A background traffic flow modeling method and system for roundabout scene autonomous driving testing

Abstract

The present invention discloses a background traffic flow modeling method and system for automatic driving test in roundabout scene. On the rounD data set, based on the mechanical model, the motion model of the test vehicle in the roundabout is constructed by using the viscoelastic-plastic constitutive theory, and the influencing factors of the test vehicle in different positions are expressed by the viscoelastic-plastic constitutive model; the motion model of the test vehicle in the roundabout includes a self-driving force model, a boundary force model and a repulsive force model; based on the self-driving force, boundary force and repulsive force, the horizontal and longitudinal force models of the test vehicle in the roundabout are determined, and the horizontal / longitudinal simulated acceleration, simulated speed and simulated displacement of the vehicle at each moment are predicted by the horizontal and longitudinal force models, so as to describe the motion trend of the vehicle. The present invention improves the efficiency and accuracy of virtual acceleration test of automatic driving in roundabout scene.

Description

Technical Field

[0001] The present invention belongs to the technical field of autonomous driving testing, and in particular to a background traffic flow dynamics modeling method for autonomous driving testing at a roundabout. Background Art

[0002] With the development of artificial intelligence technology, self-driving cars provide a new way to solve the problems of traditional cars such as "traffic congestion, environmental pollution, frequent accidents, and energy shortages". A scientific and complete test and evaluation system is crucial to improving the efficiency of self-driving car research and development, improving technical standards and laws and regulations, and promoting the innovation and development of related industries. With the improvement of the level of autonomous driving, the complexity of the vehicle system has further increased. Changeable weather, complex traffic environment, diverse driving tasks, and dynamic driving conditions have all posed new challenges to its test and evaluation system. Therefore, considering background traffic flow is crucial to the test and evaluation system of self-driving cars.

[0003] Background traffic flow reflects the behavioral characteristics of vehicles and pedestrians in the real road environment, and has an important impact on the driving decisions, path planning and interactive behaviors of autonomous vehicles. Therefore, in this case, the modeling of background traffic flow is not only a description of the flow of vehicles on the road, but also a comprehensive reflection of factors such as driver behavior characteristics, traffic environment and road facilities.

[0004] For unsignalized intersections such as roundabouts, the reasons for the complexity and diversity of background traffic flow involve many aspects, including the intersection of multi-directional traffic, lane change requirements, speed differences, guide signs, and driving behavior. These factors interact with each other to form a complex traffic dynamic system. However, the current background traffic flow modeling methods do not fully consider these factors, especially the lack of consideration of human driving behavior in actual traffic scenarios. Summary of the invention

[0005] In view of the shortcomings in the prior art, the present invention provides a background traffic flow dynamics modeling method for roundabout intersection autonomous driving testing, providing support for the testing of autonomous driving technology in complex environments.

[0006] The present invention achieves the above technical objectives through the following technical means.

[0007] A method for modeling background traffic flow dynamics for automated driving testing at roundabouts:

[0008] On the rounD dataset, based on the mechanical model, the motion model of the test vehicle at the roundabout is constructed using the viscoelastic-plastic constitutive theory, and the influencing factors of the test vehicle at different positions are represented by the viscoelastic-plastic constitutive model; the motion model of the test vehicle at the roundabout includes a self-driving force model, a boundary force model and a repulsive force model;

[0009] Determine the lateral and longitudinal force models of the test vehicle at the roundabout based on self-driving force, boundary force, and repulsive force:

[0010] The lateral force model is:

[0011]

[0012] The longitudinal force model is:

[0013] F x (t+1)=η 1 (V x -V ix (t+1))+

[0014] g(θ ij )*[k 1 (X ix (t+1)-X jx (t+1))-(k 1 α 1 +η 2 )V ix (t+1)+η 2 V jx (t+1)]

[0015] Among them, F y (t+1) is the lateral force on the test vehicle at time t+1, V y is the expected speed of the test vehicle in the y direction, V iy (t+1) is the speed of the test vehicle in the y direction at time t+1, X iy (t+1) is the position of the test vehicle in the y direction at time t+1, X jy (t+1) is the position of the surrounding vehicles in the y direction at time t+1, V jy (t+1) is the speed of the surrounding vehicles in the y direction at time t+1, a y (t) is the acceleration of the test vehicle in the y direction at time t, X iy (t) is the y-direction position of the test vehicle at time t, X ky (t) is the y-direction position of the test vehicle at time t that is closest to the inner circle boundary of the roundabout, and F x (t+1) is the longitudinal force on the test vehicle at time t+1, V x is the expected speed of the test vehicle in the x direction, V ix (t+1) is the speed of the test vehicle in the x direction at time t+1, g(θ ij ) is a function of the object’s perceived anisotropy, X ix (t+1) is the x-direction position of the test vehicle at time t+1, jx(t+1) is the x-direction position of the surrounding vehicles at time t+1, V jx (t+1) is the speed of the surrounding vehicles in the x direction at time t+1, k 1 , α 1 , η 2 , η 1 , k 2 , α 2 , η 4 , k 3 , α 4 , η 5 are all model parameters;

[0016] Fit the transverse and longitudinal force model of the test vehicle at the roundabout, determine the model parameters, and then substitute them into the transverse and longitudinal force model, and then substitute the actual position and speed of the test vehicle and surrounding vehicles at time t+1, and the actual acceleration and position of the test vehicle at time t, determine the transverse and longitudinal forces of the test vehicle at time t+1, and calculate the transverse and longitudinal simulated acceleration of the test vehicle at time t+1, and then calculate the simulated speed and displacement of the test vehicle at time t+1 from the simulated acceleration at time t+1, and use the simulation result at time t+1 to calculate the simulated acceleration, simulated speed, and simulated displacement at time t+2, and so on, to obtain the transverse / longitudinal simulated acceleration, simulated speed, and simulated displacement of the test vehicle at each time;

[0017] The simulation results of the test vehicle at each moment are compared with the actual values. If the error is within a reasonable range, the lateral and longitudinal force models are used to predict the lateral / longitudinal simulated acceleration, simulated speed, and simulated displacement of other vehicles at each moment, thereby describing the movement trend of the vehicle.

[0018] Furthermore, the constitutive model selected by the self-driving force model is the test vehicle series damper, and the self-driving force model is:

[0019]

[0020] Among them, a n+1 is the acceleration of the test vehicle under self-driving force, M n+1 is the mass of the test vehicle, v n is the expected speed of the test vehicle, v n+1 is the speed of the test vehicle, η 1 are the constitutive model parameters of the test vehicle series damper.

[0021] Furthermore, the speed and acceleration of the test vehicle are extracted within the field of view of the test vehicle using Matlab based on the rounD dataset.

[0022] Furthermore, the constitutive model selected by the boundary force model is the test vehicle series damper and spring, and the boundary force model is:

[0023]

[0024] Among them, a′ n+1 represents the acceleration of the test vehicle under the action of boundary force, x n+1 is the distance between the position of the test vehicle in the coordinate system and the center coordinate of the roundabout, x n-1 is the distance between the test vehicle and the boundary of the roundabout, T is the time of each frame in the dataset, and a n is the acceleration of the test vehicle at time n; k 1 , α 1 , η 2 are the constitutive model parameters of the test vehicle series damper and spring.

[0025] Furthermore, the distance between the test vehicle and the boundary of the roundabout is confirmed in the following way: let the distance from the position of the test vehicle to the center of the roundabout be L, the inner ring radius of the roundabout be r, the outer ring radius of the roundabout be R, D1=Lr, D2=RL, and select the smaller value of D1 and D2 as the distance between the test vehicle and the boundary of the roundabout.

[0026] Furthermore, the coordinates of the center of the roundabout and the radius of the inner and outer rings are based on the map of the rounD dataset. The position coordinates of all points on the inner and outer circles of the roundabout in the map are extracted using Open Street Map, and then the position coordinates of the extracted points are fitted.

[0027] Furthermore, the constitutive model selected for the repulsive force model is: the damper is connected in parallel with the spring and then connected in series with the test vehicle, and a function of object perception anisotropy is added. The repulsive force model is:

[0028]

[0029] Among them, g(θ ij ) is a function of the object’s perceived anisotropy, θ ij is the angle between the line connecting the test vehicle i and the surrounding vehicle j and the x-axis, △x is the distance between the test vehicle and the surrounding vehicles, and v′ n is the speed of a surrounding vehicle, M n is the mass of a surrounding vehicle, λ 1 , k 2 , α 2 , η 3 It is the constitutive model parameter of the test vehicle after the damper is connected in parallel with the spring and in series with the anisotropy function of the object is added.

[0030] Furthermore, the position coordinates and speed of the test vehicle and surrounding vehicles are extracted within the field of view of the test vehicle using Matlab based on the rounD dataset.

[0031] Furthermore, the field of view of the test vehicle is specifically as follows: with the center of the test vehicle as the origin, the x-axis perpendicular to the forward direction of the test vehicle as the x-axis, and the y-axis parallel to the forward direction of the test vehicle as the y-axis, a coordinate system is established and a 10-meter range circle is set, with 30°-150° as the field of view in front of the test vehicle, and -30° to -150° as the field of view behind the test vehicle.

[0032] A background traffic flow dynamics modeling system for roundabout autonomous driving testing, comprising:

[0033] The motion model construction module of the test vehicle at the roundabout is used to construct the self-driving force model, boundary force model and repulsive force model of the test vehicle at the roundabout based on the viscoelastic-plastic constitutive theory;

[0034] The lateral and longitudinal force model construction module determines the lateral and longitudinal force models of the test vehicle at the roundabout based on self-driving force, boundary force, and repulsive force;

[0035] The motion trend description module of the test vehicle substitutes the actual position, speed and acceleration of the test vehicle in subsequent frames and the actual position and speed of surrounding vehicles into the lateral and longitudinal force models that determine the model parameters, and determines the lateral / longitudinal forces, lateral / longitudinal simulated accelerations, simulated speeds and simulated displacements of the test vehicle in turn, and describes the motion trend of the test vehicle in the simulation scene.

[0036] The beneficial effects of the present invention are:

[0037] (1) The present invention proposes several main forces that affect driving control based on the fan-shaped field of view and the interaction range, including the vehicle's self-driving force, boundary force and repulsive force, which more comprehensively describes the driving behavior of the vehicle in the roundabout scene. The comprehensive consideration in the form of force makes the lateral and longitudinal force models closer to the actual driving situation, providing accurate and realistic traffic flow background support for the testing of autonomous driving technology in complex environments.

[0038] (2) Based on the viscoelastic-plastic constitutive model, the present invention constructs a model of each force one by one according to the characteristics of different forces, which can accurately describe the movement state of the vehicle at the roundabout and truly reflect the actual movement of a human-driven vehicle at the roundabout, thereby improving the efficiency and accuracy of virtual acceleration testing of autonomous driving in roundabout scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a schematic diagram of the process of the present invention;

[0040] Figure 2 A schematic diagram of the center and radius of the inner and outer circles of the roundabout according to an embodiment of the present invention;

[0041] Figure 3 A schematic diagram of the sectoral field of view of a test vehicle driving on a roundabout according to an embodiment of the present invention;

[0042] FIG4( a ) is a diagram showing the acceleration fitting result of the test vehicle in the x direction in an embodiment of the present invention;

[0043] FIG4( b ) is a speed comparison diagram of the test vehicle in the x direction according to an embodiment of the present invention;

[0044] FIG4( c ) is a comparison diagram of the displacement of the test vehicle in the x direction in an embodiment of the present invention;

[0045] FIG5( a ) is a diagram showing the acceleration fitting result of the test vehicle in the y direction in an embodiment of the present invention;

[0046] FIG5( b ) is a speed comparison diagram of the test vehicle in the y direction according to an embodiment of the present invention;

[0047] FIG5( c ) is a comparison diagram of the displacement of the test vehicle in the y direction in an embodiment of the present invention. DETAILED DESCRIPTION

[0048] The present invention is further described below in conjunction with the accompanying drawings and specific embodiments, but the protection scope of the present invention is not limited thereto.

[0049] like Figure 1 As shown, the present invention provides a background traffic flow dynamics modeling method for roundabout autonomous driving testing, which is specifically as follows:

[0050] The drone obtains the traffic flow situation on a certain road for a period of time and stores it in the rounD dataset, which is processed as follows:

[0051] (1) Extract the geometric dimension data of the roundabout.

[0052] The map of the rounD dataset was imported into the software Open Street Map, and the position coordinates of all points on the inner and outer circles of the roundabout in the map were extracted. Since the rounD dataset does not provide the geometric dimension data of the roundabout, the position coordinates of the extracted points were fitted using the least squares method in Matlab software to obtain the center coordinates and radius of the inner and outer circles of the roundabout, as shown in Figure 2 As shown, this is used to prepare preliminary data for the boundary force below.

[0053] (2) The field of view of the test vehicle is set, and vehicles that appear within the field of view are determined as surrounding vehicles.

[0054] With the center of the test vehicle as the origin, the x-axis perpendicular to the test vehicle's forward direction and the y-axis parallel to the test vehicle's forward direction, a coordinate system is established and a 10-meter range circle is set, with 30°-150° as the test vehicle's front field of view, and -30° to -150° as the test vehicle's rear field of view ( Figure 3 ). Matlab software is used to calculate the angle between the line connecting other vehicles and the test vehicle and the x-axis. If the angle is within the degree range of the above-mentioned field of view, the vehicle is determined as a surrounding vehicle, and it is considered that the two have an interactive relationship.

[0055] (3) Extract specific data of the test vehicle and surrounding vehicles.

[0056] According to the set field of view and surrounding vehicles, the rounD data set is used to extract the data of surrounding vehicles and test vehicles within the above field of view using Matlab software, including the vehicle's position, speed, displacement and acceleration.

[0057] This example takes vehicle ID 295 in 03_tracks in the rounD dataset folder as an example. The vehicle with ID 295 is driving on a roundabout and interacting with surrounding vehicles behind it.

[0058] (4) Use a mechanical model to express the force applied to the test vehicle at the roundabout.

[0059] The force analysis of the test vehicle driving on the roundabout shows that the forces on the test vehicle in this scenario include: self-driving force, which is determined by the characteristics of the individual driver; boundary force, which is determined by the boundary lines on both sides of the roundabout lane; and the repulsive force between the test vehicle and the surrounding vehicles, which is determined by the location of the surrounding vehicles. Therefore, the movement between vehicles is regarded as being carried out under the action of force, and the force form of the test vehicle at the roundabout is analyzed.

[0060] The force on the test vehicle when driving on a roundabout is expressed as:

[0061]

[0062] in, represents the self-driving force determined by the individual driver of the test vehicle, F iw (t) represents the boundary force between the test vehicle and the boundary of the roundabout, F ij (t) represents the repulsive force between the test vehicle and surrounding vehicles.

[0063] (5) Based on the mechanical model, the motion model of the test vehicle at the roundabout is constructed using the viscoelastic-plastic constitutive theory, and the influencing factors of the test vehicle at different positions are represented by the viscoelastic-plastic constitutive model. Since the acceleration of the second frame is calculated based on the data of the first frame, the calculation of the second frame data is taken as an example.

[0064] 1) Self-drive force describes the motivation of an individual to move at the expected speed by adjusting the actual speed within a certain relaxation time. It can be seen that the self-drive force is related to the difference between the speed of the test vehicle and the expected speed. Therefore, when using the stress-strain theory of the viscoelastic-plastic constitutive model, the constitutive model selected is the test vehicle series damper, and the self-drive force model established is as follows:

[0065]

[0066] Among them, a n+1 is the acceleration of the test vehicle under self-driving force, M n+1 is the mass of the test vehicle, v n is the expected speed of the test vehicle, v n+1 is the speed of the test vehicle, η 1 are the constitutive model parameters of the test vehicle series damper;

[0067] The speed and acceleration of the test vehicle mentioned above are the data extracted by (3).

[0068] 2) Boundary force describes the repulsive effect between an individual and the surrounding building environment. It can be seen that the boundary force is related to the acceleration of the test vehicle and the distance between the test vehicle and the boundary of the roundabout. Therefore, when using the strain stress theory of the viscoelastic plastic constitutive model, the constitutive model selected is the test vehicle series damper and spring, and the established boundary force model is as follows:

[0069]

[0070] Among them, a′ n+1 represents the acceleration of the test vehicle under the action of boundary force, x n+1 is the distance between the position of the test vehicle in the coordinate system and the center coordinate of the roundabout, x n-1 is the distance between the test vehicle and the boundary of the roundabout, T is the time of each frame in the dataset, and a n is the acceleration of the test vehicle at time n; k 1 , α 1 , η 2 are the constitutive model parameters of the test vehicle series damper and spring;

[0071] The distance between the test vehicle and the boundary of the roundabout is determined in the following way: let the distance from the position of the test vehicle to the center of the roundabout be L, the inner radius of the roundabout be r, and the outer radius of the roundabout be R, D1 = Lr, D2 = RL, and select the smaller value of D1 and D2 as the distance between the test vehicle and the boundary of the roundabout. The coordinates of the center of the roundabout and the inner and outer ring radius data are extracted from (1).

[0072] 3) Repulsive force describes the interaction force between the vehicle and surrounding vehicles. Therefore, the repulsive force is related to the driving speed of the test vehicle, the driving speed of surrounding vehicles, and the distance between the test vehicle and surrounding vehicles. Therefore, when using the strain stress theory of the viscoelastic plastic constitutive model, the constitutive model selected is: the damper and spring are connected in parallel and then in series with the test vehicle, and the function of object perception anisotropy is added. The repulsive force model established is as follows:

[0073]

[0074] Among them, g(θ ij ) is a function of the object’s perceived anisotropy, θ ij is the angle between the line connecting the test vehicle i and the surrounding vehicle j and the x-axis, △x is the distance between the test vehicle and the surrounding vehicles (calculated based on the position coordinates), v′ n is the speed of a surrounding vehicle, M n is the mass of a surrounding vehicle, λ 1 , k 2 , α 2 , η 3 It is the constitutive model parameter of the test vehicle after the damper is connected in parallel with the spring and in series, and the function of the anisotropy of the object is added;

[0075] The repulsive force model involves the position coordinates and speed of the test vehicle and surrounding vehicles, which are the data extracted from (3).

[0076] (6) Based on the self-driving force, boundary force and repulsive force, the forces on the test vehicle at the roundabout are determined in two directions: lateral and longitudinal. In the lateral direction, the forces on the test vehicle include the self-driving force, the repulsive force with the surrounding vehicles and the boundary force with the boundary of the roundabout. In the longitudinal direction, the forces on the test vehicle include the self-driving force and the repulsive force with the surrounding vehicles. The mass of the test vehicle and the mass of the surrounding vehicles are both equivalent to 1. The specific form of the force is as follows:

[0077] 1) According to formulas (2) and (5), the longitudinal force model of the test vehicle is expressed as:

[0078] F x (t+1)=η 1 (V x -Vix (t+1))+

[0079] g(θ ij )*[k 1 (X ix (t+1)-X jx (t+1))-(k 1 α 1 +η 2 )V ix (t+1)+η 2 V jx (t+1)](6)

[0080] Among them, F x (t+1) is the longitudinal force on the test vehicle at time t+1, V x is the expected speed of the test vehicle in the x direction (needs to be calibrated), V ix (t+1) is the speed of the test vehicle in the x direction at time t+1, g(θ ij ) is a function of the object’s perceived anisotropy, X ix (t+1) is the x-direction position of the test vehicle at time t+1, jx (t+1) is the x-direction position of the surrounding vehicles at time t+1, V jx (t+1) is the speed of the surrounding vehicles in the x direction at time t+1; η 1 , k 1 , α 1 , η 2 are model parameters.

[0081] Using the least squares method to fit formula (6) in Matlab, the obtained parameter value is: 1 =0.2656, λ 1 =0.8128, V x =7.8731, k 1 =6.9736, α 1 =0.6399, η 2 =0.2158.

[0082] 2) According to formulas (2), (3) and (5), the lateral force model of the test vehicle is expressed as:

[0083]

[0084] Among them, F y (t+1) is the lateral force on the test vehicle at time t+1, V y is the expected speed of the test vehicle in the y direction (needs to be calibrated), V iy (t+1) is the speed of the test vehicle in the y direction at time t+1, X iy(t+1) is the position of the test vehicle in the y direction at time t+1, X jy (t+1) is the position of the surrounding vehicles in the y direction at time t+1, V jy (t+1) is the speed of the surrounding vehicles in the y direction at time t+1, a y (t) is the acceleration of the test vehicle in the y direction at time t, X iy (t) is the y-direction position of the test vehicle at time t, X ky (t) is the y-direction position of the test vehicle at time t that is closest to the inner circle boundary of the roundabout; η 1 , k 2 , α 2 , η 4 , k 3 , α 4 , η 5 are model parameters.

[0085] Using the least squares method to fit formula (7) in Matlab, the obtained parameter value is: V y =7.8389,λ 1 =0.7065, k 2 =0.8168, α 2 =0.6168, η 4 =0.9116, k 3 =3.6661, α 4 =6.3682, η 5 =0.2383.

[0086] (7) In order to verify the effectiveness of the model, the acceleration, velocity, and displacement of the test vehicle are predicted based on the fitted parameters and compared with the actual values.

[0087] Substitute the fitted parameters into formulas (6) and (7), and then substitute the actual position and speed of the test vehicle and surrounding vehicles at time t+1, and the actual acceleration and position of the test vehicle at time t, determine the lateral and longitudinal forces of the test vehicle at time t+1, and then calculate the lateral and longitudinal simulated accelerations of the test vehicle at time t+1, and then calculate the simulated speed and displacement of the test vehicle at time t+1 from the simulated acceleration at time t+1, and use the simulation results at time t+1 to calculate the simulated acceleration, simulated speed, and simulated displacement at time t+2, and then deduce the lateral / longitudinal simulated acceleration, simulated speed, and simulated displacement of the test vehicle in sequence. Among them, the actual position, speed, and acceleration of the vehicle are all from the rounD dataset.

[0088] By calculating the root mean square error (RMSE) between the simulated acceleration, simulated speed, simulated displacement and the actual value of the test vehicle in the horizontal and vertical directions, it can be obtained that the RMSE in the horizontal direction is 0.528m / s 2, 0.272m / s, 0.369m, and the RMSE in the vertical direction are 0.061m / s 2 , 0.048m / s, 0.052m; Since the value of the root mean square error is within a reasonable range, the lateral and longitudinal force models of the test vehicle are used in the subsequent use, and the initial vehicle-related parameters (including the position and speed of the test vehicle and surrounding vehicles, and the acceleration of the test vehicle at the previous moment) are input, and the lateral / longitudinal simulated acceleration, simulated speed, and simulated displacement of the test vehicle at each moment are output to describe the movement trend of the test vehicle. This not only provides an accurate and realistic traffic flow background for autonomous driving tests, but also improves the efficiency and accuracy of virtual acceleration tests of autonomous driving at roundabouts.

[0089] Fig. 4(a), (b), (c) are respectively the acceleration fitting results, velocity comparison and displacement comparison of the test vehicle in the x direction in this embodiment, and Fig. 5(a), (b), (c) are respectively the acceleration fitting results, velocity comparison and displacement comparison of the test vehicle in the y direction in this embodiment.

[0090] A background traffic flow dynamics modeling system for roundabout autonomous driving testing, comprising:

[0091] The motion model construction module of the test vehicle at the roundabout is used to construct the self-driving force model, boundary force model and repulsive force model of the test vehicle at the roundabout based on the viscoelastic-plastic constitutive theory;

[0092] The lateral and longitudinal force model construction module determines the lateral and longitudinal force models of the test vehicle at the roundabout based on self-driving force, boundary force, and repulsive force;

[0093] The motion trend description module of the test vehicle substitutes the actual position, speed and acceleration of the test vehicle in subsequent frames and the actual position and speed of surrounding vehicles into the lateral and longitudinal force models that determine the model parameters, and determines the lateral / longitudinal forces, lateral / longitudinal simulated accelerations, simulated speeds and simulated displacements of the test vehicle in turn, and describes the motion trend of the test vehicle in the simulation scene.

[0094] The embodiments are preferred implementations of the present invention, but the present invention is not limited to the above-mentioned implementations. Any obvious improvements, substitutions or modifications that can be made by those skilled in the art without departing from the essential content of the present invention belong to the protection scope of the present invention.

Claims

1. A background traffic flow dynamics modeling method for roundabout autonomous driving testing, characterized by: On the rounD dataset, based on the mechanical model, the motion model of the test vehicle at the roundabout is constructed using the viscoelastic-plastic constitutive theory, and the influencing factors of the test vehicle at different positions are represented by the viscoelastic-plastic constitutive model; the motion model of the test vehicle at the roundabout includes a self-driving force model, a boundary force model and a repulsive force model; The lateral and longitudinal force models of the test vehicle at the roundabout are determined based on self-driving force, boundary force, and repulsive force: The lateral force model is: The longitudinal force model is: F x (t+1)=η1(V x -V ix (t+1))+ g(θ ij )*[k1(X ix (t+1)-X jx (t+1))-(k1α1+η2)V ix (t+1)+η2V jx (t+1)] Among them, F y (t+1) is the lateral force on the test vehicle at time t+1, V y is the expected speed of the test vehicle in the y direction, V iy (t+1) is the speed of the test vehicle in the y direction at time t+1, X iy (t+1) is the position of the test vehicle in the y direction at time t+1, X jy (t+1) is the position of the surrounding vehicles in the y direction at time t+1, V jy (t+1) is the speed of the surrounding vehicles in the y direction at time t+1, a y (t) is the acceleration of the test vehicle in the y direction at time t, X iy (t) is the y-direction position of the test vehicle at time t, X ky (t) is the y-direction position of the test vehicle at time t that is closest to the inner circle boundary of the roundabout, and F x (t+1) is the longitudinal force on the test vehicle at time t+1, V x is the expected speed of the test vehicle in the x direction, V ix (t+1) is the speed of the test vehicle in the x direction at time t+1, g(θ ij ) is a function of the object’s perceived anisotropy, X ix (t+1) is the x-direction position of the test vehicle at time t+1. jx (t+1) is the x-direction position of the surrounding vehicles at time t+1, V jx (t+1) is the speed of the surrounding vehicles in the x direction at time t+1, T is the time of each frame in the data set, k1, α1, η2, η1, k2, α2, η4, k3, α4, and η5 are all model parameters; Fit the transverse and longitudinal force model of the test vehicle at the roundabout, determine the model parameters, and then substitute them into the transverse and longitudinal force model, and then substitute the actual position and speed of the test vehicle and surrounding vehicles at time t+1, and the actual acceleration and position of the test vehicle at time t, determine the transverse and longitudinal forces of the test vehicle at time t+1, and calculate the transverse and longitudinal simulated acceleration of the test vehicle at time t+1, and then calculate the simulated speed and displacement of the test vehicle at time t+1 from the simulated acceleration at time t+1, and use the simulation result at time t+1 to calculate the simulated acceleration, simulated speed, and simulated displacement at time t+2, and so on, to obtain the transverse / longitudinal simulated acceleration, simulated speed, and simulated displacement of the test vehicle at each time; The simulation results of the test vehicle at each moment are compared with the actual values. If the error is within a reasonable range, the lateral and longitudinal force models are used to predict the lateral / longitudinal simulated acceleration, simulated speed, and simulated displacement of other vehicles at each moment, thereby describing the movement trend of the vehicle.

2. The background traffic flow dynamics modeling method according to claim 1, characterized in that: The constitutive model selected by the self-driving force model is the test vehicle series damper, and the self-driving force model is: Among them, a n+1 is the acceleration of the test vehicle under self-driving force, M n+1 is the mass of the test vehicle, v n is the expected speed of the test vehicle, v n+1 is the driving speed of the test vehicle, and η1 is the constitutive model parameter of the series damper of the test vehicle.

3. The background traffic flow dynamics modeling method according to claim 2, characterized in that: The speed and acceleration of the test vehicle are extracted within the field of view of the test vehicle using Matlab based on the rounD dataset.

4. The background traffic flow dynamics modeling method according to claim 1, characterized in that: The constitutive model selected by the boundary force model is the test vehicle series damper and spring, and the boundary force model is: Among them, a′ n+1 represents the acceleration of the test vehicle under the action of boundary force, x n+1 is the distance between the position of the test vehicle in the coordinate system and the center coordinate of the roundabout, x n-1 is the distance between the test vehicle and the boundary of the roundabout, a n is the acceleration of the test vehicle at time n; k1, α1, η2 are the constitutive model parameters of the series damper and spring of the test vehicle.

5. The background traffic flow dynamics modeling method according to claim 4, characterized in that: The distance between the test vehicle and the boundary of the roundabout is confirmed in the following way: let the distance from the position of the test vehicle to the center of the roundabout be L, the inner ring radius of the roundabout be r, the outer ring radius of the roundabout be R, D1=Lr, D2=RL, and select the smaller value of D1 and D2 as the distance between the test vehicle and the boundary of the roundabout.

6. The background traffic flow dynamics modeling method according to claim 5, characterized in that: The coordinates of the center of the roundabout and the radius of the inner and outer rings are obtained based on the map of the rounD dataset. The Open Street Map is used to extract the position coordinates of all points on the inner and outer circles of the roundabout in the map, and then the position coordinates of the extracted points are fitted.

7. The background traffic flow dynamics modeling method according to claim 1, characterized in that: The constitutive model selected for the repulsive force model is: the damper is connected in parallel with the spring and then connected in series with the test vehicle, and a function of object perception anisotropy is added. The repulsive force model is: Among them, g(θ ij ) is a function of the object’s perceived anisotropy, θ ij is the angle between the line connecting the test vehicle i and the surrounding vehicle j and the x-axis, △x is the distance between the test vehicle and the surrounding vehicles, and v′ n is the speed of a surrounding vehicle, M n is the mass of a surrounding vehicle, λ1, k2, α2, η3 are the constitutive model parameters of the test vehicle after the damper is connected in parallel with the spring, and the function of the anisotropy of the object perception is added.

8. The background traffic flow dynamics modeling method according to claim 7, characterized in that: The position coordinates and speed of the test vehicle and surrounding vehicles are extracted within the field of view of the test vehicle using Matlab based on the rounD dataset.

9. The background traffic flow dynamics modeling method according to claim 3 or 8, characterized in that: The field of view of the test vehicle is specifically as follows: with the center of the test vehicle as the origin, the x-axis perpendicular to the forward direction of the test vehicle as the x-axis, and the y-axis parallel to the forward direction of the test vehicle as the y-axis, a coordinate system is established and a 10-meter range circle is set, with 30°-150° as the field of view in front of the test vehicle, and -30° to -150° as the field of view behind the test vehicle.

10. A system for implementing the background traffic flow dynamics modeling method according to any one of claims 1 to 9, characterized in that: include: The motion model construction module of the test vehicle at the roundabout is used to construct the self-driving force model, boundary force model and repulsive force model of the test vehicle at the roundabout based on the viscoelastic-plastic constitutive theory; The lateral and longitudinal force model construction module determines the lateral and longitudinal force models of the test vehicle at the roundabout based on self-driving force, boundary force, and repulsive force; The motion trend description module of the test vehicle substitutes the actual position, speed and acceleration of the test vehicle in subsequent frames and the actual position and speed of surrounding vehicles into the lateral and longitudinal force models that determine the model parameters, and determines the lateral / longitudinal forces, lateral / longitudinal simulated accelerations, simulated speeds and simulated displacements of the test vehicle in turn, and describes the motion trend of the test vehicle in the simulation scene.

Images