus indicating its presence. The road congestion factor has a preset average vehicle speed threshold of 30 km / h. The current vehicle speed is 35 km / h, not exceeding this threshold, thus indicating its absence.

[0116] Specifically, the system has a built-in factor threshold table, with each factor corresponding to one or a pair of thresholds, such as upper and lower limits. For each dimension of induced factor data collected in real time, the system calls a comparison function. If the factor value is within the normal range, it is marked as non-existent; if it exceeds the threshold, it is marked as present, and the induced factor data is used as the input feature of the scoring model.

[0117] In this embodiment, an anomaly detection threshold is preset for each triggering factor and compared with real-time collected data. Only factors exceeding the threshold are identified as currently existing triggering factors. This screening mechanism effectively filters out inactive triggering factors within the normal range, avoids interference from a large amount of data missing in subsequent evaluations, significantly reduces computational resource consumption, and improves the real-time performance and accuracy of the evaluation.

[0118] In some embodiments, calculating the comprehensive precipitating score of various precipitating factors includes the following steps.

[0119] S31: Calculate the average of the factor induction scores corresponding to the currently existing induction factors in each category of induction factors, and use it as the comprehensive induction score of that category of induction factors.

[0120] It should be understood that in step S31, the average value is used instead of the sum in order to eliminate the influence of the quantity of similar factors on the comprehensive score.

[0121] For example, a driver's own physiological and internal state factors may include five triggering factors, while external road traffic environment factors may only include two. If the scores are summed, even if the scores for each factor are low, the category with more factors will receive a higher overall score, causing bias. The average score, however, ensures that the overall score reflects the average triggering intensity of each factor in that category, thus achieving a fair comparison between categories.

[0122] For example, if at a certain moment, the evoked factor scores of each evoked factor output by the scoring model are as shown in Table 2.

[0123] The formula for calculating the combined induced score T1 of the driver's own physiological and internal state factors is as follows: T1 = (S1 + S2 + S3 + S4 + S5) / 5.

[0124] Table 2. Inducing Factor Scores

[0125] In this embodiment, the average value of the factor induced score corresponding to the current induced factor under each type of induced factor is calculated, and this average value is used as the comprehensive induced score of that type of induced factor. This makes the comprehensive induced score of each type of induced factor no longer affected by the number of induced factors within it, effectively eliminating the calculation bias caused by the difference in the number of induced factors under different categories, and providing an accurate data basis for the subsequent weighted calculation of the road rage risk index.

[0126] In some embodiments, calculating the dynamic weights corresponding to various inducing factors includes the following steps.

[0127] S32.1: Calculate the sum of the factor induction scores corresponding to the currently existing induction factors in each category of induction factors, and use it as the total score of that category of induction factors.

[0128] S32.2: Calculate the sum of the factor induction scores corresponding to the currently existing induction factors among various induction factors, and use it as the total score of various induction factors.

[0129] S32.3: Calculate the ratio of the total score of each type of inducing factor to the total score of all types of inducing factors, and use it as the dynamic weight corresponding to that type of inducing factor.

[0130] In this embodiment, the total score for each type of inducing factor refers to the sum of the factor inducing scores of all currently existing inducing factors under a certain category, which reflects the overall contribution of all inducing factors in terms of score intensity within each type of inducing factor. The total score for each type of inducing factor refers to the sum of the total scores of all categories, which represents the total energy of all abnormal factors at the current moment and serves as the denominator for the normalization weights.

[0131] It should be understood that in step S32.1, the summation preserves the information on the number of factors within the same type of precipitating factors. That is, the more precipitating factors there are and the higher the score of each precipitating factor, the greater the total score of that type of precipitating factors. If the driver simultaneously feels fatigued, sick, and irritable, the overall influence of the driver's own physiological and internal state factors is significantly greater than when there is only one fatigue factor.

[0132] In step S32.3, adaptive weight allocation is implemented. The higher the sum of the activity factor scores of a certain type of triggering factor, the higher its weight, meaning that the triggering factor contributes more to the risk of road rage at the current moment. For example, in a congested scenario, the total score of external road traffic environment factors may increase significantly, leading to an increase in their weight. Conversely, when the in-vehicle air conditioning malfunctions, the weights of vehicle and in-vehicle environment factors increase.

[0133] For example, data collected at a certain moment revealed that the following five factors may trigger road rage: fatigued driving (score: 3.2), personality type A (score: 4.2), malicious cutting in front of others (score: 5.1), bad weather (score: 4.6), and excessive noise inside the vehicle (score: 3.8).

[0134] The formulas for calculating the weights of the major categories of inducing factors at that moment are as follows: Driver's own physiological and internal state factors: W1=(3.2+4.2) / (3.2+4.2+5.1+4.6+3.8)=0.354.

[0135] Other traffic participant violation factors W2=5.1 / (3.2+4.2+5.1+4.6+3.8)=0.244.

[0136] External road traffic environmental factors W3=4.6 / (3.2+4.2+5.1+4.6+3.8)=0.220.

[0137] Vehicle and in-vehicle environmental factors W4 = 3.8 / (3.2 + 4.2 + 5.1 + 4.6 + 3.8) = 0.182.

[0138] The weights corresponding to each category of inducing factors are shown in Table 3.

[0139] Table 3 Weight Table

[0140] In this embodiment, the sum of the scores corresponding to the current triggering factors under each type of triggering factor is taken as the total score of each type of triggering factor, and the ratio of the total score of each type of triggering factor to the total score of all types of triggering factors is taken as the dynamic weight of each type of triggering factor. This allows the weight of each category to be automatically adjusted according to the risk intensity of the triggering factors triggered in real time, overcoming the defect of fixed weights in the prior art that cannot adapt to changes in the scenario. This enables the weight allocation to be adaptively adjusted in real time according to changes in the current driving scenario and driver state, thereby making the final road rage risk index closer to the actual driving situation and improving the accuracy of the assessment results.

[0141] In some embodiments, the evaluation method further includes the following steps.

[0142] The road rage risk index is compared with a preset threshold range to determine the current road rage risk level.

[0143] Based on the risk level of road rage, corresponding intervention actions are triggered.

[0144] In this embodiment, the preset threshold range refers to multiple continuous sub-ranges divided according to the numerical range of the road rage risk index, such as 0-10 points, with each range corresponding to a risk level. For example, it can be divided into [1,3) for ultra-low risk, [3,5) for low risk, [5,7) for medium risk, [7,9) for medium-high risk, and [9,10] for high risk.

[0145] In this embodiment, intervention refers to prompting, alarming, or proactive control actions performed by the vehicle system based on the risk level.

[0146] Different levels of risk correspond to different levels of intervention intensity. For example, ultra-low risk involves no intervention; low risk involves gentle voice prompts, such as asking you to remain calm; medium risk involves initiating cabin environment adjustments, such as playing soothing music, lowering the air conditioning temperature, and providing voice reminders; medium-high risk involves forcibly interrupting the entertainment system, issuing an alarm, and suggesting that you pull over to rest; and high risk involves directly intervening in driver assistance systems, such as limiting vehicle speed, automatically activating hazard lights, and calling emergency contacts.

[0147] In this embodiment, after calculating the road rage risk index, the index is further compared with multiple preset threshold ranges to determine the specific level of the current risk. Based on the level, a matching intervention operation is triggered, which achieves a precise match between the risk level and the intensity of intervention. This avoids excessive intervention that could lead to a decline in user experience, while ensuring the timeliness and effectiveness of intervention measures in high-risk scenarios.

[0148] This application also provides a risk assessment system for use in vehicles. Figure 4 The diagram shown is a schematic representation of the evaluation system.

[0149] The assessment system includes a data acquisition module 100, a scoring module 200, a weight and score calculation module 300, and a risk index calculation module 400.

[0150] The data acquisition module 100 is used to collect data on multiple road rage triggering factors in real time. Each triggering factor includes at least one inducing agent.

[0151] The scoring module 200 is connected to the data acquisition module 100 and is used to receive inducing factor data, determine the existing inducing factors through a pre-trained scoring model, and output the factor induction score corresponding to each inducing factor. The weight and score calculation module 300 is connected to the scoring module 200 and is used to calculate the comprehensive induction score and corresponding dynamic weight of each type of induction factor based on the existing induction factors and their corresponding factor induction scores.

[0152] The risk index calculation module 400 is connected to the weight and score calculation module 300. It is used to perform weighted summation calculation based on the comprehensive induction score of various inducing factors and the corresponding dynamic weights to obtain the current road rage risk index.

[0153] For example, the data acquisition module 100 can be a computing platform or a data acquisition circuit, processing circuit, processor, or controller mounted on the computing platform. Taking the data acquisition module 100 as a processor in the computing platform as an example, the data acquisition module 100 can collect data on various road rage triggering factors in real time through sensors or other acquisition devices configured on the vehicle.

[0154] For example, the scoring module 200 can be a computing platform or a data acquisition circuit, processing circuit, processor, or controller mounted on the computing platform. Taking the scoring module 200 as a processor in the computing platform as an example, the scoring module 200 is configured to: receive the inducing factor data sent by the data acquisition module, load the pre-trained scoring model to determine the inducing factors present at the current moment, and output the factor inducing score corresponding to each inducing factor.

[0155] For example, the weight and score calculation module 300 can be a computing platform or a data acquisition circuit, processing circuit, processor, or controller mounted on the computing platform. Taking the weight and score calculation module 300 as a processor in the computing platform as an example, the weight and score calculation module 300 is configured to: obtain the factor induced score of each induced factor from the scoring module, and calculate the comprehensive induced score and corresponding dynamic weight of each type of factor according to the preset classification rules.

[0156] For example, the risk index calculation module 400 can be a computing platform or a data acquisition circuit, processing circuit, processor, or controller mounted on the computing platform. Taking the risk index calculation module as a processor in the computing platform as an example, the risk index calculation module 400 is configured to: receive the comprehensive induced scores and dynamic weights of various factors output by the weight and score calculation module 400, perform a weighted summation operation to obtain the road rage risk index at the current moment, and send the index to the vehicle bus or display unit for subsequent graded early warning and intervention decision-making.

[0157] This application also provides a vehicle, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the risk assessment method described in the first aspect above, or the vehicle includes the risk assessment system described above.

[0158] This application also provides a computer program product, which includes computer program code. When the computer program code is run on a computer, it causes the computer to execute the risk assessment method involved in the above embodiments. This computer program can be installed in a vehicle system.

[0159] This application also provides a computer-readable storage medium storing program code that is executed by one or more processors. When the program code runs on the processor, it causes a device including one or more processors to perform the risk assessment method described in the above embodiments. The processor running this computer-readable storage medium can be installed in a vehicle system.

[0160] It should be understood that when the modules or units described herein are implemented using software, they can be implemented in whole or in part as a computer program product. A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state disk (SSD)).

[0161] This application provides a chip system including a processor, or the chip system including a memory and a processor, for calling computer programs or computer instructions stored in the memory to cause the processor to execute the risk assessment method involved in the above embodiments. The chip system can be a single chip or a chip module composed of multiple chips. This chip system can be installed in a vehicle system.

[0162] This application provides an electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the electronic device implements the risk assessment method described in the above embodiments. This electronic device can be installed in a vehicle system.

[0163] Those skilled in the art will recognize that the modules, units, and method steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0164] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be covered. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A risk assessment method, characterized by, The evaluation method includes: Real-time data collection of multiple road rage triggering factors; each triggering factor includes at least one inducing agent; The collected triggering factor data is input into a pre-trained scoring model, which determines the currently existing triggering factors and outputs the factor triggering score corresponding to each triggering factor. Based on the existing inducing factors and their corresponding inducing scores, the comprehensive inducing scores and corresponding dynamic weights of each inducing factor are calculated. Based on the comprehensive induced scores of various triggering factors and their corresponding dynamic weights, a weighted summation is performed to obtain the current road rage risk index.

2. The risk assessment method of claim 1, wherein, The pre-training includes: The scoring model learns the mapping relationship between triggering factors and road rage behavior; The output of factor-induced scores corresponding to each induced factor includes: The scoring model, based on the mapping relationship, predicts the road rage behavior that each of the currently existing triggering factors may induce. Based on the road rage behaviors that the triggering factors may induce, and the preset behavior scores for each type of road rage behavior, the factor induction score corresponding to each triggering factor is determined.

3. The risk assessment method of claim 2, wherein, When the number of road rage behaviors that each of the aforementioned triggering factors may induce is greater than the first number; The determination of the factor-induced score corresponding to each induced factor includes: Calculate the probability of each type of rage disorder behavior that may be induced by each triggering factor; The second number of road rage behaviors with the highest probability of occurrence is selected, and the average of the corresponding preset scores is calculated as the factor induction score corresponding to the inducing factor; the second number is less than the first number.

4. The risk assessment method according to claim 2 or 3, characterized in that, The step of enabling the scoring model to learn the mapping relationship between triggering factors and road rage behavior includes: Acquire historical driving behavior data; the historical driving behavior data includes the driver's road rage behavior and the sequence of triggering factors present at the time of the behavior; The historical driving behavior data is used to iteratively train the scoring model. During the training process, the triggering factor sequence is used as the input feature of the scoring model, and the road rage behavior corresponding to the triggering factor sequence is used as the output target of the scoring model, so that the scoring model learns the mapping relationship from triggering factors to road rage behavior.

5. The risk assessment method of claim 1, wherein, The determination of the currently existing triggering factors includes: Compare each input trigger factor data with the preset anomaly detection threshold for that trigger factor; If the data of a certain triggering factor exceeds its corresponding anomaly detection threshold, then the triggering factor is determined to be a currently existing triggering factor.

6. The risk assessment method of claim 1, wherein, Calculate the comprehensive precipitating score for various precipitating factors, including: Calculate the average factor induced score corresponding to the currently existing induced factors in each category of induced factors, and use it as the comprehensive induced score for that category of induced factors.

7. The risk assessment method of claim 1, wherein, Calculate the dynamic weights corresponding to various triggering factors, including: Calculate the sum of the factor induction scores corresponding to the currently existing induction factors in each category, and use it as the total score for that category of induction factors; Calculate the sum of the factor induction scores corresponding to the currently existing inducing factors among various inducing factors, and use it as the total score of various inducing factors; The ratio of the total score of each type of triggering factor to the total score of all types of triggering factors is calculated and used as the dynamic weight corresponding to that type of triggering factor.

8. The risk assessment method of claim 1, wherein, The evaluation method further includes... The road rage risk index is compared with a preset threshold range to determine the current road rage risk level; Based on the road rage risk level, the corresponding level of intervention is triggered.

9. A risk assessment system, characterized by, The evaluation system includes: The data acquisition module is used to collect data on multiple road rage triggering factors in real time; each triggering factor contains at least one triggering agent. The scoring module is used to receive the inducing factor data, determine the currently existing inducing factors through a pre-trained scoring model, and output the factor induction score corresponding to each inducing factor. The weight and score calculation module is used to calculate the comprehensive induction score and corresponding dynamic weight of each induction factor based on the existing induction factors and their corresponding factor induction scores. The risk index calculation module is used to perform a weighted summation calculation based on the comprehensive induced scores of various inducing factors and their corresponding dynamic weights to obtain the current road rage risk index. The computer-readable storage medium stores program code that is executed by one or more processors, and when the program code is run on the processor, causes a device including the one or more processors to perform the risk assessment method as described in any one of claims 1 to 8.

10. A vehicle characterized by comprising: The vehicle includes a memory, a processor, and a computer program stored in the memory and executable on the processor; when the processor executes the computer program, it causes the processor to implement the risk assessment method according to any one of claims 1 to 8, or the vehicle includes the risk assessment system according to claim 9.

Images