A vehicle driving risk field modeling method under a networked environment

Abstract

The application provides a vehicle driving risk field modeling method in a network environment. The environment information around a target vehicle is collected through a vehicle networking device, a millimeter wave radar, a GPS and the like; the target networked vehicle is selected as a reference system, and the speed and distance of surrounding objects relative to the target vehicle are calculated; the distance between two vehicles is corrected as a pseudo distance according to the speed component, so that the change of the safety risk degree of the vehicle approaching the target vehicle from different angles is better described; the target vehicle is taken as a stationary reference system, the relative speed between the surrounding objects and the target vehicle is used to unify the moving objects and the stationary objects, the driving risk of each object to the target vehicle is calculated in combination with the equivalent mass, a road environment influence factor and the like, and is superposed, so that the driving risk field modeling and real-time dynamic risk evaluation in the network environment are realized, and theoretical support is provided for the active safety control and trajectory planning method of the intelligent networked vehicle.

Description

Technical Field

[0001] This invention relates to the field of intelligent connected vehicles and driving risk assessment, specifically to a method for modeling vehicle driving risk fields in a connected environment. Background Technology

[0002] In recent years, with the rapid development of artificial intelligence and the Internet of Things (IoT) technologies, vehicle-to-everything (V2X) technology has become a major direction for the intelligent and connected development of the global automotive and transportation sectors. In a connected environment, vehicles are connected to the internet, and communication between vehicles and between vehicles and infrastructure will significantly change traffic patterns on the road, providing new solutions to the critical issue of traffic safety. Fully utilizing the advantages of information sharing in a connected environment and appropriately assessing vehicle driving risks can help prevent and reduce traffic accidents and improve road traffic safety.

[0003] Existing vehicle driving risk assessment models are typically based on vehicle kinematics and dynamics theories. The expression of vehicle driving safety is based on vehicle state information (speed, acceleration, yaw rate, etc.) and information on the relative motion relationship between two vehicles (relative speed, relative distance, etc.). Tao Pengfei et al. summarized various vehicle behaviors as the interaction between efficiency and safety factors, and based on the concept of APF (Advanced Persistent Factors), abstracted these two factors as the driving force and resistance experienced by the driver, thus establishing a car-following model. Yang considered the vehicle as an independent charge unit in a potential energy field, simplifying the influence of various factors on the vehicle's following behavior as the attraction and repulsion forces between vehicles, and studied the influence of lateral offset on following behavior accordingly. Sattel designed a cooperative vehicle path planning algorithm using the concept of elastic order in robotics, and applied it to lane keeping and collision avoidance systems of autonomous vehicles. This algorithm considers the risk field formed by the road centerline and its boundaries. Rossetter et al. used artificial potential field theory to study lane keeping systems. They assumed that lane lines could generate potential energy fields that varied depending on the road alignment. They established a Lyapunov function using the vehicle's potential energy in the potential energy field and its own kinetic energy to control the vehicle's lateral movement and achieve lane keeping. Wang Jianqiang et al. proposed a new concept, "driving risk field," which characterizes the degree of risk to vehicle safety posed by various human-vehicle-road elements. This includes the "kinetic energy field" determined by moving objects such as motor vehicles and non-motor vehicles on the road, the "potential energy field" determined by road environmental factors, and the "behavioral field" determined by the driver's individual characteristics. In summary, most existing risk assessment models suffer from limitations in application scenarios. More complex models categorize risk fields, requiring independent calculations for each field, leading to cumbersome computations and limiting the methods' practicality, accuracy, and scalability, especially in connected environments where multiple vehicles need to be considered. Furthermore, these methods use actual distance to characterize the impact of distance on vehicle driving risk; that is, regardless of the angle at which other vehicles approach the target vehicle, as long as the distance and speed are the same, their contribution to the vehicle's driving risk is identical. However, in real-world scenarios, the risks posed by a vehicle passing from an adjacent lane and following in the same lane at the same distance and speed are clearly different, indicating that the representation of distance's impact on driving risk using actual distance is somewhat inaccurate.

[0004] Therefore, addressing the shortcomings of existing vehicle driving risk assessment models, this paper proposes a new method for assessing vehicle driving risk in a connected environment. The method selects the target connected vehicle as the reference frame, acquires environmental information surrounding the target connected vehicle through communication technology, and uses velocity components to correct distance into pseudo-distance, thus better describing the impact of vehicles approaching the target connected vehicle from different angles on driving risk. Using the target connected vehicle as a stationary reference, and combining the velocity information of surrounding objects relative to the target connected vehicle, the method unifies moving and stationary objects in the field to calculate the risk of the target connected vehicle, establishing a driving risk field for that target connected vehicle. This enables real-time dynamic risk assessment of the target connected vehicle, providing theoretical support for active safety control and trajectory planning methods for intelligent connected vehicles. Summary of the Invention

[0005] To address the shortcomings of existing vehicle driving risk assessment models, this invention aims to correct distance into pseudo-distance by utilizing velocity components. By uniformly processing moving and stationary objects in the field through relative velocity, a driving risk field for the target connected vehicle is established. To achieve this objective, this paper provides a method for modeling the vehicle driving risk field in a connected environment. The specific implementation steps of this method are as follows:

[0006] Step 1: Collect information about the vehicle's surrounding environment through vehicle networking devices, millimeter-wave radar, GPS and other devices. Specifically, this includes lane lines, obstacles, traffic lights and the movement data of surrounding vehicles (including vehicles in front and behind in the vehicle's own lane and vehicles in front and behind in the left and right lanes).

[0007] Step 2: Based on the data from Step 1, select vehicle j as the target connected vehicle and use it as a reference point. Assuming it is stationary, the velocity of the surrounding object i relative to the target connected vehicle is:

[0008]

[0009] Where, Δx ij Let Δt be the relative displacement of object i relative to the target connected vehicle j.

[0010] Step 3: The risk quantity is determined based on the probability and severity of an accident. Based on Step 2, when the target connected vehicle j is used as a reference point, the risk quantity E posed by object i to the surrounding environment is... v_ij for:

[0011]

[0012] Where R i M is the influencing factor of the road conditions where object i is located; i r is the equivalent mass of object i; ij This represents the distance vector between surrounding object i and target connected vehicle j; Indicates the direction of the risk originating from object i and the target connected vehicle j; ij Let be the pseudo distance between object i and target connected vehicle j.

[0013] Step 31: Establish a Cartesian coordinate system with the direction of motion of the target connected vehicle j as the horizontal axis. Let the position of object i be (x... i ,y i The location of the target connected vehicle j is (x j ,y j If ), then the actual distance between the two is:

[0014]

[0015] Assuming the target connected vehicle is traveling at a constant speed in a straight line within the lane, and a following vehicle is traveling at a higher speed, if the actual distance between the two vehicles is the same, the risk posed by the following vehicle approaching the target connected vehicle from the same lane is far greater than that of the following vehicle approaching the target connected vehicle from either side lane. Therefore, to better characterize the impact of distance on vehicle driving risk, a pseudo-distance is generated by modifying the actual distance using a velocity component. This pseudo-distance better describes the change in driving risk when a vehicle approaches the target connected vehicle from different angles. The pseudo-distance l between object i and target connected vehicle j is... ij The expression is:

[0016]

[0017] Where θ i θ is the angle between the relative velocity direction and the x-axis, with clockwise being positive, and k1 is an undetermined coefficient.

[0018] Step 32: Confirm the undetermined coefficient k1. Considering that intelligent connected vehicles will transfer driving control to the driver in emergency situations, the calculation of k1 takes into account the driver's reaction time as t. s The vehicle's initial velocity is v0, and its deceleration is a. According to the velocity-displacement formula for uniform acceleration:

[0019]

[0020] The undetermined coefficient k1 is calculated under two scenarios: one with a relatively high relative speed and the other with a relatively low relative speed. Scenario 1 involves vehicles in the same lane traveling at a relatively high relative speed v. 01 Approaching the car in front, if the distance between the two cars along the x-axis is x1, then according to the acceleration and velocity conditions mentioned above, the two cars can just remain relatively stationary before a collision; Scenario 2 involves vehicles in the same lane traveling at a relatively small relative speed v. 02Approaching the car in front, if the distance between the two cars along the x-axis is x2, then according to the acceleration and velocity conditions mentioned above, the two cars can just remain relatively stationary before a collision. The probability of a collision is the same in both scenarios, and we can further assume that the pseudo-distances are the same, that is:

[0021]

[0022] Substituting the velocity and displacement, we get:

[0023]

[0024] k1 is related to factors such as road speed limits, road conditions, and driver reaction time, and its specific value can be selected according to the actual situation.

[0025] Assume that on a certain road, the driver's reaction time is 0.5 seconds and the vehicle's deceleration is -6 m / s². 2 Scenario 1: Vehicles in the same lane move in a V-shape. ij_1 =Approaching a vehicle parked on the side of the road at a speed of 30 m / s, if the distance between the two vehicles in the x-axis direction is x ij_1 =90m, then the two cars can just remain relatively stationary before the collision; Scenario 2 is that vehicles in the same lane move at v ij_2 A car approaches another car with a relative speed of 10 m / s. If the two cars remain relatively stationary just before a collision, then the distance x between the two cars along the x-axis at that moment is... ij_2 =18m. Substituting the relative speed of vehicles, the distance difference on the x-axis, the vehicle deceleration, and the driver's reaction time for the two scenarios on the same road into the above formula, we obtain:

[0026] k1≈0.15

[0027] Step 33: The vehicle's inherent attributes mainly include its type and mass. Generally, the larger the vehicle's dimensions and the greater its mass, the more severe the collision will be. Wang Jianqiang et al. believe that the severity of a vehicle's collision is correlated with both the target connected vehicle's current speed and mass, and they derived an expression for the equivalent mass of object i by fitting highway speed and accident data as follows:

[0028] M i =m i ×(1.566×10-14×v i 6.687 +0.3345)

[0029] Where m i Let be the actual mass of object i, and v be the velocity of object i. This expression shows that at the same vehicle speed, a larger actual mass results in a larger equivalent mass; conversely, for the same actual mass, a higher vehicle speed results in a larger equivalent mass. In other words, actual mass and vehicle speed influence risk based on the severity of the accident.

[0030] Step 34: Driving risk is related to road conditions; the worse the road conditions, the greater the likelihood of an accident. Road conditions include factors such as the road's coefficient of adhesion, road curvature, road slope, and visibility. These factors can be uniformly expressed as "road condition factors." Define the location of object i (x... i ,y i The road condition influencing factors at point ) are shown in the formula, namely:

[0031]

[0032] In the formula, δ i For road visibility; μ i ρ is the road surface adhesion coefficient; i τ is the road curvature; i Let be the road slope; γ1, γ2, γ3, and γ4 are all undetermined constants, and γ1, γ2 < 0, γ3, γ4 > 0; μ * δ is the standard road surface adhesion coefficient; * Standard road visibility; ρ * For standard road curvature; τ * This is the standard road gradient. R i R represents the road condition influence factor, which describes the varying degrees of potential danger to driving caused by different road conditions. For object i, its corresponding road condition influence factor R is... i By (x) i y i The road surface adhesion coefficient, environmental visibility, road curvature, and slope at point (x) determine the road impact factor. As the road surface adhesion coefficient and road visibility decrease, while road curvature and slope increase, the road impact factor increases. With other conditions remaining constant, the interaction between the vehicle and (x) i y i The probability of a collision with object i at location i increases, thus increasing the driving hazard.

[0033] Step 4, establish the risk field. Calculate and sum the risk quantities posed by all objects on the road to the target connected vehicle:

[0034]

[0035] E j This refers to the driving risk of the target connected vehicle j at this time. Attached Figure Description

[0036] Figure 1 This is a schematic diagram of the overall concept of the present invention.

[0037] Figure 2 This is a schematic diagram of pseudo-distance. Detailed Implementation

[0038] To address the shortcomings of existing vehicle driving risk assessment models, this invention aims to correct distance into pseudo-distance by utilizing velocity components. It then uses relative velocity to uniformly process moving and stationary objects in the field, establishing a driving risk field for the target connected vehicle. The overall conceptual framework of this invention's vehicle driving risk field modeling method in a connected environment is shown in the diagram below. Figure 1 As shown, the specific implementation method includes the following steps:

[0039] Step 1: Collect information about the vehicle's surrounding environment through vehicle networking devices, millimeter-wave radar, GPS and other devices. Specifically, this includes lane lines, obstacles, traffic lights and the movement data of surrounding vehicles (including vehicles in front and behind in the vehicle's own lane and vehicles in front and behind in the left and right lanes).

[0040] Step 2: Based on the data from Step 1, select vehicle j as the target connected vehicle and use it as a reference point. Assuming it is stationary, the velocity of the surrounding object i relative to the target connected vehicle is:

[0041]

[0042] Where, Δx ij Let Δt be the relative displacement of object i relative to the target connected vehicle j.

[0043] Step 3: The risk quantity is determined based on the probability and severity of an accident. Based on Step 2, when the target connected vehicle j is used as a reference point, the risk quantity E posed by object i to the surrounding environment is... v_ij for:

[0044]

[0045] Where R i M is the influencing factor of the road conditions where object i is located; i r is the equivalent mass of object i; ij This represents the distance vector between surrounding object i and target connected vehicle j; Indicates the direction of the risk originating from object i and the target connected vehicle j; ij Let be the pseudo distance between object i and target connected vehicle j.

[0046] Step 31: Establish a Cartesian coordinate system with the direction of motion of the target connected vehicle j as the horizontal axis. Let the position of object i be (x... i ,y i The location of the target connected vehicle j is (x j ,y j If ), then the actual distance between the two is:

[0047]

[0048] If the risk level is calculated using the actual distance |r ij If we assume that other vehicles approach the target connected vehicle at the same angle, their contribution to the risk field is the same, regardless of the angle at which they approach, as long as the distance to the target connected vehicle is the same, this is inconsistent with reality. Assuming the target connected vehicle is traveling at a constant speed in a straight line within the lane, and a following vehicle is traveling at a higher speed, if the actual distance between the two vehicles is the same, the risk posed by the following vehicle approaching the target connected vehicle in the same lane is far greater than the risk posed by the following vehicle approaching the target connected vehicle from either side lane. This phenomenon occurs because there is no velocity component in the direction perpendicular to the vehicle's direction of travel, therefore no vertical displacement occurs at the current moment. This method uses the velocity component to correct the actual distance into a pseudo-distance, such as... Figure 2 The following vehicle i travels at a relative speed v ij Approaching or moving away from the vehicle in front (j, r) ij The actual distance between the two workshops, l ij This represents the pseudo-distance between the two vehicles. When the two vehicles approach each other, their relative speed is v. ij Greater than 0, pseudo distance l ij It will be much smaller than the actual distance r ij When the two vehicles are moving away from each other, their relative speed is v. ij Less than 0, pseudo distance l ij Greater than the actual distance r ij This better describes the change in the level of safety risk when a vehicle approaches a target connected vehicle from different angles, and the pseudo-distance l between object i and target connected vehicle j. ij The expression is:

[0049]

[0050] Where θ i Let be the angle between the relative velocity direction and the x-axis, with clockwise being positive, and k1 be an undetermined coefficient. If the two vehicles tend to approach each other along the x-axis, the lateral distance between them will decrease, making rear-end collisions more likely.

[0051]

[0052]

[0053] Similarly, if two vehicles tend to approach each other on the y-axis, the vertical distance between the two vehicles will be shortened, making a side collision more likely.

[0054] Step 32: Confirm the undetermined coefficient k1. Considering that intelligent connected vehicles will transfer driving control to the driver in emergency situations, the calculation of k1 takes into account the driver's reaction time as t. s The vehicle's initial velocity is v0, and its deceleration is a. According to the velocity-displacement formula for uniform acceleration:

[0055]

[0056] The undetermined coefficient k1 is calculated under two scenarios: one with a relatively high relative speed and the other with a relatively low relative speed. Scenario 1 involves vehicles in the same lane traveling at a relatively high relative speed v. 01 Approaching the car in front, if the distance between the two cars along the x-axis is x1, then according to the acceleration and velocity conditions mentioned above, the two cars can just remain relatively stationary before a collision; Scenario 2 involves vehicles in the same lane traveling at a relatively small relative speed v. 02 Approaching the car in front, if the distance between the two cars along the x-axis is x2, then according to the acceleration and velocity conditions mentioned above, the two cars can just remain relatively stationary before a collision. The probability of a collision is the same in both scenarios, and we can further assume that the pseudo-distances are the same, that is:

[0057]

[0058] Substituting the velocity and displacement, we get:

[0059]

[0060] k1 is related to factors such as road speed limits, road conditions, and driver reaction time, and its specific value can be selected according to the actual situation.

[0061] Assume that on a certain road, the driver's reaction time is 0.5 seconds and the vehicle's deceleration is -6 m / s². 2 Scenario 1: Two vehicles in the same lane approach a vehicle parked on the side of the road ahead at a speed of v1 = 30 m / s. If the distance between the two vehicles along the x-axis is x1 = 90 m, then the two vehicles will just be able to remain relatively stationary before a collision. Figure 2 (a); Scenario 2: Two vehicles in the same lane approach each other at a relative speed of v2 = 10 m / s. If the two vehicles are able to remain relatively stationary just before a collision, then the distance between the two vehicles in the x-axis direction is x2 = 18 m. Figure 2 (b) Substituting the relative vehicle speeds, x-axis distance differences, vehicle decelerations, and driver reaction times for the two scenarios on the same road into the above formula, we obtain:

[0062] k1≈0.15

[0063] Step 33: The vehicle's inherent attributes mainly include its type and mass. Generally, the larger the vehicle's dimensions and the greater its mass, the more severe the collision will be. Wang Jianqiang et al. believe that the severity of a vehicle's collision is correlated with both the target connected vehicle's current speed and mass, and they derived an expression for the equivalent mass of object i by fitting highway speed and accident data as follows:

[0064] M i =m i ×(1.566×10-14×v i6.687 +0.3345)

[0065] Where m i Let be the actual mass of object i, and v be the velocity of object i. This expression shows that at the same vehicle speed, a larger actual mass results in a larger equivalent mass; conversely, for the same actual mass, a higher vehicle speed results in a larger equivalent mass. In other words, actual mass and vehicle speed influence risk based on the severity of the accident.

[0066] Step 34: Driving risk is related to road conditions; the worse the road conditions, the greater the likelihood of an accident. Road conditions include factors such as the road's coefficient of adhesion, road curvature, road slope, and visibility. These factors can be uniformly expressed as "road condition factors." Define the location of object i (x... i ,y i The road condition influencing factors at point ) are shown in the formula, namely:

[0067]

[0068] Where, δ i For road visibility; μ i ρ is the road surface adhesion coefficient; i τ is the road curvature; i Let be the road slope; γ1, γ2, γ3, and γ4 are all undetermined constants, and γ1, γ2 < 0, γ3, γ4 > 0; μ * δ is the standard road surface adhesion coefficient; * Standard road visibility; ρ * For standard road curvature; τ * This is the standard road gradient. R i R represents the road condition influence factor, which describes the varying degrees of potential danger to driving caused by different road conditions. For object i, its corresponding road condition influence factor R is... i By (x) i y i The road surface adhesion coefficient, environmental visibility, road curvature, and slope at point (x) determine the road impact factor. As the road surface adhesion coefficient and road visibility decrease, while road curvature and slope increase, the road impact factor increases. With other conditions remaining constant, the interaction between the vehicle and (x) i y i The probability of a collision with object i at location i increases, thus increasing the driving hazard.

[0069] Step 4, establish the risk field. Calculate and sum the risk quantities posed by all objects on the road to the target connected vehicle:

[0070]

[0071] E j This refers to the driving risk of the target connected vehicle j at this time.

[0072] The above steps describe the implementation process of the present invention in detail, but the present invention is not limited to the specific details of the above embodiments. Anything within the scope of the present invention should not be excluded from the protection scope of the present invention.

Claims

1. A method for modeling vehicle driving risk field in a connected environment, characterized in that, Comprise the following steps: Step 1, through the Internet of vehicles equipment, millimeter wave radar, GPS equipment collection vehicle surrounding environment information, specifically including lane line, obstacle, signal lamp and the motion state data of surrounding vehicle; Step 2, based on the data in step 1, select self-vehicle j as the target connected vehicle, and as the reference, the speed of surrounding object i relative to the target connected vehicle is: wherein, is the relative displacement of object i with respect to target connected vehicle j within the time period. Step 3, based on the reference selected in step 2, the amount of risk generated by object i to the surrounding environment when taking target connected vehicle j as the reference is: Rj(i) = R(i) - R(i, j) (3) is: wherein is an impact factor of the road condition where object i is located; is an equivalent mass of object i; represents a distance vector between object i and target connected vehicle j; indicates the direction of the risk of target connected vehicle j from object i; is a pseudo distance between object i and target connected vehicle j; Step 31, a plane rectangular coordinate system is established with the direction of the target connected vehicle j motion as the horizontal axis, the position of the object i is set as , the position of the target connected vehicle j is set as , and the actual distance between the two is: = The actual distance is corrected to pseudo distance by borrowing the speed component to better describe the change of driving risk when the vehicle approaches the target connected vehicle from different angles. The pseudo distance of object i and target connected vehicle j is The expression is: = wherein is the angle between the relative velocity direction and the x-axis, positive in the clockwise direction, is the undetermined coefficient; Step 32, confirm pending coefficients , considering the driver reaction time as , the initial speed of the vehicle , the vehicle deceleration is a, according to the uniform acceleration speed displacement formula: Set up two scenes of large relative speed and small relative speed to calculate the undetermined coefficient ; Scene one is that the vehicles in the same lane are close to the front vehicle at a large relative speed , and if the distance between the front vehicle and the rear vehicle in the x-axis direction at this time is , according to the acceleration and speed conditions, the two vehicles can just keep relative stillness before the collision; Scene two is that the vehicles in the same lane are close to the front vehicle at a small relative speed , and if the distance between the front vehicle and the rear vehicle in the x-axis direction at this time is , according to the acceleration and speed conditions, the two vehicles can just keep relative stillness before the collision; The possibility of collision in the two scenes is the same, and it can be further considered that the pseudo distance is the same, that is: After substituting the speed and displacement, we get: wherein The road speed limit, road environment, and driver reaction time factor are related, and the specific size can be selected according to the actual situation. Step 4, establish risk field; calculate the risk formed by all objects on the road to the target connected vehicle and sum: is the driving risk of the target connected vehicle j at this time.

2. The method of claim 1, wherein, Select the target connected vehicle as the reference system, use the speed component to correct the actual distance to pseudo distance, better describe the influence of the vehicle approaching the target connected vehicle from different angles on the driving risk, and take the target connected vehicle as the static reference, combine the speed information of the surrounding environment objects relative to the target connected vehicle, unify the moving objects and stationary objects in the field, calculate the risk of the target connected vehicle, and establish the driving risk field for the target connected vehicle, realize the real-time dynamic risk assessment of the target connected vehicle, and provide theoretical support for the active safety control and trajectory planning method of intelligent connected vehicle.

Images