Vehicle passability risk determination method and system integrating driver ability and confidence

Abstract

This invention belongs to the field of road vehicle control systems and relates to a method and system for determining vehicle passability risk that integrates driver capability and confidence. The system includes a driving behavior and environmental data acquisition module, a scene matching module, a capacity estimation module, a capacity scoring module, a driving behavior feature analysis module, a passability confidence scoring module, a passability risk calculation module, and a database. The scene matching module calculates similarity based on the structural features of historical traffic sample sets and the current traffic scene, screening similar samples. The capacity estimation module performs a weighted fusion of the screened capacity and the calculated matching weights to obtain a capacity estimate. The passability risk calculation module uses the capacity score and the passability confidence score to modify the original passability risk based on structural complexity and vehicle performance. This system enables personalized and dynamic adjustment of passability risk, improving the accuracy of passability risk prediction.

Description

Technical Field

[0001] The present invention belongs to the field of road vehicle control systems and relates to the calculation or judgment of driving parameters, and specifically to a vehicle passability risk judgment method and system that integrates driver ability and confidence. Background Art

[0002] When driving in complex terrain or high-risk areas, such as steep slopes, gravel roads, and sections with concentrated obstacles, drivers often need to quickly assess the vehicle's passability and adjust their operations in a very short period of time to ensure smooth and safe passage. During this process, the driver's perception, confidence, and practical operational skills directly influence their subjective judgment of passability risk and their choice of control strategy. If the driver's abilities are not adequately matched to the current traffic scenario, or if they lack confidence, operational errors or misjudgments can easily occur, leading to safety issues such as stagnation and loss of control. Therefore, assessing the driver's abilities and status in specific passability scenarios and revising passability risk judgments accordingly are crucial for improving vehicle operation safety.

[0003] At present, the methods for judging vehicle passability risk mainly include the following two categories:

[0004] (1) Modeling of vehicle capacity based on physical performance and terrain environment: This type of method constructs models of vehicle chassis structure, power system, and tire parameters, and combines terrain information such as slope, obstacle size, and soil type to conduct theoretical analysis or simulation evaluation of whether the vehicle has physical capacity to pass. For example, Chinese patent CN 117087675 B determines the passability conditions through image recognition of terrain and ground height detection, and Chinese patent CN 114061969 B estimates whether the vehicle has the risk of scratching or getting stuck based on parameters such as approach angle and passing angle. This type of method can evaluate the adaptability of the vehicle to the environment, but generally ignores the impact of the driver's control ability and status on the actual passability results.

[0005] (2) Obstacle identification and passability prediction methods based on images, lidar, and simulation platforms: This type of method uses cameras or lidar to extract obstacle height, width, spacing, and other features, and combines vehicle parameters and traffic strategies to predict the passability probability. For example, Chinese patent CN 117087675 A combines real-time images and chassis height to make passability judgments, Chinese patent CN 116698157 A uses laser point cloud to reconstruct obstacle shapes to assist decision-making, and Chinese patent CN 115221672 A uses video scenes to reconstruct a virtual simulation environment and simulate the vehicle driving process to assess risks. These methods have advantages in static scene reconstruction and path prediction, but the driver is usually regarded as an ideal controller during the modeling process, and there is a lack of modeling of the interaction between their actual operating behavior and the scene.

[0006] In addition, some studies have attempted to incorporate the driver's psychological state into risk assessments, often using physiological signals such as heart rate, skin conductance, and blood pressure to infer emotional stress or confidence. While these methods have some value under experimental conditions, practical applications often suffer from issues such as wearing discomfort, signal delays, and environmental interference, making them difficult to meet the real-time and non-invasive requirements of high-risk scenarios.

[0007] In summary, existing vehicle passability risk assessment methods still have the following shortcomings: First, physical modeling methods ignore differences in driver control capabilities, making it difficult to characterize individual operational risk contributions based solely on driving style. Second, obstacle identification and simulation prediction methods fail to model the dynamic impact of driving behavior on the passage process, making it impossible to predict risks at the level of human-vehicle-road interaction, resulting in insufficient risk prediction accuracy. Third, psychological state recognition methods based on physiological signals are limited by perception methods and deployment conditions, and their practical application value is limited. Therefore, there is an urgent need to develop a method that can non-invasively acquire driving behavior characteristics during actual driving and dynamically assess driver confidence and ability based on scenario requirements and passage task requirements, thereby achieving a more individually adaptable passability risk assessment method. Summary of the Invention

[0008] In view of the shortcomings and deficiencies of the existing technology, the purpose of the present invention is to provide a vehicle passability risk assessment method that integrates driver ability and confidence. This method non-invasively obtains driving behavior characteristics during actual driving, and dynamically evaluates the driver's confidence and ability in combination with scenario requirements and traffic task requirements. The passability and driver's confidence score results are then used as risk adjustment factors, thereby achieving personalized and dynamic adjustment of traffic risk and improving the accuracy of passability risk prediction.

[0009] To achieve the above object, the present invention adopts the following technical solutions:

[0010] A vehicle passability risk determination method integrating driver ability and confidence, the method comprising the following steps:

[0011] Step 1. Collect driving behavior and environment data;

[0012] Step 2. Evaluate the driver's traffic capability in the current traffic scenario;

[0013] The structural features of the current traffic scenario are dynamically constructed based on the collected environmental data. The current traffic scenario is then matched with the structural features of historical traffic scenarios in the historical traffic sample set, and traffic scenarios with a similarity greater than a set threshold are selected. The selected similarity data is then normalized to obtain the matching weight vector required for weighted fusion. Finally, a weighted fusion is performed based on the traffic capacity vector in the selected historical traffic sample set and the calculated matching weight vector to obtain the traffic capacity estimation result for the current traffic scenario.

[0014] Step 3. Calculate the driver's traffic capability score for the current traffic scenario:

[0015] A basic capability scoring function is constructed based on the driver's static attributes. This is combined with the capability estimation results and structural characteristics in the current traffic scenario, and a structure-aware logistic regression model is used to predict the probability of successful passage. The final capability scoring function integrates these two parts and outputs the driver's capability score in the current traffic scenario.

[0016] Step 4. Extracting driving behavior characteristic indicators based on the collected driving behavior data, wherein the driving behavior characteristic indicators include the four-dimensional grip force characteristic vectors of the left and right hands, the micro-operation behavior characteristic vectors, and the back posture characteristic vectors;

[0017] Step 5. Based on the driving behavior characteristic indicators extracted in step 4, calculate the driver's driving confidence score for the current driving scenario;

[0018] Step 6. Integrate the driver's status to correct the passability risk:

[0019] The original passability risk based on structural complexity and vehicle performance is modified using the passability score from step 3 and the passability confidence score from step 5 to obtain the final passability risk.

[0020] As a preferred embodiment of the present invention, the historical traffic sample set in step 2 is ;in, is the capacity vector of the i-th traffic scenario in history; is the structural feature vector of the i-th traffic scene; is the pass result label, success = 1, failure = 0; , For longitudinal operation stability; For lateral control stability; for lane change continuity; For space utilization ability; for hill responsiveness; , The effective width of the road; is the road curvature; The degree of line of sight obstruction; is the road adhesion coefficient; is the road longitudinal slope angle, N represents the number of representative traffic scenarios the driver has experienced;

[0021] In step 2, the similarity is calculated using the Euclidean similarity function with a smoothing term.

[0022] As a preferred embodiment of the present invention, the expression of the basic ability scoring function in step 3 is:

[0023] ;

[0024] in, , indicating that a neutral driving style is best; is the Sigmoid function; Set as learnable parameters or experience; For driving experience, is the driving style label, output , represents the driver’s general traffic ability score;

[0025] The structure-aware logistic regression model includes a structure-aware weight generation model and logistic regression models, structure-aware weight generation models Taking the structural characteristics of the traffic scene as input, automatically generate the ability score weight under the current structural environment , the expression of the scoring function of the logistic regression model is:

[0026] ;

[0027] in, is the Sigmoid function, which is used to compress the output to interval, forming a probability score, the superscript T represents the transpose, and b represents the bias parameter in the model; The capability vector estimation result of the current traffic scene is output , represents the predicted probability of successful passage of the driver under the current ability and structural scenario conditions;

[0028] The expression of the capacity scoring function is:

[0029] ;

[0030] in, is the weight parameter; Indicates the driver's trafficability score in the current structural scenario.

[0031] As a preferred embodiment of the present invention, the method for extracting the four-dimensional grip force feature vectors of the left and right hands in step 4 is as follows: the grip force of the driver's left and right hands is collected in real time by flexible resistive grip force sensors installed at the 3 o'clock and 9 o'clock positions of the steering wheel, and then the original signal is preprocessed using a two-stage filtering strategy. The intensity deviation, fluctuation stability, normalized mutation index, and change sharpness are extracted from the preprocessed data to construct the four-dimensional grip force feature vectors of the left and right hands. and , the expression is:

[0032] ;

[0033] ;

[0034] in, represents the four-dimensional grip force feature vector of the left hand, represents the four-dimensional grip force feature vector of the right hand, and Represent the intensity deviation of the left hand and right hand respectively, and represent the fluctuation stability of the left and right hands respectively, and represent the normalized mutation index of the left and right hands, respectively, and They represent the sharpness of change for the left and right hands respectively, and the superscript T stands for transpose.

[0035] As a preferred embodiment of the present invention, the method for extracting the micro-operation behavior feature vector in step 4 is: a (t), brake pedal pressure P b (t) and steering wheel angle The signal is processed by sliding window mean filtering, and then the frequency domain features are extracted by short-time Fourier transform of the filtered signal to obtain the power spectrum density of the signal, and the main frequency energy ratio is calculated based on the power spectrum density. , peak frequency , spectrum entropy If the judgment condition is met ,and <1.2, that is, the current time window is identified as the micro-operation behavior segment, and then the signal of the micro-operation behavior segment is extracted, and the damped oscillation function is used for nonlinear fitting, and the physical parameters after fitting are extracted. Based on the obtained physical parameters, the characteristic vector of the micro-operation behavior is constructed. , the expression is:

[0036] in, is the initial amplitude, is the damping ratio, is the oscillation frequency, is the amplitude ratio.

[0037] As a preferred embodiment of the present invention, the method for extracting the back posture feature vector in step 4 is: collecting the three-dimensional coordinate information of the driver's left shoulder key point, right shoulder key point and the middle point of the back spine, respectively comparing them with the three-dimensional coordinate information of the corresponding position reference point of the left shoulder of the seat, the corresponding position reference point of the right shoulder of the seat, and the center point of the seat back, and calculating the Euclidean distance to obtain the back posture distance index; then calculating the driver correction factor according to the driver's height and body index, and using the driver correction factor to standardize the back posture distance index to obtain the back distance ; Calculate the shoulder angle based on the driver's left shoulder key point and right shoulder key point According to the The back posture stability is calculated by the position of the key points on the back of the frame and the average coordinates within the time window , construct a standardized back posture feature vector , the expression is:

[0038] ;

[0039] The superscript T stands for transpose.

[0040] As a preferred embodiment of the present invention, the driving confidence score of the driver in the current driving scene in step 5 is The expression is:

[0041] ;

[0042] in: 、 、 are the normalized sub-scores of grip strength feature, micro-manipulation behavior feature, and back posture feature respectively; 、 、 is the corresponding importance weight, satisfying the constraint .

[0043] As a preferred embodiment of the present invention, the final passing risk in step 6 The expression is:

[0044] ;

[0045] in, and are the risk sensitivity coefficients of ability bias and self-confidence bias respectively; Characterizes the relative degree of inadequacy of the driver's abilities; Indicates the degree of the driver's lack of confidence in the driving task, It is the original passability risk based on structural complexity and vehicle performance; DCCI represents the passability confidence score.

[0046] As a further preferred embodiment of the present invention, the intensity deviation The expression is:

[0047] ;

[0048] in, is the grip strength value at the i-th sampling point in the time window, and are the mean and standard deviation of the baseline period, respectively, and n represents the number of samples;

[0049] Fluctuation stability The expression is:

[0050] ;

[0051] Normalized mutation index The expression is:

[0052] ;

[0053] in, The grip strength in the baseline period was extremely poor; is the maximum grip strength of the current window; is the minimum grip strength of the current window;

[0054] Sharpness of change The expression is:

[0055] ;

[0056] in, and Represents the grip strength value of two adjacent sampling points.

[0057] As a further preferred embodiment of the present invention, the main frequency energy ratio The expression is:

[0058] ;

[0059] in, is the signal frequency The power spectral density at ;

[0060] Peak frequency The expression is:

[0061] ;

[0062] Spectral entropy The expression is:

[0063] ;

[0064] in, Representative frequency The power spectral density at the point, Representative frequency The probability distribution of the normalized power spectral density at the point, N is the window size.

[0065] As a further preferred embodiment of the present invention, the expression of the damped oscillation function in step 4 is:

[0066] ;

[0067] in, is the initial amplitude, is the damping ratio, is the natural frequency of the system, is the actual oscillation frequency after damping; is the initial phase angle, set to a constant or zero; is the time variable after the micro-operation starts, signals representing the extracted micro-operation behavior segments;

[0068] Oscillation frequency ; Amplitude ratio is the ratio of the two peaks.

[0069] As a further preferred embodiment of the present invention, the expression of the normalized sub-score of the grip strength feature is:

[0070] ;

[0071] in, Input vector for left or right hand grip force feature; Symmetric weight matrix, modeling the interaction between features; is the first-order linear weight vector; is a scalar bias term; is the Sigmoid activation function, represent The transposed vector of represent The transposed vector of

[0072] The expression of the normalized sub-score of the micro-operation behavior feature is:

[0073] ;

[0074] in, is the original micro-operation behavior feature vector, and are the hidden feature vector and the gating vector respectively; and is the hidden layer parameter, is the weight matrix of the fully connected layer, is the corresponding bias term; the gate unit is composed of the weight matrix of the linear fully connected layer and bias The gate vector generated by Control feature fusion strategy channel by channel; 、 is the output linear combination weight; is the nonlinear Tanh activation function; is the Sigmoid function; ∘ represents element-by-element multiplication, represents the fused micro-operation behavior feature vector, represents the micro-operation behavior characteristic score;

[0075] The expression of the normalized sub-score of the back posture feature is:

[0076] ;

[0077] in, is the spatial posture feature; It is the dynamic posture fluctuation feature; and are the fully connected layer parameters of each branch; is the nonlinear Tanh activation function of each branch; It is the intermediate feature representation of the output of the two branches; To finally fuse the weights and biases, act on the overall representation after branch concatenation; is the Sigmoid function.

[0078] As a further preferred embodiment of the present invention, in step 6, a maximum upper limit is set for the corrected passability risk value. , when the calculated passability risk value is greater than the maximum upper limit When the output passability risk value is the maximum upper limit .

[0079] The present invention also provides a vehicle passability risk determination system that integrates driver ability and confidence. The system is used to implement the above-mentioned vehicle passability risk determination method that integrates driver ability and confidence. The system includes a driving behavior and environment data acquisition module, a scene matching module, a passability estimation module, a passability scoring module, a driving behavior feature analysis module, a passability confidence scoring module, a passability risk calculation module, and a database.

[0080] The driving behavior and environment data acquisition module is used to collect the driver's behavior data, driver static characteristics and historical traffic data, and structural characteristics of the current traffic scene in real time. The driver's behavior data includes grip strength, micro-operations, and posture;

[0081] The scene matching module calculates similarity based on the structural features of the historical traffic sample set and the current traffic scene, and selects the top-K similar samples;

[0082] The capacity estimation module is used to normalize the filtered similarity data to obtain a matching weight vector; perform weighted fusion based on the capacity vectors in the filtered historical traffic sample set and the calculated matching weight vector to obtain a capacity estimation result for the current traffic scenario;

[0083] The traffic capability scoring module is used to calculate the traffic capability score of the driver's current traffic scenario;

[0084] The driving behavior feature analysis module is used to filter and perform time-frequency analysis on the original behavior signals of grip strength, micro-operation, and posture, and construct four-dimensional grip strength feature vectors, micro-operation behavior feature vectors, and back posture feature vectors for the left and right hands;

[0085] The passing confidence scoring module includes a grip strength feature scoring module, a micro-operation behavior feature scoring module, a back posture feature scoring module, and an overall confidence scoring module;

[0086] The passability risk calculation module is used to obtain the original passability risk based on structural complexity and vehicle performance, and use the passability score and the passability confidence score to correct the original passability risk based on structural complexity and vehicle performance, and output the passability risk considering individual status;

[0087] The database stores a driver's historical traffic sample set and personal feature data. The historical traffic sample set stores a structural feature vector and a traffic capability vector of a historical traffic scene.

[0088] Advantages and beneficial effects of the present invention:

[0089] (1) In order to solve the problem that the existing passability assessment results fail to consider the influence of the driver's subjective state, the present invention proposes to use the passability and driver's passability confidence score results as risk adjustment factors. By introducing the driver's confidence and ability status and the current structure matching results, the original passability risk index based on the vehicle's physical performance is corrected to achieve personalized and dynamic adjustment of the passability risk, and enhance the system's safety perception ability in unstructured complex scenarios.

[0090] (2) In order to solve the problem that it is difficult to accurately identify the current driver's driving confidence state, the present invention proposes a state perception method that integrates grip strength, micro-operation frequency and posture stability. This method can achieve low-invasive and non-perceptual joint recognition of the driver's operating state and psychological state without relying on interfering physiological sensors, and then construct a driver's driving confidence index to achieve both practicality and real-time performance of state recognition.

[0091] (3) In response to the problem that the existing passability risk judgment method fails to fully consider the individual differences and ability status of drivers, the present invention proposes an ability scoring method based on structure perception, which matches the driver's passability score with the terrain structure characteristics. This method enhances the individual adaptability of the system to passability judgment in complex scenarios and improves the personalization and accuracy of passability assessment.

[0092] (4) In order to improve the generalization effect of driver ability assessment in complex scenarios, the ability transfer mechanism based on structural similarity proposed in this paper performs Top-K screening and weighted fusion through the matching degree between historical traffic samples and current structural scenarios, effectively solving the problem that driver ability cannot be directly transferred to new scenarios, enhancing the adaptability of the model to unseen structural combinations, and improving the stability and accuracy of ability estimation.

[0093] (5) In order to achieve a rapid response link from risk perception to control execution, the present invention constructs a multi-level auxiliary control strategy system based on risk level triggering. Through voice prompts, throttle restrictions, speed limit control, path replanning and other measures, corresponding interventions for different risk levels from L0 to L4 are achieved, ensuring that the system provides effective compensation when the driver's cognition is insufficient, and significantly improving the vehicle's safety assurance capabilities in complex traffic scenarios.

[0094] (6) The present invention introduces a damped oscillation function to model micro-operation behavior, which can accurately reflect the periodic adjustment, operation hesitation and amplitude attenuation characteristics exhibited by the driver under specific traffic pressure, thereby improving the model's ability to depict driving behavior and the sensitivity of traffic risk judgment.

[0095] (7) The present invention achieves posture standardization processing without additional calibration conditions through spatial comparison between posture points and seat reference points, thereby improving the comparability of posture characteristics among different drivers. BRIEF DESCRIPTION OF THE DRAWINGS

[0096] By referring to the following description in conjunction with the accompanying drawings, and with a more complete understanding of the present invention, other objects and results of the present invention will become more clear and easy to understand. In the accompanying drawings:

[0097] Figure 1 A flow chart of the vehicle passability risk determination method that integrates driver ability and confidence in the present invention;

[0098] Figure 2 This is a structural block diagram of the vehicle passability risk determination system that integrates driver ability and confidence in the present invention. DETAILED DESCRIPTION

[0099] In order to enable those skilled in the art to better understand the technical solutions and advantages of the present invention, the present application is described in detail below with reference to the accompanying drawings, but this is not intended to limit the scope of protection of the present invention.

[0100] Example 1:

[0101] The present invention proposes a vehicle passability risk determination method that integrates driver ability and confidence. Figure 1 This is a flow chart of the vehicle passability risk determination method integrating driver ability and confidence in this embodiment. Figure 1 As shown, the vehicle passability risk determination method integrating driver ability and confidence includes the following steps:

[0102] Step 1. Collect driving behavior and environment data:

[0103] Step 1.1. Collect short-term driving status data:

[0104] In this embodiment, flexible resistive grip force sensors are embedded at the 3 o'clock and 9 o'clock positions of the steering wheel. The flexible resistive grip force sensors are used to collect the grip force of the driver's left and right hands in real time. The collection frequency is f s =10Hz, specifically including the left hand grip force F L (t), right hand grip strength F R (t).

[0105] In this embodiment, the longitudinal and lateral micro-operation data are collected by the original sensors of the vehicle, and the collection frequency f s =10Hz, including the accelerator pedal pressure P a (t), brake pedal pressure P b (t), steering wheel angle .

[0106] The RGB-D camera and human posture recognition algorithm are used to collect the three-dimensional coordinate information of the three key skeleton points of the driver's upper body, including the key point of the driver's left shoulder, which is expressed as ; The key point of the driver's right shoulder is expressed as ; The midpoint of the driver's spine, represented by .

[0107] In order to accurately judge the driving posture, three fixed spatial reference points corresponding to the above points are set on the seat back in this embodiment, including the reference point corresponding to the left shoulder of the seat, which is represented by ; The reference point of the right shoulder of the seat is expressed as The center point of the seat back is located on the vertical axis and at the same height as the middle of the driver's back, which is expressed as .

[0108] Step 1.2. Collect vehicle historical driving data:

[0109] To model and assess a driver's ability in specific traffic scenarios, we collect their traffic records from multiple historical scenarios and construct their traffic capability profile. Based on this, we combine the structural information of the current traffic scenario to calculate the driver's traffic capability vector for that scenario, which is then used to determine traffic risk.

[0110] Step 1.2.1. Record the structural features of historical scenes:

[0111] Assume the driver has experienced A representative traffic scene, building its historical scene collection :

[0112] ;

[0113] in, For the i-th traffic scene, each scene records five structural features. The effective width of the road; is the road curvature; The degree of line of sight obstruction; is the road adhesion coefficient; is the road longitudinal slope angle; each group Describes the structural requirements for driving behavior in this scenario. Data sources include maps, visual perception, and ground state recognition modules.

[0114] Step 1.2.2. Historical scene behavior recording and capability extraction:

[0115] For the driver in The driving process in each historical scene records its complete trajectory and control behavior data to form a time series , using CNN-LSTM network to extract five types of driving behavior indicators as the basis of special ability, and construct its historical scenario ability set :

[0116] ;

[0117] in, is the traffic capacity vector of the i-th traffic scenario, For longitudinal operation stability; For lateral control stability; is the lane-changing continuity, i.e. the smoothness of the angular velocity curve during the lane-changing section; is the space utilization ability, that is, the mean and variance of the closest distance to obstacles; is the ramp response capability, that is, the average acceleration adjustment rate in the ramp section.

[0118] At the same time, record the driver's passing result label in the scene , used to indicate whether the passage is successful (1 means smooth passage, 0 means failure, jam or detour).

[0119] Step 1.3. Collect driver's personal characteristic data:

[0120] Specifically, the driver's basic information is preset and collected through the vehicle identification system or the driver login interface, including the driver's height. ; Driver's BMI body shape factor ; Driving experience ; Driving style label (conservative = 0, neutral = 0.5, aggressive = 1), the collected data is stored in the vehicle user profile for subsequent posture normalization and risk judgment model call.

[0121] Step 2. Evaluate the driver's current traffic capability:

[0122] Step 2.1. Structural matching between the current traffic scene and the historical scene:

[0123] Extract the structural features of the current traffic scene from sensor perception information or high-definition maps , including road width , road curvature , degree of line of sight obstruction , road adhesion coefficient Angle with road longitudinal slope , which correspond to the five dimensions in the structure vector. These structural feature values will show differences for different scene types or road environments. In this embodiment, the current traffic scene structure vector is dynamically constructed and used as the core matching basis.

[0124] The driver's historical traffic sample set is ;in, Represents the i-th traffic scene, records five structural features, represents the traffic capability vector of the i-th traffic scenario, The result label of the pass.

[0125] To measure the degree of structural matching of the current traffic scene, the Euclidean similarity function with a smoothing term is used for calculation:

[0126] ;

[0127] in, is the similarity between the i-th traffic scene and the current traffic scene; To avoid small constants with zero denominators, the value is usually .

[0128] Step 2.2. Top-K similar scenarios screening and weighting:

[0129] In order to ensure that the historical samples used for driver capacity estimation are representative and credible in terms of structural attributes, this embodiment introduces a structural similarity threshold control mechanism in the similar scene sample screening stage. This mechanism sets a matching threshold. , only keep The historical traffic scene samples are used to estimate the participation ability, exclude samples with too large structural differences, and improve the effectiveness and generalization stability of the judgment model.

[0130] Based on all similarity scores Sort the historical samples in descending order and select the first The most similar traffic scene numbers constitute a set ,in ; Represents the most similar traffic scenario The similarity score of Represents the most similar traffic scenario The similarity score of Represents the most similar traffic scenario The similarity score of the Top-K subset is considered to be the closest to the current traffic scene in terms of structural attributes and can be used as a reference for capability migration. ≤ As a rule of thumb, avoid over-averaging.

[0131] In order to ensure the numerical stability and interpretability during the capability fusion process, it is necessary to normalize the similarity of the above Top-K samples to obtain the matching weight coefficient required for weighted fusion. , the weight vector satisfy , so that the more similar historical scenes will get greater weights, Represents the most similar traffic scenario The weight coefficient of Represents the most similar traffic scenario The weight coefficient of .

[0132] Step 2.3. Calculate the traffic capacity of the current traffic scenario:

[0133] The historical traffic capacity vectors of the drivers in the selected Top-K samples are used for weighted fusion to obtain the capacity estimation result under the current traffic scenario, which is expressed as:

[0134] ;

[0135] in, Represents the most similar traffic scenario The traffic capacity vector is obtained, and finally the capacity vector under the current traffic scenario is obtained. ,in , They respectively represent the longitudinal operation smoothness, lateral control stability, lane change continuity, space utilization ability, and slope response ability in the current traffic scenario.

[0136] Step 3. Calculate the driver's traffic capability score for the current traffic scenario:

[0137] To quantitatively evaluate a driver's ability in a traffic scenario with a specific structure, this paper proposes a capability scoring function that integrates individual attribute characteristics with the capability-structure matching relationship. This capability scoring function constructs a basic capability scoring function based on the driver's static attributes, combines the capability estimation results and structural characteristics in the current traffic scenario, and uses a structure-aware logistic regression model to predict the probability of successful passage. Finally, the two results are integrated to output the driver's ability score for the traffic scenario.

[0138] Specifically, in this embodiment, the static individual attributes of the driver are used, including driving experience and driving style labels , construct a basic ability scoring function to characterize the general ability tendency of the driver. The expression of the basic ability scoring function is:

[0139] ;

[0140] in, , indicating that a neutral driving style is best; is the Sigmoid function; is a learnable parameter or experience setting; output , which represents the driver's general traffic ability score.

[0141] In this embodiment, the structure-aware logistic regression model includes a structure-aware weight generation model and logistic regression models, structure-aware weight generation models Taking the structural characteristics S of the traffic scene as input, the ability score weight under the current structural environment is automatically generated, and then a logistic regression model is constructed based on the ability score weight.

[0142] Specifically, according to the structural feature vector of the current traffic scene , , , , and the capability vector estimation result of the current traffic scene , first bring the structural feature vector into the structure-aware weight generation model , generating capability score weights from structural features , then build a logistic regression model. The expression of the scoring function of the logistic regression model is:

[0143] ;

[0144] in, is the Sigmoid function, which is used to compress the output to interval, forming a probability score, the superscript T represents the transposition, and b represents the bias parameter in the model, which is used to adjust the pass score benchmark value under the current input conditions; the model score output , which represents the predicted probability of successful passage of the driver under the current ability and structural scenario conditions.

[0145] Finally, the capability scoring function integrates the basic capability and scenario scoring results, and the expression is:

[0146] ;

[0147] in, is a weight parameter, which can be fixed or dynamically adjusted based on the complexity of the structure; Indicates the driver's trafficability score in the current structural scenario.

[0148] In this embodiment, a structure-aware weight generation model is introduced, which takes the structural characteristics S of the traffic scene as input and automatically generates the capability score weight vector under the current structural environment, thereby constructing a mapping mechanism between capability, structure, and results. This structure-aware weight generation model has the ability to generalize to unseen structural conditions, so that it can still output reasonable scoring results when faced with new structural combinations, improving scenario adaptability and reasoning robustness. Specifically,

[0149] ;

[0150] ;

[0151] in, to They represent the weights of road width, road curvature, visibility obstruction, road adhesion coefficient and road longitudinal slope angle respectively. Represents the capability score weight generated based on structural characteristics, to They represent road width, road curvature, visibility obstruction, road adhesion coefficient and road longitudinal slope angle respectively;

[0152] The driving ability requirements vary significantly across different traffic scenarios. For example, lateral control is particularly critical on complex curves, while speed control and micro-manipulation precision are even more important on slopes or in narrow lanes. If the scoring function relies solely on fixed empirical weights, it will fail to accurately represent the relative importance of ability indicators in different structural features, resulting in a lack of adaptability and predictive power in the evaluation.

[0153] To solve this problem, this embodiment trains the model and introduces the pass result label in the training phase. The model automatically learns the joint influence between the capability vector and structural features as a supervisory signal. The training process enables the model to identify which capability combinations are more likely to lead to successful passage and which are more risky under different structural characteristics.

[0154] During the training phase, historical traffic sample data is used Supervised learning of the capacity scoring function. is the capacity vector of the i-th traffic scenario in history; is the structural feature vector of the i-th traffic scene; is the pass result label (success = 1, failure = 0); represents the passing probability calculated for the i-th passing scenario in history, where b is the bias parameter, and the training goal is to minimize the cross entropy loss function:

[0155] ;

[0156] Step 4. Extract driving behavior characteristic indicators based on the collected driving behavior data:

[0157] Step 4.1. Extract grip force features:

[0158] Assume that the system is in the time window Sampling frequency Collect grip force signals and obtain samples, denoted as .

[0159] The system requires the driver to collect grip strength data for a certain period of time under normal driving conditions. At this time, the driver should be in a normal and relaxed driving posture, without any emotional fluctuations or stress reactions; by calculating the average grip strength under normal conditions , standard deviation and range , represents the maximum value of grip strength collected, It represents the minimum value of grip strength collected, and can more accurately identify abnormal situations where grip strength exceeds the normal fluctuation range, thereby effectively distinguishing abnormal grip strength performance in specific situations and further improving the assessment accuracy of driving confidence.

[0160] In this example, to improve the smoothness of the grip force signal and the accuracy of feature extraction, a two-stage filtering strategy is used to preprocess the raw signal. First, a first-order IIR low-pass filter is used to suppress high-frequency noise above 10 Hz and remove interference caused by outliers. Subsequently, a Kalman filter is introduced to recursively smooth the signal. By constructing a state transition equation and an observation model, historical information is effectively integrated with current observations to dynamically estimate the grip force trend.

[0161] On this basis, the system collects the left hand grip force F in real time. L (t), right hand grip strength F R (t) Data, extract the following four types of characteristic indicators:

[0162] 1. Intensity deviation : Measures the degree of deviation of the overall grip strength during a period relative to the driver's normal grip strength level.

[0163] ;

[0164] in, is the grip strength value at the i-th sampling point in the time window, and are the mean and standard deviation of the baseline period (normal state), respectively.

[0165] 2. Fluctuation stability : Describes the stability of the fluctuation amplitude of the current grip force signal relative to the baseline period. When it is close to 1, it means the fluctuation is consistent with the normal state; when When the grip is kept highly stable, it may be a highly concentrated or rigid grip; when Significantly increased volatility during trading may indicate hesitation, lack of confidence, or overcorrection.

[0166] ;

[0167] 3. Normalized Mutation Index : Measures the instantaneous extreme change in grip force within the current time window, reflecting whether there is an obvious short-term mutation or jump operation. The indicator is particularly sensitive to capturing local sudden psychological changes.

[0168] ;

[0169] in, The grip strength was extremely poor during the baseline period (normal state); is the maximum grip strength of the current window; The minimum grip strength of the current window.

[0170] 4. Change Sharpness : Measures the relative amplitude of change between adjacent sampling points in the grip force signal during its temporal evolution, used to assess the smoothness of the change process. A larger value for sharpness of change, r, indicates a more discontinuous change, potentially indicating abrupt movements, continuous fine-tuning, or mental stress. This indicates a driver's confidence level or potential risk.

[0171] ;

[0172] in, and Represents the grip strength values of two adjacent sampling points;

[0173] Based on the above calculation process, the left and right hand grip force signal sequences within the set time window are recorded as , construct the four-dimensional grip force feature vectors of the left and right hands respectively:

[0174] ;

[0175] ;

[0176] Each term is defined in the same way as the original feature, except that the signal Replace with the grip sequence of the corresponding hand or , the subscript L represents the left hand and R represents the right hand.

[0177] Step 4.2. Extract micro-operation frequency features:

[0178] The original accelerator pedal pressure P a (t), brake pedal pressure P b (t) and steering wheel angle The signal is filtered using a sliding window mean filter with a window length of 0.2 seconds (i.e., 2 sampling points) to suppress high-frequency noise. The specific formula is as follows:

[0179] ;

[0180] in, is the filtered signal; is the window size, indicating that each time window contains 2 sampling points; is the sampling interval; is the original signal, 、 or One of them is to perform the first-order difference on the filtered signal and calculate the gradient , used to capture instantaneous changes in micro-operations and identify sudden adjustments in acceleration, braking or steering.

[0181] In order to further analyze the frequency components of micro-operation behavior, the present invention performs time domain-frequency domain conversion on the smoothed signal and uses short-time Fourier transform (STFT) to extract frequency domain features of the filtered signal with a window length of 5 seconds (i.e., 50 sampling points) and an overlap rate of 50%, so as to analyze the frequency changes of the signal in different time periods.

[0182] ;

[0183] in, It is the result of short-time Fourier transform, which indicates that the signal and frequency The frequency components at is a window function, which is used to limit the time window of each transformation; is the signal sequence, at time The signal value at time observed values).

[0184] Through Fourier transform, the power spectral density of the signal is obtained , which represents the energy distribution of the signal at different frequencies.

[0185] ;

[0186] in, is the frequency The power spectral density at , which represents the energy of the frequency component; Is the signal at the moment Observed values of is the frequency in Hz.

[0187] After converting these features into frequency domain features, the present invention further extracts key frequency domain judgment indicators to determine whether the signal exhibits micro-operation characteristics. The present invention sets the micro-operation frequency range to 0.5Hz to 3.0Hz, based on the understanding of the controllable movement frequency of human hands and feet in ergonomics. This frequency band covers the stable adjustment frequency range that can be achieved by humans during continuous active operation. It can not only reflect frequent small-amplitude correction behaviors, but also effectively distinguish between uncontrolled noise and normal manipulation. On this basis, the system calculates the following micro-operation judgment indicators.

[0188] 1. Main frequency energy ratio : , defines the ratio of the signal energy in the 0.5Hz to 3Hz frequency band to the total energy as a key indicator for determining the frequency component of micro-operations. This frequency band usually reflects frequent small-amplitude acceleration or steering wheel correction and other micro-operations; among them, is the signal at frequency Power spectral density at 0.5Hz≤ ≤ 3Hz is the micro-operation frequency band, indicating frequent small adjustments.

[0189] 2. Peak frequency : Used to identify the frequency corresponding to the maximum amplitude in the power spectrum. If If it falls within the frequency range of 0.5 Hz to 3 Hz, it is determined to be the dominant frequency of micro-operation, indicating that there are frequent small-amplitude operations in the signal, such as continuous small-amplitude acceleration, braking or steering adjustments. , determine the frequency point corresponding to the maximum spectral energy , , that is, to find The frequency corresponding to the maximum value ; If the peak frequency , then it is considered that the current operation signal is characterized by micro-operation, that is, frequent small-amplitude repeated adjustments; if If it is not within this range, the signal is considered to have no typical micro-operation characteristics.

[0190] 3. Spectral entropy : In order to exclude the mixed behavior or noise dominance, the power spectral density at each frequency point is Normalized to a probability distribution ,like Close to uniform distribution, that is, the energy of all frequencies is almost the same, then the entropy value is close to the maximum; if Highly concentrated, that is, only a few frequencies dominate, then the entropy value is small; use the Shannon entropy formula to calculate the entropy value of the distribution , is the spectrum entropy value, which measures the "discreteness" or "chaos" of frequency energy distribution.

[0191] ;

[0192] In the present invention, if the judgment condition is met ,and <1.2, the current time window is considered to be a micro-operation behavior segment.

[0193] The present invention uses the damped oscillation function to extract the driver's micro-operation characteristics. The core advantage of the damped oscillation function is to simulate the behavior of "gradual attenuation of amplitude and periodic adjustment", which is highly consistent with the actual control action of the driver during the micro-operation process. , use the following damped oscillation function for nonlinear fitting:

[0194] ;

[0195] in, is the initial amplitude, which is used to reflect the initial adjustment amplitude of the micro-operation; is the damping ratio, which measures the decay rate of the operation and reflects the convergence and decisiveness of the operation; is the natural frequency of the system, reflecting the potential operational response rhythm; is the actual oscillation frequency after damping; is the initial phase angle, set to a constant or zero; It is the time variable after the micro-operation starts.

[0196] The fitting method is implemented using the least squares nonlinear optimization tool to extract the physical parameters after fitting.

[0197] Based on the damping model, the key behavioral characteristic indicators, initial amplitude Damping ratio ; Oscillation frequency ; Amplitude ratio Indicates the ratio of the two peak values before and after, and determines whether there is repeated fine-tuning.

[0198] Based on the above frequency determination and dynamic modeling process, the characteristic vector of micro-operation behavior is constructed as follows:

[0199] ;

[0200] Step 4.3. Extract the driver's back posture features:

[0201] 1. Driver back - backrest distance:

[0202] Calculate the three-dimensional Euclidean distances between the three pairs of points based on the three-dimensional coordinate information extracted in step 1.1:

[0203] ;

[0204] in, Represents the Euclidean distance operation, which is used to measure the absolute displacement between point pairs in three-dimensional space; Represents the Euclidean distance between the key point of the driver's left shoulder and the corresponding position of the backrest, Represents the Euclidean distance between the right shoulder key point and the corresponding position of the backrest, Represents the Euclidean distance between the middle point of the driver's back spine and the corresponding position of the backrest. Combining the above three items, the overall backrest posture distance index is constructed:

[0205] ;

[0206] in, It is an indicator of the overall backrest posture distance; , , The weighting coefficient of , parameters can be adjusted according to feature stability.

[0207] In order to avoid the influence of individual body shape differences on backrest posture characteristics, a driver correction factor is introduced , according to height , body size index The driver's back distance is standardized and the calculation formula is as follows:

[0208] ;

[0209] The parameter value is =170cm, =22, =0.15, =0.10, calculated For standardizing back distance .

[0210] ;

[0211] 2. Shoulder angle Shoulder Posture: Reflects the angle between the line connecting the left and right shoulders and the horizontal, determining whether there is a "shrugging" or asymmetrical twisting motion. Changes in shoulder posture provide nonverbal feedback to the driver's assessment of environmental risks. Under normal driving conditions, the driver's shoulders should maintain a high degree of left-right symmetry. However, when navigating narrow roads, uphill or downhill slopes, or encountering sudden obstacles, the pressure increases, and some drivers may experience stress reactions such as shoulder tension, a slight shrugging, or unilateral sinking.

[0212] ;

[0213] 3. Back posture stability : Analyze the stability of the back point within the time window to reflect whether the driver's posture changes drastically in a continuous time period; define the sliding window size as ,but:

[0214] ;

[0215] in, For the Frame back keypoint positions; is the average coordinate within the time window.

[0216] Based on the above calculation process, a standardized back posture feature vector is constructed , the expression is:

[0217] ;

[0218] Step 5. Based on the driving behavior characteristic indicators extracted in step 4, calculate the driver's driving confidence score for the current driving scenario:

[0219] Step 5.1. Driver grip strength characteristic score:

[0220] There is a significant interaction effect between the dimensions of grip strength characteristics. "High grip strength + high volatility" is often more risk-indicative than a single factor. To capture this nonlinear interaction, a quadratic model combined with a neural network is used to model the grip strength characteristic score. and , find the average value :

[0221] ;

[0222] in, Input vector for left or right hand grip force feature; Symmetric weight matrix, modeling the interaction between features; is a first-order linear weight vector that preserves the independent linear influence of each feature on the score; is a scalar bias term used to offset the overall score benchmark; is the Sigmoid activation function, represent The transposed vector of represent The transposed vector of .

[0223] Step 5.2. Driver micro-operation behavior characteristic score:

[0224] Aiming at the coexistence of characteristics such as "amplitude, frequency, stability and repeated adjustment" in micro-operation behavior, a gated residual network structure is used as a scoring function modeling method for micro-operation behavior characteristics. In this structure, on the one hand, the original micro-operation characteristics are retained, and on the other hand, potential nonlinear combinations are extracted and the hidden feature vector is used to obtain the score function of the micro-operation behavior characteristics. and the gate vector To adjust the information flow for fusion, in order to improve the balance between the model's stability modeling and generalization ability, the structure of the scoring function is as follows:

[0225] ;

[0226] in, is the original micro-operation behavior feature vector, which reflects the initial response, damping convergence, frequency component and amplitude fluctuation behavior of the operation; and is the hidden layer parameter, the former is the weight matrix of the fully connected layer, and the latter is the corresponding bias term. The high-order combination features are extracted by the activation function to obtain the hidden feature vector , the two together determine the projection method of the input feature in the latent space; the gate unit is composed of the weight matrix of the linear fully connected layer and bias The gate vector generated by Control feature fusion strategy channel by channel; 、 is the output linear combination weight; It is a nonlinear Tanh activation function that can extract the potential high-order combination structure in micro-operation behavior; is the Sigmoid function; ∘ represents element-by-element multiplication, represents the fused micro-operation behavior feature vector, Represents the micro-operation behavior feature score.

[0227] Step 5.3. Driver back posture feature score:

[0228] To accurately model the impact of the driver's back posture on traffic conditions, this paper divides posture features into two categories: spatial posture features and dynamic stability features. Based on this, a lightweight two-branch fusion neural structure is designed. This structure can model static position features and dynamic fluctuation features separately while ensuring interpretability, and fuse them to generate a final score. The structure of the scoring function is as follows:

[0229] ;

[0230] in, are spatial posture characteristics, representing the distance between the back and the backrest, and the left and right deflection angles of the shoulders; is the dynamic posture fluctuation feature, which indicates the stability of the back key points within the set time window; and are the fully connected layer parameters of each branch, which perform linear mapping and nonlinear extraction on different types of features respectively; is the nonlinear Tanh activation function of each branch; It is the intermediate feature representation of the output of the two branches; To finally fuse the weights and biases, act on the overall representation after branch concatenation; is the Sigmoid function.

[0231] Step 5.4. Calculate the driver's overall confidence:

[0232] To integrate the comprehensive effects of grip strength, micro-manipulation frequency, and back posture on driver confidence, the following nonlinear enhanced scoring function is constructed to calculate the driver's driving confidence index (DCCI):

[0233] ;

[0234] in: 、 、 are the normalized sub-scores of grip strength feature, micro-manipulation behavior feature, and back posture feature respectively; 、 、 is the importance weight of the corresponding module, satisfying the constraint .

[0235] This function has an exponential nonlinear smooth growth characteristic. When the driver performs well in multiple dimensions, his or her driving confidence will show a significant increasing trend; however, when any key dimension is seriously insufficient, the overall score will be significantly suppressed, effectively reflecting the sensitivity of the driver's driving ability to the combined effect of multiple ability factors.

[0236] Step 6. Integrate the driver's status to correct the passability risk:

[0237] In order to effectively integrate the driver status into the passability risk assessment, the original passability risk based on structural complexity and vehicle performance is assumed to be , introducing driver capability scores and traffic confidence score , the modified risk model is constructed as follows:

[0238] ;

[0239] in, and are the risk sensitivity coefficients of ability bias and self-confidence bias respectively; Characterizes the relative degree of inadequacy of the driver's abilities; Indicates the degree of driver's lack of confidence in the driving task.

[0240] To prevent the risk value from expanding abnormally due to a low driver status score or excessive parameter settings, a maximum upper limit is set for the corrected risk value. To ensure the stability of the assessment results and the controllability of the engineering system, the maximum boundary constraint of the risk value is set as follows:

[0241] ;

[0242] This revised model is both interpretable and adjustable, and can dynamically reflect the impact of changes in driver status on traffic risks while ensuring the accuracy of basic risk assessment.

[0243] In this embodiment, the original passability risk value based on structural complexity and vehicle performance is , the existing traffic capacity assessment model can be directly called for calculation; this type of model is usually based on the adaptation relationship between vehicle chassis parameters and scene structural characteristics, by setting traffic thresholds, building traffic feasible domains, or reconstructing physical scenes and replaying traffic processes in a simulation platform to determine whether the vehicle has the ability to pass under standard driving conditions. Therefore, without reconstructing the underlying physical model, the present invention directly uses the risk calculation framework formed by the above-mentioned existing research to evaluate the current vehicle and structural scene and obtain the basic traffic risk value. , as a subsequent fusion of driver factors for personalized risk correction.

[0244] In this embodiment, after obtaining the corrected passability risk value Then, hierarchical management and auxiliary automatic control strategy triggering are implemented according to the risk value to improve the safety assurance capability during the actual passage process.

[0245] Specifically, the risk levels are divided into: L0 (completely safe): ; L1 (low risk): ; L2 (medium risk): ; L3 (high risk): ; L4 (No Passing): .

[0246] According to different risk levels, the trigger control strategy is as follows:

[0247] (1) If the risk level is L1 (low risk), the system will remind the driver through voice prompts: "The structure ahead is complex, please proceed with caution", and display a blue low-risk logo on the instrument interface. The system will not interfere with vehicle control, maintain the current throttle response and steering settings, and will only continue to monitor and prompt risks without actively intervening in driving operations.

[0248] (2) If the risk level is L2 (medium risk), the system will issue a voice reminder: "The risk of traffic ahead is high, please operate with caution." At the same time, the instrument interface will display a yellow medium risk icon and a suggestion window will pop up. At the same time, the throttle opening limit will be compressed to 80% of the current value, and the fine steering assist mode will be enabled to increase the steering response sensitivity by about 15% to assist the driver in maintaining smooth control in complex areas.

[0249] (3) If the risk level is L3 (high risk), the system will issue a voice prompt: "The traffic environment ahead is complex. It is recommended to slow down and observe." The instrument panel will display an orange high-risk warning and flash. At the same time, the throttle opening limit will be automatically reduced to 70% of the current value. The speed limit function will be enabled, limiting the maximum speed to 20 km / h. The high-sensitivity brake warning mechanism will be activated to monitor and respond to possible obstacles or sudden changes in slope in advance, prompting the driver to enter manual low-speed cautious operation mode.

[0250] (4) If the risk level is L4 (No Passing), the system will continuously announce: "The risk of passing is too high, please do not proceed." The red No Passing sign will flash on the instrument panel. At the same time, the throttle opening will be forced to 50% of the current value, the output of the sub-hydraulic brake system will be increased to 1.5 times the current value, and the speed limit control will be activated (maximum speed 10 km / h, only for escape). If necessary, the emergency stop mechanism will be triggered, and the route re-planning will be recommended to prevent the driver from continuing to pass into the high-risk area.

[0251] During the passage process, if the system continuously detects changes in risk levels, the control strategy will be updated only when the risk level rises to a higher level, and the control instructions of the same level will only be triggered once to avoid repeated intervention and frequent control.

[0252] Example 2:

[0253] like Figure 2 As shown, the present invention also provides a vehicle passability risk assessment system that integrates driver ability and confidence. The system includes a driving behavior and environment data acquisition module, a scene matching module, a passability estimation module, a passability scoring module, a driving behavior feature analysis module, a passability confidence scoring module, a passability risk calculation module, and a database.

[0254] The driving behavior and environment data acquisition module is used to collect the driver's grip strength, micro-operation, posture and other behavioral data, the driver's static characteristics and historical traffic data, and the structural characteristics of the current traffic scene in real time;

[0255] The scene matching module calculates similarity based on the structural features of the historical traffic sample set and the current traffic scene, and selects the top-K similar samples;

[0256] The capacity estimation module is used to normalize the filtered similarity data to obtain a matching weight vector; perform weighted fusion based on the capacity vectors in the filtered historical traffic sample set and the calculated matching weight vector to obtain a capacity estimation result for the current traffic scenario;

[0257] The traffic capability scoring module is used to calculate the traffic capability score of the driver's current traffic scenario;

[0258] The driving behavior feature analysis module is used to filter and perform time-frequency analysis on the original behavior signals of grip strength, micro-operation, and posture, and construct four-dimensional grip strength feature vectors, micro-operation behavior feature vectors, and back posture feature vectors for the left and right hands;

[0259] The passing confidence scoring module includes a grip strength feature scoring module, a micro-operation behavior feature scoring module, a back posture feature scoring module, and an overall confidence scoring module;

[0260] The passability risk calculation module is used to obtain the original passability risk based on structural complexity and vehicle performance, and use the passability score and the passability confidence score to correct the original passability risk based on structural complexity and vehicle performance, and output the passability risk considering individual status;

[0261] The database stores data such as the driver's historical traffic sample set, personal characteristics, etc., and the historical traffic sample set stores the structural feature vector and traffic capability vector of the historical traffic scene.

[0262] Furthermore, the system described in this embodiment also includes a risk assessment and strategy decision module, which is used to implement L0~L4 graded control strategies based on the risk level of the corrected passability risk value, and transmit the risk level and control strategy to the driver and the vehicle automatic control execution unit through the vehicle system, and perform voice reminders, instrument prompts and automatic control execution to ensure the synchronization and effectiveness of the system response.

[0263] The present invention also provides an electronic device comprising: one or more processors and a memory; wherein the memory is used to store one or more programs, and when the one or more programs are executed by the one or more processors, the one or more processors implement the above-mentioned vehicle passability risk assessment method that integrates driver ability and confidence.

[0264] The present invention also provides a computer-readable medium having a computer program stored thereon, and when the computer program is executed by a processor, the vehicle passability risk assessment method that integrates driver ability and confidence as described above is implemented.

[0265] Those skilled in the art will appreciate that all or part of the functions of the various methods / modules in the above embodiments may be implemented via hardware or via computer programs. When all or part of the functions in the above embodiments are implemented via computer programs, the program may be stored in a computer-readable storage medium, which may include a read-only memory, random access memory, a magnetic disk, an optical disk, a hard disk, etc., and the program is executed by a computer to implement the above functions. For example, the program may be stored in a memory of a device, and when the program in the memory is executed by a processor, all or part of the above functions may be implemented.

[0266] In addition, when all or part of the functions in the above-mentioned embodiments are implemented by means of a computer program, the program can also be stored in a storage medium such as a server, another computer, a disk, an optical disk, a flash drive or a mobile hard disk, and saved to the memory of a local device by downloading or copying, or the system of the local device is updated. When the program in the memory is executed by the processor, all or part of the functions in the above-mentioned embodiments can be implemented.

[0267] The above description of the present invention using specific examples is intended only to facilitate understanding of the present invention and is not intended to limit the present invention. A person skilled in the art of the present invention may make several simple deductions, modifications, or substitutions based on the principles of the present invention. Therefore, the scope of protection of the present invention shall be subject to the scope of protection of the claims.

Claims

1. A vehicle passability risk assessment method that integrates driver ability and confidence, characterized by: The following steps are involved: Step 1. Collect driving behavior and environment data; Step 2. Evaluate the driver's traffic capability in the current traffic scenario; The structural features of the current traffic scenario are dynamically constructed based on the collected environmental data. The current traffic scenario is then matched with the structural features of historical traffic scenarios in the historical traffic sample set, and traffic scenarios with a similarity greater than a set threshold are selected. The selected similarity data is then normalized to obtain the matching weight vector required for weighted fusion. Finally, a weighted fusion is performed based on the traffic capacity vector in the selected historical traffic sample set and the calculated matching weight vector to obtain the traffic capacity estimation result for the current traffic scenario. Step 3. Calculate the driver's traffic capability score for the current traffic scenario: A basic capability scoring function is constructed based on the driver's static attributes. This is combined with the capability estimation results and structural characteristics in the current traffic scenario, and a structure-aware logistic regression model is used to predict the probability of successful passage. The final capability scoring function integrates these two parts and outputs the driver's capability score in the current traffic scenario. Step 4. Extracting driving behavior characteristic indicators based on the collected driving behavior data, wherein the driving behavior characteristic indicators include the four-dimensional grip force characteristic vectors of the left and right hands, the micro-operation behavior characteristic vectors, and the back posture characteristic vectors; Step 5. Based on the driving behavior characteristic indicators extracted in step 4, calculate the driver's driving confidence score for the current driving scenario; Step 6. Integrate the driver's status to correct the passability risk: The original passability risk based on structural complexity and vehicle performance is modified using the passability score from step 3 and the passability confidence score from step 5 to obtain the final passability risk.

2. The vehicle passability risk determination method integrating driver ability and confidence according to claim 1 is characterized in that: The historical traffic sample set in step 2 is ;in, is the capacity vector of the i-th traffic scenario in history; is the structural feature vector of the i-th traffic scene; is the pass result label, success = 1, failure = 0; , For longitudinal operation stability; For lateral control stability; for lane change continuity; For space utilization ability; for hill responsiveness; , The effective width of the road; is the road curvature; The degree of line of sight obstruction; is the road adhesion coefficient; is the road longitudinal slope angle, N represents the number of representative traffic scenarios the driver has experienced; In step 2, the similarity is calculated using the Euclidean similarity function with a smoothing term.

3. The vehicle passability risk determination method integrating driver ability and confidence according to claim 1 is characterized in that: The expression of the basic ability scoring function in step 3 is: ； in, , indicating that a neutral driving style is best; is the Sigmoid function; Set as learnable parameters or experience; For driving experience, is the driving style label, output , represents the driver’s general traffic ability score; The structure-aware logistic regression model includes a structure-aware weight generation model and logistic regression models, structure-aware weight generation models Taking the structural characteristics of the traffic scene as input, automatically generate the ability score weight under the current structural environment , the expression of the scoring function of the logistic regression model is: ； in, is the Sigmoid function, which is used to compress the output to interval, forming a probability score, the superscript T represents the transpose, and b represents the bias parameter in the model; The capability vector estimation result of the current traffic scene is output , represents the predicted probability of successful passage of the driver under the current ability and structural scenario conditions; The expression of the capacity scoring function is: ； in, is the weight parameter; Indicates the driver's trafficability score in the current structural scenario.

4. The vehicle passability risk determination method integrating driver ability and confidence according to claim 1 is characterized in that: The method for extracting the four-dimensional grip force feature vectors of the left and right hands in step 4 is as follows: the grip force of the driver's left and right hands is collected in real time by flexible resistive grip force sensors installed at the 3 o'clock and 9 o'clock positions of the steering wheel. Then, a two-stage filtering strategy is used to preprocess the original signal. The intensity deviation, fluctuation stability, normalized mutation index, and change sharpness are extracted from the preprocessed data to construct the four-dimensional grip force feature vectors of the left and right hands. and , the expression is: ； ； in, represents the four-dimensional grip force feature vector of the left hand, represents the four-dimensional grip force feature vector of the right hand, and Represent the intensity deviation of the left hand and right hand respectively, and represent the fluctuation stability of the left and right hands respectively, and represent the normalized mutation index of the left and right hands, respectively, and They represent the sharpness of change for the left and right hands, respectively, and the superscript T stands for transposition; The method for extracting the micro-operation behavior feature vector in step 4 is: the original accelerator pedal pressure P a (t), brake pedal pressure P b (t) and steering wheel angle The signal is processed by sliding window mean filtering, and then the frequency domain features are extracted by short-time Fourier transform of the filtered signal to obtain the power spectrum density of the signal, and the main frequency energy ratio is calculated based on the power spectrum density. , peak frequency , spectrum entropy If the judgment condition is met ,and <1.2, that is, the current time window is identified as the micro-operation behavior segment, and then the signal of the micro-operation behavior segment is extracted, and the damped oscillation function is used for nonlinear fitting, and the physical parameters after fitting are extracted. Based on the obtained physical parameters, the characteristic vector of the micro-operation behavior is constructed. , the expression is: ； in, is the initial amplitude, is the damping ratio, is the oscillation frequency, is the amplitude ratio; The method for extracting the back posture feature vector in step 4 is as follows: collect the three-dimensional coordinate information of the driver's left shoulder key point, right shoulder key point and the middle point of the back spine, compare them with the three-dimensional coordinate information of the corresponding position reference point of the seat left shoulder, the corresponding position reference point of the seat right shoulder, and the center point of the seat back, and calculate the Euclidean distance to obtain the back posture distance index; then calculate the driver correction factor based on the driver's height and body index, and use the driver correction factor to standardize the back posture distance index to obtain the back distance ; Calculate the shoulder angle based on the driver's left shoulder key point and right shoulder key point According to the The back posture stability is calculated by the position of the key points on the back of the frame and the average coordinates within the time window , construct a standardized back posture feature vector , the expression is: ； The superscript T stands for transpose.

5. The vehicle passability risk determination method integrating driver ability and confidence according to claim 1 is characterized in that: The driver's confidence score for the current traffic scenario in step 5 The expression is: ； in: 、 、 are the normalized sub-scores of grip strength feature, micro-manipulation behavior feature, and back posture feature respectively; 、 、 is the corresponding importance weight, satisfying the constraint .

6. The vehicle passability risk determination method integrating driver ability and confidence according to claim 1 is characterized in that: Final pass risk in step 6 The expression is: ； in, and are the risk sensitivity coefficients of ability bias and self-confidence bias respectively; Characterizes the relative degree of inadequacy of the driver's abilities; Indicates the degree of the driver's lack of confidence in the driving task, It is the original passability risk based on structural complexity and vehicle performance; DCCI represents the passability confidence score.

7. The vehicle passability risk determination method integrating driver ability and confidence according to claim 4 is characterized in that: Intensity deviation The expression is: ； in, is the grip strength value at the i-th sampling point in the time window, and are the mean and standard deviation of the baseline period, respectively, and n represents the number of samples; Fluctuation stability The expression is: ； Normalized mutation index The expression is: ； in, The grip strength in the baseline period was extremely poor; is the maximum grip strength of the current window; is the minimum grip strength of the current window; Sharpness of change The expression is: ； in, and Represents the grip strength values of two adjacent sampling points; Main frequency energy ratio The expression is: ； in, is the signal frequency The power spectral density at ; Peak frequency The expression is: ； Spectral entropy The expression is: ； ； in, Representative frequency The power spectral density at the point, Representative frequency The probability distribution of the normalized power spectrum density at the point, N is the window size; The expression of the damped oscillation function in step 4 is: ； in, is the initial amplitude, is the damping ratio, is the natural frequency of the system, is the actual oscillation frequency after damping; is the initial phase angle, set to a constant or zero; is the time variable after the micro-operation starts, signals representing the extracted micro-operation behavior segments; Oscillation frequency ; Amplitude ratio is the ratio of the two peaks.

8. The vehicle passability risk determination method integrating driver ability and confidence according to claim 5 is characterized in that: The expression of the normalized subscore of the grip strength feature is: ； in, Input vector for left or right hand grip force feature; Symmetric weight matrix, modeling the interaction between features; is the first-order linear weight vector; is a scalar bias term; is the Sigmoid activation function, represent The transposed vector of represent The transposed vector of The expression of the normalized sub-score of the micro-operation behavior feature is: ； in, is the original micro-operation behavior feature vector, and are the hidden feature vector and the gating vector respectively; and is the hidden layer parameter, is the weight matrix of the fully connected layer, is the corresponding bias term; the gate unit is composed of the weight matrix of the linear fully connected layer and bias The gate vector generated by Control feature fusion strategy channel by channel; 、 is the output linear combination weight; is the nonlinear Tanh activation function; is the Sigmoid function; ∘ represents element-by-element multiplication, represents the fused micro-operation behavior feature vector, represents the micro-operation behavior characteristic score; The expression of the normalized sub-score of the back posture feature is: ； in, is the spatial posture feature; It is the dynamic posture fluctuation feature; and are the fully connected layer parameters of each branch; is the nonlinear Tanh activation function of each branch; It is the intermediate feature representation of the output of the two branches; To finally fuse the weights and biases, act on the overall representation after branch concatenation; is the Sigmoid function.

9. The vehicle passability risk determination method integrating driver ability and confidence according to claim 6 is characterized in that: In step 6, set the maximum upper limit for the corrected passability risk value , when the calculated passability risk value is greater than the maximum upper limit When the output passability risk value is the maximum upper limit .

10. A vehicle passability risk assessment system that integrates driver ability and confidence is characterized by: The system is used to implement the vehicle passability risk determination method integrating driver ability and confidence as described in any one of claims 1 to 9, and the system includes a driving behavior and environment data acquisition module, a scene matching module, a passability estimation module, a passability scoring module, a driving behavior feature analysis module, a passability confidence scoring module, a passability risk calculation module, and a database; The driving behavior and environment data acquisition module is used to collect the driver's behavior data, driver static characteristics and historical traffic data, and structural characteristics of the current traffic scene in real time. The driver's behavior data includes grip strength, micro-operations, and posture; The scene matching module calculates similarity based on the structural features of the historical traffic sample set and the current traffic scene, and selects the top-K similar samples; The capacity estimation module is used to normalize the filtered similarity data to obtain a matching weight vector; perform weighted fusion based on the capacity vectors in the filtered historical traffic sample set and the calculated matching weight vector to obtain a capacity estimation result for the current traffic scenario; The traffic capability scoring module is used to calculate the traffic capability score of the driver's current traffic scenario; The driving behavior feature analysis module is used to filter and perform time-frequency analysis on the original behavior signals of grip strength, micro-operation, and posture, and construct four-dimensional grip strength feature vectors, micro-operation behavior feature vectors, and back posture feature vectors for the left and right hands; The passing confidence scoring module includes a grip strength feature scoring module, a micro-operation behavior feature scoring module, a back posture feature scoring module, and an overall confidence scoring module; The passability risk calculation module is used to obtain the original passability risk based on structural complexity and vehicle performance, and use the passability score and the passability confidence score to correct the original passability risk based on structural complexity and vehicle performance, and output the passability risk considering individual status; The database stores a driver's historical traffic sample set and personal feature data. The historical traffic sample set stores a structural feature vector and a traffic capability vector of a historical traffic scene.

Images