CN-122019379-A - Test scene library construction method, vehicle and storage medium

Abstract

The embodiment of the application provides a test scene library construction method, a vehicle and a storage medium, which comprise the steps of constructing a layered parameterized scene model based on a target test scene, wherein the layered parameterized scene model is used for dividing the target test scene into a functional layer, a logic layer and a specific layer, the functional layer is used for defining a functional module corresponding to the target test scene, the logic layer is used for determining scene element parameters corresponding to the target test scene, the specific layer is used for generating a test scene case based on the scene element parameters, acquiring historical parking data in the target test scene, wherein the historical parking data comprises multidimensional sensor data and network communication data when the vehicle performs parking operation, constructing a test scene library based on the layered parameterized scene model and the historical parking data, and performing scene test according to the test scene library to generate a scene test report. The method solves the technical problems of low construction efficiency and low scene coverage rate of the test scene library construction method provided in the related technology.

Inventors

• Zhang Juanhang

Assignees

• 安徽智界新能源汽车有限公司
• 奇瑞汽车股份有限公司

Dates

Publication Date

20260512

Application Date

20260127

Claims (10)

1. 1. The method for constructing the test scene library is characterized by comprising the following steps of: Constructing a layered parameterized scene model based on a target test scene, wherein the layered parameterized scene model is used for dividing the target test scene into a functional layer, a logic layer and a specific layer, the functional layer is used for defining a functional module corresponding to the target test scene, the logic layer is used for determining scene element parameters corresponding to the target test scene, and the specific layer is used for generating a test scene case based on the scene element parameters; Acquiring historical parking data in the target test scene, wherein the historical parking data comprises multidimensional sensor data and network communication data when a vehicle performs parking operation; Constructing the test scene library based on the hierarchical parameterized scene model and the historical parking data; and performing scene test according to the test scene library to generate a scene test report.
2. 2. The method of claim 1, wherein the constructing the test scene library based on the hierarchical parameterized scene model and the historical parking data comprises: Analyzing and processing the historical parking data to obtain an analysis result, wherein the analysis result comprises a plurality of clusters and outliers, and the clusters and the outliers are used for determining scene types corresponding to the historical parking data; And constructing the test scene library based on the hierarchical parameterized scene model and the analysis result.
3. 3. The method of claim 2, wherein the analyzing the historical parking data to obtain an analysis result comprises: preprocessing the historical parking data to obtain target parking data, wherein the target parking data is used for representing standardized parking data in the target test scene; Extracting features of the target parking data to obtain event feature parameters corresponding to the functional module; And carrying out cluster analysis on the event characteristic parameters to obtain the plurality of clusters and the outliers.
4. 4. The method of claim 3, wherein the constructing the test scene library based on the hierarchical parameterized scene model and the analysis results comprises: Determining a target cluster in the plurality of clusters based on a first preset threshold, wherein the frequency corresponding to the target cluster is greater than the first preset threshold; and constructing the test scene library based on the hierarchical parameterized scene model, the target cluster and the outlier.
5. 5. The method of claim 4, wherein constructing the test scene library based on the hierarchical parameterized scene model, the target cluster, and the outliers comprises: Performing generalization processing on the target cluster based on the hierarchical parameterized scene model to obtain a logic scene, wherein the logic scene comprises a static element parameter set corresponding to the target cluster and a dynamic element parameter set corresponding to the target cluster; performing generalization processing on the outliers based on the hierarchical parameterized scene model to obtain long-tail scenes, wherein the frequencies corresponding to the long-tail scenes are smaller than a second preset threshold value, and the second preset threshold value is smaller than the first preset threshold value; and constructing the test scene library based on the logic scene and the long tail scene.
6. 6. The method of claim 5, wherein the scene element parameters comprise a static element parameter set and a dynamic element parameter set, the logical scene further comprising a first target scene, the method further comprising: Constructing an evaluation function by using the event characteristic parameters and the preset offset, wherein the evaluation function is used for determining a risk level corresponding to the logic scene, and the risk level is in direct proportion to an evaluation function value corresponding to the evaluation function; And determining the first target scene based on the evaluation function, the static element parameter set corresponding to the target cluster and the dynamic element parameter set corresponding to the target cluster, wherein the risk level corresponding to the first target scene meets a first preset condition.
7. 7. The method of claim 6, wherein the determining the first target scene based on the evaluation function, the static element parameter set corresponding to the target cluster, and the dynamic element parameter set corresponding to the target cluster comprises: Generating a plurality of parameter combinations based on the static element parameter set corresponding to the target cluster and the dynamic element parameter set corresponding to the target cluster; generating a plurality of test scene cases according to the plurality of parameter combinations, and performing scene test based on the plurality of test scene cases to obtain event characteristic parameters corresponding to the plurality of test scene cases; determining a plurality of initial parameter combinations based on event characteristic parameters and the evaluation functions corresponding to the plurality of test scene use cases, wherein a plurality of initial evaluation function values corresponding to the plurality of initial parameter combinations are larger than a third preset threshold; The first target scene is determined based on the plurality of initial parameter combinations.
8. 8. The method of claim 7, wherein the determining the first target scene based on the plurality of initial parameter combinations comprises: performing cross variation on the plurality of initial parameter combinations to obtain a plurality of parameter combinations; Selecting a target parameter combination from the plurality of parameter combinations based on the evaluation function, wherein an evaluation function value corresponding to the target parameter combination meets a second preset condition; The first target scene is determined based on the target parameter combination.
9. 9. A vehicle, characterized by comprising: A memory storing an executable program; A processor for executing the program, wherein the program when run performs the method of any one of claims 1 to 8.
10. 10. A computer readable storage medium, characterized in that the computer readable storage medium comprises a stored executable program, wherein the executable program when run controls a device in which the storage medium is located to perform the method of any one of claims 1 to 8.

Description

Test scene library construction method, vehicle and storage medium Technical Field The embodiment of the application relates to the technical field of intelligent driving test of vehicles, in particular to a test scene library construction method, a vehicle and a storage medium. Background In the related art, the test scene construction is generally performed in a manner depending on empirical data and manual design. However, the above-described method in the related art has a technical problem of low test scene construction efficiency and scene coverage, and thus it is difficult to comprehensively evaluate the performance of the automatic parking (Automated Parking, AP) system in a complex and diverse real environment. There is currently no good solution to the above problems. Disclosure of Invention The embodiment of the application provides a test scene library construction method, a vehicle and a storage medium, which are used for at least solving the technical problems of low construction efficiency and low scene coverage rate of the test scene library construction method provided in the related technology. According to one aspect of the embodiment of the application, a test scene library construction method is provided, which comprises the steps of constructing a layered parameterized scene model based on a target test scene, wherein the layered parameterized scene model is used for dividing the target test scene into a functional layer, a logic layer and a specific layer, the functional layer is used for defining a functional module corresponding to the target test scene, the logic layer is used for determining scene element parameters corresponding to the target test scene, the specific layer is used for generating a test scene use case based on the scene element parameters, acquiring historical parking data in the target test scene, wherein the historical parking data comprises multidimensional sensor data and network communication data when a vehicle performs parking operation, constructing a test scene library based on the layered parameterized scene model and the historical parking data, and performing scene test according to the test scene library to generate a scene test report. Further, constructing the test scene library based on the hierarchical parameterized scene model and the historical parking data comprises the steps of analyzing and processing the historical parking data to obtain an analysis result, wherein the analysis result comprises a plurality of clusters and outliers, the clusters and outliers are used for determining scene types corresponding to the historical parking data, and constructing the test scene library based on the hierarchical parameterized scene model and the analysis result. Further, the historical parking data is analyzed and processed to obtain an analysis result, wherein the analysis result comprises preprocessing of the historical parking data to obtain target parking data, the target parking data is used for representing standardized parking data in a target test scene, feature extraction of the target parking data is conducted to obtain event feature parameters corresponding to the functional module, and cluster analysis is conducted on the event feature parameters to obtain a plurality of clusters and outliers. Further, constructing the test scene library based on the hierarchical parameterized scene model and the analysis result comprises determining a target cluster in the plurality of clusters based on a first preset threshold, wherein the frequency corresponding to the target cluster is greater than the first preset threshold, and constructing the test scene library based on the hierarchical parameterized scene model, the target cluster and the outlier. The method comprises the steps of establishing a test scene library, wherein the test scene library comprises a hierarchical parameterized scene model, target clusters and outliers, performing generalization processing on the target clusters based on the hierarchical parameterized scene model to obtain a logic scene, the logic scene comprises a static element parameter set corresponding to the target clusters and a dynamic element parameter set corresponding to the target clusters, performing generalization processing on the outliers based on the hierarchical parameterized scene model to obtain long-tail scenes, the frequency corresponding to the long-tail scenes is smaller than a second preset threshold value, the second preset threshold value is smaller than the first preset threshold value, and the test scene library is established based on the logic scene and the long-tail scenes. Further, the scene element parameters comprise a static element parameter set and a dynamic element parameter set, the logic scene further comprises a first target scene, and the test scene library construction method in the embodiment of the application further comprises the steps of constructing an evaluation function by