Automobile driving safety simulation test method and device and electronic equipment

Abstract

The invention provides an automobile driving safety simulation test method and device and electronic equipment, and relates to the technical field of automatic driving, and the method comprises the steps: obtaining an anthropomorphic driving model at the beginning of a test, controlling a target background vehicle through the anthropomorphic driving model, and testing a tested vehicle based on the target background vehicle. According to the method, the interactivity between the target background vehicle and the tested vehicle and the randomness of the driving decision of the target background vehicle can be controlled in a personification manner, so that potential risk events can be triggered more effectively, and the tested vehicle can be tested better.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of automatic driving, in particular to a driving car safety simulation test method and device and electronic equipment. BACKGROUND

[0002] With the booming development of autonomous vehicle industry, algorithm safety testing becomes increasingly important in the development process of autonomous vehicles. In order to reduce development costs, virtual simulation testing is widely used to test the safety of autonomous vehicles. In the test scheme of enterprises, inherent matrix scene testing is a common simulation test method, which is mainly carried out in a set of predefined scenes, which usually include but are not limited to: following scene, lane changing scene, emergency braking scene, etc. By changing various parameters such as speed, vehicle distance, environmental conditions, etc. in these inherent scenes, testers can evaluate the performance and safety of autonomous driving algorithms in specific situations. This method has been widely used in virtual simulation platforms such as Simulink, etc. However, the limitations of inherent matrix scene testing have become increasingly apparent, especially in dealing with complex, variable and unpredictable real traffic conditions.

[0003] For example: patent CN115344503B discloses a traffic flow simulation system and simulation method for autonomous driving planning and control testing. In order to provide interactive high-fidelity background traffic flow in simulation testing, the sensor simulation module and vehicle dynamics module are linked to the simulation scene engine. The autonomous driving system is composed of multiple modules, including perception algorithm module, positioning algorithm module, planning algorithm module and control algorithm module. These modules interact through message middleware. Using this method can improve the realism of testing and accelerate the discovery and optimization of deficiencies in autonomous driving system planning and control algorithms. This test method well establishes the motion characteristics of the vehicle in simulation, but does not consider how the behavior characteristics of human drivers are reflected in the background vehicles.

[0004] The patent CN114398251A discloses an automatic driving simulation test method and system based on a scene database. The method uses a scene database, which includes various legal test scenes. A legal test scene refers to a set of data units that meet certain conditions. These data units are generated by input data, including main vehicle data units, interactive vehicle data units, road data units, and weather data units. The method can retrieve all possible and test-demand-compliant test scenes based on input data, thereby avoiding the problem of single scene construction and possible omissions in the prior art. At the same time, the scene library construction of the method supports unlimited number of tests, greatly improving the test efficiency. The method comprehensively constructs test scenes, covering common scene elements, but does not depict the behavior characteristics of human drivers, such as competitive gaming, in the scene, resulting in the background vehicle actions in the test scene being a repetition of the behavior in the data set, without constructing the behavior interaction between the background vehicle and the vehicle under test, and without embodying the randomness in the scene.

[0005] In actual driving scenarios, driving behavior is influenced by the driving habits, emotions, and specific situations of the driver, and has certain behavior randomness. Therefore, even if the automatic driving vehicle is trained through simulation testing, it still lacks good decision-making ability in actual driving scenarios. SUMMARY

[0006] Therefore, the embodiments of the present application provide a driving car safety simulation test method, device and electronic equipment to realize testing of the vehicle under test by simulating actual driving scenarios, so that the vehicle after testing still has good decision-making ability in actual driving scenarios.

[0007] To achieve the above object, the embodiments of the present application provide the following technical solutions:

[0008] A driving car safety simulation test method, comprising:

[0009] Obtaining a personified driving model;

[0010] Controlling a target background vehicle using the personified driving model;

[0011] Using the target background vehicle as a background vehicle to test the driving car safety of the vehicle under test.

[0012] Optionally, in the driving car safety simulation test method, the personified driving model and the target background vehicle are one-to-one corresponding, and there are at least one personified driving model and one target background vehicle; controlling the target background vehicle using the personified driving model comprises:

[0013] Obtaining the control of the target background vehicle corresponding to the personified driving model using the personified driving model.

[0014] Optionally, in the driving safety simulation test method, the target background vehicle is controlled by the humanized driving model, and the method comprises the following steps.

[0015] Obtaining inherent scene variables and value ranges of the inherent scene variables;

[0016] Obtaining behavior-related variables and value ranges of the behavior-related variables;

[0017] Constructing a test matrix based on the behavior-related variables, the inherent scene variables and the value ranges of the variables;

[0018] Controlling the target background vehicle based on the test matrix by using the humanized driving model.

[0019] Optionally, in the driving safety simulation test method, the behavior-related variables are obtained, and the method comprises the following steps.

[0020] Obtaining a target test scene;

[0021] Obtaining behavior-related variables matched with the target test scene;

[0022] Before the target background vehicle is controlled by using the humanized driving model, the method further comprises the following steps.

[0023] Selecting a background vehicle matched with the target test scene as the target background vehicle.

[0024] Optionally, in the driving safety simulation test method, the inherent scene variables are at least one or more combinations of the following target parameters:

[0025] The target parameters comprise:

[0026] At the moment when the target background vehicle starts to change lanes, the time absolute value at the initial moment, the longitudinal speed of the target background vehicle, the longitudinal distance of the target background vehicle relative to the front vehicle, and the longitudinal speed of the front vehicle of the target background vehicle;

[0027] At the end of the simulation, the time absolute value at the initial moment, the speed change amount of the front vehicle of the target background vehicle, and the acceleration of the front vehicle of the target background vehicle;

[0028] When the target test scene is a lane-changing scene, the behavior-related variables comprise a following tolerance, a lane-changing courtesy and an expected driving state.

[0029] Optionally, in the driving safety simulation test method, the target background vehicle is controlled by using the humanized driving model, and the method comprises the following steps.

[0030] obtaining a behavior-related variable and a value range of the behavior-related variable;

[0031] constructing a behavior-related variable test matrix based on the behavior-related variable and the value range of the behavior-related variable;

[0032] configuring the anthropomorphic driving model based on the behavior-related variable test matrix;

[0033] obtaining an inherent scene variable and a value range of the inherent scene variable;

[0034] constructing an inherent test matrix based on the inherent scene variable and the value range of the inherent scene variable;

[0035] controlling the target background vehicle based on the inherent test matrix by using the configured anthropomorphic driving model.

[0036] Optionally, before the anthropomorphic driving model is used to control the target background vehicle in the driving vehicle safety simulation test method, the method further includes:

[0037] preheating each vehicle in the test scene, so that the measured vehicle and the target background vehicle in the test scene reach a target state, and the target state includes that the vehicle speed of the measured vehicle and the target background vehicle reaches a set vehicle speed, and the vehicle distance of the measured vehicle and the target background vehicle reaches a set vehicle distance.

[0038] Optionally, in the driving vehicle safety simulation test method, when the test matrix is constructed, the continuous Latin hypercube sampling method is used to construct the test matrix.

[0039] A driving vehicle safety test device includes:

[0040] a model obtaining unit configured to obtain an anthropomorphic driving model;

[0041] a target background vehicle control unit configured to control a target background vehicle by using the anthropomorphic driving model;

[0042] a simulation test unit configured to perform a driving vehicle safety test on a measured vehicle by using the target background vehicle as a background vehicle.

[0043] An electronic device includes at least one processor and a memory connected to the processor, wherein:

[0044] the memory is configured to store a computer program;

[0045] the processor is configured to execute the computer program, so that the electronic device can implement the driving vehicle safety simulation test method described in any one of the above.

[0046] Based on the technical solution, the method provided by the embodiment of the application can control the interaction between the target background vehicle and the vehicle under test and the randomness of the driving decision of the target background vehicle, so as to more effectively trigger potential risk events and better test the vehicle under test. BRIEF DESCRIPTION OF DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the application or the prior art, the drawings needed to be used in the embodiments or the prior art description will be briefly introduced. Obviously, the drawings in the following description are only embodiments of the application, and other drawings can be obtained by those skilled in the art without creative labor on the basis of the provided drawings.

[0048] Figure 1 The flowchart of the driving car safety simulation test method disclosed by the embodiment of the application is shown in the figure.

[0049] Figure 2 The inherent scene diagram for driving car safety test is shown in the figure.

[0050] Figure 3 The interface diagram of the simulation software and the humanized driving model for driving car safety test is shown in the figure.

[0051] Figure 4 The lane changing scene diagram under the inherent scene is shown in the figure.

[0052] Figure 5 The lane changing mechanism diagram of the target background vehicle under the behavior-related lane changing scene is shown in the figure.

[0053] Figure 6 The Latin hypercube sampling diagram is shown in the figure.

[0054] Figure 7 The lane changing scene design diagram with preheating scheme is shown in the figure.

[0055] Figure 8 The driving trajectory cluster of the target background vehicle in the lane changing process is shown in the figure.

[0056] Figure 9 The event extraction diagram of the inherent scene collision under the condition that p=-1.5 and the action time of the following vehicle tolerance mechanism is 3.2s is shown in the figure.

[0057] Figure 10 The event extraction diagram of the behavior-related scene without collision under the condition that p=-1.5 and the action time of the following vehicle tolerance mechanism is 8s is shown in the figure.

[0058] Figure 11 Event extraction graph of inherent scene non-collision under the condition that p=-1.5 and the action time of the car-following patience mechanism is 4.8s;

[0059] Figure 12 Event extraction graph of behavior-related scene collision under the condition that p=-1.5 and the action time of the car-following patience mechanism is 6.9s;

[0060] Figure 13 Event extraction graph of inherent scene collision under the condition that p=-1.5 and the action time of the car-following patience mechanism is 4.6s;

[0061] Figure 14 Event extraction graph of behavior-related scene collision under the condition that p=-1.5 and the action time of the car-following patience mechanism is 4.9s;

[0062] Figure 15 Structure schematic diagram of the driving car safety test device disclosed by the embodiment of the application;

[0063] Figure 16 Structure schematic diagram of the electronic device disclosed by the embodiment of the application. DETAILED DESCRIPTION

[0064] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative work fall within the protection scope of the application.

[0065] The application provides a driving car safety simulation test method with humanized behavior correlation characteristics. The method aims to solve the limitations of automatic driving vehicle safety testing, especially the differences between existing inherent scene testing and actual driving environment. The method uses a humanized driving model of background vehicles to realize the humanized behavior correlation behavior of background vehicles by augmenting behavior correlation variables in inherent scene variables. In the lane changing environment, the background vehicles that have an impact on the tested vehicle in the inherent scene variables are added with behavior correlation variables such as "following tolerance", "lane changing courtesy" and "expected driving state". Following tolerance refers to the tolerance or psychological endurance of the driver when following the front vehicle in terms of the front vehicle's driving speed, acceleration, deceleration and other behaviors. In traffic, following driving is a common driving situation, and each driver's response and acceptance of the front vehicle's driving behavior may be different. Maintaining appropriate following tolerance is crucial for driving safety. If the driver's following tolerance is too low, it may lead to frequent overtaking, sudden braking and other behaviors. Lane changing courtesy mainly describes the degree of courtesy and attitude of the driver during lane changing. This involves the driver's understanding of traffic rules, driving skills and the importance of road safety. The higher the lane changing courtesy, the lower the probability of traffic accidents and the higher the road traffic efficiency. The expected driving state refers to the driver's expectation or desired state of the vehicle's driving state in a specific traffic environment. In the test part, the adaptive cruise control (ACC) module of the tested vehicle is tested, and the background vehicle lane changing scene test on the virtual simulation platform tests the performance of the tested vehicle. This method can accurately simulate the interactivity between the background vehicle and the tested vehicle and the randomness of the background vehicle behavior, thereby more effectively triggering potential risk events and better testing the tested vehicle.

[0066] Referring to Figure 1 The driving car safety simulation test method disclosed in the embodiments of the application can include:

[0067] Step S101: Obtain a humanized driving model.

[0068] The humanized driving model refers to a technology or method that integrates human driving behavior and decision-making process into a vehicle driving model. This model simulates human perception, decision-making and execution process, enabling the vehicle to make driving decisions similar to humans in various traffic scenarios, so that the vehicle's driving state is more in line with actual driving behavior. In addition, the humanized driving model can also consider factors such as human driver's emotions, fatigue level, attention distraction, etc. to more comprehensively simulate human driving behavior. This helps the vehicle better adapt to and integrate into complex traffic environments during autonomous driving, improving driving comfort and acceptability.

[0069] In this application, the anthropomorphic driving model can be pre-configured based on behavioral correlation variables to obtain an anthropomorphic driving model for controlling a background vehicle. Alternatively, the behavioral correlation variables can be used as input data for the anthropomorphic driving model during simulation testing. In this case, the specific configuration of the behavioral correlation variables can be adjusted arbitrarily according to testing requirements during the simulation testing process. The behavioral correlation variables can be user attribute data related to the user's driving decisions, such as the user's tolerance, courtesy, and desired driving state.

[0070] After acquiring the behavioral correlation variables and the inherent scenario variables during the testing process, the anthropomorphic driving model automatically generates corresponding driving strategies. These driving strategies can include decisions such as lane changing, acceleration, deceleration, overtaking, and parking. The inherent scenario variables can include vehicle status data that can influence the user's driving decisions, such as the vehicle's speed, the speeds of surrounding vehicles, the distance between the vehicle and surrounding vehicles, and the type of lane the vehicle is in.

[0071] In the technical solution disclosed in this embodiment, different test scenarios are required for different test needs when testing the vehicle under test. These test scenarios may include highway scenarios, urban road scenarios, mountain road scenarios, congestion scenarios, lane-changing scenarios, overtaking scenarios, etc. In this application, the required test scenario can be referred to as the target test scenario. Since users' mindsets differ in different test scenarios, the types and value ranges of behavioral correlation variables affecting user driving behavior will differ depending on the target test scenario. Therefore, after obtaining the target test scenario, the behavioral correlation variables matching the target test scenario and their value ranges can be obtained. Of course, the target background vehicle selected can also be different in different test scenarios. For example, in a lane-changing scenario, the vehicle in front of the test vehicle can be used as the target background vehicle; in a congestion scenario, all vehicles around the test vehicle can be used as the target background vehicle. Therefore, after determining the target test scenario, it is also necessary to select a background vehicle matching the target test scenario as the target background vehicle.

[0072] Step S102: Use an anthropomorphic driving model to control the target background vehicle.

[0073] In this step, the vehicle controlled by the anthropomorphic driving model in the background vehicle is denoted as the target background vehicle. The anthropomorphic driving model will acquire the inherent scene variables of other vehicles that affect the driving decisions of the target background vehicle, as well as the aforementioned behavioral correlation variables, and generate corresponding driving decisions. Based on the driving decisions, the driving state of the target background vehicle is controlled. At this time, the driving behavior of the target background vehicle is the anthropomorphic driving behavior.

[0074] In this embodiment, the target background vehicle controlled by the anthropomorphic driving model can be a single background vehicle or several background vehicles in the test scenario. Users can select and configure the target background vehicles controlled by the anthropomorphic driving model, as well as the behavioral correlation variables matched by the anthropomorphic driving model for each target background vehicle, based on testing requirements. This allows each target background vehicle in the test scenario to correspond to a virtual person with an independent personality, making the driving behavior of each target background vehicle more closely resemble the real-world scenario. For example, the target background vehicle controlled by the anthropomorphic driving model can be a background vehicle located in front of and closest to the vehicle under test, or each background vehicle can be used as a target background vehicle.

[0075] Step S103: Conduct a driving safety test based on the test vehicle and the background vehicle.

[0076] The test can be conducted using a conventional test procedure, except that the anthropomorphic driving model is used to control the target background vehicle during the test.

[0077] Figure 2 For example, a lane-changing scene with background vehicles. Figure 2 As shown, an anthropomorphic driving model is used to control the target background vehicle. The driver's psychological motivations during lane changing, such as "tolerance for following other vehicles," "caution during lane changing," and "desired driving state," are used as behavioral correlation variables. Based on the inherent scenario variables collected during the test and the correlation variables, the anthropomorphic driving model determines the lane changing decision that the target background vehicle should make. During the lane changing process of the target background vehicle, the test vehicle is in a straight-going state. At this time, the lane changing behavior of the background vehicle is a random event relative to the test vehicle, which can more effectively verify the safety of the control strategy of the test vehicle under this situation.

[0078] As can be seen from the above scheme, at the start of the test, an anthropomorphic driving model is obtained, and the target background vehicle is controlled using the anthropomorphic driving model. Based on the target background vehicle, the adaptive cruise control (ACC) module of the vehicle under test is tested. This method can anthropomorphically control the interaction between the target background vehicle and the vehicle under test, as well as the randomness of the driving decisions of the target background vehicle, thereby more effectively triggering potential risk events and conducting better testing of the vehicle under test.

[0079] In this application, the scheme of using an anthropomorphic driving model to control the background vehicle is based on the inherent scenario of the existing testing process. To ensure the smooth operation of the anthropomorphic driving model in the simulation software, the signal input / output interface between the anthropomorphic driving model and the software must be defined. See [link to relevant documentation]. Figure 3 The interface needs to include simulation time (t), modeling vehicle information (i.e., information about the target background vehicle from which the anthropomorphic driving model is to be built), and environment vehicle information (here, environment vehicle refers to the environment vehicle relative to the modeling vehicle information). Specifically, simulation time (t) represents the current simulation moment; modeling vehicle information includes coordinates... speed acceleration and direction (heading, yaw rate) Environmental vehicle information describes the coordinates of the environmental vehicles in the simulation. speed acceleration Direction (heading, yaw rate) Relative distance to this vehicle (modeled vehicle) Relative speed difference with this vehicle The anthropomorphic driving model uses the acquired modeled vehicle information and environmental vehicle information as vehicle condition data to predict the next driving decision of the modeled vehicle controlled by the anthropomorphic driving model. The prediction result is the modeled vehicle information at the next moment. The modeled vehicle information at the next moment is sent to the simulation software. The simulation software then updates the environmental vehicle information based on the acquired modeled vehicle information at the next moment. This forms a control loop, realizing the anthropomorphic control of the modeled vehicle.

[0080] The testing process disclosed in the above embodiments of this application is conducted in existing inherent testing scenarios. Inherent testing scenarios refer to the testing scenarios of existing simulation testing schemes. The difference between this application and existing schemes is that this application requires one or more background vehicles in the testing environment that affect the driving state of the vehicle under test as control vehicles of an anthropomorphic driving model, making the actions of these background vehicles more consistent with human driving habits. For example, the lane-changing / cut-in scenario in the inherent testing scenario mainly consists of a background vehicle (PV), a lead vehicle (LV), and the vehicle under test (AV). During the testing process, the driving state of the background vehicle (PV) is mainly changed to perform corresponding tests on the vehicle under test (AV). Its main testing scenarios are as follows: Figure 4 The scene shown.

[0081] In existing, inherent testing scenarios, Figure 4 The key time points in the inherent test scenario shown can be divided into three stages: t0 (initial time), t1 (time when the background vehicle begins lane changing), and t2 (time when the simulation ends). The scenario variables include the following: at the initial time t0, the main test variables are the AV speed, PV speed, and the relative distance between AV and PV, denoted by v0. AV v0 PV and Δx0 PV At the moment t1 when the PV begins lane changing, the main test variables are the absolute value of time t1 and the PV speed v1. PV and PV acceleration a1 PV At the end of the simulation at time t2, the main test variables are the absolute value of time t2 and the PV velocity v2. PV and PV acceleration a2 PV .

[0082] The matrix design method in inherent scenario testing is as follows:

[0083] Based on the characteristics of the development model and the analysis of testing requirements, the testing function primarily focuses on the adaptive cruise control (ACC) performance of the vehicle under test. By simulating more test scenarios, the impact of different scenarios, methods, and factors on the ACC function of the vehicle under test is examined. In the inherent test scenarios, the vehicle state in the simulated scenarios is configured using an inherent test matrix. The inherent test matrix contains inherent scenario variables that include at least one or more of the following target parameters: the absolute value at time t1, the absolute value at time t2, and the longitudinal speed of the vehicle at time t1. Longitudinal distance between the preceding vehicle and the current vehicle at time t1 longitudinal velocity of the vehicle in front at time t1 Change in speed of the vehicle ahead at time t2 and the acceleration of the vehicle in front The aforementioned variables form a matrix for inherent scenario testing. By providing a certain range of values ​​for each element in the inherent test matrix and sampling initial parameters within the corresponding range, the inherent scenario testing function is completed. In a specific example provided in this application, the variables in the specific inherent test scenario are shown in Table 1:

[0084]

[0085] Table 1

[0086] The test scenarios disclosed in this application are behavior association test scenarios based on inherent scenarios. The following is a scenario introduction using the lane-changing behavior association test scenario as an example:

[0087] The lane-changing / cut-in scenario test under behavioral correlation conditions mainly designed a behavioral correlation test scenario incorporating anthropomorphic factors. The anthropomorphic driving model achieves anthropomorphic vehicle control by combining inherent scenario variables and behavioral correlation variables. The vehicles in this scenario mainly consist of a target background vehicle (PV), a lead vehicle (LV), and the test vehicle (AV), where the target background vehicle (PV) is controlled by the anthropomorphic driving model. In this embodiment, the specific type of the anthropomorphic driving model can be set according to user needs. It can be generated from a large model. This model controls the vehicle state of the controlled vehicle based on quantified anthropomorphic factors, the state information of the controlled vehicle, and the state information of surrounding vehicles. For example, in this embodiment, the anthropomorphic factors that the anthropomorphic driving model can identify may include following tolerance, lane-changing courtesy, and desired driving state. Here, the quantified anthropomorphic factors are used as the behavioral correlation variables. A schematic diagram of the control mechanism of the target background vehicle (AV) is shown below. Figure 5 As shown, the target background vehicle HV is controlled by an anthropomorphic driving model, supplemented by a virtual lead vehicle LV. By controlling the target background vehicle HV through the anthropomorphic driving model, anthropomorphic control of the target background vehicle HV is achieved, thus making the test scenario more human-like. Figure 5 The test scenario shown primarily tests the vehicle under test (AV) by altering the relevant states of the initial scenario to stimulate changes in the driving decisions of the target background vehicle (HV), incorporating certain anthropomorphic elements to test the AV. The main test scenario includes three key moments: t0 (initial moment), t1 (moment when the background vehicle begins lane changing), and t2 (moment when the simulation ends).

[0088] In the technical solution disclosed in this embodiment, the anthropomorphic driving model can be a model configured based on the behavioral correlation variables. In this case, when the anthropomorphic driving model controls the target background vehicle, it can directly use the variables in the inherent test scenario as input data. Of course, to facilitate testers in freely adjusting and configuring the behavioral correlation variables according to their own needs, the behavioral correlation variables can also be used as input data for the anthropomorphic driving model. In this case, when the anthropomorphic driving model controls the target background vehicle, it needs to use both the behavioral correlation variables and the variables in the inherent test scenario as input data. When testing the vehicle under test in this scenario, the main approach is to change the relevant states of the background vehicles in the initial scenario (which can refer to other background vehicles besides those controlled by the anthropomorphic driving model) to stimulate changes in the relevant driving decisions of the target background vehicle (HV) and to exhibit certain anthropomorphic factors, thereby conducting anthropomorphic scenario testing on the vehicle under test (AV). Its main test scenario also includes three main moments: the initial moment, the moment when the preceding vehicle changes, and the simulation end moment.

[0089] During the testing process, in addition to designing a test matrix for inherent scenarios, a test matrix for behavior-related scenarios also needs to be designed. The inherent scenario test matrix stores inherent scenario variables, and the behavior-related scenario test matrix stores behavior-related variables. During behavior-related scenario testing, the inherent scenario variables and behavior-related variables are combined to generate an augmented matrix of behavior-related variables to complete the test for that scenario. The inherent scenario variables are consistent with Table 1. For example, in this embodiment, during lane-changing scenario testing, the behavior-related variables of the background vehicle lane-changing mainly include three anthropomorphic factors (behavior-related variables): following tolerance, lane-changing courtesy, and desired driving state. Tolerance includes following tolerance α. p Lane changing courtesy includes the courtesy value P and the safety threshold Δa. th The desired driving state includes the desired speed v des This study investigates lane-changing needs of different personality types in the same scenario by adjusting following tolerance, yielding values, safety threshold values, and desired speed. For example, in a given scenario, people with high tolerance values ​​will not make lane-changing decisions, while those with low following tolerance values ​​will. Similarly, people with high yielding values ​​will not make lane-changing decisions, while those with low yielding values ​​will. This embodiment adds behavioral correlation variables such as tolerance, yielding values, and desired driving state to the existing testing scheme, forming a behavioral correlation test matrix for lane-changing scenarios. Table 2 shows the table of behavioral correlation variables for lane-changing scenarios.

[0090]

[0091] When testing a vehicle under test using a predefined scenario, a predefined scenario test matrix needs to be constructed based on predefined scenario variables and their value ranges. The specific values ​​of each predefined scenario variable are then extracted from this test matrix. Based on these extracted values, each virtual vehicle in the test scenario is configured, and the vehicle under test is then tested. In the technical solution disclosed in this embodiment, when constructing the test matrix for controlling the target background vehicle, in addition to considering predefined scenario variables, behavioral correlation variables also need to be considered. That is, first, the predefined scenario variables and their value ranges are obtained; then, behavioral correlation variables and their value ranges are obtained; then, a test matrix is ​​constructed based on the behavioral correlation variables, predefined scenario variables, and their value ranges; finally, the anthropomorphic driving model is used to control the target background vehicle based on the test matrix. Of course, when constructing the test matrix, a humanoid driving model can be configured based on behavioral correlation variables. In this case, the process of controlling the target background vehicle can be as follows: first, obtain the behavioral correlation variables and their value ranges, and construct a behavioral correlation variable test matrix based on the behavioral correlation variables and their value ranges; then, configure the humanoid driving model based on the behavioral correlation variable test matrix; next, obtain the inherent scene variables and their value ranges, and construct an inherent test matrix based on the inherent scene variables and their value ranges; finally, use the configured humanoid driving model to control the target background vehicle based on the inherent test matrix.

[0092] In this embodiment, taking the construction of a test matrix based on inherent scene variables and behavior-related variables as an example, the specific process of constructing the test matrix includes: sampling the high-dimensional test space, generating a test matrix of inherent scene and behavior-related scene based on the sampling results, and designing a test matrix generation method based on Latin hypercube. The test matrix can be generated by continuous Latin hypercube sampling method, and each variable in the test scene is configured based on the test matrix.

[0093] The test matrix generation method based on Latin hypercube is as follows:

[0094] Suppose an uncertainty problem contains M random variables X1, X2, X3…X4. M-1 X M M is a positive integer greater than 1, and the variables are independent of each other. When conducting safety tests on the vehicle under test, multiple independent random variables are also required, such as the driving states of the background vehicle at multiple moments in the inherent test scenario, and following tolerance, lane-changing courtesy, and desired driving states in the behavior-related test scenario. During the test, all the above variables need to be sampled and analyzed N times, where N is a positive integer greater than 1, X...m Let X be any random variable from M random variables. m The probability distribution function is: F m (X m ) = P(x < X m ).

[0095] Latin hypercube sampling (LHS) is a commonly used random sampling method that can minimize the number of sampling points while ensuring a uniform distribution of sample points, thereby improving computational efficiency. The main steps of Latin hypercube sampling are as follows:

[0096] (1) Divide the vertical axis of the probability distribution function of each variable into N equal intervals, with the width of each interval being 1 / N;

[0097] (2) Generate an M*N dimensional matrix B M×N Each row of the matrix is ​​a random sequence of integers 1 and N. mn For B M×N The element in the m-th row and n-th column of the matrix, b mn It can represent the probability distribution value corresponding to the nth sample value of the mth variable;

[0098] (3) Generate an M*N dimensional random number matrix C M×N element c mn Let c represent the nth random value of the mth random variable. From (2), we know that c mn satisfy In the formula, m = 1, 2, ..., M, and n = 1, 2, ..., N.

[0099] Element CMN is present The above is randomly generated according to a uniform distribution. If x mn Let F represent the nth sample value of the mth variable. Then, the following relationship exists: F m (x mn ) = C mn .

[0100] The random number matrix C generated from step (3) of the Latin hypercube sampling M×N By using the probability distribution function to solve in reverse, the sampling matrix X can be obtained. M×N Each row of this matrix represents all sampled values ​​of a certain state variable, and each column constitutes a sample point in a subspace of the equally divided sample space of the state variable. Therefore, all sampled samples of the state variable can cover the entire sample space, making the method more efficient and stable. A sampling diagram is shown below. Figure 6 As shown, in Figure 6In the example shown, the uncertainty problem involves two random variables, X1 and X2. Latin hypercube sampling is used to collect samples, and random sampling within each interval forms the sample points. Figure 6 (The black solid dots in the middle).

[0101] This application can generate a test matrix for inherent test scenarios and a test matrix for behavioral correlation variables based on the Latin hypercube sampling process described above. In this process, the test variables for the two scenarios are first stratified, and then the variable stratification sampling and sample value combination are performed to generate the corresponding test matrix.

[0102] Latin hypercube sampling can be further divided into discrete sampling and continuous sampling. Discrete Latin hypercube sampling requires manual input of test variable values ​​in the program implementation, which cannot cover all test scenarios and has certain limitations. Continuous Latin hypercube sampling, on the other hand, only requires determining the range of test variables to automatically generate the test matrix, saving testing costs. Furthermore, uniform sampling within the value range of each test variable ensures comprehensive testing. Therefore, this test primarily utilizes continuous Latin hypercube sampling to generate the test matrix.

[0103] That is, in step S102, when using the anthropomorphic driving model to control the background vehicle, a test matrix needs to be constructed first using the continuous Latin hypercube sampling method based on the inherent scene variables, behavioral correlation variables, and the value range of each variable. Then, the anthropomorphic driving model is used to control the target background vehicle based on this test matrix. In this scheme, the test matrices corresponding to different target background vehicles may be different. Therefore, when using the anthropomorphic driving model to control based on its corresponding test matrix, the driving states of different target background vehicles may be different, thus making the test scenario more closely resemble the actual driving environment.

[0104] Furthermore, to better test the vehicle under test, this application can also construct parameter indicators for the adaptive cruise control function of the vehicle under test. In this embodiment, when setting the parameter indicators, tracking, safety (collision rate), and comfort need to be set as core indicators.

[0105] The following should be considered when designing and tracking performance parameters:

[0106] ① Vehicle speed stability: The control system of the vehicle under test should be able to maintain the stability of the vehicle speed and avoid excessive or sudden speed changes;

[0107] ② Accuracy of forward vehicle recognition: The control system of the tested vehicle should be able to accurately identify the position and speed of the vehicle in front so as to automatically adjust the distance and speed.

[0108] ③ Vehicle acceleration: The control system of the vehicle under test should be able to accelerate and decelerate smoothly, avoiding unnecessary bumps and swaying.

[0109] ④ Distance control of the vehicle in front: The control system of the vehicle under test should be able to automatically control the distance between the vehicle and the vehicle in front to avoid getting too close or too far apart.

[0110] ⑤ Human-machine interaction: The control system of the vehicle under test should have good human-machine interaction, and the driver should be able to easily set, adjust and turn off the system to meet different driving needs.

[0111] The specific indicators and limitations of these parameters can be adjusted and optimized according to the specific design and application scenario. For example, on highways, vehicle speed stability and the accuracy of forward vehicle recognition may be more important; while on urban roads, human-machine interaction may be more critical. Tracking performance is evaluated based on the relative speed deviation between the actual and expected speeds, and quantitative indicators for tracking performance are designed. The physical meaning of the tracking performance indicator is that the cumulative value of the relative speed deviation reflects transient tracking performance. Within a simulation cycle, the larger the tracking indicator value, the worse the controller's tracking performance.

[0112] The vehicle uses sensors such as radar or cameras to monitor the driving status of vehicles ahead, automatically adjusting its speed and distance to maintain a safe following distance. To ensure the safety performance of the vehicle under test, the following factors need to be considered during the design process:

[0113] ① Maximum following speed difference: This is the maximum speed difference between the tested vehicle and the vehicle in front. To avoid rear-end collisions, the maximum following speed difference should be kept within a safe range, generally recommended not to exceed 30 km / h.

[0114] ② Maximum Deceleration: This refers to the maximum rate of deceleration that the tested vehicle can achieve. If the vehicle in front brakes suddenly, the tested vehicle needs to decelerate rapidly to avoid a rear-end collision. Therefore, the maximum deceleration needs to be sufficiently large, generally recommended to be no less than 3 m / s². 2 .

[0115] ③ Distance Control Range: This refers to the minimum safe distance between the tested vehicle and the vehicle in front. To ensure driving safety, the following distance needs to be adjusted according to the driving status of the vehicle in front. Generally, it is recommended to keep the distance control range between 2 and 5 seconds.

[0116] ④ Speed ​​Control Range: This refers to the range within which the vehicle's speed can be controlled. To avoid speeding and driving at low speeds, the speed control range needs to be adjusted according to factors such as road type and speed limits. Generally, the speed control range should be kept between 30 and 180 km / h.

[0117] When designing safety performance parameters, safety performance can be evaluated based on both Time-To-Collision (TTC) and Time-to-Headway (THW), thus creating quantitative safety indicators. The specific physical meaning of these safety performance indicators is to record the average number of safety hazards occurring within a simulation cycle. The higher the number of hazards, the lower the safety level.

[0118] To verify the safety of the tested vehicle and evaluate its safety based on the test results, this application also constructs a quantitative evaluation system for the safety of the tested vehicle:

[0119] ① Tracking performance quantification index design:

[0120] Tracking performance is evaluated based on the relative velocity deviation between the actual and expected values, and a quantitative index for tracking performance is designed accordingly. The physical meaning of the tracking performance index is that the cumulative value of the relative velocity deviation reflects the transient tracking performance. Within a simulation cycle, a larger value of the tracking performance index ltp indicates a worse tracking effect of the controller. The formula for the tracking performance index is as follows:

[0121]

[0122] V erro =v AV -v PV

[0123] Among them, V erro The relative velocity error is given by T, where T is the total simulation time of the test process, and v is the velocity error. AV v represents the speed of the vehicle being measured. PV The speed of the target background vehicle.

[0124] ② Design of safety quantifiable indicators:

[0125] Safety performance is evaluated based on both Time to Collision (TTC) and Time to Headway (THW), and this is used to design quantitative safety indicators.

[0126] The specific physical meaning of the safety performance index is to record the average number of safety hazards that occur within a simulation cycle. The higher the number of hazards, the lower the safety. The formula for the safety performance index is as follows:

[0127]

[0128]

[0129]

[0130] Where T is the total simulation time, l ttc For TTC indicators, l thw The THW (Thickness, Weight, and Width) indicators are all determined by threshold values. TTC (Total Cost safe For a safe TTC threshold, THW safe The safe THW threshold.

[0131] ③ Design of quantitative indicators for comfort performance:

[0132] Comfort performance is evaluated based on both acceleration and the rate of change of acceleration, and a quantitative index for comfort performance is designed accordingly.

[0133] The physical meaning of the comfort performance index is the average value of acceleration and its rate of change over a simulation cycle. The larger the value, the worse the comfort performance. The comfort performance formula is designed as follows:

[0134]

[0135]

[0136] Where a is acceleration, j is the rate of change of acceleration, and T is the sampling time. s The unit is the sampling time.

[0137] After the test is completed, the tracking performance index ltp, safety performance index ls, and comfort performance index lc of the tested vehicle can be calculated based on the tracking performance index formula, safety performance index formula, and comfort performance index formula, respectively. Then, the control system of the tested vehicle is optimized and improved based on the tracking performance index ltp, safety performance index ls, and comfort performance index lc.

[0138] In the technical solution disclosed in this embodiment, during the Simulink simulation environment test, each virtual vehicle in the test scenario can immediately reach the set speed at the initial stage. However, in actual road conditions, vehicles do not suddenly appear, and the initial speed does not suddenly reach the set speed. Therefore, in order to further restore the real-world scenario, this application adds a warm-up scenario before the formal simulation test of the vehicle under test begins. In the warm-up scenario, the vehicle under test is able to reach the set speed at the start of the simulation test. The background vehicles and target background vehicles in the test scenario then generate action changes (inherent matrix test variables), and anthropomorphic factors (behavioral correlation variables) are incorporated into the control process of the target background vehicle. That is, before using the anthropomorphic driving model to control the target background vehicle, the application further includes: warming up the driving state of each vehicle in the test scenario (including the vehicle under test, the target background vehicle, and other background vehicles) so that the vehicle under test and the target background vehicle in the test scenario reach the target state. The target state includes: the speed of the vehicle under test and the target background vehicle reaches the set speed, and the distance between the vehicle under test and the target background vehicle reaches the set distance.

[0139] See Figure 7 In the lane-changing test scenario shown, the test scheme after introducing the warm-up scenario is as follows: t0 is still the initial time, t1 is the time when the target background vehicle PV is traveling, t2 is the time when the two vehicles are stable, which is the time when the warm-up scenario ends, t3 is the time when the target background vehicle PV changes its action (lane-changing time), and t4 is the time when the simulation ends.

[0140] During the warm-up period, from t0 to t2, the main purpose is to ensure that the speeds of both the tested vehicle and the target background vehicle reach the set speeds and the set relative distance. Once the tested vehicle (AV) reaches the set speed and stably follows the target vehicle, it sends a trigger signal, indicating that the test can begin, and the test process is initiated at time t1. The test process, from t2 to t4, occurs at time t2 when the anthropomorphic factors of the target background vehicle (PV) are activated, and the PV's actions change (e.g., ...). Figure 7 The lane-changing maneuver shown completes the overall testing process for the vehicle under test. The introduction of a preheating process allows each AV system of the vehicle under test to reach the set speed after a period of stabilization, which is more conducive to bench testing.

[0141] In the technical solutions disclosed in this application, behavioral correlation variables may include: following tolerance, lane-changing courtesy, and desired driving state. To verify the effectiveness of this solution, the following simulation tests were also conducted, and the test results were compared:

[0142] Regarding tolerance for following other vehicles:

[0143] Among them, following vehicle tolerance α pThe calculation formula is:

[0144] Among them, v c Let T1 be the speed of the target background vehicle when it changes lanes, and v be the speed of the vehicle when it changes lanes. In this formula, T1 is the duration of the following tolerance mechanism. des The desired speed of the target background vehicle (the stable speed 5 seconds after the lane change ends).

[0145] When the lane-changing scenario is automatically triggered under the inherent test scenario, a trigger signal is fed back when the distance between the two vehicles reaches the set relative distance, and the test begins. Under this condition, the background vehicle begins to change lanes. The inherent scenario matrix is ​​set as shown in the table below under the above test scenario:

[0146] Inherent scene matrix parameters

[0147]

[0148] Based on the existing scenario matrix, a following tolerance mechanism is introduced to delay the lane-changing start time of background vehicles, thereby increasing the collision rate of the tested vehicle. The behavior correlation test matrix is ​​designed as follows:

[0149] Behavior-related scene matrix parameters

[0150]

[0151] The applicant conducted 1000 trials in the inherent scenario matrix test and 1000*10 trials in the behavioral association scenario. All variables involved in these trials were uniformly sampled from the intervals of each variable within the given matrix dimensions to ensure the generality of the experiments. Based on the above matrix settings, the collision rates for the two test scenarios are shown in the table below:

[0152] Collision rate results of inherent and behaviorally related test scenarios

[0153]

[0154] The collision rate was 4% in the inherent scenario and 27.6% in the behavior-related test scenario. In the behavior-related scenario, the probability of collision events was significantly increased due to the influence of anthropomorphic factors on the tolerance of the target background vehicle.

[0155] Regarding lane-changing courtesy:

[0156] In this embodiment, the lane-changing courtesy level is divided into three courtesy coefficient ranges: rude, normal, and conservative. Drivers within the same courtesy coefficient range are considered to have the same driver characteristic, as shown below:

[0157] Courtesy coefficient range division

[0158]

[0159] Finally, all trajectories for each courtesy coefficient range are obtained, along with the trajectory features corresponding to each trajectory, to generate a cluster of driver trajectories related to the courtesy coefficient P, such as... Figure 8 As shown.

[0160] In the above test scenario, the inherent scene matrix is ​​set as follows:

[0161] Inherent scene matrix parameters

[0162]

[0163] Based on the inherent scenario matrix dimensions, design a behavior correlation test matrix:

[0164] Behavior-related scene matrix parameters

[0165]

[0166] Since the tolerance range for following other vehicles is difficult to determine, we take the inherent value of 0, and set the courtesy coefficient P to three different intervals: [-2, -0.5], [-0.5, 0.5], and [0.5, 2], and take three fixed values: -1.5, 0, and 1.5.

[0167] In the inherent scenario matrix test, a total of 1000 trials were conducted, and in the behavioral association scenario test, a total of 1000*10 trials were conducted. All variables involved in these trials were uniformly sampled from the intervals of each variable within the given matrix dimensions to ensure the generality of the trials. The results for the two test scenarios, based on the above matrix settings, are shown in the table below.

[0168] Collision rate results of inherent and behaviorally related test scenarios

[0169]

[0170] The collision rate in the inherent scenario was 33%. In the behavioral correlation test scenario: when P = -1.5, the collision rate was 56.88% and the lane-changing rate was 100%; when P = 0, the collision rate was 50.24% and the lane-changing rate was 98%; when P = 1.5, the collision rate was 7.32% and the lane-changing rate was 16%. The smaller the yield coefficient, the higher the lane-changing rate and the higher the collision rate. Under the condition of P = -1.5, the events of collisions in the inherent scenario and non-collisions in the behavioral correlation variable scenario are extracted as follows: Figure 9 and Figure 10 As shown.

[0171] Analysis of the causes reveals that the short lane-changing time led to a collision in the vehicle under the given scenario.

[0172] Under the condition of P=-1.5, the inherent scene has no collisions, and the event extraction of scene collisions for behavior-related variables is as follows: Figure 11 As shown in Figure 12.

[0173] Under the condition of P=-1.5, the event extraction of inherent scene collision and behavior-related variable scene collision, such as... Figure 13 , Figure 14 As shown.

[0174] As can be seen from the above schemes, under the same initial scenario, the lane change time has a relatively significant impact on the collision.

[0175] As can be seen from the above testing processes, the technical solution disclosed in this embodiment can simulate actual driving scenarios, and the simulation results are more reliable.

[0176] This embodiment discloses a vehicle safety testing device. For the specific working content of each unit in the device, please refer to the above method embodiment.

[0177] The following describes the driving safety simulation test device provided in the embodiments of the present invention. The driving safety test device described below and the driving safety simulation test method described above can be referred to in correspondence.

[0178] See Figure 15 The vehicle safety testing device may include:

[0179] The model acquisition unit 10, which corresponds to step S101 in the above method, is used to acquire an anthropomorphic driving model;

[0180] The target background vehicle control unit 20, which corresponds to step S102 in the above method, is used to control the target background vehicle using an anthropomorphic driving model;

[0181] The simulation test unit 30, which corresponds to step S103 in the above method, is used to conduct driving safety tests on the test vehicle using the target background vehicle as the background vehicle.

[0182] Corresponding to the above method, there is a one-to-one correspondence between the anthropomorphic driving model and the target background vehicle, and there is at least one anthropomorphic driving model and a target background vehicle. In this case, the target background vehicle control unit 20 is specifically used to acquire the control of the target background vehicle corresponding to it using the anthropomorphic driving model.

[0183] Corresponding to the above method, in this embodiment, the target background vehicle control unit 20 is specifically used to: obtain inherent scene variables and the value range of the inherent scene variables; obtain behavioral correlation variables and the value range of the correlation variables; construct a test matrix based on the behavioral correlation variables, inherent scene variables and the value range of each variable; and use the anthropomorphic driving model to control the target background vehicle based on the test matrix.

[0184] Corresponding to the above method, in this embodiment, when the target background vehicle control unit 20 acquires behavioral correlation variables, it is specifically used to: acquire the target test scenario; acquire behavioral correlation variables that match the target test scenario; and before using the anthropomorphic driving model to control the target background vehicle, it further includes: selecting a background vehicle that matches the target test scenario as the target background vehicle.

[0185] Corresponding to the above method, in this embodiment, when the target background vehicle control unit 20 controls the target background vehicle using a humanoid driving model, it specifically performs the following steps: obtaining behavioral correlation variables and their value ranges; constructing a behavioral correlation variable test matrix based on the behavioral correlation variables and their value ranges; configuring the humanoid driving model based on the behavioral correlation variable test matrix; obtaining inherent scene variables and their value ranges; constructing an inherent test matrix based on the inherent scene variables and their value ranges; and controlling the target background vehicle using the configured humanoid driving model based on the inherent test matrix.

[0186] Corresponding to the above method, the above device may further include a preheating unit for preheating each vehicle in the test scenario so that the vehicle under test and the target background vehicle in the test scenario reach a target state, the target state including: the speed of the vehicle under test and the target background vehicle reaches a set speed, and the distance between the vehicle under test and the target background vehicle reaches a set distance.

[0187] In summary, when using the device disclosed in this embodiment to conduct a driving safety simulation test, at the beginning of the test, a preheating unit is first used to preheat each vehicle in the test scenario, so that the vehicle under test and the background vehicle reach the target state. Then, a model acquisition unit is used to acquire an anthropomorphic driving model. The target background vehicle control unit then uses the anthropomorphic driving model to control the target background vehicle. Finally, a simulation test unit uses the target background vehicle to test the vehicle under test. This device can simulate a more realistic driving scenario by preheating the scenario, and by controlling the interaction between the target background vehicle and the vehicle under test and the randomness of the target background vehicle's driving decisions through the anthropomorphic driving model, it can more realistically restore the actual driving scenario, thereby more effectively triggering potential risk events and better testing the vehicle under test.

[0188] This application also provides an electronic device in its embodiments. (See reference...) Figure 16 The diagram illustrates a structural schematic suitable for implementing the electronic device in the embodiments of this application. The electronic device in the embodiments of this application may include, but is not limited to, fixed terminals such as mobile phones, laptops, PDAs (personal digital assistants), PADs (tablet computers), desktop computers, etc. Figure 16 The electronic device shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of this application.

[0189] like Figure 16 As shown, the electronic device may include a processing unit (e.g., a central processing unit, a graphics processing unit, etc.) 601, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 602 or a program loaded from a storage device 608 into a random access memory (RAM) 603. When the electronic device is powered on, the RAM 603 also stores various programs and data required for the operation of the electronic device. The processing unit 601, ROM 602, and RAM 603 are interconnected via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0190] Typically, the following devices can be connected to I / O interface 605: input devices 606 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 607 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 608 including, for example, memory cards, hard drives, etc.; and communication devices 609. Communication device 609 allows electronic devices to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 16Electronic devices with various devices are shown, but it should be understood that it is not required to implement or have all of the devices shown. More or fewer devices may be implemented or have alternatively.

[0191] It should also be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. In addition, in the device embodiment drawings provided in this application, the connection relationship between modules indicates that they have a communication connection, which can be implemented as one or more communication buses or signal lines.

[0192] Through the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware, or it can be implemented by special-purpose hardware including application-specific integrated circuits, special-purpose CPUs, special-purpose memory, special-purpose components, etc. Generally, any function performed by a computer program can be easily implemented by corresponding hardware, and the specific hardware structure used to implement the same function can also be diverse, such as analog circuits, digital circuits, or special-purpose circuits. However, for this application, software program implementation is more often the preferred implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a readable storage medium, such as a computer floppy disk, USB flash drive, mobile hard disk, ROM, RAM, magnetic disk, or optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, training equipment, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0193] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product.

[0194] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, training device, or data center to another website, computer, training device, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can store or a data storage device such as a training device or data center that integrates one or more available media. The available media may be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid-state drives (SSDs)).

Claims

1. A method for simulating and testing the safety of driving a car, characterized in that, include: Obtain an anthropomorphic driving model; A human-like driving model is used to control the target background vehicle; The target background vehicle is used as the background vehicle to conduct driving safety tests on the test vehicle.

2. The driving safety simulation test method according to claim 1, characterized in that, There is a one-to-one correspondence between the anthropomorphic driving model and the target background vehicle, and there is at least one anthropomorphic driving model and a target background vehicle; A human-like driving model is used to control the target background vehicle, including: The system employs an anthropomorphic driving model to control the target background vehicle.

3. The driving safety simulation test method according to claim 1, characterized in that, A human-like driving model is used to control the target background vehicle, including: Obtain the inherent scene variables and the value range of the inherent scene variables; Obtain the behavioral correlation variables and the value range of the correlation variables; A test matrix is ​​constructed based on the aforementioned behavioral correlation variables, inherent scenario variables, and the value range of each variable; The anthropomorphic driving model is used to control the target background vehicle based on the test matrix.

4. The driving safety simulation test method according to claim 3, characterized in that, Obtain behavioral-related variables, including: Obtain the target test scenario; Obtain behavioral correlation variables that match the target test scenario; Before using an anthropomorphic driving model to control the target background vehicle, the following steps are also included: Select a background vehicle that matches the target test scenario as the target background vehicle.

5. The driving safety simulation test method according to claim 4, characterized in that, The inherent scene variables are at least a combination of one or more of the following objective parameters: The target parameters include: At the moment when the target background vehicle begins to change lanes, the absolute value of the time compared to the initial time, the longitudinal speed of the target background vehicle, the longitudinal distance of the target background vehicle relative to the vehicle in front, and the longitudinal speed of the vehicle in front of the target background vehicle; At the end of the simulation, the absolute value of the time compared to the initial time, the change in the speed of the vehicle in front of the target background vehicle, and the acceleration of the vehicle in front of the target background vehicle; When the target test scenario is a lane-changing scenario, the behavioral correlation variables include: following tolerance, lane-changing courtesy, and desired driving state.

6. The driving safety simulation test method according to claim 1, characterized in that, A human-like driving model is used to control the target background vehicle, including: Obtain the behavioral correlation variables and the value range of the correlation variables; Construct a test matrix of behavioral association variables based on the behavioral association variables and the range of values ​​of the association variables; Configure the anthropomorphic driving model based on the aforementioned behavioral correlation variable test matrix; Obtain the inherent scene variables and the value range of the inherent scene variables; An inherent test matrix is ​​constructed based on inherent scenario variables and the value range of the inherent scenario variables; The configured anthropomorphic driving model is used to control the target background vehicle based on the inherent test matrix.

7. The driving safety simulation test method according to claim 1, characterized in that, Before using an anthropomorphic driving model to control the target background vehicle, the following steps are also included: Each vehicle in the test scenario is preheated so that the test vehicle and the target background vehicle in the test scenario reach the target state, which includes: the speed of the test vehicle and the target background vehicle reaches the set speed, and the distance between the test vehicle and the target background vehicle reaches the set distance.

8. The driving safety simulation test method according to claim 3 or 6, characterized in that, When constructing the test matrix, the continuous Latin hypercube sampling method is specifically used.

9. A vehicle safety testing device, characterized in that, include: The model acquisition unit is used to acquire an anthropomorphic driving model; The target background vehicle control unit is used to control the target background vehicle using an anthropomorphic driving model; The simulation test unit is used to conduct driving safety tests on the test vehicle using the target background vehicle as the background vehicle.

10. An electronic device, characterized in that, include: At least one processor and a memory connected to the processor, wherein: The memory is used to store computer programs; The processor is used to execute the computer program to enable the electronic device to implement the driving vehicle safety simulation test method as described in any one of claims 1 to 8.

Images