CN116223056A

CN116223056A - Virtual collision test method, apparatus, device, storage medium, and program product

Info

Publication number: CN116223056A
Application number: CN202211605116.9A
Authority: CN
Inventors: 聂冰冰; 刘斯源; 李泉; 周青
Original assignee: Tsinghua University
Current assignee: Tsinghua University
Priority date: 2022-12-14
Filing date: 2022-12-14
Publication date: 2023-06-06
Anticipated expiration: 2042-12-14
Also published as: CN116223056B

Abstract

The application relates to a virtual collision test method, a virtual collision test apparatus, a virtual collision test device, a virtual collision test storage medium and a virtual collision test program product. The method comprises the following steps: acquiring a test scene, generating digital pedestrians based on a preset pedestrian generation model, acquiring collision risk levels of the digital pedestrians under the condition that the digital pedestrians are detected to actively avoid the target vehicle, and controlling the target vehicle to execute anti-collision processing according to the collision risk levels to obtain a test result. By adopting the method, the accuracy of the virtual collision test can be improved.

Description

Virtual collision test method, apparatus, device, storage medium, and program product

Technical Field

The present disclosure relates to the field of automotive safety technologies, and in particular, to a virtual collision test method, apparatus, device, storage medium, and program product.

Background

With the development of automobile safety technology, safety test systems for the weak road traffic participants and pedestrians are more mature, such as a collision test and an automatic emergency braking system (Autonomous Emergency Braking, AEB) pedestrian anti-collision test which are more mature at present.

At present, pedestrian virtual collision test is to randomly generate pedestrians in front of a vehicle, and then the vehicle-mounted risk early warning protection system changes the vehicle speed according to the randomly generated pedestrians to avoid collision.

However, the test scenario in the above test scheme is different from the actual scenario, so there is a problem of low test accuracy.

Disclosure of Invention

In view of the foregoing, it is desirable to provide a virtual collision test method, apparatus, device, storage medium, and program product that can improve the accuracy of virtual collision test.

In a first aspect, the present application provides a virtual collision test method. The method comprises the following steps:

acquiring a test scene, and generating digital pedestrians based on a preset pedestrian generation model;

under the condition that the digital pedestrian actively avoids the target vehicle is detected, acquiring the collision risk level of the digital pedestrian;

controlling the target vehicle to execute anti-collision processing according to the collision risk level to obtain a test result; the test result is used for representing whether the target vehicle collides with the digital pedestrian after the anti-collision treatment is performed.

In a second aspect, the present application further provides a virtual collision test apparatus. The device comprises:

the generation module is used for acquiring a test scene and generating digital pedestrians based on a preset pedestrian generation model;

the first acquisition module is used for acquiring the collision risk level of the digital pedestrian under the condition that the digital pedestrian is detected to actively avoid the target vehicle;

The second acquisition module is used for controlling the target vehicle to execute anti-collision processing according to the collision risk level to obtain a test result; the test result is used for representing whether the target vehicle collides with the digital pedestrian after the anti-collision treatment is performed.

In a third aspect, the present application also provides a computer device. The computer device comprises a memory storing a computer program and a processor which when executing the computer program performs the steps of:

In a fourth aspect, the present application also provides a computer-readable storage medium. The computer readable storage medium having stored thereon a computer program which when executed by a processor performs the steps of:

In a fifth aspect, the present application also provides a computer program product. The computer program product comprises a computer program which, when executed by a processor, implements the steps of:

According to the virtual collision test method, the virtual collision test device, the virtual collision test equipment, the storage medium and the virtual collision test program product, the digital pedestrians are generated based on the preset pedestrian generation model by acquiring the test scene, the collision risk level of the digital pedestrians is acquired under the condition that the digital pedestrians are detected to actively avoid the target vehicle, and the target vehicle is controlled to execute anti-collision processing according to the collision risk level, so that the test result is obtained. Compared with the prior art, the method and the device have the advantages that the digital pedestrians with the active behavior attribute are generated based on the virtual dangerous traffic scene and the preset pedestrian generation model, so that the active behavior capability of the digital pedestrians when facing dangerous vehicles is considered in the virtual collision test scene, the digital pedestrians with high fidelity are provided for the digital test platform of the automatic driving vehicle aiming at the pedestrians, the gap that the digital pedestrians lack the active avoidance capability is supplemented, and the accuracy of the virtual collision test algorithm is improved; further, under the condition that the digital pedestrians actively avoid the target vehicle, the collision risk level of the digital pedestrians is obtained, so that when the digital pedestrians avoid the target vehicle under the condition that different active avoidance modes are adopted by the digital pedestrians and the relative position relations between the digital pedestrians and the target vehicle are different, the collision risk level of the digital pedestrians is different, the influence of the avoidance actions adopted by the digital pedestrians on the collision risk level is further considered, and the virtual collision test scene is more in accordance with the collision condition of the pedestrians and the vehicles in the real collision scene; further, the target vehicle is controlled to execute anti-collision processing according to the collision risk level to obtain a test result, so that the target vehicle executes different anti-collision processing for different collision risk levels, namely, the target vehicle executes lower-level anti-collision processing for low risk levels, and the target vehicle executes higher-level anti-collision processing for high risk levels, so that the target vehicle adopts different-level anti-collision processing to minimize pedestrian collision damage under the condition of not wasting vehicle resources, and interactive logic among dangerous traffic scenes, digital pedestrians and the target vehicle is comprehensively considered, thereby improving accuracy of virtual collision test.

Drawings

FIG. 1 is a schematic illustration of a pedestrian and vehicular interaction scenario in one embodiment;

FIG. 2 is a schematic illustration of an intersection scenario in one embodiment;

FIG. 3 is a schematic illustration of a T-junction scenario in one embodiment;

FIG. 4 is a schematic diagram of a "ghost probe" scenario in one embodiment;

FIG. 5 is a schematic diagram of a cross road scenario in one embodiment;

FIG. 6 is an application environment diagram of a virtual collision detection method in one embodiment;

FIG. 7 is a flow chart of a virtual collision detection method in one embodiment;

FIG. 8 is a flow chart of determining a collision risk level for a digital pedestrian in one embodiment;

FIG. 9 is a schematic diagram of a dangerous traffic scenario in one embodiment;

FIG. 10 is a flow chart of determining a collision risk level for a digital pedestrian in one embodiment;

FIG. 11 is a schematic diagram of an avoidance mode decision model in one embodiment;

FIG. 12 is a flow chart illustrating determining a collision risk level in one embodiment;

FIG. 13 is a schematic diagram of a standard avoidance risk level model in one embodiment;

FIG. 14 is a flow chart illustrating determining a collision risk level in one embodiment;

FIG. 15 is a schematic representation of risk domains of young male volunteers in one embodiment;

FIG. 16 is a schematic of a mid-year male volunteer risk domain in one embodiment;

FIG. 17 is a schematic representation of risk domains of middle-aged and elderly male volunteers, according to one embodiment;

FIG. 18 is a schematic illustration of pedestrian collision risk levels based on collision risk domains in one embodiment;

FIG. 19 is a schematic diagram of a method of screening for false decision pedestrians in one embodiment;

FIG. 20 is a schematic diagram of a method of whistling early warning and pedestrian avoidance activation in one embodiment;

FIG. 21 is a schematic diagram of a young man deceleration avoidance speed limit in one embodiment;

FIG. 22 is a schematic diagram of an acceleration avoidance speed limit for a young man in one embodiment;

FIG. 23 is a graph illustrating a single deceleration avoidance speed profile for a pedestrian in one embodiment;

FIG. 24 is a graph illustrating a single acceleration avoidance speed profile of a pedestrian in one embodiment;

FIG. 25 is a graph illustrating a speed profile before pedestrian avoidance in one embodiment;

FIG. 26 is a graph showing the speed of a pedestrian reaction time in one embodiment;

FIG. 27 is a schematic diagram of a speed profile for pedestrian avoidance in one embodiment;

FIG. 28 is a graph illustrating velocity profile after pedestrian avoidance in one embodiment;

FIG. 29 is a schematic diagram of an autopilot test platform for digital pedestrian initiative in one embodiment;

FIG. 30 is a block diagram of a virtual collision test apparatus in one embodiment;

FIG. 31 is a block diagram of a virtual collision test apparatus in one embodiment.

Detailed Description

In order to make the objects, technical solutions and advantages of the present application more apparent, the present application will be further described in detail with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the present application.

Vehicles and pedestrians act as road traffic participants, and both interact and collide during road use. Limited by pedestrian life safety, real collision tests of intelligent automobiles for pedestrian safety cannot be carried out, so dangerous traffic scenes are simulated through Virtual Reality technology (VR). FIG. 1 shows an example of a VR scenario for a vehicle-to-pedestrian interaction, wherein a pedestrian is assumed to be within the detection range of the vehicle radar, the pedestrian is positioned in the left front of the vehicle, the pedestrian's travel trajectory is perpendicular to the vehicle's travel trajectory, the potential "collision zone" is the region where the pedestrian's travel trajectory coincides with the vehicle's travel trajectory, and the vehicle speed is

The walking speed of the pedestrian is v ₀ The longitudinal distance of the pedestrian from the central axis of the vehicle is +.>

The vehicle width is L _vw The potential "collision zone" is located a distance +. >

The nearest distance between the pedestrian and the potential collision area is D _p,ns The furthest distance between the pedestrian and the potential collision zone is D _p,fs 。

According to the statistical analysis of domestic and foreign pedestrian safety test scenes, typical dangerous traffic test scenes can comprise but are not limited to three types of test scenes which are easy to generate collision accidents, namely a crosswalk, a crossing road and a ghost probe. By way of example, at the crossroad shown in fig. 2 and the t-intersection shown in fig. 3, pedestrians and vehicles interact at the crosswalk position, and as no other factors interfere with the perception capability of the pedestrians and the vehicles, the pedestrians and the vehicles can perceive the other side earlier, and the interaction mode can be called long-distance interaction; the two interactive scenes shown in the figure 4 and the figure 5 are abrupt, wherein the pedestrian cannot notice dangerous coming vehicles all the time before the pedestrian leaves the range of shielding vehicles because other vehicles interfere the sensing range of the pedestrian and the vehicles in the figure, and the radar detection of the automatic driving vehicle is shielded by other vehicles, so that the pedestrian cannot be detected, and the decision of the pedestrian and the vehicle in the scene is hysteresis; the pedestrians crossing the road are suddenly in front of the running of the vehicle, and the detection failure of the scene simulation vehicle-mounted radar can not timely detect the pedestrians with potential collision risk, so that the vehicle can not start safety protection measures such as risk zones, AEB emergency braking and the like on time.

Under the influence of inertia during the movement of the vehicle and the pedestrian, the movement states of the vehicle and the pedestrian cannot be controlled instantaneously to avoid collision when the dangerous traffic test scene appears, which results in that the direct collision between the vehicle and the pedestrian is unavoidable in a certain time and space range of the interaction between the pedestrian and the vehicle, that is, the area of the vehicle-pedestrian relative position where the collision risk possibly occurs during the interaction between the pedestrian and the vehicle is called a pedestrian-vehicle collision risk area. The pedestrian-vehicle collision risk area depends on the information such as the current traveling speed, traveling direction, and relative position of the vehicle and the pedestrian.

After the simulation of the dangerous traffic scene, the typical dangerous traffic test scene and the pedestrian-vehicle collision risk domain are described above, the implementation environment related to the virtual collision test method provided by the embodiment of the application will be briefly described below.

The anti-collision test method provided by the embodiment of the application can be applied to an application environment shown in fig. 6. The computer device comprises a processor, a memory, and a computer program stored in the memory, wherein the processor is connected through a system bus, and when executing the computer program, the processor can execute the steps of the method embodiments described below. Optionally, the computer device may further comprise a network interface, a display screen and an input means. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium, which stores an operating system and a computer program, an internal memory. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface of the computer device is used for communicating with an external terminal through a network connection. Optionally, the computer device may be a server, may be a personal computer, may also be a personal digital assistant, may also be other terminal devices, such as a tablet computer, a mobile phone, etc., and may also be a cloud or remote server.

After the application scenario of the virtual collision test method provided in the embodiment of the present application is described above, the determination process of the virtual collision test method is mainly described below.

In one embodiment, as shown in fig. 7, a virtual collision test method is provided, and the method is applied to the computer device in fig. 6 for illustration, and includes the following steps:

s701, acquiring a test scene, and generating digital pedestrians based on a preset pedestrian generation model.

The test scenario may be the typical dangerous traffic test scenario described above.

The preset pedestrian generation model can be a risk perception boundary model of pedestrians aiming at intelligent automobiles in a virtual dangerous traffic scene through relevant statistical data of the pedestrians avoiding dangerous vehicles in the early virtual dangerous traffic scene; the decision-making model of the pedestrian when facing the dangerous vehicle can be obtained according to decision-making logic of an active avoidance mode of the pedestrian in a virtual dangerous traffic scene; the method can also be an avoidance kinematic model of the fitted pedestrian when facing the dangerous vehicle according to the active avoidance data of the pedestrian in the virtual dangerous traffic scene, and the avoidance speed curve of the pedestrian when facing the dangerous vehicle can be obtained through the model.

The digital pedestrians can be digital pedestrians which are defined in the computer equipment and have the preset pedestrian generation model according to the active avoidance characteristics of young male volunteers in the virtual dangerous traffic scene.

In the embodiment of the application, in a virtual collision test of a complete dangerous traffic scene, the computer equipment can randomly generate a typical dangerous traffic test scene in front of the running of the automatic driving vehicle, and generate a digital pedestrian coupled with the typical dangerous traffic test scene based on a pedestrian generation model preset in the computer equipment. Before a typical dangerous traffic test scenario is generated, it is also necessary to define initial attributes of the autonomous vehicle, for example, as shown in table 1, the vehicle width of the autonomous vehicle may be defined as 2m, the vehicle length may be defined as 4m, the initial coordinates of the autonomous vehicle may be defined as (0, 0), and the initial speed may be defined as 60km/h.

Table 1 automated driving vehicle attributes

S702, acquiring the collision risk level of the digital pedestrian under the condition that the digital pedestrian is detected to actively avoid the target vehicle.

The pedestrian actively avoiding target vehicle can be an active avoiding behavior adopted when the pedestrian finds dangerous coming, and according to traffic accident investigation and experimental research on the pedestrian actively avoiding capability under a virtual dangerous traffic scene, the pedestrian generally adopts two active avoiding modes of forward acceleration or scram and back when facing dangerous coming, and the pedestrian actively avoiding capability can effectively reduce dangerous traffic accidents. The target vehicle may be a vehicle closest to the pedestrian in front of the traveling locus of the pedestrian.

The collision risk level of the digital pedestrian may be that the collision risk of the digital pedestrian is divided into 5 risk levels according to the relative positional relationship between the digital pedestrian and the pedestrian-vehicle collision risk domain at the moment of active avoidance: the risk level is 0 (no collision risk), 1 (low collision risk), false avoidance risk level (false avoidance decision), 2 (high collision risk), 3 (unavoidable collision risk).

In the embodiment of the application, under the condition that the computer equipment detects that the digital pedestrians avoid the target vehicle in an active avoidance mode, different collision risk levels of the digital pedestrians can be determined according to the different active avoidance modes of the pedestrians.

S703, controlling the target vehicle to execute anti-collision processing according to the collision risk level to obtain a test result; the test result is used for representing whether the target vehicle collides with the digital pedestrian after the anti-collision treatment is performed.

The anti-collision process may be different braking processes performed by the target vehicle according to the different collision risk levels determined in S702, and for example, if the collision risk level determined in S702 is level 3, the target vehicle performs the highest level braking to reduce the vehicle speed maximally and minimize the pedestrian collision damage.

The test result may be that the target vehicle collides with the digital pedestrian after the anti-collision process is performed, or that the target vehicle does not collide with the digital pedestrian after the anti-collision process is performed.

In this embodiment of the present application, the computer device may control the target vehicle to perform different anti-collision processes according to the different collision risk levels determined in S702, and obtain the test result after the target vehicle performs different anti-collision processes. The test result comprises that the target vehicle collides with the digital pedestrian and also comprises that the target vehicle does not collide with the digital pedestrian.

According to the virtual collision test method, the test scene is obtained, the digital pedestrians are generated based on the preset pedestrian generation model, the collision risk level of the digital pedestrians is obtained under the condition that the digital pedestrians are detected to actively avoid the target vehicle, and the target vehicle is controlled to execute anti-collision processing according to the collision risk level, so that the test result is obtained. Compared with the prior art, the method and the device have the advantages that the digital pedestrians with the active behavior attribute are generated based on the virtual dangerous traffic scene and the preset pedestrian generation model, so that the active behavior capability of the digital pedestrians when facing dangerous vehicles is considered in the virtual collision test scene, the digital pedestrians with high fidelity are provided for the digital test platform of the automatic driving vehicle aiming at the pedestrians, the gap that the digital pedestrians lack the active avoidance capability is supplemented, and the accuracy of the virtual collision test algorithm is improved; further, under the condition that the digital pedestrians actively avoid the target vehicle, the collision risk level of the digital pedestrians is obtained, so that when the digital pedestrians avoid the target vehicle under the condition that different active avoidance modes are adopted by the digital pedestrians and the relative position relations between the digital pedestrians and the target vehicle are different, the collision risk level of the digital pedestrians is different, the influence of the avoidance actions adopted by the digital pedestrians on the collision risk level is further considered, and the virtual collision test scene is more in accordance with the collision condition of the pedestrians and the vehicles in the real collision scene; further, the target vehicle is controlled to execute anti-collision processing according to the collision risk level to obtain a test result, so that the target vehicle executes different anti-collision processing for different collision risk levels, namely, the target vehicle executes lower-level anti-collision processing for low risk levels, and the target vehicle executes higher-level anti-collision processing for high risk levels, so that the target vehicle adopts different-level anti-collision processing to minimize pedestrian collision damage under the condition of not wasting vehicle resources, and interactive logic among dangerous traffic scenes, digital pedestrians and the target vehicle is comprehensively considered, thereby improving accuracy of virtual collision test.

In one embodiment, based on the embodiment shown in fig. 7, the process of acquiring the collision risk level of the digital pedestrian and the vehicle in the case that the vehicle detects that the digital pedestrian actively dodges, as shown in fig. 8, S702 "the process of acquiring the collision risk level of the digital pedestrian in the case that the digital pedestrian actively dodges the target vehicle" may include the following steps:

s801, acquiring a relative position relation between a digital pedestrian and a target vehicle; the relative positional relationship includes a lateral distance and a longitudinal distance between the digital pedestrian and the target vehicle.

In the embodiment of the application, the computer device can acquire the relative position relationship between the digital pedestrian and the target vehicle based on the virtual dangerous traffic scene. For example, as shown in fig. 9, in a virtual dangerous traffic scenario, the relative positional relationship between the digital pedestrian and the target vehicle may include a lateral distance l of the pedestrian from the vehicle, and a longitudinal distance d between the pedestrian and the central axis of the vehicle.

S802, determining the avoidance mode of the digital pedestrians according to the pedestrian generation model and the relative position relation.

In this embodiment of the present application, in a virtual dangerous traffic scenario, the computer device may generate a model based on a preset pedestrian, and after the pedestrian realizes the risk, the computer device may select an appropriate avoidance mode according to a space-time relationship (collision time, longitudinal distance) between the pedestrian and the target vehicle, for example, forward acceleration avoidance and scram and reverse avoidance. The collision time (TimetoCollision, TTC) may be a time when the vehicle collides with the vehicle, and may be obtained from a ratio of a relative distance between the vehicle and the hazard to a current vehicle speed.

Optionally, the pedestrian generation model includes a boundary model and an avoidance decision model, as shown in fig. 10, and determining an avoidance mode of the digital pedestrian according to the relationship between the pedestrian generation model and the relative position includes:

s1001, acquiring a risk perception boundary condition based on a boundary model.

The boundary model can be a model which is preset in the computer equipment and is determined to be avoided by stopping suddenly and then retreating or accelerating forwards according to the space-time relationship between the pedestrian and the target vehicle. For example, in a virtual dangerous traffic scene, if the subject is 2.2m (standard deviation 0.25) away from the central axis of the vehicle on the premise that the vehicle width of the target vehicle is 2m, the collision time is greater than 1.6s (standard deviation 0.38), and the subject can avoid collision by means of scram and back-off; if the subject is 1.2m (standard deviation 0.97) from the central axis of the vehicle, the collision time is more than 1.8s (standard deviation 0.64), and the subject can avoid collision by accelerating forward to avoid collision.

The risk perception boundary condition can be that the collision time is less than 2.44s, and the longitudinal distance between the pedestrian and the central axis of the vehicle is less than 3.84m. From the perspective of pedestrian safety protection and active avoidance capability test of pedestrians under dangerous working conditions, it is considered that if the pedestrians are aware of risks and take active avoidance measures, the pedestrians are free from collision risks when the collision time is longer than 2.44 s. If the pedestrian does not notice the coming vehicle all the time, the pedestrian can walk 2.84m at the maximum distance in 2.44s according to the normal maximum average walking speed of the subject, and the longitudinal distance between the pedestrian and the central axis of the vehicle is considered to be greater than 3.84m because of the vehicle width of 2m, so that the pedestrian does not need to avoid the vehicle.

In the embodiment of the application, the computer device may acquire the risk perception boundary condition based on a preset boundary model and a space-time relationship between the pedestrian and the target vehicle.

S1002, determining the avoidance mode of the digital pedestrians according to the risk perception boundary conditions and the avoidance decision model.

The avoidance decision model may be a model that, in a dangerous traffic scene, after a pedestrian realizes risk, the computer device may choose to make a back-off or accelerate forward to avoid after sudden stop according to a spatial-temporal relationship (collision time, longitudinal distance) between the pedestrian and the target vehicle. It should be noted that the avoidance decision model includes a standard avoidance decision model and an erroneous avoidance decision model.

For example, as shown in fig. 11, the standard avoidance decision model may include:

(1) The longitudinal distance (d) between the central axis of the pedestrian and the central axis of the vehicle is larger than 3.84m, and the pedestrian is considered not to have interactive risks with the vehicle, so that the pedestrian does not have avoidance behavior;

(2) When the distance between the pedestrian and the vehicle is 2.17m < d <3.84m, the horizontal distance between the pedestrian and the vehicle is considered to be far, and the pedestrian adopts an emergency stop and back avoidance mode;

(3) When 1.95m < d <2.17m, the pedestrian is in the overlapping area of acceleration avoidance and scram back avoidance at this moment, so that the ideal avoidance mode of the pedestrian needs to be judged according to the collision time, and the method comprises the following steps: when TTC is less than 2.3s, the collision risk of the pedestrians adopting an acceleration avoidance mode is higher than that of the backward avoidance mode, so that the pedestrians adopt sudden stop backward avoidance; when TTC is more than 2.3s, the pedestrians adopt accelerating avoidance and backing avoidance, and have high probability of safely avoiding the vehicles, but the pedestrians are considered to tend to adopt accelerating avoidance due to inertia;

(4) When 1m < d <4.95m, collision risks exist in acceleration avoidance and backward avoidance, the boundary model considers that when 0< TTC <1.67s, the collision time cannot support the risk of pedestrian safety acceleration avoidance, and the pedestrian adopts an emergency stop backward avoidance mode; when TTC is more than 1.67s, the pedestrians adopt an acceleration avoidance mode;

(5) When d is less than 1m, the safety of the pedestrian in the vehicle width range is higher than that of the pedestrian in the back-off mode, so that the pedestrian adopts the acceleration to avoid.

It should be noted that, when a pedestrian notices a vehicle in a real dangerous traffic scene, the pedestrian does not make a complete ideal decision, and space-time pressure can cause part of pedestrians to make an erroneous avoidance decision, so that on the basis of the standard avoidance decision model, the pedestrians are considered to have 10% probability of erroneous judgment when making a decision on the avoidance mode, and the part of decision model is called an erroneous avoidance decision model. The pedestrian standard avoidance decision model and the false avoidance decision model jointly form a pedestrian avoidance decision model in a dangerous traffic scene.

In this embodiment of the present application, the computer device may determine, according to the risk perceived boundary condition and the avoidance decision model determined in S1001, a manner of avoiding the digital pedestrian based on a collision time and a relative positional relationship between the pedestrian and the target vehicle.

S1003, determining the collision risk level of the digital pedestrians according to the avoidance mode through a preset collision risk level model.

The preset collision risk level model can be a mean risk domain model, a minimum risk domain model and a maximum risk domain model of different age groups, which are obtained based on the information such as the current running speed, running direction and relative position of the vehicle and the pedestrian. It should be noted that the mean risk domain, the minimum risk domain, and the maximum risk domain models of different age groups may be different.

In the embodiment of the application, the computer equipment can determine the collision risk level of the digital pedestrian through a preset collision risk level model according to different avoidance modes of the pedestrian in the virtual dangerous traffic scene.

Optionally, the determining, by the preset collision risk level model, the collision risk level of the digital pedestrian according to the avoidance mode in S1003 includes: under the condition that the avoidance mode is a standard avoidance mode, determining a collision risk level through a collision risk level model, a relative position relationship and collision time; wherein the collision time is determined based on the vehicle speed and the relative positional relationship.

In the embodiment of the present application, if the pedestrian avoidance mode is the standard avoidance mode, the computer device may determine the collision risk level through the collision risk level model according to the relative positional relationship and the collision time. The relative positional relationship may include a longitudinal distance between the center axes of the pedestrian and the target vehicle, or may be a lateral distance between the pedestrian and the target vehicle. It should be noted that the collision time may be determined based on a relationship between the speed of the target vehicle and the relative position of the pedestrian and the target vehicle, and, for example, the collision time may be obtained based on a ratio of the lateral distance of the target vehicle from the hazard to the current speed.

Optionally, the collision risk level model includes a plurality of distance ranges and time thresholds, as shown in fig. 12, and determining the collision risk level according to the relative positional relationship and the collision time includes:

and S1201, determining that the collision risk level is the first level under the condition that the longitudinal distance and the collision time meet the first preset conditions.

In a dangerous traffic scene, according to the space-time relationship between pedestrians and dangerous vehicles, based on a pedestrian standard avoidance decision model, the most suitable avoidance mode in the current scene can be made under the condition that no erroneous judgment occurs. However, in the real human-vehicle interaction scene, the pedestrian notices that the moment of coming vehicles and the transverse position of the relative vehicle have larger differences with the pedestrian risk perception boundary 2.44s and the space boundary 3.83m, so that collision is possible under a proper avoidance mode, and therefore, the collision risk level needs to be refined according to the relative position relation and the collision time of the pedestrian and the target vehicle.

In the embodiment of the application, under the condition that the longitudinal distance between the central axis of the pedestrian and the central axis of the target vehicle and the collision time accord with the first preset condition, the collision risk level is determined to be the first level. The first preset condition may be that a longitudinal distance between the pedestrian and a central axis of the target vehicle is within a certain distance range, and the collision time is within a certain time range.

For example, as shown in fig. 13, the first preset condition may be 2.45m < d <3.84m, and TTC <1.98s, where the pedestrian adopts an avoidance mode of retreating avoidance; or the collision avoidance mode that pedestrians back and avoid is adopted by the pedestrians with the TTC of less than 1.65s and 2.17m < d <2.45 m; or 1.95m < d <2.17m, and TTC >2.3s or TTC <1.98s, the pedestrians adopt an avoidance mode of accelerating avoidance; the first preset condition can be 1m < d <1.95m, TTC <1.67s, and the pedestrian adopts a retreating avoidance mode or adopts an accelerating avoidance mode; the first preset condition can also be 0.23m < d <1m, TTC <1.67s, and pedestrians adopt an avoidance mode of accelerating avoidance; the first preset condition can be d <0.23m, TTC <1.16s, and the pedestrian adopts an avoidance mode of accelerating avoidance.

S1202, determining that the collision risk level is risk-free under the condition that the longitudinal distance and the collision time meet the second preset conditions.

In the embodiment of the application, the collision risk level is determined to be risk-free under the condition that the longitudinal distance between the central axis of the pedestrian and the central axis of the target vehicle and the collision time meet the second preset condition. The first preset condition may be that a longitudinal distance between the pedestrian and a central axis of the target vehicle is within a certain distance range, and the collision time is within a certain time range.

Illustratively, as shown in fig. 13, the first preset condition may be d >3.84m; or 2.45m < d <3.84m, and TTC >1.98s, and adopting an avoidance mode of retreating and avoiding by pedestrians; the collision avoidance mode can be that the collision avoidance mode is adopted by pedestrians, wherein the collision avoidance mode is that 2.17m < d <2.45m, and the TTC is more than 1.65 s; the first preset condition may be 1.95m < d <2.17m, and 1.98s < TTC <2.3s, and the pedestrian adopts an avoidance mode of retreating avoidance; the first preset condition can also be 0.23m < d <1m, TTC is more than 1.67s, and pedestrians adopt an avoidance mode of accelerating avoidance; the first preset condition can be d <0.23m, TTC >1.16s, and the pedestrian adopts an avoidance mode of accelerating avoidance.

Optionally, the collision risk level model includes a plurality of risk domains, as shown in fig. 14, and determining, according to an avoidance mode, a collision risk level of a digital pedestrian through a preset collision risk level model includes:

s1401, determining a target risk domain where the digital pedestrians are located according to an avoidance mode, a relative position relation and a preset risk domain model.

The preset risk domain model may be a model determined according to information such as the current running speed, running direction, and relative position of the target vehicle and the pedestrian.

Exemplary, operating conditions are as follows: running speed v of vehicle _v =60 km/h, pedestrian detected by vehicle is atThe left front of the vehicle, the motion direction of the vehicle and the pedestrian is vertical, and the walking speed of the pedestrian can be divided into 4 stages according to the avoidance state, namely: the risk perception avoidance decision stage, the first avoidance period, the second avoidance period and the post-avoidance uniform velocity stage, wherein the pedestrian walking centroid velocity is shown in formula (1):

wherein v is _p.Ba Representing the instantaneous speed of the mass center in the decision stage of pedestrian risk perception avoidance; v _p.A1 Representing the instantaneous speed of the mass center of the first avoidance cycle of the pedestrian; v _p.A2 Representing the instantaneous speed of the mass center of the second avoidance cycle of the pedestrian; v _p.Aa Representing a uniform speed stage after pedestrian avoidance; t represents the duration of the first avoidance period; t' represents the duration of the second avoidance period; Δt (delta t) _r Indicating the response time of the pedestrian.

The judgment of collision risk of any pedestrian and target vehicle interaction scene can be converted into the moment t when the vehicle reaches the potential collision area _v At the time, the pedestrian is longitudinally spaced from the central axis of the vehicle

Width L of track of vehicle _vw The space position judgment of the (a) is that the following working conditions have potential collision risks, and the conditions met by the working conditions are expressed by a formula (2):

wherein d _p (t _v ) Representing t _v And the longitudinal movement distance is the time when people are avoided.

In one case, if the pedestrian finds the target vehicle and takes avoidance measures, and the pedestrian takes an acceleration avoidance mode, calculating the instantaneous speed v of the stage of the pedestrian in the acceleration avoidance stage according to the formula (3) _p.FA ：

Wherein v is _rR Representing an average walking speed function of pedestrians in the reaction time of adopting an acceleration avoidance mode; v _p.A1R Representing a first acceleration avoidance gait average speed function; v _p.A2R Representing a second acceleration avoidance gait average speed function; v _p.AaR And representing the average walking speed function of the pedestrians after the pedestrians finish accelerating and avoiding.

At this time, calculating the interactive risk boundary of the pedestrian and the target vehicle in the acceleration avoidance mode according to the formula (2), wherein the interactive risk boundary of the pedestrian and the target vehicle satisfies the formula (4):

wherein d _p.FA (t _v ) Representing t _v And accelerating avoidance longitudinal movement distances of pedestrians in different age groups in time.

In another case, if the pedestrian finds the target vehicle and adopts the avoidance measure, and the pedestrian adopts the deceleration avoidance mode, the interaction risk boundary of the pedestrian and the target vehicle, which selects the deceleration avoidance mode, can be calculated according to the above formulas (3) - (4), and the interaction risk boundary of the pedestrian and the target vehicle satisfies the formula (5):

wherein d _p.BA (t _v ) Representing t _v And the longitudinal movement distance of the pedestrian in different age groups is reduced and avoided in time.

Under a real dangerous traffic scene, a pedestrian notices that the vehicle comes and then selects a proper active avoidance mode according to a relative space-time position relation with a target vehicle and an avoidance decision model, so that an actual pedestrian and vehicle interaction risk boundary is formed by an intersection of an acceleration avoidance risk boundary and a back-off risk boundary after sudden stop, the pedestrian and vehicle interaction risk boundary in the real dangerous scene can be calculated according to a formula (3) and a formula (5), and the pedestrian and vehicle interaction risk boundary in the real dangerous scene is shown as a formula (6):

and (3) obtaining the mean value and the standard deviation of the formula (3), taking the mean value and the standard deviation into the formula (6), calculating a minimum risk domain, and similarly, taking the mean value and the standard deviation into the formula (6), calculating a maximum risk domain, taking the mean value into the formula (6), and calculating a mean risk domain.

In yet another case, if the pedestrian does not find the target vehicle, the pedestrian will always keep traveling forward at a constant speed, and the risk boundary of the interaction between the pedestrian and the vehicle can be expressed by the formula (7):

here, d _p.N (t _v ) Representing t _v Normal walking longitudinal movement distance of pedestrians in different age groups in time.

According to the formula (7), the avoidance-free risk domain of pedestrians at all ages can be determined. The visual result of the collision risk domain of the young male population and the target vehicle is shown in fig. 15, the visual result of the collision risk domain of the middle-aged male population and the target vehicle is shown in fig. 16, and the visual result of the collision risk domain of the old male population and the target vehicle is shown in fig. 17.

In the embodiment of the application, the computer device may determine the target risk domain where the digital pedestrian is located according to different avoidance modes of the pedestrian, the relative positional relationship between the pedestrian and the target vehicle, and the preset risk domain model.

S1402, determining collision risk level of the digital pedestrians according to the avoidance mode and the target risk domain.

In the embodiment of the present application, the computer device may determine the collision risk level of the digital pedestrian according to the different avoidance manners selected by the digital pedestrian and the target risk domain determined in S1401. The collision risk level of the digital pedestrian may be level 3 (unavoidable collision) or level 2 (high risk collision). It should be noted that, the crowd with the collision risk level of the digital pedestrians being the 3 rd level is a direct attention object of the computer device, when the pedestrians enter the pedestrian-vehicle average collision risk domain, the target vehicle needs to execute the pedestrian safety active and passive cooperative protection measures such as whistling early warning, emergency braking, and an external safety airbag, so as to minimize the risk of pedestrian injury. The collision risk level of the digital pedestrians is pedestrians with high collision risk level, the avoidance success of the digital pedestrians has high randomness, the computer equipment can hardly judge whether the pedestrians are always collided, and the people are considered to be important attention objects of the computer equipment from the safety perspective.

The following will be an explanation of the case where the collision risk level of the digital pedestrian is level 3 (unavoidable collision):

based on the pedestrian-vehicle average collision risk domain obtained in S1401, as shown in fig. 18, the functional relationships between the upper and lower boundaries of the pedestrian-vehicle average collision risk domain and the longitudinal direction are defined as F _Top 、F _Down Assume that the moment when a pedestrian notices a dangerous coming vehicle and executes active avoidance is t _r At this time, the lateral relative position of the pedestrian and the vehicle is l (t _r ) The speed of the automatic driving vehicle is 60km/h, the average active avoidance risk domain of young men is the furthest transverse distance A _Fmax When the pedestrian and the risk domain meet the following formula (8), the unavoidable collision risk of the pedestrian can be judged according to the relative position relation between the pedestrian and the pedestrian-vehicle average collision risk domain when the pedestrian notices the vehicle.

l(t _r )<A _Fmax (8)

(1) If the pedestrian notices an incoming vehicle and executes an avoidance action in the pedestrian-vehicle average collision risk domain, and the pedestrian satisfies the formula (9) in the pedestrian-vehicle average collision risk domain, no matter what avoidance mode is adopted by the pedestrian, the pedestrian cannot safely avoid the vehicle, namely:

F _Down (l(t _r ))<D _p (t _r )<F _Top (l(t _r )) (9)

here, D _p (t _r ) Representing t _r Longitudinal relative position between central axis of pedestrian and target vehicle at moment, F _Down (l(t _r ) At the moment of active avoidance of the pedestrian, the corresponding longitudinal distance of the lateral relative position of the pedestrian and the vehicle on the lower boundary of the pedestrian-vehicle average collision risk domain, F _Top (l(t _r ) And) represents the corresponding longitudinal distance of the lateral relative position of the pedestrian and the vehicle on the upper boundary of the pedestrian-vehicle average collision risk domain at the moment when the pedestrian performs active avoidance.

(2) If the pedestrian notices an incoming vehicle and executes an avoidance action outside the average collision risk of the pedestrian and the vehicle, the avoidance mode and D adopted by the pedestrian are needed at the moment _p (t _r ) And F is equal to _Top 、F _Down And judging whether the pedestrian can successfully avoid the vehicle according to the relative position relation.

In one case, if the pedestrian adopts the acceleration avoidance method and the relative positional relationship expressed by the following equation (10) and equation (11) is satisfied, it is considered that the pedestrian cannot avoid collision with the vehicle in the acceleration avoidance method.

F _Down (l(t _r ))<D _p (t _r ) (11)

Here, t _v Representing the time required for the vehicle to reach a potential "collision zone" at the current speed; d, d _p.FA (t _v ) Indicating that the pedestrian adopts an accelerating avoidance mode at t _v Distance moved during the time period.

In another case, if the pedestrian adopts the avoidance mode of sudden stop and reverse, and the relative positional relationship expressed by the following equation (12) and equation (13) is satisfied, it is considered that the pedestrian cannot avoid collision with the vehicle in the manner of deceleration avoidance.

D _p (t _r )<F _Top (l(t _r )) (13)

Here, d _p.BA (t _v ) Indicating that the pedestrian adopts a deceleration avoidance mode to be at t _v Distance moved during the time period.

The following will be an expanded description of the case where the collision risk level of a digital pedestrian is level 2 (high risk collision):

if the relative position between the pedestrians and the vehicles and the average collision risk domain of the pedestrians and the vehicles do not meet the formula (8), the computer equipment cannot distinguish whether the pedestrians can inevitably collide, meanwhile, the pedestrian avoidance decision model shows that when the pedestrians enter the vehicle running track, dangerous vehicles are noticed, and unless the pedestrians are affected by space-time pressure to make wrong judgment, the pedestrians can choose to accelerate to avoid the vehicles. At this time, the running speed of the target vehicle is defined as 60km/h, and the intersection lateral distance between the pedestrian-vehicle maximum acceleration collision risk zone and the leftmost side of the vehicle running track is A as shown in FIG. 18 _RunFmax Defining the lower boundary function of the maximum acceleration collision risk domain of the pedestrian-vehicle as F _RDown And respectively screening whether the following two types of pedestrians have high collision risks:

(1) If the pedestrian notices an incoming vehicle in the vehicle running track and executes the avoiding action, and when the maximum acceleration collision risk domain of the pedestrian and the pedestrian-vehicle and the average collision risk domain of the pedestrian-vehicle meet the condition of the following formula (14), whether the pedestrian has high collision risk or not can be judged.

A _Fmax <l(t _r )<A _RunFmax (14)

Under one condition, if the pedestrian adopts an avoidance mode of accelerating avoidance and satisfies the relative position relationship described in the following formula (15), the pedestrian is considered to have a high collision risk.

In another case, if the pedestrian adopts the avoidance mode of sudden stop and backward, it is determined that the pedestrian makes an erroneous judgment on the avoidance mode according to the standard avoidance decision model described in S1402, so that the collision risk level is an erroneous avoidance risk level.

(2) If the pedestrian notices the target vehicle at the leftmost side far away from the vehicle driving track, and when the pedestrian and the pedestrian-vehicle average collision risk domain meets the condition of the following formula (16), whether the pedestrian has high collision risk can be screened.

A _Fmax <l(t _r ) (16)

In one case, if the pedestrian adopts a back-off avoidance mode, the collision risk level of the pedestrian depends on the collision risk level of the pedestrian under the standard decision.

In another case, if the pedestrian adopts an acceleration avoidance mode, defining the moment when the pedestrian reaches the leftmost side of the vehicle running track as t _left At this time, the lateral relative position of the center axis of the pedestrian and the subject vehicle is l (t _left ) If at t _left The relative positional relationship between the pedestrian and the pedestrian-vehicle maximum acceleration collision risk domain at the moment satisfies the condition satisfied by the following equation (17), and the pedestrian is considered to have a high collision risk.

l(t _left )<A _RunFmax (17)

Optionally, the determining, by using a preset collision risk level model, the collision risk level of the digital pedestrian according to the avoidance mode further includes: under the condition that the avoidance mode is an error avoidance mode, determining that the collision risk level is the error avoidance risk level according to the collision risk level model.

The pedestrian avoidance decision model in S1402 assumes that the pedestrian has 10% probability of erroneously selecting the current avoidance mode under the dangerous working condition, and that the incorrect avoidance mode may cause the pedestrian to have a higher collision risk, and defines the collision risk level of the pedestrian as the incorrect avoidance risk level. A method flow for screening for false decision pedestrians is shown in fig. 19.

The standard avoidance decision model in S1402 is an idealized pedestrian decision model, and the model considers that the pedestrian will select the most appropriate avoidance mode under dangerous working conditions. Therefore, the decision model of the standard avoidance mode is used as a judgment basis for the computer equipment to judge whether the pedestrian makes an error decision. When the pedestrian notices the target vehicle and takes the active avoidance measures, the target vehicle can judge the avoidance measures taken by the pedestrian according to the change of the speed of the pedestrian, and compare the actual avoidance measures taken by the pedestrian with the decision result of the standard avoidance mode, if the pedestrian is consistent, the pedestrian is considered to select an appropriate avoidance mode, if the pedestrian is inconsistent, the pedestrian is considered to be influenced by space-time pressure to select an improper avoidance mode, and under the condition that the pedestrian selects an error avoidance mode, the computer equipment determines that the collision risk level is the error avoidance risk level according to the collision risk level model.

Further, two types of pedestrians with high collision risk and unavoidable collision risk are assisted and screened based on a pedestrian-vehicle collision risk domain and a vehicle running track. The other pedestrians with the risk level of 1 have lower collision risk, and compared with three groups of people with wrong avoidance, high collision risk and unavoidable collision risk, the computer equipment can properly reduce the attention degree of the pedestrians with the risk level of 1.

In one case, for a pedestrian that selects a sudden stop and a reverse avoidance, when it satisfies the condition of the following equation (18) -equation (20) at time t, the collision risk level of the pedestrian is cleared.

t<t _v (18)

0<v _p.BA (t) (19)

Wherein v is _p.BA (t) represents the instantaneous speed of the pedestrian after scram and back-off at the moment t; d (D) _p And (t) represents the longitudinal distance of the pedestrian relative to the central axis of the vehicle at the moment t. This indicates that the pedestrian has retracted to the far left of the vehicle path before the vehicle reaches the potential "collision zone" at which time the pedestrian must not collide with the vehicle, thus bringing this categoryThe collision risk level of pedestrians is adjusted to 0 and the computer device is no longer concerned with such pedestrians.

In another case, for a pedestrian whose acceleration avoidance is selected, when it satisfies the condition of the following formulas (21) to (23) at time t, the collision risk level of the pedestrian is cleared.

t<t _v (21)

v _p.FA (t)>0 (22)

Wherein v is _p.FA (t) represents the instantaneous speed of pedestrian acceleration avoidance at the moment t; d (D) _p And (t) represents the longitudinal distance of the pedestrian relative to the central axis of the vehicle at the moment t. This indicates that the pedestrian has been far to the far right of the vehicle travel path before the vehicle reaches the potential "collision zone", at which point the pedestrian must not collide with the vehicle, so the collision risk level for such pedestrian is adjusted to 0 and the computer device is no longer focusing on such pedestrian.

According to the method for acquiring the collision risk level of the digital pedestrian, the relative position relation between the digital pedestrian and the target vehicle is acquired, the avoidance mode of the digital pedestrian is determined according to the pedestrian generation model and the relative position relation, and the collision risk level of the digital pedestrian is determined according to the avoidance mode through the preset collision risk level model. In the implementation, the avoidance mode of the digital pedestrians is determined based on the pedestrian generation model and the relative position relation between the pedestrians and the target vehicle, so that the determined avoidance mode of the digital pedestrians is more accurate; further, through a preset collision risk level model, determining the collision risk level of digital pedestrians according to an avoidance mode, wherein for the pedestrians which cannot avoid collision (the risk level is 3 rd level), the crowd is a direct attention object of a target vehicle, and after the pedestrians enter a mean value risk domain of the pedestrians-vehicles, the target vehicle needs to execute pedestrian safety active and passive cooperative protection measures such as whistling early warning, emergency braking, an outside-vehicle safety airbag and the like, so that the risk of pedestrian injury is minimized; for pedestrians with high collision risk, the avoidance success of the pedestrians is highly random, and the crowds are considered as important attention objects of computer equipment from the safety perspective, so that the risk of pedestrian injury is further minimized.

In one embodiment, on the basis of the embodiment shown in fig. 14, the controlling the target vehicle to execute the anti-collision process according to the collision risk level, to obtain the test result includes: controlling the target vehicle to execute braking processing corresponding to the collision risk level; and under the condition that the collision risk level accords with the preset level, starting the pedestrian safety airbag.

Wherein the preset level may be an unavoidable collision level. The braking treatment can be vehicle primary deceleration braking, vehicle secondary deceleration braking, and vehicle tertiary deceleration braking. The first-level deceleration of the vehicle

The braking satisfies the following formula (24), the two-stage deceleration degree of the vehicle +.>

The braking satisfies the following formula (25), three-stage deceleration +.>

Braking satisfies the following formula (26):

in the embodiment of the application, the computer device may control the target vehicle to execute different braking processes corresponding to different collision risk levels, and the computer device may further directly start the pedestrian airbag when the collision risk level meets a preset level.

The following describes in detail the different braking processes corresponding to the different collision risk classes:

(1) And if the pedestrian does not notice the dangerous coming vehicle and enters the avoidance-free risk domain, giving a vehicle braking measure according to the potential collision time TTC of the pedestrian and the vehicle at the moment.

In one case, if the TTC satisfies the following equation (27), the vehicle assumes three levels of braking with a whistle warning. At this time, the highest risk of collision exists between the pedestrians and the vehicles under the most dangerous working conditions, and the vehicles need to adopt the highest-grade braking to maximally reduce the speed of the vehicles and minimize the collision damage of the pedestrians.

0<TTC<0.5 s (27)

In another case, if the TTC satisfies the following formula (28), the target vehicle adopts a secondary braking, and simultaneously whistles for early warning, the secondary braking is a buffer braking, and the target vehicle adjusts the braking level in real time according to the interaction position with the pedestrian.

0.5 s<TTC<1.8 s (28)

In yet another case, if the TTC satisfies the following equation (29), the target vehicle assumes primary braking while whistling warning. At the moment, pedestrians and vehicles interact in a long distance, and the vehicles can avoid collision by lightly stepping on the brakes.

1.8 s<TTC<2.44 s (29)

(2) If the pedestrian notices the target vehicle and enters the avoidance risk domain, the pedestrian and the vehicle interaction at this time is considered to belong to short-distance interaction (TTC <1.8 s), and the vehicle braking logic is basically consistent with the above, namely: if the TTC is 0.5s <1.8s, the vehicle adopts secondary braking; if 0< TTC <0.5s, the vehicle adopts three-stage braking.

According to the braking processing method for the target vehicle, the target vehicle can be controlled to execute the braking processing corresponding to the collision risk level, and the pedestrian safety airbag is started under the condition that the collision risk level accords with the preset level. The method and the device enable different braking processes to be corresponding to different collision risk levels, so that the target vehicle can adjust the braking level in real time according to the interaction position with the pedestrian, and interaction between the pedestrian and the vehicle in a dangerous traffic scene is further simulated to the greatest extent.

In one embodiment, the method further comprises, based on the embodiment shown in fig. 14: outputting prompt information under the condition that the digital pedestrians do not actively avoid; the prompt information is used for prompting the digital pedestrians to avoid.

In this application embodiment, under the circumstances that the digital pedestrian does not actively avoid and enter the risk domain without avoiding, the target vehicle may take the whistle measure to prompt the pedestrian to judge the risk for the second time, so as to prompt the digital pedestrian to avoid, namely: in a single operation interval of the test platform (the test platform is generally at intervals of 0.01 s), as shown in fig. 20, an interference item I is randomly set in the intervals of [0,1] again, if the current I <0.1 considers that the vehicle whistling measures force pedestrians to pay attention to the vehicles and enter an avoidance decision stage, if the current I >0.1 considers that the pedestrians do not pay attention to dangerous vehicles, the judgment is carried out again in the next operation interval.

According to the prompting method provided by the embodiment of the application, under the condition that the digital pedestrians do not actively avoid, prompting information for prompting the digital pedestrians to avoid is output. Based on the method, the activation of the pedestrian avoiding state in the avoiding-free risk area is coupled with the vehicle whistle early warning, so that the interaction between the pedestrian and the vehicle in the dangerous traffic scene is truly simulated.

In one embodiment, based on the embodiment shown in fig. 14, the pedestrian generation model includes a boundary model, an avoidance decision model and a kinematic model, where the boundary model is used to define a risk perception boundary condition of the digital pedestrian, the avoidance decision model is used to determine an avoidance mode of the digital pedestrian, and the kinematic model is used to generate an avoidance speed curve of the digital pedestrian.

The kinematic model is a model for describing the transient speed of the pedestrian in the active avoidance process, and the model can fully reflect the randomness of the pedestrian in the active avoidance process. It should be noted that, the active avoidance behavior of a pedestrian has high randomness, for example, there are differences in the moment when different pedestrians realize potential risks, the selected avoidance mode and the instantaneous speed of executing the avoidance behavior in the same dangerous traffic scene, or the same pedestrians realize potential risks at the same moment, and the instantaneous speed of executing the avoidance behavior of the same avoidance mode is also dependent on the current instantaneous speed of the pedestrian.

In this embodiment of the present application, the boundary model and the avoidance decision model have been specifically described in the foregoing embodiment, and will not be described in detail in this embodiment. The kinematic model is described in detail below.

(1) Simplifying the original single-step transient kinematic equation:

in the process of fitting a single avoidance speed curve of a pedestrian by computer equipment, parameters limt of a single-step transient kinematic equation ₁ 0, so that the fitting parameter t of the standard avoidance gait cycle can be ignored ₁ An active avoidance function in each standard avoidance gait of the pedestrian is obtained, and the expression of the active avoidance function is shown in the following formula (30):

wherein v is _A (t) the transient speed in each standard avoidance gait, and C the mass center speed at the moment when the pedestrian starts to execute the active avoidance action; a, a ₁ The average deceleration at the end of the single-side swing of the pedestrian; a, a ₂ Early average acceleration for pedestrian contralateral swing; t is t ₂ The boundary time for the transformation of the end phase of the unilateral swing and the early phase of the contralateral swing of the pedestrian.

(2) And (3) determining the active avoidance function parameters:

defining the time when the pedestrian starts to execute the active avoidance action as a zero moment point, wherein the mass center speed of the pedestrian at the moment is v ₀ . Let t be ₂ The speed at the moment of time is v ₁ The speed of the single step avoidance ending time T is v ₂ Namely (0, v) ₀ ) Starting point of active avoidance function (t) ₂ ,v ₁ ) Is the inflection point of the active avoidance function, (T, v) ₂ ) Is the end point of the active avoidance function.

Single step transient equation of motionThe medium parameter C is the swing medium-term speed of the standard gait cycle of the pedestrian, and the medium parameter C is the swing medium-term speed of the standard gait cycle of the pedestrian, which is in limt ₁ Under the condition of 0, C is the mass center speed v at the moment when the pedestrian starts to execute the avoidance action ₀ ；a ₁ 、a ₂ The two parameters can be solved according to the following equations (31) and (32):

as shown in fig. 21 and 22, the boundary point t between the end of unilateral swing and the early of contralateral swing in two standard avoidance gait cycles of the crowd average avoidance speed curve ₂ 、t' ₂ And an avoidance gait cycle T, T', wherein the first standard avoidance gait cycle start point of the pedestrian is taken as a zero moment point, and the zero moment point speed v is randomly generated according to the uniform distribution principle ₀ Same t ₂ Instantaneous speed v at time T ₁ 、v ₂ Are randomly generated in the crowd avoiding speed corridor at corresponding moment according to the uniform distribution principle, v ₀ Generated according to the following formula (33), v ₁ Generated according to the following formula (34), v ₂ Generated according to the following formula (35):

v ₀ ＝v _0min +(v _0max -v _0min )×rand (33)

v ₁ ＝v _1min +(v _1max -v _1min )×rand (34)

v ₂ ＝v _2min +(v _2max -v _2min )×rand (35)

wherein v is _0min To minimum initial avoidance speed, v _0max For maximum initial avoidance speed, rand is [0,1]Random number between v _1min For minimum mid-back support velocity, v _1max For maximum mid-back support velocity, v _2min To minimize the early speed of swing, v _2max The early speed of swing is the maximum avoidance.

Bringing equations (33) - (35) into equations (31) and (32), respectively, yields a ₁ And a ₂ The first standard avoidance function of the single avoidance of the pedestrian can be obtained, and the first standard avoidance function of the single avoidance of the pedestrian is represented by the following formula (36):

the starting point speed of the second standard avoidance gait is the end point speed v of the first standard avoidance gait ₂ The starting point of the gait of the second standard avoidance is taken as a zero moment point, and the rest parameter determination method is consistent with the first standard avoidance function parameter determination method, so that a second standard avoidance function is determined, and the single avoidance of the pedestrian is represented by the following formula (37):

as shown in fig. 23 and 24, the single avoidance speed curve of the pedestrian is obtained by splicing the formula (36) and the formula (37) according to the time sequence, fig. 23 is a single deceleration avoidance speed curve of the pedestrian, and fig. 24 is a single acceleration avoidance speed curve of the pedestrian.

In the real human-vehicle interaction process, pedestrians have corresponding speeds in the front, middle and rear stages of avoidance, and meanwhile, the speed curves of the pedestrians in the three stages are required to accord with the kinematic characteristics of people and have physical continuity in time. The following specifically describes the running speeds of the young men in the three stages of before, during and after avoidance.

(1) Pedestrian operation curve before avoidance:

define initializing young male speed as

The average walking speed of the speed corridor before the young men avoid is

The mapping factor of the young man pre-avoidance speed profile is expressed by the following equation (38):

here the number of the elements is the number,

for the average speed of the young man before avoiding the speed corridor at zero moment, the mapping factor determined by the formula (38) is utilized, as shown in fig. 25, and the average speed curve before avoiding the crowd is mapped by the following formula (39) to obtain the speed curve in the period before avoiding the young man:

(2) Avoidance middleman running curve:

the young men periodically run in the pre-avoidance speed curve until the dangerous coming vehicles are noticed at the time t, and the instantaneous speed is that

The pedestrian has a fixed reaction time delta t from noticing a dangerous coming vehicle to executing the avoidance action _r Defining the average walking speed of young male response time speed corridor as +.>

The mapping factor of the young man reaction time-velocity curve is shown in the following formula (40):

wherein, the liquid crystal display device comprises a liquid crystal display device,

for the average speed of the young man in the response time and speed corridor at zero time, the mapping factor in the formula (40) is utilized, as shown in fig. 26, and the average speed curve of the crowd response time is mapped by the following formula (41) to obtain the speed curve of the young man in the fixed response time: / >

Further, according to the digital pedestrian active avoidance behavior kinematic model, taking the moment of completing a response time speed curve of a young man as a new time zero point, generating a first standard avoidance gait speed function, wherein the expression of the function is shown in a formula (42):

wherein v is _YB1 Representing a first standard avoidance gait speed for young men; v _YR Representing walking speed function in young man reaction time;

representing the standard avoiding gait cycle demarcation point speed of young men; />

Representing the standard avoidance gait cycle demarcation point time of young men; />

Representing a first standard avoidance gait endpoint speed; t (T) _Y Representing the first standard avoidance gait cycle duration for young men.

Likewise, the moment of completing the first standard avoiding gait speed function is taken as a new time zero point,

as a second avoidance gait starting speed, a second standard avoidance gait speed function is generated, the expression of which is shown in formula (43):

wherein v is _YB2 Representing a second standard avoidance gait speed for young men;

representing the second standard avoiding gait cycle demarcation point speed of young men; />

Representing the second standard avoiding gait cycle demarcation point time of young men; / >

Representing a second standard avoidance gait endpoint speed; t'. _Y Representing the length of the second standard avoidance gait cycle for young men.

As shown in fig. 27, the response time speed curve and the two standard avoidance gait speed curves are spliced according to time sequence, namely, the current avoidance speed curve of the young man.

(3) Pedestrian operation curve after avoidance:

after the young men finish the fixed active avoidance curve, the young men continue to run with the speed curve after avoidance until dangerous vehicles arrive at a potential collision area. Defining the average walking speed of the speed corridor after the young men avoid as

The mapping factor of the speed profile after the young man dodges is shown in the following formula (44):

for the average speed of the young man population in the zero moment of the speed corridor after avoidance, the mapping factor in the formula (44) is utilized, as shown in fig. 28, and the average speed curve after avoidance of the population is mapped by the following formula (45) to obtain the speed curve in a period after avoidance of the young man:

and splicing the speed curves before, during and after the avoidance of the pedestrians according to the time sequence to obtain a complete speed curve of the pedestrians.

The boundary model, the avoidance decision model and the kinematic model provided by the embodiment of the application are respectively used for defining the risk perception boundary condition of the digital pedestrians, determining the avoidance mode of the digital pedestrians, generating the avoidance speed curve of the digital pedestrians, considering the initiative behavior capability of the digital pedestrians when the digital pedestrians face dangerous vehicles, and further improving the accuracy of the virtual collision test.

In one embodiment, as shown in fig. 29, an intelligent automobile safety digital test platform for digital pedestrian active behavior is constructed according to the operation logic among a typical dangerous traffic test scene, a digital pedestrian and an automatic driving vehicle. The architecture of the automatic driving platform comprises a test scene, digital pedestrians and intelligent automobiles, wherein the test scene comprises typical traffic scenes such as an intersection, a crossing road, a ghost probe, a T-shaped intersection and the like in the scene introduction of the embodiment, the digital pedestrians have risk perception, avoidance decision and active avoidance capability, and the intelligent automobiles store collision risk domains, vehicle-mounted AEB and collision risk grades of pedestrians of different age groups.

Taking a complete virtual dangerous interaction scene test as an example, the operation logic of the intelligent automobile safety digital test platform for digital pedestrian active behavior is described below. The one-time complete dangerous interaction scene has three test factors of automatic driving vehicles, digital pedestrians and dangerous traffic conditions. First, defining initial attributes of the automatic driving vehicle:

table 2 autopilot vehicle attributes

Initial coordinates (0, 0) for the autonomous vehicle, initial speed 60km/h, vehicle length 4m, vehicle width 2m, t _v ∈[0,2.44s]Dangerous test conditions and pedestrians coupled with the dangerous test conditions and having potential collision risks are randomly generated in the range. At this time, the three parts together form a complete dangerous interaction test scene. After the test scene is predefined, the vehicle end detects the digital pedestrians in real time through the simulation radar, and if the digital pedestrians are not in the radar detection range, no interaction occurs between the vehicle and the pedestrians; when the radar detects a pedestrian and the pedestrian enters the non-avoidance risk domain, the vehicle adopts graded braking according to the relative position relation with the pedestrian and whistles to remind the pedestrian, so that the pedestrian is prompted to perceive the risk again until the pedestrian notices the dangerous vehicle.

After the digital pedestrian realizes the potential risk, combining the relative position relationship between the digital pedestrian and the vehicle, selecting the avoiding mode of the interaction according to the avoiding decision model, generating the active avoiding speed curve, and giving the collision risk grade of the pedestrian at the initial avoiding moment according to the collision risk grade model; meanwhile, the vehicle risk early warning system detects that pedestrians are converted from normal walking to active avoidance walking, and the risk domain is switched from the avoidance-free risk domain to the average avoidance risk domain corresponding to the ages of the pedestrians; based on the switched average value avoidance risk domain and the maximum acceleration avoidance risk domain of the corresponding pedestrians, screening whether the pedestrians are at high collision risk or unavoidable collision risk, and clearing the collision risk level of the pedestrians away from the running track of the vehicle; if the pedestrian enters the mean risk domain in the avoidance process, the vehicle-mounted risk early warning system starts AEB hierarchical braking, and if the collision risk level of the digital pedestrian is the unavoidable collision risk level, the pedestrian safety airbag is started. If the pedestrian safety protection algorithm fails, when the collision time TTC=50 ms between the pedestrian and the vehicle, directly starting the pedestrian safety airbag to minimize the risk of pedestrian injury; finally, when the vehicle reaches a potential collision area and collides with a pedestrian, the pedestrian number is recorded, and if the pedestrian does not collide, the automatic driving vehicle stops moving, and continues to run to interact with the next pedestrian.

It should be understood that, although the steps in the flowcharts related to the embodiments described above are sequentially shown as indicated by arrows, these steps are not necessarily sequentially performed in the order indicated by the arrows. The steps are not strictly limited to the order of execution unless explicitly recited herein, and the steps may be executed in other orders. Moreover, at least some of the steps in the flowcharts described in the above embodiments may include a plurality of steps or a plurality of stages, which are not necessarily performed at the same time, but may be performed at different times, and the order of the steps or stages is not necessarily performed sequentially, but may be performed alternately or alternately with at least some of the other steps or stages.

Based on the same inventive concept, the embodiment of the application also provides a virtual collision test device for realizing the virtual collision test method. The implementation of the solution provided by the device is similar to the implementation described in the above method, so the specific limitation of one or more embodiments of the virtual collision test device provided below may be referred to the limitation of the virtual collision test method hereinabove, and will not be repeated herein.

In one embodiment, as shown in fig. 30, there is provided a virtual collision test apparatus, comprising: a generating module 3001, a first acquiring module 3002, and a second acquiring module 3003, wherein:

the generating module 3001 is configured to obtain a test scene, and generate a digital pedestrian based on a preset pedestrian generating model.

The first obtaining module 3002 is configured to obtain a collision risk level of the digital pedestrian when the digital pedestrian is detected to actively avoid the target vehicle.

A second module 3003, configured to control the target vehicle to perform anti-collision processing according to the collision risk level, so as to obtain a test result; the test result is used for representing whether the target vehicle collides with the digital pedestrian after the anti-collision treatment is performed.

In one embodiment, the first obtaining module 3002 includes: an acquisition unit, a first determination unit, and a second determination unit. Wherein:

the acquisition unit is specifically used for acquiring the relative position relationship between the digital pedestrian and the target vehicle; the relative positional relationship includes a lateral distance and a longitudinal distance between the digital pedestrian and the target vehicle.

The first determining unit is specifically configured to determine an avoidance mode of the digital pedestrian according to the pedestrian generation model and the relative position relationship.

The second determining unit is specifically configured to determine, according to a preset collision risk level model, a collision risk level of the digital pedestrian according to an avoidance mode.

In one embodiment, the second determining unit is specifically configured to determine, when the avoidance mode is a standard avoidance mode, a collision risk level according to a relative position relationship and a collision time through a collision risk level model; wherein the collision time is determined based on the vehicle speed and the relative positional relationship.

In one embodiment, the collision risk level model includes a plurality of distance ranges and time thresholds, and the second determining unit is specifically configured to determine that the collision risk level is the first level if the longitudinal distance and the collision time meet a first preset condition; and under the condition that the longitudinal distance and the collision time meet the second preset conditions, determining the collision risk level as no risk.

In one embodiment, the collision risk level model includes a plurality of risk domains, and the second determining unit is specifically configured to determine, according to an avoidance manner, a relative position relationship, and a preset risk domain model, a target risk domain in which the digital pedestrian is located; and determining the collision risk level of the digital pedestrians according to the avoidance mode and the target risk domain.

In an embodiment, the second determining unit is specifically configured to determine, according to the collision risk level model, that the collision risk level is an error avoidance risk level when the avoidance mode is the error avoidance mode.

In one embodiment, the pedestrian generation model includes a boundary model and an avoidance decision model, and the first determining unit is specifically configured to obtain a risk perception boundary condition based on the boundary model; and determining the avoidance mode of the digital pedestrians according to the risk perception boundary conditions and the avoidance decision model.

In one embodiment, the second obtaining module 3003 is configured to control the target vehicle to perform a braking process corresponding to the collision risk level; and under the condition that the collision risk level accords with the preset level, starting the pedestrian safety airbag.

In one embodiment, as shown in fig. 31, the apparatus further includes: the output module 3004 is configured to output prompt information when the digital pedestrian does not actively avoid; the prompt information is used for prompting the digital pedestrians to avoid.

In one embodiment, the pedestrian generation model includes a boundary model, an avoidance decision model and a kinematic model, wherein the boundary model is used for defining a risk perception boundary condition of the digital pedestrian; the avoidance decision model is used for determining the avoidance mode of the digital pedestrians; the kinematic model is used for generating an avoidance speed curve of the digital pedestrians.

The respective modules in the virtual collision test apparatus described above may be implemented in whole or in part by software, hardware, and a combination thereof. The above modules may be embedded in hardware or may be independent of a processor in the computer device, or may be stored in software in a memory in the computer device, so that the processor may call and execute operations corresponding to the above modules.

In one embodiment, a computer device is provided, which may be a server, the internal structure of which may be as shown in fig. 6. The computer device includes a processor, a memory, an Input/Output interface (I/O) and a communication interface. The processor, the memory and the input/output interface are connected through a system bus, and the communication interface is connected to the system bus through the input/output interface. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The database of the computer device is used for storing collision result data. The input/output interface of the computer device is used to exchange information between the processor and the external device. The communication interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is executed by a processor to implement a virtual collision test method.

It will be appreciated by those skilled in the art that the structure shown in fig. 6 is merely a block diagram of some of the structures associated with the present application and is not limiting of the computer device to which the present application may be applied, and that a particular computer device may include more or fewer components than shown, or may combine certain components, or have a different arrangement of components.

In an embodiment, there is also provided a computer device comprising a memory and a processor, the memory having stored therein a computer program, the processor implementing the steps of the method embodiments described above when the computer program is executed.

In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored which, when executed by a processor, carries out the steps of the method embodiments described above.

In an embodiment, a computer program product is provided, comprising a computer program which, when executed by a processor, implements the steps of the method embodiments described above.

Those skilled in the art will appreciate that implementing all or part of the above described methods may be accomplished by way of a computer program stored on a non-transitory computer readable storage medium, which when executed, may comprise the steps of the embodiments of the methods described above. Any reference to memory, database, or other medium used in the various embodiments provided herein may include at least one of non-volatile and volatile memory. The nonvolatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical Memory, high density embedded nonvolatile Memory, resistive random access Memory (ReRAM), magnetic random access Memory (Magnetoresistive Random Access Memory, MRAM), ferroelectric Memory (Ferroelectric Random Access Memory, FRAM), phase change Memory (Phase Change Memory, PCM), graphene Memory, and the like. Volatile memory can include random access memory (Random Access Memory, RAM) or external cache memory, and the like. By way of illustration, and not limitation, RAM can be in the form of a variety of forms, such as static random access memory (Static Random Access Memory, SRAM) or dynamic random access memory (Dynamic Random Access Memory, DRAM), and the like. The databases referred to in the various embodiments provided herein may include at least one of relational databases and non-relational databases. The non-relational database may include, but is not limited to, a blockchain-based distributed database, and the like. The processors referred to in the embodiments provided herein may be general purpose processors, central processing units, graphics processors, digital signal processors, programmable logic units, quantum computing-based data processing logic units, etc., without being limited thereto.

The technical features of the above embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.

The above examples only represent a few embodiments of the present application, which are described in more detail and are not to be construed as limiting the scope of the present application. It should be noted that it would be apparent to those skilled in the art that various modifications and improvements could be made without departing from the spirit of the present application, which would be within the scope of the present application. Accordingly, the scope of protection of the present application shall be subject to the appended claims.

Claims

1. A virtual collision test method, the method comprising:

controlling the target vehicle to execute anti-collision processing according to the collision risk level to obtain a test result; the test result is used for representing whether the target vehicle collides with the digital pedestrian after the anti-collision processing is executed.

2. The method of claim 1, wherein the acquiring the collision risk level of the digital pedestrian if the digital pedestrian is detected to actively avoid the target vehicle comprises:

acquiring the relative position relation between the digital pedestrian and the target vehicle; the relative positional relationship includes a lateral distance and a longitudinal distance between the digital pedestrian and the target vehicle;

determining the avoidance mode of the digital pedestrians according to the pedestrian generation model and the relative position relation;

and determining the collision risk level of the digital pedestrian according to the avoidance mode through a preset collision risk level model.

3. The method according to claim 2, wherein the determining, by a preset collision risk level model, the collision risk level of the digital pedestrian according to the avoidance mode includes:

under the condition that the avoidance mode is a standard avoidance mode, determining the collision risk level according to the relative position relation and the collision time through the collision risk level model; wherein the collision time is determined based on a vehicle speed and the lateral relative positional relationship.

4. A method according to claim 3, wherein the collision risk level model comprises a plurality of distance ranges and time thresholds, and wherein said determining the collision risk level from the relative positional relationship and collision time comprises:

under the condition that the longitudinal distance and the collision time meet a first preset condition, determining the collision risk level as a first level;

and under the condition that the longitudinal distance and the collision time meet a second preset condition, determining the collision risk level as no risk.

5. The method according to claim 2, wherein the collision risk level model includes a plurality of risk domains, and the determining the collision risk level of the digital pedestrian according to the avoidance mode through the preset collision risk level model includes:

determining a target risk domain where the digital pedestrian is located according to the avoidance mode, the relative position relationship and a preset risk domain model;

and determining the collision risk level of the digital pedestrian according to the avoidance mode and the target risk domain.

6. The method according to claim 2, wherein the determining, by a preset collision risk level model, the collision risk level of the digital pedestrian according to the avoidance mode includes:

And under the condition that the avoidance mode is an error avoidance mode, determining that the collision risk level is an error avoidance risk level according to the collision risk level model.

7. The method of claim 2, wherein the pedestrian generation model includes a boundary model and an avoidance decision model, and wherein the determining the manner of avoidance of the digital pedestrian based on the relative positional relationship between the pedestrian generation model and the pedestrian comprises:

acquiring a risk perception boundary condition based on the boundary model;

and determining the avoidance mode of the digital pedestrians according to the risk perception boundary condition and the avoidance decision model.

8. The method according to claim 1, wherein the controlling the target vehicle to perform the anti-collision process according to the collision risk level to obtain the test result includes:

controlling the target vehicle to execute braking processing corresponding to the collision risk level; and

and under the condition that the collision risk level accords with a preset level, starting the pedestrian safety airbag.

9. The method according to claim 1, wherein the method further comprises:

outputting prompt information under the condition that the digital pedestrians do not actively avoid; the prompt information is used for prompting the digital pedestrians to avoid.

10. The method of claim 1, wherein the pedestrian generation model comprises a boundary model, an avoidance decision model, and a kinematic model, the boundary model for defining risk-aware boundary conditions for the digital pedestrian;

the avoidance decision model is used for determining the avoidance mode of the digital pedestrians;

the kinematic model is used for generating an avoidance speed curve of the digital pedestrian.

11. A virtual collision test apparatus, the apparatus comprising:

the second acquisition module is used for controlling the target vehicle to execute anti-collision processing according to the collision risk level to obtain a test result; the test result is used for representing whether the target vehicle collides with the digital pedestrian after the anti-collision processing is executed.

12. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor implements the steps of the method of any one of claims 1 to 10 when the computer program is executed.

13. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the steps of the method of any of claims 1 to 10.

14. A computer program product comprising a computer program, characterized in that the computer program, when being executed by a processor, implements the steps of the method of any one of claims 1 to 10.