A method, device and equipment for generating autonomous driving high-value scenario use cases

Abstract

The application discloses a method for generating high-value scene use cases of automatic driving, and the method comprises the following steps: acquiring scene characteristic parameters, constructing a scene use case library according to the scene characteristic parameters; running the scene use case library by using an automatic driving algorithm to obtain an actual weight ratio of the scene characteristic parameters, wherein the weight ratio refers to the ratio of the weight of a scene characteristic parameter running successfully to the weight of a scene characteristic parameter running unsuccessfully; predicting the weight ratio of the scene characteristic parameters to obtain a predicted weight ratio; constructing a characteristic evaluation model according to the actual weight ratio and the predicted weight ratio; and performing targeted optimization on the scene use case library by using the characteristic evaluation model to obtain high-value scene use cases.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of automatic driving, in particular to a method, device and equipment for generating high-value scenario use cases of automatic driving. BACKGROUND

[0002] The simulation test work of cloud control networked vehicle automatic driving needs to build a large-scale scenario use case that can continuously improve the value of existing algorithms. Based on this goal, a networked cloud control simulation test platform is currently built, which can build test scenarios of scenario vehicles and networked automatic driving vehicles according to vehicle running scripts. At the same time, a large number of use cases are verified by changing the script use case parameters. The purpose of mass production of test cases can be achieved. However, the current production use cases have a certain homogeneity, and it is difficult to define the quality of the use cases. SUMMARY

[0003] The embodiments of the present specification provide a method, device and equipment for generating high-value scenario use cases of automatic driving, to solve the problem that the current production use cases have a certain homogeneity and it is difficult to define the quality of the use cases.

[0004] The embodiments of the present specification adopt the following technical solutions:

[0005] In a first aspect, the embodiments of the present specification provide a method for generating high-value scenario use cases of automatic driving, the method comprising the following steps:

[0006] Obtaining scenario feature parameters, and constructing a scenario use case library according to the scenario feature parameters;

[0007] Running the scenario use case library by using an automatic driving algorithm to obtain an actual weight ratio of the scenario feature parameters, wherein the weight ratio refers to the ratio of the weight of a scenario feature parameter that runs successfully to the weight of a scenario feature parameter that runs unsuccessfully;

[0008] Predicting the weight ratio of the scenario feature parameters to obtain a predicted weight ratio;

[0009] Constructing a feature evaluation model according to the actual weight ratio and the predicted weight ratio;

[0010] Optimizing the scenario use case library by using the feature evaluation model to obtain high-value scenario use cases;

[0011] Adjusting the scenario feature parameters to obtain a new scenario use case library;

[0012] Repeating the above steps for the new scenario use case library to obtain new high-value scenario use cases again.

[0013] In a second aspect, the embodiments of the present specification provide a device for generating high-value scenario use cases of automatic driving, comprising:

[0014] a scene module, configured to acquire scene characteristic parameters, and construct a scene use case library according to the scene characteristic parameters;

[0015] an execution module, configured to run the scene use case library by using an automatic driving algorithm, and obtain an actual weight ratio of the scene characteristic parameters, the weight ratio being a ratio of a weight of a scene characteristic parameter that is successfully run to a weight of a scene characteristic parameter that is unsuccessfully run;

[0016] a prediction module, configured to predict the weight ratio of the scene characteristic parameters, and obtain a predicted weight ratio;

[0017] a model construction module, configured to construct a characteristic evaluation model according to the actual weight ratio and the predicted weight ratio;

[0018] a high-quality use case module, configured to perform targeted optimization on the scene use case library by using the characteristic evaluation model, and obtain high-value scene use cases;

[0019] an adjustment module, configured to adjust the scene characteristic parameters, and obtain a new scene use case library;

[0020] a circulation module, configured to repeat the above steps for the new scene use case library, and obtain new high-value scene use cases again.

[0021] In a third aspect, an electronic device is provided, including at least one processor and a memory, the memory storing a program and being configured to cause the at least one processor to execute any of the methods for generating high-value scene use cases for automatic driving.

[0022] In a fourth aspect, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing the computer to execute any of the methods for generating high-value scene use cases for automatic driving.

[0023] The above at least one technical solution adopted by the embodiments of the present specification can achieve the following beneficial effects: high-quality test case characteristics can be found, and high-value scene use cases can be continuously produced dynamically. The execution results are used as feedback to continuously optimize the characteristic weight coefficients, and it is predicted whether newly generated use cases are high-value use cases, so that the generation mode of use cases is continuously optimized and improved, and the purpose of producing high-quality use cases is achieved. BRIEF DESCRIPTION OF DRAWINGS

[0024] The accompanying drawings, which are included to provide a further understanding of the embodiments of the present specification, constitute a part of the present specification, and the illustrative embodiments of the present specification and their descriptions serve to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0025] Figure 1 A flowchart of a method for generating high-value scenario use cases for automatic driving provided by an embodiment of the present specification is shown in the figure;

[0026] Figure 2 A schematic diagram of the structure of an apparatus for generating high-value scenario use cases for automatic driving provided by an embodiment of the present specification is shown in the figure. DETAILED DESCRIPTION

[0027] To make the purposes, technical solutions, and advantages of the present application clearer, the technical solutions of the present application will be described in detail below with reference to the specific embodiments of the present specification and the corresponding drawings. Obviously, the described embodiments are only some of the embodiments of the present specification, not all the embodiments. Based on the embodiments in the present specification, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of protection of the present application.

[0028] In the prior art, because the use cases are generated on a large scale in the prior art, the produced use cases have certain homogeneity, and it is difficult to define the quality of the use cases.

[0029] Therefore, the present specification provides a method, an apparatus, and a device for generating high-value scenario use cases for automatic driving, which can discover the characteristics of high-quality test use cases and continuously produce high-value scenario use cases. The execution results are used as feedback to continuously optimize the feature weight coefficients, and it is predicted whether the newly generated use cases are high-value use cases, so as to realize the purpose of continuously optimizing and improving the generation method of the use cases and producing high-quality use cases.

[0030] The technical solutions provided by the embodiments of the present specification will be described in detail below with reference to the drawings.

[0031] Embodiment 1

[0032] Figure 1 A flowchart of a method for generating high-value scenario use cases for automatic driving provided by an embodiment of the present specification is shown in the figure.

[0033] Referring to Figure 1 The method for generating high-value scenario use cases for automatic driving provided by Embodiment 1 includes the following steps:

[0034] S101, acquiring scenario feature parameters, and constructing a scenario use case library according to the scenario feature parameters;

[0035] In specific implementation, the scene characteristic parameters include, but are not limited to, factors relied on driving environment and ranges supported by script commands, such as map, vehicle, group relationship and other elements. The way of constructing the scene use case library according to the scene characteristic parameters includes, but is not limited to, inputting the obtained scene characteristic parameters into an existing scene use case library to obtain a new scene use case library, so that the new scene use case library is used for running of the automatic driving algorithm.

[0036] The map includes, but is not limited to, highway, ramp, urban road, crossroad, road curvature, slope, number of lanes and the like. The vehicle includes, but is not limited to, vehicle type, initial position, vehicle speed, lane change, acceleration, angular acceleration and the like. The group relationship includes, but is not limited to, relative speed, speed maintenance, lane maintenance, total number of vehicles, number of controlled vehicles, inter-vehicle distance and the like. The other elements include, but are not limited to, non-motor vehicle, pedestrian, obstacle and the like. According to the scene characteristic parameters, a conventional use case library construction method can be used to construct the scene use case library. The scene use case library contains a plurality of scene use cases. In use, according to an existing scene construction script (divided into five categories of simulation, control, constraint, tracking and perception), a basic scene is configured with different characteristic maps, so that scene use cases are generated for different scene algorithms and driving design of the automatic driving controlled vehicle based on the basic scene. For example, on the basis, initial parameters, running parameters, constraint objects, trigger conditions and other elements of the basic scene vehicle and the automatic driving controlled vehicle are adjusted according to a test matrix model to generate large-scale scene use cases.

[0037] It should be understood that the above-mentioned specific related contents are only used for exemplary illustration, and should not limit the present application in any way.

[0038] S103, running the scene use case library by using the automatic driving algorithm to obtain an actual weight ratio of the scene characteristic parameters, the weight ratio being a ratio of a weight of a successfully run scene characteristic parameter to a weight of a failed run scene characteristic parameter;

[0039] In specific implementation, the way of running the scene use case library by using the automatic driving algorithm to obtain the actual weight ratio of the scene characteristic parameters includes, but is not limited to, executing the automatic driving algorithm in the scene use case library to obtain a failed execution scene use case library; obtaining a failed execution weight of the scene characteristic parameters by using a logistic regression model on the failed execution scene use case library; and obtaining the actual weight ratio of the scene characteristic parameters according to the failed execution weight.

[0040] For example, in the execution of the automatic driving algorithm, in the scene case library, multiple scene characteristic parameters in multiple scene cases cannot be executed successfully, the scenes that cannot pass are marked, and then the total number of scene cases that fail to execute due to scene characteristic parameters is counted, so as to count the weight of each scene characteristic parameter execution failure. Taking the slope in the scene characteristic as an example, when the automatic driving algorithm is executed, the system cannot pass when it runs to the scene constructed by the slope, then the total number of all scene cases in the scene case library that fail to execute successfully due to the slope is counted, so as to obtain the execution failure weight of the scene characteristic parameter of the slope. Of course, each scene case contains multiple scene characteristics, and each scene case also contains multiple scene characteristics that fail to execute, and the counting method is the same. No repeated description.

[0041] It should be understood that the above-mentioned specific related contents are only used for exemplary illustration, and should not limit the present application in any way.

[0042] S105, predicting the weight ratio of the scene characteristic parameter, and obtaining a predicted weight ratio;

[0043] In specific implementation, predicting the weight ratio of the scene characteristic parameter can be understood as predicting the weight ratio of the scene characteristic parameter in advance, that is, predicting the weight ratio of the scene characteristic parameter of the constructed scene case library in advance. The way of predicting the weight ratio of the scene characteristic parameter and obtaining the predicted weight ratio includes but is not limited to constructing a prediction model; predicting the weight ratio of the scene characteristic parameter by the prediction model to obtain the predicted weight ratio. Wherein, the way of constructing the prediction model includes but is not limited to using the sigmoid function to construct the prediction model by fitting the decision boundary, establishing the decision boundary and combining the probability after the logistic regression model.

[0044] For example, continue to use the scene characteristic slope in S103 as an example, in the scene case library constructed by the slope scene characteristic, predict how many scene cases fail to execute after the constructed scene case library is executed, and then predict how many scene cases fail to execute due to the slope in the scene cases that fail to execute. So as to obtain the predicted failure weight of the slope scene characteristic parameter.

[0045] It should be understood that the above-mentioned specific related contents are only used for exemplary illustration, and should not limit the present application in any way.

[0046] S107, constructing a feature evaluation model according to the actual weight ratio and the predicted weight ratio;

[0047] In specific embodiments, a logistic regression model is used for classification, and the comparison result is labeled after comparing the simulation vehicle operation result simulated by the autonomous driving algorithm with the expected result of the scenario use case library. The label can be understood as a pre-prepared feature, such as failure or success, and only the weight, such as the weight of failure, is calculated. The weight of the scenario feature is obtained by the logistic regression model, and a feature evaluation model is established.

[0048] It should be understood that the above-mentioned specific related content is only used for exemplary illustration, and should not limit the present application.

[0049] S109, the scenario use case library is optimized by the feature evaluation model, and a high-value scenario use case is obtained;

[0050] In specific embodiments, the way of optimizing the scenario use case library by the feature evaluation model to obtain a high-value scenario use case includes but is not limited to adjusting the scenario feature parameter of the scenario use case library by the feature evaluation model to the direction with high execution failure weight, so as to obtain a high-value scenario use case.

[0051] For example, continuing to use the slope in S103 as an example of a scenario feature, the slope in the scenario use case with successful execution is adjusted in the scenario use case library constructed by the slope, and the direction of execution failure is adjusted, so as to obtain a high-value scenario use case.

[0052] Further, the scenario feature parameter is adjusted to obtain a new scenario use case library;

[0053] In specific embodiments, the way of adjusting the scenario feature parameter includes but is not limited to inputting new scenario feature parameters after manually adjusting the scenario feature parameters, or inputting new scenario feature parameters after automatically selecting the feature parameters.

[0054] The new scenario use case library is repeated in the above steps, and a new high-value scenario use case is obtained again.

[0055] In this embodiment, for the scenario use case with execution failure, such as failure of the autonomous driving algorithm, for example, vehicle collision, failure to generate expected events, etc., the autonomous driving algorithm will re-perform scenario training after fixing the related problems. The system can repeatedly iterate the scenario training process periodically, can find the characteristics of high-quality test cases, and can dynamically and continuously produce high-quality test cases. According to the execution result as feedback to continuously optimize the feature weight coefficient, it is predicted whether the newly generated use case is a high-value use case, the generation mode of the use case is continuously optimized and improved, and the purpose of producing high-quality use cases that can challenge the cloud autonomous driving algorithm is achieved.

[0056] Furthermore, the process of adjusting the scene feature parameters to obtain new scene use cases further includes: obtaining an accuracy evaluation of the feature evaluation model by comparing the actual weight ratio and the predicted weight ratio.

[0057] In practice, the autonomous driving algorithm is executed in the training scenario use case library to obtain a library of scenario use cases that failed to execute. The library of scenario use cases that failed to execute is then used to obtain the execution failure weights of scenario feature parameters through a logistic regression model. The predicted weight ratio and the actual weight ratio are compared to determine whether the comparison result exceeds a preset value. If it does not exceed the preset value, the accuracy of the feature evaluation model is high; if it does exceed the preset value, the accuracy of the feature evaluation model is low.

[0058] Example 2

[0059] Figure 2 This is a schematic diagram of a device structure for generating high-value use cases for autonomous driving, provided as an embodiment of this specification.

[0060] Please see Figure 2 As shown, Embodiment 2 provides an apparatus for generating high-value use cases for autonomous driving scenarios, comprising:

[0061] Scene module 301 is used to acquire scene feature parameters and construct a scene use case library based on the scene feature parameters;

[0062] The execution module 302 is used to run the scenario use case library using the autonomous driving algorithm to obtain the actual weight ratio of the scenario feature parameters. The weight ratio refers to the ratio of the weight of the scenario feature parameters that run successfully to the weight of the scenario feature parameters that fail to run.

[0063] Prediction module 303 predicts the weight ratio of the scene feature parameters and obtains the prediction weight ratio;

[0064] Model building module 304 is used to build a feature evaluation model based on the actual weight ratio and the predicted weight ratio;

[0065] The high-quality use case module 305 is used to optimize the scenario use case library through the feature evaluation model to obtain high-value scenario use cases;

[0066] The adjustment module is used to adjust the scene feature parameters to obtain a new scene use case library;

[0067] The loop module is used to repeat the above steps to obtain new high-value scenario use cases again from the new scenario use case library.

[0068] The scenario module 301 is also used to input the acquired scenario feature parameters into the scenario use case library to obtain a new scenario use case library, so that the new scenario use case library can be used for the operation of the autonomous driving algorithm.

[0069] The execution module 302 is further configured to execute the automatic driving algorithm in the scenario use case library, and obtain an execution failure scenario use case library; the execution failure scenario use case library is subjected to a logic regression model to obtain an execution failure weight of a scenario characteristic parameter; and the execution failure weight is used to obtain an actual weight ratio of the scenario characteristic parameter.

[0070] The model construction module 304 is further configured to construct a prediction model.

[0071] The adjustment module 306 is further configured to adjust the scenario characteristic parameter of the scenario use case library to a direction with a high execution failure weight by using the characteristic evaluation model, so as to obtain a high-value scenario use case.

[0072] The accuracy judgment module is configured to compare the actual weight ratio with the prediction weight ratio to obtain an accuracy evaluation of the characteristic evaluation model.

[0073] Embodiment 3

[0074] Embodiment 3 provides an electronic device, which comprises at least one processor and a memory, the memory stores a program and is configured to make the at least one processor execute the method for generating an automatic driving high-value scenario use case according to any one of the embodiments.

[0075] Embodiment 4

[0076] Embodiment 4 provides a computer readable storage medium, which stores computer instructions for making the computer execute the method for generating an automatic driving high-value scenario use case according to any one of the embodiments.

[0077] In the 1990s, it was relatively easy to distinguish whether an improvement in a technology was a hardware improvement (e.g., an improvement in the circuit structure of a diode, transistor, switch, etc.) or a software improvement (an improvement in a method flow). However, as technology has evolved, many improvements in method flows today can be considered as direct improvements in hardware circuit structures. Designers almost always obtain the corresponding hardware circuit structures by programming the improved method flows into hardware circuits. Therefore, it cannot be said that an improvement in a method flow cannot be implemented using hardware entity modules. For example, a programmable logic device (PLD) (e.g., a field programmable gate array (FPGA)) is an integrated circuit whose logic function is determined by user programming of the device. A digital system is "integrated" on a PLD by the designer programming the PLD, rather than by ordering a chip manufacturer to design and fabricate a custom integrated circuit chip. Moreover, instead of manually fabricating integrated circuit chips, this programming is now mostly implemented using "logic compiler" software, which is similar to software compilers used in program development, and the original code to be compiled is written in a specific programming language, which is called a hardware description language (HDL), and there are many such languages, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, RHDL (Ruby Hardware Description Language), etc., and the most commonly used are VHDL (Very-High-Speed Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should be aware that, as long as the method flow is logically programmed using the above-mentioned hardware description languages and programmed into an integrated circuit, a hardware circuit implementing the logical method flow can be easily obtained.

[0078] The controller can be implemented in any suitable way, for example, the controller can take the form of a microprocessor or processor and a computer readable medium storing computer readable program code, such as software or firmware, executable by the (micro)processor, logic gates, switches, an application specific integrated circuit (ASIC), a programmable logic controller and an embedded microcontroller, examples of which include but are not limited to the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20 and Silicone Labs C8051F320, the memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art will also know that, in addition to implementing the controller purely in computer readable program code, it is possible to implement the controller to perform the same functions in the form of logic gates, switches, application specific integrated circuits, programmable logic controllers and embedded microcontrollers, etc. by logically programming the method steps. Such a controller can therefore be considered to be a hardware component, and the means included within it for performing the various functions can also be considered to be structures within the hardware component. Alternatively, the means for performing the various functions can even be considered to be both a software module implementing the method and a structure within a hardware component.

[0079] The systems, apparatuses, modules or units illustrated by the above-described embodiments can be implemented by computer chips or entities, or by products with certain functions. Among them, a typical implementation device is a computer. Specifically, the computer can be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0080] For the convenience of description, the above apparatuses are described in various modules or units respectively according to functions. Of course, the functions of each module or unit can be implemented in the same or multiple software and / or hardware in the implementation of the present application.

[0081] Those skilled in the art will appreciate that embodiments of the present application can be provided as a method, a system, or a computer program product. Accordingly, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application can take the form of a computer program product on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage devices, etc.) embodying computer readable program code.

[0082] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more functions specified in the flowchart block or blocks. Figure 1 one or more functions specified in the flowchart block or blocks.

[0083] These computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture including instructions which implement the Figure 1 one or more functions specified in the flowchart block or blocks. Figure 1 one or more functions specified in the flowchart block or blocks.

[0084] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more functions specified in the flowchart block or blocks. Figure 1 one or more functions specified in the flowchart block or blocks.

[0085] In one typical configuration, the computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0086] The memory can include non-transitory memory in the form of a computer readable medium, random access memory (RAM), and / or non-volatile memory (NVM) such as read only memory (ROM) or flash memory. The memory is an example of a computer readable medium.

[0087] Computer-readable media includes permanent and non-permanent, movable and non-movable media that can be implemented by any method or technology to store information. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassette, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information accessible to a computing device. According to the definition herein, computer-readable media does not include transitory media such as modulated data signals and carriers.

[0088] It should also be noted that the terms "comprising", "containing", or any other variant thereof are intended to encompass non-exclusive inclusion, such that processes, methods, articles or devices that include a list of elements not only include those elements, but also include other elements not explicitly listed or inherent to such processes, methods, articles or devices. Without more limitations, the element defined by the statement "comprising a" does not exclude the presence of additional identical elements in the process, method, article or device including the element.

[0089] The present application can be described in the general context of computer-executable instructions, such as program modules, executed by computers. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. The present application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in both local and remote computer storage media, including storage devices.

[0090] Each embodiment in the specification is described in a progressive manner, and the same or similar parts between each embodiment can be referred to each other, and each embodiment focuses on the difference from other embodiments. In particular, for system embodiments, since they are basically similar to method embodiments, the description is relatively simple, and the relevant parts can be referred to the part of the method embodiment.

[0091] The above merely provides a description of the embodiments of the present application, but should not be taken as a limitation. Various changes and modifications can be made to the present application by those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall fall within the scope of claims of the present application.

Claims

1. A method for generating high-value use cases for autonomous driving scenarios, characterized in that, The method includes the following steps: Obtain scene feature parameters and construct a scene use case library based on the scene feature parameters; The scenario use case library is run using an autonomous driving algorithm to obtain the actual weight ratio of the scenario feature parameters. The weight ratio refers to the ratio of the weight of the scenario feature parameters that run successfully to the weight of the scenario feature parameters that fail to run. Predict the weight ratio of the scene feature parameters to obtain the predicted weight ratio; Based on the actual weight ratio and the predicted weight ratio, a feature evaluation model is constructed; The scenario use case library is optimized using the feature evaluation model to obtain high-value scenario use cases. The step of optimizing the scenario use case library using the feature evaluation model to obtain high-value scenario use cases includes: The feature evaluation model is used to adjust the scenario feature parameters of the scenario use case library in a direction with higher execution failure weights in order to obtain high-value scenario use cases.

2. The method for generating high-value use cases for autonomous driving according to claim 1, characterized in that, After obtaining the scene feature parameters, the process also includes: Adjust the scene feature parameters to obtain a new scene use case library; The new scenario use case library is run using an autonomous driving algorithm until the scenario use case library is optimized in a targeted manner through the feature evaluation model, thereby obtaining new high-value scenario use cases again.

3. The method for generating high-value use cases for autonomous driving according to claim 1, characterized in that, The step of constructing a scenario use case library based on the scenario feature parameters includes: The acquired scene feature parameters are input into the scene use case library to obtain a new scene use case library, which can then be used for the operation of the autonomous driving algorithm.

4. The method for generating high-value use cases for autonomous driving according to claim 1, characterized in that, The step of using the autonomous driving algorithm to run the scenario use case library and obtain the actual weight ratio of the scenario feature parameters includes: Execute the autonomous driving algorithm in the scenario use case library to obtain the scenario use case library that failed to execute; The failure weights of the scenario feature parameters are obtained by using a logistic regression model to analyze the library of scenario test cases that failed to execute. Based on the execution failure weight, the actual weight ratio of the scene feature parameters is obtained.

5. The method for generating high-value use cases for autonomous driving according to claim 1, characterized in that, The process of predicting the weight ratio of the scene feature parameters, and obtaining the prediction weight ratio, includes: Build a predictive model; The prediction weight ratio is obtained by predicting the weight ratio of the scene feature parameters using the prediction model.

6. The method for generating high-value use cases for autonomous driving according to claim 5, characterized in that, The construction of the prediction model includes fitting the decision boundary to the logistic regression model, establishing the decision boundary and combining it with probability, and then using the sigmoid function to construct the prediction model.

7. The method for generating high-value use cases for autonomous driving according to claim 1, characterized in that, The process of adjusting the scene feature parameters to obtain new scene use cases further includes: The accuracy evaluation of the feature evaluation model is obtained by comparing the actual weight ratio and the predicted weight ratio.

8. An apparatus for generating high-value use cases for autonomous driving scenarios, characterized in that, The apparatus is used to perform the method for generating high-value autonomous driving scenario use cases as described in claim 1, the apparatus comprising: The scenario module is used to acquire scenario feature parameters and construct a scenario use case library based on the scenario feature parameters. The execution module is used to run the scenario use case library using the autonomous driving algorithm to obtain the actual weight ratio of the scenario feature parameters. The weight ratio refers to the ratio of the weight of the scenario feature parameters that run successfully to the weight of the scenario feature parameters that fail to run. The prediction module is used to predict the weight ratio of the scene feature parameters and obtain the predicted weight ratio. The model building module is used to build a feature evaluation model based on the actual weight ratio and the predicted weight ratio; The high-quality use case module is used to optimize the scenario use case library through the feature evaluation model to obtain high-value scenario use cases; The adjustment module is used to adjust the scene feature parameters to obtain a new scene use case library; The loop module is used to repeat the above steps to obtain new high-value scenario use cases again from the new scenario use case library. The step of optimizing the scenario use case library using the feature evaluation model to obtain high-value scenario use cases includes: The feature evaluation model is used to adjust the scenario feature parameters of the scenario use case library in a direction with higher execution failure weights in order to obtain high-value scenario use cases.

9. An electronic device, characterized in that, It includes at least one processor and a memory, the memory storing a program and configured such that at least one processor executes a method for generating high-value use cases for autonomous driving as described in any one of claims 1-7.

Images