Automatic driving high-risk fuzzy test scene generation method and device and storage medium

Abstract

The invention discloses an automatic driving high-risk fuzzy test scene generation method, and the method comprises the steps: carrying out the deconstruction of a scene according to the construction demands of an automatic driving test scene, and extracting the static environment information needed by the test scene and the dynamic parameter distribution of traffic participants; static environment information and dynamic parameters of traffic participants are sampled, and a scene is extracted into a plurality of scene sections; scene section generalization generation is realized by using a conditional table adversarial generative network; through each generated scene section, according to scene variable parameter distribution, behavior uncertainty is captured by using a triangular fuzzy number, and a complete fuzzy track sequence is expanded; constructing an automatic driving high-risk fuzzy test scene in combination with the static environment information and the fuzzy trajectory sequence; therefore, not only is the coverage rate of the test scene improved, but also the pertinence and authenticity of the test are enhanced, and more effective support is provided for safety performance evaluation of the automatic driving system in a complex environment.

Description

Technical Field

[0001] The present application relates to the field of autonomous driving, and specifically to a method, device, computer equipment and computer-readable storage medium for generating high-risk fuzzy test scenarios for autonomous driving. Background Art

[0002] High-risk test scenarios are particularly important for building high-risk test scenarios for autonomous vehicles. In scenario-based testing and evaluation, the fuzzy decisions of background vehicles directly affect the performance and experimental results of autonomous vehicles. The actual data collected is limited by the sample size and diversity, and may not contain key attribute combinations that are critical to autonomous driving testing. In addition, due to the determinism of a single collision trajectory, the fuzzy decision-making characteristics of background vehicles cannot be reflected, and thus the safety performance of autonomous driving cannot be tested in interactive vehicle decision-making uncertainty test scenarios. Therefore, it is crucial to consider the uncertainty of interactive vehicle behavior when generating scenarios to improve the richness of test scenarios. Summary of the invention

[0003] In response to the above technical problems, the present application provides a method, device, computer equipment and computer-readable storage medium for generating high-risk fuzzy test scenarios for autonomous driving, which can improve the accuracy of robustness assessment of autonomous driving vehicles.

[0004] The present application provides a method for generating a high-risk fuzzy test scenario for autonomous driving, the method comprising:

[0005] Step S1: Deconstruct the scene according to the requirements of the autonomous driving test scene, and extract the static environment information required for the test scene and the dynamic parameter distribution of the traffic participants;

[0006] Step S2, sampling static environmental information and dynamic parameters of traffic participants, and extracting the scene into a number of scene sections;

[0007] Step S3, using the conditional table to counter the generative network to achieve generalized generation of scene sections;

[0008] Step S4, through each generated scene section, according to the distribution of scene variable parameters, using triangular fuzzy numbers to capture behavioral uncertainty and expand it into a complete fuzzy trajectory sequence;

[0009] Step S5: Combine static environment information with fuzzy trajectory sequences to construct a high-risk fuzzy test scenario for autonomous driving.

[0010] Preferably, the static environmental information includes: vehicle type of traffic participants, road type, road width, number of lanes, lighting conditions, weather conditions, traffic control method, and number of background vehicles.

[0011] Preferably, the dynamic parameters of the traffic participant are the state of the traffic participant at each time step, including: the longitudinal coordinate, the transverse coordinate, the speed, acceleration, direction, turn, and relative position difference at each coordinate within each step.

[0012] Preferably, the sampling of the static environment information and the dynamic parameters of the traffic participants extracts the scene into a plurality of scene sections, including:

[0013] The static environment information and the dynamic parameters of the traffic participants are sampled at a preset sampling interval to construct a scene section; wherein the static environment information is represented by a categorical variable (s1, s2, s3...s l ) description, and dynamic parameters are continuous variables with time labels (d1, d2, d3...d n )describe.

[0014] Preferably, the method of using the trained condition table to generate adversarial network to realize the generalization generation of scene sections includes:

[0015] The scene sections are generated by setting the conditional vector, generator loss and sampling training for the conditional table adversarial generative network;

[0016] The loss is penalized by averaging all scenes in the batch, and sampling training is performed according to the set conditional vector. When the training loss values ​​of the generator and discriminator reach convergence, the trained conditional table adversarial generation network is obtained.

[0017] Preferably, each scene section generated by the method captures the behavioral uncertainty by using triangular fuzzy numbers according to the distribution of scene variable parameters, and is expanded into a complete fuzzy trajectory sequence, including:

[0018] Through each generated scene section, a complete trajectory is constructed using dynamic variable expansion. The target vehicle trajectory is expanded until the simulation duration is reached with the preset time as the time step. The parameters of each time are updated as follows:

[0019] x t+1 =x t +v t cosh t

[0020] y t+1 =y t +v t sinh t

[0021] v t+1 =v t +a t T t

[0022]

[0023] x t ,y t ,v t ,a t ,h t They represent the x-coordinate, y-coordinate, speed, acceleration, and direction of the target vehicle at time t; T is the time step, Δh t is the rate of change of the orientation at time t, expressed using triangular fuzzy numbers: in Respectively expressed as Δh t The lower quartile, mean and upper quartile of Δh in the database at time t; Jerk t is the derivative of acceleration at time t, also expressed by triangular fuzzy numbers: Respectively represent the Jerk in the database t are the lower quartile, mean and upper quartile at time t; the fuzzy position of the next step is continuously updated through the above formula to obtain a set of fuzzy trajectory sequences.

[0024] Preferably, the combining of static environment information and fuzzy trajectory sequence to construct a high-risk fuzzy test scenario for autonomous driving includes:

[0025] The fuzzy trajectory sequence is placed into the static environment information of each generated scene to form a complete high-risk fuzzy test scenario for autonomous driving.

[0026] An embodiment of the present invention further provides a device for generating a high-risk fuzzy test scenario for autonomous driving, comprising:

[0027] The acquisition module is used to build requirements according to the autonomous driving test scenario, deconstruct the scenario, and extract the static environment information required for the test scenario and the dynamic parameter distribution of traffic participants;

[0028] The sampling module is used to sample static environmental information and dynamic parameters of traffic participants and extract the scene into several scene sections;

[0029] A generation module, which is used to achieve generalized generation of scene sections using a conditional table adversarial generative network;

[0030] An expansion module is used to capture the behavior uncertainty by using triangular fuzzy numbers according to the distribution of scene variable parameters through each generated scene section, and expand it into a complete fuzzy trajectory sequence;

[0031] A building block is used to combine static environment information with fuzzy trajectory sequences to construct high-risk fuzzy testing scenarios for autonomous driving.

[0032] The present application also provides a computer device, including: a memory and a processor, wherein the memory stores computer program instructions for execution on the processor, and when the processor executes the computer program instructions, the method for generating high-risk fuzzy test scenarios for autonomous driving as described above is implemented.

[0033] The present application also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the steps of the method for generating a high-risk fuzzy test scenario for autonomous driving are implemented as described above.

[0034] As described above, the method for generating high-risk fuzzy test scenarios for autonomous driving of the present application includes: S1, deconstructing the scenario according to the requirements for building the autonomous driving test scenario, and extracting the static environmental information and dynamic parameter distribution of traffic participants required for the test scenario; by deconstructing the scenario and extracting information, the key elements required for the test can be accurately obtained, including static environmental information such as road layout and traffic signals, and dynamic parameter distributions such as speed, acceleration, and position of various traffic participants. This refined extraction method makes the test scenario construction more flexible, and can quickly adjust the scene elements according to specific test requirements, thereby improving the pertinence and efficiency of the test. Step S2, sampling the static environmental information and the dynamic parameters of traffic participants, and extracting the scene into several scene sections; by sampling the scene information, the complex scene can be decomposed into multiple easy-to-process sections, and each section can be used as an independent test unit. This segmented processing method can reduce the complexity of the entire scene, making subsequent generation and calculation more efficient, while ensuring the full expression of scene diversity and improving the flexibility of scene generation. Step S3, using conditional table adversarial generative network to achieve generalized generation of scene sections; using conditional table adversarial generative network (CTGAN) to achieve generalized generation of sections, it can generate more scene sections under different conditions, greatly expanding the coverage of test scenarios. CTGAN can capture scene changes under different conditions while ensuring that the generated scenes conform to the distribution of real scenes, thereby greatly improving the adaptability of the autonomous driving system in various complex scenes. Step S4, through each generated scene section, according to the distribution of scene variable parameters, using triangular fuzzy numbers to capture behavioral uncertainty and expand it into a complete fuzzy trajectory sequence; triangular fuzzy numbers can effectively capture the uncertainty in the behavior of traffic participants and expand it into a complete fuzzy trajectory sequence. This method not only improves the authenticity of the generated scene, but also can better simulate the random behavior that may be encountered in the actual driving process, so as to conduct more rigorous tests on the response ability of the autonomous driving system. Step S5, combining static environment information with fuzzy trajectory sequence to construct a high-risk fuzzy test scene for autonomous driving. Through the combination of static environment information and fuzzy trajectory sequence, the constructed scene can realistically restore the complex interaction in high-risk traffic environment, especially in multi-vehicle interaction and emergency situations, which can better test the decision-making and response capabilities of the autonomous driving system. Compared with traditional scenario generation methods, this method is more flexible and scalable, and can generate more diverse high-risk scenarios, thereby improving the test coverage and effectiveness of autonomous driving systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] The drawings herein are incorporated into the specification and constitute a part of the specification, illustrate embodiments consistent with the present application, and are used together with the specification to explain the principles of the present application. In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the description of the embodiments are briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor.

[0036] Figure 1 A schematic diagram of a process for generating a high-risk fuzzy test scenario for autonomous driving provided by an embodiment of the present invention;

[0037] Figure 2 A schematic diagram of the structure of a device for generating high-risk fuzzy test scenarios for autonomous driving provided in one embodiment of the present invention.

[0038] The realization of the purpose, functional features and advantages of this application will be further described in conjunction with the embodiments and with reference to the accompanying drawings. The above-mentioned drawings have shown clear embodiments of this application, which will be described in more detail later. These drawings and textual descriptions are not intended to limit the scope of the concept of this application in any way, but to illustrate the concept of this application to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0039] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the attached claims. It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application.

[0040] Before further describing the present invention in detail, the nouns and terms involved in the embodiments of the present invention are explained. The nouns and terms involved in the embodiments of the present invention are subject to the following explanations.

[0041] See also Figure 1 This embodiment provides a method for generating a high-risk fuzzy test scenario for autonomous driving, the method comprising:

[0042] Step S1: Deconstruct the scene according to the requirements of the autonomous driving test scene, and extract the static environment information required for the test scene and the dynamic parameter distribution of the traffic participants;

[0043] Here, by deconstructing the scene and extracting information, we can accurately obtain the key elements required for the test, including static environmental information such as road layout and traffic signals, as well as dynamic parameter distributions such as speed, acceleration, and position of various traffic participants. This refined extraction method makes the test scene construction more flexible and can quickly adjust the scene elements according to specific test requirements, thereby improving the pertinence and efficiency of the test.

[0044] Step S2, sampling static environmental information and dynamic parameters of traffic participants, and extracting the scene into a number of scene sections;

[0045] Here, by sampling the scene information, the complex scene can be decomposed into multiple easy-to-process sections, each of which can be used as an independent test unit. This segmented processing method can reduce the complexity of the entire scene, making subsequent generation and calculation more efficient, while ensuring the full expression of scene diversity and improving the flexibility of scene generation.

[0046] Step S3, using the conditional table to counter the generative network to achieve generalized generation of scene sections;

[0047] Here, the Conditional Tabular Adversarial Generative Network (CTGAN) is used to achieve the generalization generation of cross sections, which can generate more scene cross sections under different conditions, greatly expanding the coverage of test scenarios. While ensuring that the generated scenes conform to the distribution of real scenes, CTGAN can also capture scene changes under different conditions, thereby greatly improving the adaptability of the autonomous driving system in various complex scenarios.

[0048] Step S4, through each generated scene section, according to the distribution of scene variable parameters, using triangular fuzzy numbers to capture behavioral uncertainty and expand it into a complete fuzzy trajectory sequence;

[0049] Here, triangular fuzzy numbers can effectively capture the uncertainty in the behavior of traffic participants and expand it into a complete fuzzy trajectory sequence. This method not only improves the realism of the generated scene, but also better simulates the random behavior that may be encountered in actual driving, thereby conducting more rigorous tests on the response capabilities of the autonomous driving system.

[0050] Step S5: Combine static environment information with fuzzy trajectory sequences to construct a high-risk fuzzy test scenario for autonomous driving.

[0051] Here, by combining static environment information and fuzzy trajectory sequences, the constructed scenario can realistically restore the complex interactions in high-risk traffic environments, especially in multi-vehicle interactions and emergencies, which can better test the decision-making and response capabilities of the autonomous driving system. Compared with traditional scenario generation methods, this method is more flexible and scalable, and can generate more diverse high-risk scenarios, thereby improving the test coverage and effectiveness of the autonomous driving system.

[0052] In the above-mentioned embodiment of the present invention, the test scenario generation method that takes into account behavioral uncertainty can improve the accuracy of robustness assessment of autonomous driving vehicles. This helps to improve the safety and reliability of autonomous driving vehicles and promote the development of their practical applications. An efficient, scalable and uncertainty-capturing scenario generation tool is provided for the safety testing of autonomous driving systems. Compared with traditional test methods based on rules or fixed scenario libraries, the proposed method can generate richer and more diverse high-risk scenarios by introducing fuzzy uncertainty and generative adversarial network technology. In this way, not only the coverage of the test scenarios is improved, but also the pertinence and authenticity of the tests are enhanced, providing more effective support for the safety performance evaluation of autonomous driving systems in complex environments.

[0053] In one embodiment, the static environment information includes: vehicle type of traffic participants, road type, road width, number of lanes, lighting conditions, weather conditions, traffic control method, and number of background vehicles.

[0054] In one embodiment, the dynamic parameters of the traffic participant are the state of the traffic participant at each time step, including: longitudinal coordinates, transverse coordinates, speed, acceleration, direction, turn, and relative position difference at each coordinate within each step.

[0055] In one embodiment, the sampling of the static environment information and the dynamic parameters of the traffic participants, extracting the scene into a plurality of scene sections, includes:

[0056] The static environment information and the dynamic parameters of the traffic participants are sampled at a preset sampling interval to construct a scene section; wherein the static environment information is represented by a categorical variable (s1, s2, s3...s l ) description, and dynamic parameters are continuous variables with time labels (d1, d2, d3...d n ) description. Here, static elements and dynamic elements are necessary information of the scene; when generating and digitizing the scene, they need to be coupled with different types of vectors and need to be separated and described. For example, dynamic vectors are time-varying and need to be expanded later, while static variables only provide non-time-varying information such as maps and weather, and do not require time-varying expansion, only sampling coupling is required.

[0057] In one embodiment, the method of using a trained condition table to generate a generalized scene section against a generative network includes:

[0058] The scene sections are generated by setting the conditional vector, generator loss and sampling training for the conditional table adversarial generative network;

[0059] The loss is penalized by averaging all scenes in the batch, and sampling training is performed according to the set conditional vector. When the training loss values ​​of the generator and discriminator reach convergence, the trained conditional table adversarial generation network is obtained.

[0060] Here, the conditional table adversarial generative network (CTGAN) is used to achieve generalized generation of scene sections. Specifically, each scene variable is processed independently. Each value is represented as a unique hot vector of a specified mode and a scalar, specifically:

[0061] 1) For each continuous variable column D i , use the variational Gaussian mixture model to estimate the number of Gaussian distribution modes and fit the Gaussian mixture. Suppose m modes are found, named η1, η2…η m The learned Gaussian mixture is where μ k and φ k are the weight and bias of the pattern, respectively.

[0062] 2) For the continuous variable D in i Each value d in i,j Calculate the probability of each Gaussian distribution mode, that is,

[0063] 3) Sample a pattern from the given probability density and normalize the value using the sampled pattern. For example, for m patterns, select one of them and then normalize d i,j Represented as a one-hot vector β i,j = [0,0,…,1] (the position of 1 indicates the selected mode, here the mth one), and a scalar Represents the value within the pattern. Then a scene section can be represented as:

[0064]

[0065] where s i,j is the one-hot encoding representation of the i-th static variable.

[0066] 4) Generate scene sections by setting conditional vectors, generator losses, and sampling training. Specifically: introduce conditional variables As a specified condition (S i* =k *) method. Among them, k * is the i-th discrete variable column S i* The value of the i-th unique hot vector is s i =[s i (k) ],fork=1,…,|S i |, let m i =[m i (k) ],fork=1,…,|S i | is the i-th mask vector of the one-hot vector. The condition can be expressed using the mask vector as:

[0067]

[0068] Conditional generator generates any set of one-hot discrete vectors By increasing and m i* The cross entropy between the two is averaged over all the scenes in the batch to penalize the loss. Then, sampling training is performed according to the set conditional vector. When the training loss of the generator and the discriminator converges, it means that the generator has been trained.

[0069] In one embodiment, each scene section generated by the method captures the behavioral uncertainty using triangular fuzzy numbers according to the scene variable parameter distribution and is expanded into a complete fuzzy trajectory sequence, including:

[0070] Through each generated scene section, a complete trajectory is constructed using dynamic variable expansion. The target vehicle trajectory is expanded until the simulation duration is reached with the preset time as the time step. The parameters of each time are updated as follows:

[0071] x t+1 =x t +v t cosh t

[0072] y t+1 =y t +v t sinh t

[0073] v t+1 =v t +a t T t

[0074]

[0075] x t ,y t ,v t ,a t,h t They represent the x-coordinate, y-coordinate, speed, acceleration, and direction of the target vehicle at time t; T is the time step, Δh t is the rate of change of the orientation at time t, expressed using triangular fuzzy numbers: in Respectively expressed as Δh t The lower quartile, mean and upper quartile of Δh in the database at time t; Jerk t is the derivative of acceleration at time t, also expressed by triangular fuzzy numbers: Respectively represent the Jerk in the database t are the lower quartile, mean and upper quartile at time t; the fuzzy position of the next step is continuously updated through the above formula to obtain a set of fuzzy trajectory sequences.

[0076] In one embodiment, the combining of static environment information and fuzzy trajectory sequence to construct a high-risk fuzzy test scenario for autonomous driving includes:

[0077] The fuzzy trajectory sequence is placed into the static environment information of each generated scene to form a complete high-risk fuzzy test scenario for autonomous driving.

[0078] See also Figure 2 , an embodiment of the present invention further provides a device for generating a high-risk fuzzy test scenario for autonomous driving, comprising:

[0079] The acquisition module 21 is used to deconstruct the scene according to the requirements of the autonomous driving test scene, and extract the static environment information required for the test scene and the dynamic parameter distribution of the traffic participants;

[0080] A sampling module 22, for sampling static environmental information and dynamic parameters of traffic participants, and extracting the scene into a plurality of scene sections;

[0081] A generation module 23, used to achieve scene section generalization generation by using a conditional table adversarial generation network;

[0082] An expansion module 24 is used to capture the behavior uncertainty by using triangular fuzzy numbers according to the distribution of scene variable parameters through each generated scene section, and expand it into a complete fuzzy trajectory sequence;

[0083] The construction module 25 is used to combine the static environment information with the fuzzy trajectory sequence to construct a high-risk fuzzy test scenario for autonomous driving.

[0084] The present application also provides a computer device, including: a memory and a processor, wherein the memory stores computer program instructions for execution on the processor, and when the processor executes the computer program instructions, the method for generating high-risk fuzzy test scenarios for autonomous driving as described above is implemented.

[0085] The present application also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the steps of the method for generating a high-risk fuzzy test scenario for autonomous driving are implemented as described above.

[0086] By integrating the functions of these modules, the constructed device can not only achieve efficient and accurate generation of high-risk fuzzy test scenarios for autonomous driving, but also improve the accuracy of robustness assessment of autonomous driving vehicles, which helps to improve the safety and reliability of autonomous driving vehicles and promote the development of their practical applications.

[0087] The above examples are only reference examples. In order to avoid redundancy, they are not listed one by one here. In actual development or application, they can be flexibly combined according to actual needs, but any combination belongs to the technical solution of this application and is covered within the scope of protection of this application.

[0088] An embodiment of the present application also provides a computer device, including a processor and a memory, wherein the memory stores computer program instructions for execution on the processor, and when the processor executes the computer program instructions, the method for generating high-risk fuzzy test scenarios for autonomous driving as described above is implemented.

[0089] An embodiment of the present application also provides a computer-readable storage medium storing computer instructions. When the computer instructions are executed by a processor, the method for generating a high-risk fuzzy test scenario for autonomous driving described above is implemented.

[0090] It is understood that the above scenarios are only examples and do not constitute a limitation on the application scenarios of the technical solutions provided in the embodiments of the present application. The technical solutions of the present application can also be applied to other scenarios. For example, it is known to those skilled in the art that with the evolution of the system architecture and the emergence of new business scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.

[0091] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0092] The steps in the method of the embodiment of the present application can be adjusted in order, combined and deleted according to actual needs.

[0093] The units in the device of the embodiment of the present application can be merged, divided and deleted according to actual needs.

[0094] In the present application, the same or similar terminology concepts, technical solutions and / or application scenario descriptions are generally described in detail only the first time they appear. When they appear again later, they are generally not repeated for the sake of brevity. When understanding the technical solutions and other contents of the present application, for the same or similar terminology concepts, technical solutions and / or application scenario descriptions that are not described in detail later, reference can be made to the previous related detailed descriptions.

[0095] In the present application, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0096] The various technical features of the technical solution of the present application can be arbitrarily combined. In order to make the description concise, not all possible combinations of the various technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of the present application.

[0097] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as above, and includes a number of instructions for a terminal device (which can be a mobile phone, a computer, a server, a controlled terminal, or a network device, etc.) to execute the method of each embodiment of the present application.

[0098] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When loading and executing computer program instructions on a computer, the process or function according to the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. Computer instructions can be stored in a storage medium, or transmitted from one storage medium to another storage medium. For example, computer instructions can be transmitted from a website site, computer, server or data center to another website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) mode. The storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integration. Available media can be magnetic media, (e.g., floppy disk, storage disk, tape), optical media (e.g., DVD), or semiconductor media (e.g., solid-state storage disk Solid State Disk (SSD)), etc.

[0099] The above are only preferred embodiments of the present application, and are not intended to limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A method for generating high-risk fuzzy test scenarios for autonomous driving, characterized in that: The method comprises: Step S1: Deconstruct the scene according to the requirements of the autonomous driving test scene, and extract the static environment information required for the test scene and the dynamic parameter distribution of the traffic participants; Step S2, sampling static environmental information and dynamic parameters of traffic participants, and extracting the scene into a number of scene sections; Step S3, using the conditional table to counter the generative network to achieve generalized generation of scene sections; Step S4, through each generated scene section, according to the distribution of scene variable parameters, using triangular fuzzy numbers to capture behavioral uncertainty and expand it into a complete fuzzy trajectory sequence; Step S5: Combine static environment information with fuzzy trajectory sequences to construct a high-risk fuzzy test scenario for autonomous driving.

2. The method for generating high-risk fuzzy test scenarios for autonomous driving according to claim 1, characterized in that: The static environment information includes: vehicle type of traffic participants, road type, road width, number of lanes, lighting conditions, weather conditions, traffic control method, and number of background vehicles.

3. The method for generating high-risk fuzzy test scenarios for autonomous driving according to claim 1, characterized in that: The dynamic parameters of the traffic participant are the state of the traffic participant at each time step, including: the longitudinal coordinate, the transverse coordinate, the speed, acceleration, direction, turn, and relative position difference at each coordinate within each step.

4. The method for generating high-risk fuzzy test scenarios for autonomous driving according to claim 1, characterized in that: The static environment information and the dynamic parameters of the traffic participants are sampled to extract the scene into several scene sections, including: The static environment information and the dynamic parameters of the traffic participants are sampled at a preset sampling interval to construct a scene section; wherein the static environment information is represented by a categorical variable (s1, s2, s3...s l ) description, and dynamic parameters are continuous variables with time labels (d1, d2, d3...d n )describe.

5. The method for generating high-risk fuzzy test scenarios for autonomous driving according to claim 1, characterized in that: The method of using the trained conditional table against the generative network to realize the generalization generation of scene sections includes: The scene sections are generated by setting the conditional vector, generator loss and sampling training for the conditional table adversarial generative network; The loss is penalized by averaging all scenes in the batch, and sampling training is performed according to the set conditional vector. When the training loss values ​​of the generator and discriminator reach convergence, the trained conditional table adversarial generation network is obtained.

6. The method for generating high-risk fuzzy test scenarios for autonomous driving according to claim 1, characterized in that: Each scene section generated by the method captures the behavioral uncertainty using triangular fuzzy numbers according to the distribution of scene variable parameters and is expanded into a complete fuzzy trajectory sequence, including: Through each generated scene section, a complete trajectory is constructed using dynamic variable expansion. The target vehicle trajectory is expanded until the simulation duration is reached with the preset time as the time step. The parameters of each time are updated as follows: x t+1 =x t +v t coSh t Y t+1 =y t +v t born t v t+1 =v t +a t T x t ,y t , v t , a t ,h t They represent the x-coordinate, y-coordinate, speed, acceleration, and direction of the target vehicle at time t; T is the time step, Δh t is the rate of change of the orientation at time t, expressed using triangular fuzzy numbers: in Respectively expressed as Δh t The lower quartile, mean and upper quartile of Δh in the database at time t; Jerk t is the derivative of acceleration at time t, also expressed by triangular fuzzy numbers: Respectively represent the Jerk in the database t are the lower quartile, mean and upper quartile at time t; the fuzzy position of the next step is continuously updated through the above formula to obtain a set of fuzzy trajectory sequences.

7. The method for generating high-risk fuzzy test scenarios for autonomous driving according to claim 6, characterized in that: The method combines static environment information with fuzzy trajectory sequences to construct a high-risk fuzzy test scenario for autonomous driving, including: The fuzzy trajectory sequence is placed into the static environment information of each generated scene to form a complete high-risk fuzzy test scenario for autonomous driving.

8. A device for generating high-risk fuzzy test scenarios for autonomous driving, characterized in that: include: The acquisition module is used to build requirements according to the autonomous driving test scenario, deconstruct the scenario, and extract the static environment information required for the test scenario and the dynamic parameter distribution of traffic participants; The sampling module is used to sample static environmental information and dynamic parameters of traffic participants and extract the scene into several scene sections; A generation module, which is used to achieve generalized generation of scene sections using a conditional table adversarial generative network; An expansion module is used to capture the behavior uncertainty by using triangular fuzzy numbers according to the distribution of scene variable parameters through each generated scene section, and expand it into a complete fuzzy trajectory sequence; A building block is used to combine static environment information with fuzzy trajectory sequences to construct high-risk fuzzy testing scenarios for autonomous driving.

9. A computer device, characterized in that: It includes a processor and a memory, the memory stores computer program instructions for execution on the processor, and when the processor executes the computer program instructions, it implements the high-risk fuzzy testing scenario method for autonomous driving as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: It stores computer instructions, which, when executed by a processor, implement the high-risk fuzzy testing scenario method for autonomous driving as described in any one of claims 1 to 7.

Images