Comprehensive driving risk quantification method based on multi-angle risk variable weight

Abstract

The invention discloses a comprehensive driving risk quantification method based on a multi-angle risk variable weight, and the method comprises the steps: calculating a multi-angle driving risk based on the real-time driving information of a vehicle, and the multi-angle driving risk comprises a vehicle instability risk, a violation risk and a collision risk; vehicle driving states are judged, wherein the vehicle driving states include the face of potential conflicts and the absence of potential conflicts; determining a variable weight value of the multi-angle driving risk according to the vehicle driving state; and constructing a comprehensive driving risk quantitative model based on the variable weight, and performing model verification. According to the method, the variable scheme of the multi-angle driving risk weight is designed by analyzing the driving state of the vehicle in the dynamic environment, and the variable weight comprehensive driving risk model is constructed, so that the real-time quantification of the driving risk in the dynamic environment is realized, and the risk quantification result is closer to the actual situation.

Description

Technical Field

[0001] The present invention relates to the field of traffic safety research, and in particular to a comprehensive driving risk quantification method based on multi-angle risk variable weights. Background Art

[0002] In a hierarchical autonomous driving system consisting of perception, cognition, decision-making, planning and control, the quantification and assessment of risks can provide important information for safe and smooth decision-making planning. In a complex and changeable driving environment, the impacts and threats to vehicles are diversified. Therefore, compared with the risk quantification of a single factor, the risk quantification method that considers multiple factors can more comprehensively express the potential threats in a dynamic and complex environment, and is also more conducive to ensuring driving safety. At present, a weighted addition method is often used to construct a multi-factor risk quantification model, but the driving environment of the vehicle is dynamically changing, and the weights of various risks should be adjusted with changes in the vehicle's driving state and environment, rather than fixed values. Based on this, this field is in urgent need of a risk quantification method that can design multi-angle driving risk weights by analyzing the driving state of the vehicle in a dynamic environment. Summary of the invention

[0003] In order to overcome the above-mentioned shortcomings and deficiencies of the prior art, an object of the present invention is to provide a comprehensive driving risk quantification method based on multi-angle risk variable weights.

[0004] The purpose of the present invention is achieved through the following technical solutions:

[0005] A comprehensive driving risk quantification method based on multi-angle risk variable weights, including:

[0006] Calculate multi-angle driving risks based on real-time vehicle driving information, including vehicle instability risk, violation risk and collision risk;

[0007] Determination of a vehicle driving state, wherein the vehicle driving state includes facing a potential conflict and having no potential conflict;

[0008] Determine variable weight values ​​of multi-angle driving risks according to the vehicle driving status;

[0009] A comprehensive driving risk quantification model based on variable weights is constructed and verified, and the verified risk quantification model is used to conduct risk quantitative assessment of the vehicle.

[0010] Furthermore, the calculation process of the vehicle instability risk is as follows:

[0011] The mapping relationship model between vehicle dynamics parameters and lateral instability risk is constructed using the yaw angular velocity of the vehicle during driving;

[0012]

[0013] Where, the actual yaw rate γ and the ideal yaw rate γ of the vehicle during driving are * and the critical yaw rate γ ** To ensure the vehicle's driving stability, the actual yaw rate γ should be between the ideal yaw rate γ * and critical yaw rate γ ** The closer γ is to γ * , the higher the stability, the smaller the risk of instability. When γ=γ * When the vehicle reaches a quasi-steady state, the risk of lateral instability is minimal.

[0014] Further, the determination of the vehicle driving state, wherein the vehicle driving state includes facing a potential conflict and having no potential conflict, is specifically performed in the following steps:

[0015] Divide the perception range of the ego vehicle: define the perception range of the vehicle to obtain surrounding information as the perception range of the autonomous driving vehicle;

[0016] Determine whether the vehicle is facing a potential conflict by combining the minimum following safety distance model and the elliptical equation. If facing a potential conflict, classify the potential conflict into longitudinal motion state and lateral motion state according to the vehicle's driving behavior.

[0017] Furthermore, if the driving vehicle is in the middle position, that is, there are lanes on both the left and right sides, the rectangular frame is centered on the driving vehicle, with a length of 100m from the front and rear of the vehicle and a width covering the left, middle and right lanes;

[0018] If the driving vehicle is in the side lane, the length of the rectangular frame is 100m from the front and rear of the vehicle, and the width is the lane where the driving vehicle is located and the adjacent lane.

[0019] Furthermore, variable weight values ​​of multi-angle driving risks are determined according to the vehicle's driving state. Specifically, the entropy weight method is used to calculate the weight values ​​of the vehicle's collision, violation and instability risks under different driving states.

[0020] Furthermore, the entropy weight method is used to calculate the weight values ​​of the collision, violation and instability risks of the vehicle under different driving conditions, specifically:

[0021] For each driving state of the vehicle, multiple vehicle driving samples are randomly selected from the NGSIM dataset. Based on the sequence values ​​of the multi-angle driving risks of the vehicles in the selected samples, the information entropy is used to evaluate the importance of each indicator and obtain the weight coefficient of each indicator.

[0022] Furthermore, to determine whether the vehicle is facing a potential conflict, the minimum following safety distance model and the elliptic equation are combined to determine whether the vehicle is facing a potential conflict, specifically:

[0023] Set the ellipse interaction area, specifically:

[0024]

[0025] Among them, a is the length of the major semi-axis of the ellipse, b is the length of the minor semi-axis of the ellipse, and x, y are the position coordinates of the surrounding vehicles;

[0026] Determine the position of the imitation point and the interaction area. If D(x,y)≤1, it means that a surrounding vehicle is in the interaction area of ​​the vehicle, and the vehicle faces a potential conflict. D(x,y)>1 means that the surrounding vehicles are outside the interaction area and the vehicle has no potential conflict with them.

[0027] Furthermore, in the elliptical interaction region, b=1.5W lane , a=MFD;

[0028] MFD=0.0029(v EV ×3.6) 2 +0.3049(v EV ×3.6);

[0029] Among them, V Ev is the vehicle speed, W lane is the width of the lane where the vehicle is located.

[0030] Furthermore, the construction of a comprehensive driving risk quantification model with variable weights is specifically as follows:

[0031] R(t)=λ1R collision (t)+λ2R violation (t)+λ3R unstable (t)

[0032] In the formula, R collision (t), R violation (t), R unstable (t) represents the risk of vehicle instability, vehicle violation and vehicle conflict faced by the vehicle at time t, and the weight coefficients λ1, λ2 and λ3 correspond to R collision (t), R violation (t), R unstable The weight coefficient in the table (t) reflects the impact of diversified risk factors on driving safety.

[0033] Furthermore, the weight coefficients of each indicator are as follows:

[0034] No potential conflict with surrounding vehicles: λ1=0, λ2=0.68, λ3=0.32;

[0035] There is a potential conflict with Zhouche.

[0036] Longitudinal motion state: λ1=0.37, λ2=0.41, λ3=0.21;

[0037] Lateral motion state: λ1=0.55, λ2=0.17, λ3=0.28.

[0038] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0039] (1) The present invention takes into account the impact of multiple factors on driving safety from the perspective of vehicle driving. At the instability angle, the real-time dynamic parameters of the vehicle are used to construct a mapping relationship model of lateral instability risk. A multi-angle driving risk quantification model is established by combining the vehicle collision and violation angles to more comprehensively express the potential threats in a dynamic and complex environment and ensure driving safety.

[0040] (2) The present invention designs a variable scheme for multi-angle driving risk weights by analyzing the driving status of vehicles in a dynamic environment. The analysis of the vehicle driving status is based on the vehicle perception range, combined with the elliptical equation and the minimum safe following distance to determine the interaction range, so as to judge the potential conflict state of the vehicle based on the real-time running speed of the vehicle and the driving position of the surrounding vehicles. Compared with the prior art, a method for determining the interaction range based on the real-time running speed of the vehicle and the driving position of the surrounding vehicles is provided, which is more adaptable to the time-varying nature of the dynamic environment.

[0041] (3) The present invention adopts a variable weight method to construct a driving risk quantification model. The model can adjust the risk weights of vehicle collision, violation and instability angle according to changes in the vehicle's motion state and driving environment. Compared with the risk quantification model in the prior art that only adopts fixed weights, the risk quantification model constructed by the present invention can adapt to the dynamic and time-varying driving environment, making the risk assessment results more in line with the actual situation. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 It is a work flow chart of the present invention;

[0043] Figure 2 It is a schematic diagram of the specific determination of the vehicle driving state of the present invention;

[0044] Figure 3 It is a schematic diagram of a variable scheme of multi-angle driving risk weights of the present invention;

[0045] Figure 4 Schematic diagram of the quantification result of vehicle driving risk verified by the model in an embodiment of the present invention;

[0046] Figure 5 It is a schematic diagram comparing the risk quantification effects of the method of the present invention and the risk fixed weight value model. DETAILED DESCRIPTION

[0047] The present invention will be further described in detail below in conjunction with the examples, but the embodiments of the present invention are not limited thereto.

[0048] Example

[0049] In this embodiment, the highway driving environment is taken as an example. The road where the vehicle is driving is a highway, and the vehicle driving environment is a normal traffic environment. Extreme working conditions and extreme weather environments are not considered. In order to facilitate the description of the relationship between various traffic participants in the highway driving environment, such as Figure 2 As shown, the object vehicle to be quantified for driving risk is the ego vehicle (EV, ego vehicle, which can be a future self-driving car), and the other vehicles include the surrounding vehicles on the road (highway) where the ego vehicle is traveling, and the surrounding vehicles include the leading vehicle (LV, leading vehicle, referring to the first vehicle in front of the ego vehicle EV) in the lane where the ego vehicle EV is located on the driving road, and the lane change vehicle (LCV, lane change vehicle) with the intention of changing lanes in the adjacent lane of the lane where the ego vehicle EV is located. In addition, in addition to the above three, the other surrounding vehicles on the road where the ego vehicle EV is traveling are the dynamic traffic vehicles (SV) around the ego vehicle, and the environmental information refers to the road information including the road width, signs and markings, and speed limit signs.

[0050] like Figure 1 As shown, a comprehensive driving risk quantification method based on multi-angle risk variable weights includes the following steps:

[0051] S1 calculates multi-angle driving risks based on the real-time driving information of the vehicle, wherein the multi-angle driving risks include the risk of vehicle instability, the risk of violation and the risk of collision.

[0052] (1) Calculate the instability risk of the vehicle from the perspective of vehicle driving stability. The specific process is as follows:

[0053] The mapping relationship model between vehicle dynamics parameters and lateral instability risk is constructed using the yaw angular velocity of the vehicle during driving, as follows:

[0054]

[0055] The formula includes the actual yaw rate γ and the ideal yaw rate γ when the vehicle is traveling. * and the critical yaw rate γ ** To ensure the vehicle's driving stability, the actual yaw rate γ should be between the ideal yaw rate γ * and critical yaw rate γ ** The closer γ is to γ * , the higher the stability, the smaller the risk of instability. γ *is the ideal yaw rate, which is used to describe the driver's desired steering characteristics. Theoretical studies have shown that the yaw rate of the car determined by the linear two-degree-of-freedom vehicle model is the most stable ideal yaw rate. The closer γ is to γ * , the higher the vehicle stability, the lower the possibility of instability. When γ=γ * When γ, the car reaches a quasi-steady state, and the risk of lateral instability of the car is minimal. * It can be calculated by the following vehicle two-degree-of-freedom model:

[0056]

[0057] In the formula, δ is the front wheel turning angle, L is the vehicle wheelbase, v is the instantaneous speed of the vehicle, μ is the road adhesion coefficient, and g is 9.80m / s 2 , K is the stability factor of the vehicle.

[0058] γ ** is the critical yaw rate. Research and practice show that among the three steady-state responses of the vehicle, a car with moderate understeering characteristics is more suitable for the driver to control and has good handling stability. According to the relationship between the vehicle yaw rate gain and the vehicle speed during neutral steering, the critical yaw rate γ can be obtained. ** as follows:

[0059]

[0060] If γ<γ ** , indicating that the vehicle is in an understeering state with good handling stability; if γ>γ ** , indicating that the vehicle is in an oversteering state with the lowest stability and there is a greater risk of instability.

[0061] From the above records, it can be seen that the critical yaw rate can be understood as the upper limit calculated based on the driving state of the vehicle; if the actual yaw rate reaches the critical yaw rate, the instability risk is 1. The ideal yaw rate can be considered as the best yaw rate value of the vehicle at this time. If this value is reached, it means that the driver's expected steering characteristics can be achieved, and the instability risk is 0.

[0062] (2) Under road constraints, the violation risk of the vehicle is calculated based on the common adverse driving conditions of the vehicle, which is specifically:

[0063] Based on the driving information, the lane departure risk amount and the speeding risk amount of the vehicle are obtained. Based on the lane departure risk amount and the speeding risk amount of the vehicle, the violation risk of the vehicle driving risk representing the road constraint angle is obtained, which is specifically:

[0064] Based on the driving information, the lane departure risk of the vehicle is obtained and calculated as follows:

[0065] R line =ke -Δd , where R line represents the risk of lane deviation of the vehicle; k represents the constraint coefficient. When the lane line is a solid line, k = 1; when the lane line is a dotted line, k = 1-p(lc), where p(lc) represents the probability of vehicle lane change; Δd ​​represents the minimum lateral distance between the edge of the vehicle and the lane lines on both sides.

[0066] That is, △d=min(d left ,d right ).

[0067] Based on the driving information, the speeding risk of the vehicle is obtained, which is calculated by the following formula:

[0068]

[0069] Among them, R limitspeed represents the risk of speeding of the vehicle, V EV represents the total speed of the vehicle, V limit Indicates the speed limit.

[0070] According to the lane departure risk and speeding risk of the vehicle, the violation risk of the vehicle is obtained and calculated by the following formula: R road =0.5*R line +0.5*R limitspeed , where R road Indicates risk of noncompliance.

[0071] R line represents the lane departure risk of the vehicle, R limitspeed Indicates the risk of speeding of the vehicle.

[0072] (3) From the perspective of vehicle-to-vehicle collision, the vehicle collision risk is calculated by combining the vehicle collision urgency and collision severity, specifically:

[0073] Obtain vehicle data of the own vehicle and other vehicles on the road where the own vehicle is traveling, the vehicle data including, in a preset coordinate system, the lateral position of the own vehicle, the longitudinal position of the own vehicle, the lateral speed of the own vehicle, the longitudinal speed of the own vehicle, the width of the own vehicle, the length of the own vehicle, the lateral position of the front vehicle, the longitudinal position of the front vehicle, the lateral speed of the front vehicle, the longitudinal speed of the front vehicle, the width of the front vehicle, the length of the front vehicle, the lateral position of the lane-changing vehicle, the longitudinal position of the lane-changing vehicle, the lateral speed of the lane-changing vehicle, the longitudinal speed of the lane-changing vehicle, the width of the lane-changing vehicle, the length of the lane-changing vehicle, the longitudinal position of the lane-changing vehicle, the heading angle of the own vehicle, the heading angle of the front vehicle, the heading angle of the lane-changing vehicle, the lateral position of the left lane line of the lane where the own vehicle is located, and the lateral position of the right lane line of the lane where the own vehicle is located.

[0074] Based on the vehicle data, a lateral collision emergency coefficient and a longitudinal collision emergency coefficient between the vehicle and the other vehicle are obtained;

[0075] ① Based on the lateral collision urgency coefficient and the longitudinal collision urgency coefficient, the lateral and longitudinal collision urgency degrees of the vehicle from other vehicles are obtained through a piecewise function;

[0076] ② Based on the collision severity prediction model, the collision severity between the vehicle and other vehicles is predicted to obtain the collision severity S;

[0077] ③ Based on the severity of the collision between the vehicle and the other vehicle and the collision urgency, a collision risk value representing the driving risk of the vehicle is obtained.

[0078] Rvehicle=U 1 / S ,

[0079] Among them, Rvehicle represents the quantitative value of driving risk, and U represents the degree of collision urgency, which is equal to the collision risk value.

[0080] The collision severity S is divided into three levels, S=1 indicates a mild collision, S=2 indicates a moderate collision, and S=3 indicates a severe collision.

[0081] For the specific acquisition process, please refer to the record of a method and system for quantifying the driving risk of a vehicle considering collision between the vehicle and other vehicles in publication number CN 116238483A.

[0082] It should be noted that in this embodiment, the vehicle instability risk, violation risk and collision risk are all quantitative values.

[0083] S2: Determine the driving state of the vehicle, wherein the driving state of the vehicle includes facing a potential conflict and having no potential conflict, such as Figure 2 As shown, specifically:

[0084] Divide the perception range of the ego vehicle: define the perception range of the vehicle to obtain surrounding information as the perception range of the autonomous driving vehicle;

[0085] There are several situations:

[0086] If the driving vehicle is in the middle position, that is, there are lanes on both the left and right sides, the rectangular frame is centered on the driving vehicle, with a length of 100m from the front and rear of the vehicle and a width covering the left, middle and right lanes;

[0087] If the driving vehicle is in the side lane, the length of the rectangular frame is 100m from the front and rear of the vehicle, and the width is the lane where the driving vehicle is located and the adjacent lane.

[0088] To determine whether the vehicle is facing a potential conflict, the minimum safe following distance model (MFD) and the elliptic equation are combined to determine whether the vehicle is facing a potential conflict. If facing a potential conflict, the potential conflict is classified into longitudinal motion state and lateral motion state according to the vehicle's driving behavior.

[0089] Specifically:

[0090] The elliptical interaction area can be defined by the following formula:

[0091]

[0092] By imitating the position judgment method of points and circles, the position coordinates of any vehicle are brought into the D(x,y) formula, and it can be judged whether the surrounding vehicles are in the interaction area of ​​the autonomous driving vehicle based on D(x,y). If D(x,y)≤1, it means that the vehicle is in the interaction area of ​​the EV, and the vehicle faces potential conflict; D(x,y)>1 means that the vehicle is outside the interaction area of ​​the EV, and the vehicle has no potential conflict with other vehicles. In the formula, b is the length of the minor semi-axis of the ellipse, which represents the lateral interaction area of ​​the autonomous driving vehicle, and b=1.5W lane , where W lane is the width of the lane where the vehicle is located; a is the length of the major semi-axis of the ellipse, which represents the length of the longitudinal interaction area of ​​the autonomous driving vehicle. This value is related to the motion state of the vehicle. The value of a is determined by the fitting formula of the minimum safe following distance (MFD) of the intelligent vehicle obtained by Japanese scholar T. Kuroda through experiments. a=MFD, as shown below:

[0093] MFD=0.0029(v EV ×3.6) 2 +0.3049(v EV ×3.6).

[0094] When a vehicle faces a potential conflict, it is further divided into longitudinal motion state and lateral motion state according to the vehicle.

[0095] Furthermore, the longitudinal motion state includes acceleration and deceleration of the vehicle, and the lateral motion state includes lane changing and turning.

[0096] S3 determines the variable weight value of multi-angle driving risk according to the vehicle driving state, specifically:

[0097] like Figure 3As shown in the figure, in a dynamic and time-varying environment, the risk factors faced by a vehicle in different driving states are different, and the degree of influence of each risk factor on driving safety also varies. In order to make the risk assessment more in line with the actual situation, this method designs the weights of driving risks at various angles according to the vehicle's driving state.

[0098] Based on the determination of the vehicle driving state in S2, in order to avoid the influence of subjective factors, the entropy weight method is used to calculate the weight values ​​of the vehicle's collision, violation and instability risks under different driving conditions. The entropy weight method is a mathematical method that uses the concept of information entropy to comprehensively consider and analyze the amount of information provided by each indicator and determine the weight of each indicator according to the amount of information. It is widely used because it avoids the influence of subjective factors. The larger the information entropy value of an indicator, the lower the importance of the indicator and the smaller its weight value; conversely, the larger the weight value of the indicator.

[0099] In order to obtain the information entropy of each risk, for each driving state of the vehicle, 20 vehicle driving samples are randomly selected from the NGSIM dataset, and the sequence value R of the multi-angle driving risk of the vehicle in the selected samples is used to calculate the risk information entropy. collision (t), R violation (t) and R unstable (t), information entropy is used to evaluate the importance of each indicator, and the weight coefficient of each indicator is obtained. The specific calculation process is as follows:

[0100] Step a: Standardize the continuous data corresponding to the indicator. collision (t) as an example, let R(t) = R collision (t), assuming the standardized value is R'(t):

[0101]

[0102] Step b: Calculate the information entropy of the indicator. Calculate R according to the definition of information entropy collision Information entropy of (t):

[0103]

[0104] where p t is the variability index:

[0105]

[0106] Step c: Determine the weight of each indicator. According to S2, we can get R collision (t), R violation (t) and R unstable (t) The information entropy of the indicators are E1, E2, and E3 respectively, and the weights of the indicators are as follows:

[0107]

[0108] According to steps ac, the weight values ​​of the risks at each angle under different driving conditions of the vehicle can be obtained, as shown in Table 1 below.

[0109] Table 1 Weight combinations of different driving states

[0110]

[0111] S4 constructs a comprehensive driving risk quantification model based on variable weights, verifies the model, and uses the verified model to conduct risk quantification assessment.

[0112] Building a quantitative model:

[0113] The entropy weight method is used to analyze the actual driving data of vehicles to determine the weights of multi-angle risks in the comprehensive risk assessment. These weights objectively reflect the impact of different risk factors on vehicle driving safety in a high-speed environment to a certain extent.

[0114] Finally, the quantitative model of comprehensive driving risk was determined as follows:

[0115] R(t)=λ1R collision (t)+λ2R violation (t)+λ3R unstable (t)

[0116] In the formula, R collision (t), R violation (t), R unstable (t) represents the risk of vehicle instability, vehicle violation and vehicle conflict faced by the vehicle at time t, and the weight coefficients λ1, λ2 and λ3 correspond to R collision (t), R violation (t), R unstable The weight coefficient in the table (t) reflects the impact of diversified risk factors on driving safety.

[0117] Model Validation:

[0118] In order to analyze the effect of the driving risk quantification method with multi-angle risk variable weights in actual risk assessment, a 30-second vehicle driving process is taken as an example, and the leftmost vehicle is selected as the research object. The figure below shows the three driving states of the vehicle and the driving risk values ​​obtained by different risk quantification models.

[0119] like Figure 4As shown in the figure, the red curve shows the result of the variable weight risk quantification model, which adjusts the risk weight according to the vehicle driving state to quantify the vehicle driving risk. The blue curve represents the result of the fixed weight risk quantification model, in which the weights of collision, violation and instability risks are calculated from the vehicle simulation data by the entropy weight method, which are 0.76, 0.11 and 0.13 respectively, and these weights will not change with the change of vehicle driving state.

[0120] The comparison shows that the two models are significantly different in cases ② and ③. Figure 5 Analyze the driving status of the vehicle.

[0121] In situation ②, the vehicle has no potential conflict, but speeding will become a major safety hazard, which is likely to lead to a single-vehicle traffic accident. The variable weight risk model adjusts the weight of the violation risk to more reasonably reflect the impact of the vehicle's own factors on driving safety in the absence of potential conflict. In situation ③, although the vehicle has a potential conflict with the vehicle cutting in from the right, the vehicle is still in a longitudinal motion state and the speed is high, so the risk obtained by the variable weight risk model is higher than that of the fixed weight risk model.

[0122] The above results show that the variable weight risk quantification model can adjust the weights of different risk angles according to changes in the vehicle driving environment. When the vehicle faces a potential conflict, the model can reasonably allocate the weights of collision, violation and instability risks according to the vehicle's driving state; when there is no potential conflict, it pays more attention to the risks that may be caused by the vehicle's driving state, so that the risk quantification results are closer to the actual situation.

[0123] The above embodiments are preferred implementation modes of the present invention, but the implementation modes of the present invention are not limited to the embodiments. Any other changes, modifications, substitutions, combinations, and simplifications that do not deviate from the spirit and principles of the present invention should be equivalent replacement methods and are included in the protection scope of the present invention.

Claims

1. A comprehensive driving risk quantification method based on multi-angle risk variable weights, characterized in that: include: Calculate multi-angle driving risks based on real-time vehicle driving information, including vehicle instability risk, violation risk and collision risk; Determination of a vehicle driving state, wherein the vehicle driving state includes facing a potential conflict and having no potential conflict; Determine variable weight values ​​of multi-angle driving risks according to the vehicle driving status; A comprehensive driving risk quantification model based on variable weights is constructed and verified, and the verified risk quantification model is used to conduct risk quantitative assessment of the vehicle.

2. The comprehensive driving risk quantification method according to claim 1 is characterized in that: The calculation process of the vehicle instability risk is as follows: The mapping relationship model between vehicle dynamics parameters and lateral instability risk is constructed using the yaw angular velocity of the vehicle during driving; In the formula, the actual yaw rate γ and the ideal yaw rate γ of the vehicle during driving are * and the critical yaw rate γ ** To ensure the vehicle's driving stability, the actual yaw rate γ should be between the ideal yaw rate γ * and the critical yaw rate γ ** The closer γ is to γ * , the higher the stability, the smaller the risk of instability. When γ=γ * When the vehicle reaches a quasi-steady state, the risk of lateral instability is minimal.

3. The comprehensive driving risk quantification method according to claim 1 is characterized in that: The determination of the vehicle driving state, wherein the vehicle driving state includes facing a potential conflict and having no potential conflict, is specifically performed in the following steps: Dividing the perception range of the ego vehicle: The perception range of the vehicle to obtain surrounding information is defined as the perception range of the autonomous driving vehicle; Determine whether the vehicle is facing a potential conflict by combining the minimum following safety distance model and the elliptical equation. If facing a potential conflict, classify the potential conflict into longitudinal motion state and lateral motion state according to the vehicle's driving behavior.

4. The comprehensive driving risk quantification method according to claim 3 is characterized in that: If the driving vehicle is in the middle position, that is, there are lanes on both the left and right sides, the rectangular frame is centered on the driving vehicle, with a length of 100m from the front and rear of the vehicle and a width covering the left, middle and right lanes; If the driving vehicle is in the side lane, the length of the rectangular frame is 100m from the front and rear of the vehicle, and the width is the lane where the driving vehicle is located and the adjacent lane.

5. The comprehensive driving risk quantification method according to claim 1 is characterized in that: The variable weight values ​​of multi-angle driving risks are determined according to the vehicle's driving status. Specifically, the entropy weight method is used to calculate the weight values ​​of the vehicle's collision, violation and instability risks under different driving conditions.

6. The comprehensive driving risk quantification method according to claim 5 is characterized in that: The entropy weight method is used to calculate the weight values ​​of collision, violation and instability risks of vehicles under different driving conditions, specifically: For each driving state of the vehicle, multiple vehicle driving samples are randomly selected from the NGSIM dataset. Based on the sequence values ​​of the multi-angle driving risks of the vehicles in the selected samples, the information entropy is used to evaluate the importance of each indicator and obtain the weight coefficient of each indicator.

7. The comprehensive driving risk quantification method according to claim 2 is characterized in that: Determine whether the vehicle is facing a potential conflict by combining the minimum following safety distance model and the elliptic equation. Specifically: Set the ellipse interaction area, specifically: Among them, a is the length of the major semi-axis of the ellipse, b is the length of the minor semi-axis of the ellipse, and x, y are the position coordinates of the surrounding vehicles; Determine the position of the imitation point and the interaction area. If D(x,y)≤1, it means that a surrounding vehicle is in the interaction area of ​​the vehicle, and the vehicle faces a potential conflict. D(x,y)>1 means that the surrounding vehicles are outside the interaction area and the vehicle has no potential conflict with them.

8. The comprehensive driving risk quantification method according to claim 7 is characterized in that: In the elliptical interaction region, b = 1.5W lane , a=MFD; MFD=0.0029(v EV ×3.6) 2 +0.3049(v EV ×3.6); Among them, V EV is the vehicle speed, W lane It refers to the width of the lane in which the vehicle is located.

9. The comprehensive driving risk quantification method according to any one of claims 1 to 8, characterized in that: The construction of the comprehensive driving risk quantification model with variable weights is specifically as follows: R(t)=λ1R collision (t)+λ2R violation (t)+λ3R unstable (t) In the formula, R collision (t), R violation (t), R unstable (t) represents the risk of vehicle instability, vehicle violation and vehicle conflict faced by the vehicle at time t, and the weight coefficients λ1, λ2 and λ3 correspond to R collision (t), R violation (t), R unstable (t) Weight coefficient of the table.

10. The comprehensive driving risk quantification method according to claim 9, characterized in that: The weight coefficients of each indicator are as follows: No potential conflict with surrounding vehicles: λ1=0, λ2=0.68, λ3=0.32; Potential conflicts with Zhouche: Longitudinal motion state: λ1=0.37, λ2=0.41, λ3=0.21; Lateral motion state: λ1=0.55, λ2=0.17, λ3=0.28.

Images