Traffic flow simulation scene extraction method and device, electronic equipment and storage medium

Abstract

The invention provides a traffic flow simulation scene extraction method and device, electronic equipment and a storage medium. The method comprises the following steps: extracting static scene elements and dynamic scene elements from a data source after data preprocessing; performing data fusion on the static scene elements and the dynamic scene elements to generate a plurality of structured reference simulation scenes; performing weight analysis processing and state probability distribution processing on each scene element in each reference simulation scene by adopting an analytic hierarchy process and an information entropy theory, and determining scene complexity of the reference simulation scene; and screening out a high-complexity simulation scene based on the scene complexity, and outputting a target simulation test scene set for testing the automatic driving system. Information of various data sources can be fully utilized, and the authenticity and complexity of a simulation scene are improved. Meanwhile, through the analytic hierarchy process and the information entropy theory, the characteristics of the actual traffic environment can be reflected more accurately, and more effective support is provided for testing and verification of the automatic driving automobile.

Description

Technical Field

[0001] This application relates to the field of simulation scene construction technology, and in particular to methods, apparatus, electronic devices and storage media for extracting traffic flow reference simulation scenes. Background Technology

[0002] With the rapid development of autonomous driving technology, how to efficiently and safely verify its reliability in complex traffic environments has become a core issue of concern for both industry and academia. Simulation testing, as a low-cost, repeatable, and efficient alternative, has been widely adopted. Existing simulation platforms (such as CARLA, PreScan, and VTD) support the construction of virtual traffic environments and perform closed-loop verification of autonomous driving algorithms through software-in-the-loop (SIL) and hardware-in-the-loop (HIL) methods. However, current mainstream methods for generating traffic flow simulation scenarios still have significant limitations: most solutions extract scenario elements based solely on natural driving data or standard regulatory data, resulting in insufficient representativeness of the generated test scenarios and difficulty in covering extreme or critical conditions; the selection of key scenarios relies heavily on engineers' experience and lacks a quantitative evaluation mechanism, leading to low efficiency in identifying high-value test cases. Therefore, improving the realism and complexity of traffic flow simulation scenarios has become a significant technical challenge. Summary of the Invention

[0003] In view of this, the purpose of this application is to provide a method, apparatus, electronic device, and storage medium for extracting traffic flow simulation scenarios, which enables full utilization of information from multiple data sources and improves the realism and complexity of the simulation scenarios. Furthermore, through the Analytic Hierarchy Process (AHP) and information entropy theory, it can more accurately reflect the characteristics of the actual traffic environment, providing more effective support for the testing and verification of autonomous vehicles.

[0004] This application provides a method for extracting traffic flow simulation scenarios, the extraction method including: Multiple types of data sources are acquired and preprocessed to extract static and dynamic scene elements from the traffic flow simulation scenario from the preprocessed data sources. Data fusion is performed on the static scene elements and the dynamic scene elements from different data sources to generate multiple structured reference simulation scenes; The analytic hierarchy process (AHP) and information entropy theory are used to perform weight analysis and state probability distribution processing on each scene element in each of the reference simulation scenarios to determine the scene complexity of each of the reference simulation scenarios. Based on the complexity of the scenarios, high-complexity simulation scenarios are selected, and a set of target simulation test scenarios for testing autonomous driving systems is output.

[0005] In one possible implementation, extracting static and dynamic scene elements from the traffic flow simulation scenario from the data source after data preprocessing includes: The road feature analysis, natural environment condition analysis, and static obstacle analysis are performed on multiple traffic flow simulation scenarios in the data source after data preprocessing to determine the static scene elements. Collision scene identification processing, traffic flow feature analysis processing, traffic participant behavior analysis processing, and intersection infrastructure analysis processing are performed on multiple traffic flow reference simulation scenarios to determine the dynamic scene elements.

[0006] In one possible implementation, the data fusion of the static scene elements and the dynamic scene elements from different data sources to generate multiple structured reference simulation scenes includes: Spatiotemporal alignment and semantic unification processing are performed on the static scene elements and dynamic scene elements from different data sources to determine the processed static scene elements and the processed dynamic scene elements. Based on a neural network model, the processed static scene elements and the processed dynamic scene elements are mapped to a unified feature space to generate a joint feature vector. The joint feature vector is then subjected to feature fusion processing to generate multiple structured reference simulation scenes.

[0007] In one possible implementation, the step of using the analytic hierarchy process (AHP) and information entropy theory to perform weight analysis and state probability distribution processing on each scene element in each reference simulation scene to determine the scene complexity of each reference simulation scene includes: The analytic hierarchy process is used to perform weight analysis on each scene element in each of the reference simulation scenarios to determine the weight value of each scene element in each of the reference simulation scenarios. For any of the aforementioned reference simulation scenarios, the weight values ​​of each scenario element in the reference simulation scenario are used as prior knowledge to assist the information entropy theory in calculating the state probability distribution of each scenario element. The state probability distributions are then summarized and processed to determine the scenario complexity of the reference simulation scenario.

[0008] In one possible implementation, for any of the aforementioned reference simulation scenarios, the step of using the analytic hierarchy process (AHP) to perform weight analysis on each scene element in each reference simulation scenario to determine the weight value of each scene element in each reference simulation scenario includes: A hierarchical model is constructed, comprising a target layer, a criterion layer, and an indicator layer; wherein, the target layer is the decision objective of the reference simulation scenario, the criterion layer includes various scenario elements, and the indicator layer consists of scenario variables for each scenario element; A judgment matrix is ​​constructed by comparing the importance of multiple scene elements in each level; The judgment matrix is ​​column normalized, and the arithmetic mean of each row of elements in the normalized judgment matrix is ​​calculated to determine the weight value of each scene element. The judgment matrix is ​​subjected to a consistency check. If the judgment matrix passes the consistency check, the judgment matrix is ​​not adjusted. If the judgment matrix fails the consistency test, the judgment matrix is ​​adjusted and the weight values ​​of each scene element in the judgment matrix are redefined.

[0009] In one possible implementation, the consistency check of the judgment matrix includes: Based on the maximum weight value and the matrix order in the judgment matrix, the target consistency index value of the judgment matrix is ​​determined. The consistency ratio is determined based on the ratio between the target consistency index value and the random consistency index value corresponding to the matrix order. If the consistency ratio is less than or equal to a preset threshold, the judgment matrix passes the consistency test; if the consistency ratio is greater than a preset age threshold, the judgment matrix fails the consistency test.

[0010] This application embodiment also provides a traffic flow simulation scene extraction device, the extraction device comprising: The data source processing module is used to acquire multiple types of data sources, perform data preprocessing on multiple data sources, and extract static scene elements and dynamic scene elements in the traffic flow simulation scenario from the data sources after data preprocessing. The fusion module is used to fuse the static scene elements and the dynamic scene elements from different data sources to generate multiple structured reference simulation scenes. The analysis module is used to perform weight analysis and state probability distribution processing on each scene element in each of the reference simulation scenarios using the analytic hierarchy process and information entropy theory, so as to determine the scene complexity of each of the reference simulation scenarios. The filtering module is used to filter out high-complexity simulation scenarios based on the scenario complexity and output a set of target simulation test scenarios for testing autonomous driving systems.

[0011] In one possible implementation, the data source processing module is used to extract static and dynamic scene elements from the traffic flow reference simulation scenario in the data source after data preprocessing: The road feature analysis, natural environment condition analysis, and static obstacle analysis are performed on multiple traffic flow reference simulation scenarios in the data source after data preprocessing to determine the static scene elements. Collision scene identification processing, traffic flow feature analysis processing, traffic participant behavior analysis processing, and intersection infrastructure analysis processing are performed on multiple traffic flow reference simulation scenarios to determine the dynamic scene elements.

[0012] This application embodiment also provides an electronic device, including: a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the electronic device is running, the processor communicates with the memory via the bus. When the machine-readable instructions are executed by the processor, the steps of the traffic flow simulation scenario extraction method described above are performed.

[0013] This application also provides a computer-readable storage medium storing a computer program, which, when run by a processor, performs the steps of the traffic flow simulation scenario extraction method described above.

[0014] This application provides a method, apparatus, electronic device, and storage medium for extracting traffic flow simulation scenarios. The extraction method includes: acquiring multiple types of data sources and preprocessing the data from these data sources; extracting static and dynamic scene elements from the traffic flow simulation scenario within the preprocessed data sources; fusing the static and dynamic scene elements from different data sources to generate multiple structured reference simulation scenarios; using the analytic hierarchy process (AHP) and information entropy theory to perform weight analysis and state probability distribution processing on each scene element in each reference simulation scenario to determine the scenario complexity of each reference simulation scenario; and filtering out high-complexity simulation scenarios based on the scenario complexity to output a target simulation test scenario set for testing autonomous driving systems. This approach fully utilizes information from multiple data sources, improving the realism and complexity of the simulation scenarios. Furthermore, the AHP and information entropy theory more accurately reflect the characteristics of the actual traffic environment, providing more effective support for the testing and verification of autonomous vehicles.

[0015] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is one of the flowcharts for a traffic flow simulation scene extraction method provided in an embodiment of this application; Figure 2 A second flowchart illustrating a method for extracting a traffic flow simulation scenario provided in an embodiment of this application; Figure 3 A schematic diagram of the structure of a traffic flow simulation scene extraction device provided in an embodiment of this application; Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. Based on the embodiments of this application, every other embodiment obtained by those skilled in the art without inventive effort falls within the scope of protection of this application.

[0019] First, the applicable application scenarios of this application will be introduced. This application can be applied to the field of simulation scene construction technology.

[0020] Research has revealed that with the rapid development of autonomous driving technology, how to efficiently and safely verify its reliability in complex traffic environments has become a core issue of concern for both industry and academia. Simulation testing, as a low-cost, repeatable, and highly efficient alternative, is widely adopted. Existing simulation platforms (such as CARLA, PreScan, and VTD) support the construction of virtual traffic environments and perform closed-loop verification of autonomous driving algorithms through software-in-the-loop (SIL) and hardware-in-the-loop (HIL) methods. However, current mainstream methods for generating traffic flow simulation scenarios still have significant limitations: most solutions extract scenario elements based solely on natural driving data or standard regulatory data, resulting in insufficient representativeness of the generated test scenarios and difficulty in covering extreme or critical conditions; the selection of key scenarios relies heavily on engineers' experience and lacks a quantitative evaluation mechanism, leading to low efficiency in identifying high-value test cases. Therefore, improving the realism and complexity of traffic flow simulation scenarios has become a significant technical challenge.

[0021] Based on this, embodiments of this application provide a method for extracting traffic flow simulation scenarios, which can fully utilize information from multiple data sources to improve the realism and complexity of the simulation scenarios. Simultaneously, through the Analytic Hierarchy Process (AHP) and information entropy theory, it can more accurately reflect the characteristics of the actual traffic environment, providing more effective support for the testing and verification of autonomous vehicles.

[0022] Please see Figure 1 , Figure 1 This is one of the flowcharts for a traffic flow simulation scene extraction method provided in an embodiment of this application. For example... Figure 1 As shown in the embodiments of this application, the extraction method includes: S101. Obtain multiple types of data sources and perform data preprocessing on multiple data sources. Extract static scene elements and dynamic scene elements in the traffic flow simulation scenario from the data sources after data preprocessing.

[0023] In this step, multiple types of data sources are acquired and preprocessed. Static and dynamic scene elements in the traffic flow simulation scenario are then extracted from the preprocessed data sources.

[0024] It should be noted that static scene elements include road features, natural environmental factors such as weather conditions and lighting conditions, and static obstacles such as guardrails, stationary people, and stationary vehicles. Dynamic scene elements include collisions between straight-going and left-turning vehicles, and between left-turning and right-turning vehicles, as well as traffic flow characteristics such as density, speed, and direction.

[0025] Here, various types of data sources include natural driving scenarios (recording vehicle trajectory, speed, and other information in real driving environments), hazardous working condition scenarios (collecting data under conditions such as traffic accidents and emergency braking), standard and regulatory scenarios (referencing traffic regulations, road design specifications, and other standards), and actual demand scenarios (side collisions, left turn conflicts, simultaneous turning conflicts, and U-turn conflicts).

[0026] The methods for acquiring data in natural driving scenarios include: deploying high-tech sensing devices (such as GPS, LiDAR scanners, and high-definition cameras) in real-world road layouts to capture multi-dimensional information such as vehicle operating status, driving path, and surrounding traffic environment in real time. This data is of great value for in-depth analysis of driving habits and traffic flow patterns. The methods for acquiring data in hazardous driving scenarios involve directly using publicly available datasets, such as traffic accident statistics and traffic flow monitoring records, as research and analysis materials. This approach effectively reduces the economic and time costs of data collection while fully utilizing existing information resources, providing solid data support for research. The methods for acquiring standards and regulations involve collecting and analyzing legal provisions and regulations related to traffic management worldwide. This information is crucial for the design, optimization, and compliance assessment of autonomous driving technology. The method for obtaining data from actual demand scenarios is as follows: understand the frequency and degree of danger of different accidents, locate high-risk areas and key issues; then analyze the characteristics of traffic flow density, speed, direction and other features to clarify the traffic and pressure conditions at intersections; at the same time, explore the impact of intersection layout (geometry, lanes, etc.) and traffic signs (setting, visibility) on collision scenarios.

[0027] In one possible implementation, extracting static and dynamic scene elements from the traffic flow simulation scenario from the data source after data preprocessing includes: A: After data preprocessing, multiple traffic flow simulation scenarios from the data source are subjected to road feature analysis, natural environment condition analysis, and static obstacle analysis to determine the static scene elements.

[0028] Here, road feature analysis, natural environment condition analysis, and static obstacle analysis are performed on multiple traffic flow simulation scenarios in the data source after data preprocessing to determine the static scene elements.

[0029] Among these, the analysis of static elements is particularly important when constructing realistic and challenging autonomous driving simulation test scenarios. A thorough analysis of static elements is crucial. Road features constitute the basic framework of the simulation scenario; the number of lanes, road surface materials, and road surface conditions all need careful consideration to ensure the comprehensiveness and practicality of the test environment. Furthermore, natural environmental factors, such as weather conditions and lighting conditions, also significantly impact the perception accuracy and driving safety of autonomous vehicles and must be fully reflected in the simulation scenario. Finally, static obstacles, such as guardrails, stationary people, and stationary vehicles, also have a significant impact on the realistic construction of the autonomous driving simulation scenario.

[0030] B: Collision scene identification processing, traffic flow feature analysis processing, traffic participant behavior analysis processing, and intersection infrastructure analysis processing are performed on multiple traffic flow reference simulation scenarios to determine the dynamic scene elements.

[0031] Here, collision scene recognition processing, traffic flow feature analysis processing, traffic participant behavior analysis processing, and intersection infrastructure analysis processing are performed on multiple traffic flow reference simulation scenarios to determine dynamic scene elements.

[0032] Regarding dynamic element analysis, analyzing typical collision scenarios at intersections primarily involves identifying accident types, such as collisions between straight-going and left-turning vehicles, or left-turning and right-turning vehicles, and understanding the frequency and hazard level of each type to determine high-incidence areas and root causes. Next, traffic flow characteristics, such as density, speed, and direction, need to be determined to understand the traffic conditions and pressure at the intersection. Furthermore, intersection layout and traffic signs cannot be ignored, including intersection shape, lane configuration, pedestrian crossing facilities, and the placement and visibility of signs, all of which influence collision scenarios.

[0033] S102: Perform data fusion on the static scene elements and the dynamic scene elements from different data sources to generate multiple structured reference simulation scenes.

[0034] In this step, static and dynamic scene elements from different data sources are fused to generate multiple structured reference simulation scenarios.

[0035] In this application, by integrating data sources from different channels (including real driving trajectories, traffic accident records, traffic regulations, etc.), the simulation comprehensively covers normal driving, edge cases, and dangerous working conditions, significantly improving the ability of the simulation scenario to reproduce the real traffic environment and solving the "test blind spot" problem caused by the single data in traditional methods.

[0036] In one possible implementation, the data fusion of the static scene elements and the dynamic scene elements from different data sources to generate multiple structured reference simulation scenes includes: (1): Spatiotemporal alignment and semantic unification processing are performed on the static scene elements and dynamic scene elements from different data sources to determine the processed static scene elements and the processed dynamic scene elements.

[0037] Here, spatiotemporal alignment includes time synchronization and spatial registration. Time synchronization: Interpolation methods (linear or spline interpolation) are used to align low-frequency data (such as accident records) with high-frequency data (such as vehicle trajectories) to a unified time series. Spatial registration: A high-precision map is used as a reference frame to project all data into the same local Cartesian coordinate system (such as UTM coordinates), and the ICP (Iterative Closest Point) algorithm is used to optimize matching errors. This eliminates information bias caused by spatiotemporal misalignment and provides a reliable foundation for subsequent fusion.

[0038] (2): Based on the neural network model, the processed static scene elements and the processed dynamic scene elements are mapped to a unified feature space to generate a joint feature vector. The joint feature vector is then subjected to feature fusion processing to generate multiple structured reference simulation scenes.

[0039] Here, based on the neural network model, the processed static scene elements and the processed dynamic scene elements are mapped to a unified feature space to generate a joint feature vector. The joint feature vector is then subjected to feature fusion processing to generate multiple structured reference simulation scenes.

[0040] S103: Using the analytic hierarchy process (AHP) and information entropy theory, weight analysis and state probability distribution processing are performed on each scene element in each of the reference simulation scenarios to determine the scene complexity of each of the reference simulation scenarios.

[0041] In this step, the Analytic Hierarchy Process (AHP) and information entropy theory are used to perform weight analysis and state probability distribution processing on each scene element in each reference simulation scene, thereby determining the scene complexity of each reference simulation scene.

[0042] In this application, the hierarchical structure model containing dimensions such as road infrastructure, traffic signals, environmental conditions, and traffic participant behavior is constructed using the analytic hierarchy process. By constructing a judgment matrix and performing a consistency check, the relative weights of each scenario element are obtained, providing an interpretable and verifiable quantitative basis for subsequent scenario priority ranking.

[0043] For further details, please refer to Figure 2 , Figure 2This is a second flowchart illustrating a method for extracting traffic flow simulation scenarios provided in an embodiment of this application. Figure 2 As shown: S201: The analytic hierarchy process is used to perform weight analysis on each scene element in each of the reference simulation scenarios to determine the weight value of each scene element in each of the reference simulation scenarios.

[0044] In this step, the Analytic Hierarchy Process (AHP) is used to perform weight analysis on each scene element in each reference simulation scenario to determine the weight value of each scene element in each reference simulation scenario.

[0045] In one possible implementation, for any of the aforementioned reference simulation scenarios, the step of using the analytic hierarchy process (AHP) to perform weight analysis on each scene element in each reference simulation scenario to determine the weight value of each scene element in each reference simulation scenario includes: i: Construct a hierarchical model comprising a target layer, a criterion layer, and an indicator layer; wherein the target layer is the decision objective of the reference simulation scenario, the criterion layer includes various scenario elements, and the indicator layer comprises scenario variables for each scenario element.

[0046] Here, the core function of the analytic hierarchy process (AHP) is to calculate the weight value of each element by comparing the importance of various scenario elements such as road infrastructure, traffic signals and signs, environmental conditions, traffic participant behavior, and traffic flow characteristics, thus providing data support for the selection and optimization of subsequent simulation scenarios.

[0047] The system is divided into three layers: TopLevel (target layer): "Identifying the key elements of the urban intersection simulation scenario"; CriteriaLevel (criteria layer): including several main dimensions affecting the risk and challenge of the scenario, such as: C1: Road infrastructure (number of lanes, road surface conditions, etc.); C2: Traffic signals and signs (traffic light timing, road marking clarity, etc.); C3: Environmental conditions (weather, lighting); C4: Traffic participant behavior (pedestrian crossing, vehicle cutting in, etc.); and C5: Traffic flow characteristics (density, speed distribution). The Index / Sub-criteriaLevel (index / sub-criteria layer): further subdivided specific indicators under each criterion.

[0048] ii: Construct a judgment matrix by comparing the importance of multiple scene elements in each level; perform column normalization on the judgment matrix, and calculate the arithmetic mean of the elements in each row of the normalized judgment matrix to determine the weight value of each scene element.

[0049] Here, for a specific layer, when comparing the importance of the i-th scene element and the j-th scene element relative to a certain factor in the previous layer, a quantitative relative importance aij is used. Assuming there are n scene elements involved in the comparison, the judgment matrix... A for:

[0050] Where, if a ij A score of 1 indicates equal importance, 3 indicates slightly important importance, and 5 indicates significant importance.

[0051] Here, the judgment matrix is ​​normalized, and the weights of each evaluation index are calculated. Normalization involves dividing each column element of the judgment matrix by the sum of its elements to obtain a normalized matrix. Calculating the weights involves averaging the elements in each row of the normalized matrix to obtain the weight value of each scene element.

[0052] ii: Perform a consistency check on the judgment matrix. If the judgment matrix passes the consistency check, the judgment matrix is ​​not adjusted. If the judgment matrix fails the consistency check, the judgment matrix is ​​adjusted and the weight values ​​of each scene element in the judgment matrix are re-determined.

[0053] Here, a consistency check is performed on the judgment matrix. If the judgment matrix passes the consistency check, it is not adjusted; if the judgment matrix fails the consistency check, it is adjusted and the weight values ​​of each scene element in the judgment matrix are re-determined.

[0054] In one possible implementation, the consistency check of the judgment matrix includes: Based on the maximum weight value and the matrix order in the judgment matrix, the target consistency index value of the judgment matrix is ​​determined; based on the ratio between the target consistency index value and the random consistency index value corresponding to the matrix order, the consistency ratio is determined; if the consistency ratio is less than or equal to a preset threshold, the judgment matrix passes the consistency test; if the consistency ratio is greater than a preset threshold, the judgment matrix fails the consistency test.

[0055] Here, the target consistency index is determined using the following formula. CI :

[0056] in, The maximum weight value, n is the order of the matrix.

[0057] Here, the random consistency index is obtained from the matrix order table. If the consistency ratio is less than or equal to 0.1, the consistency of the judgment matrix is ​​acceptable; otherwise, the judgment matrix needs to be adjusted.

[0058] In this application, the Analytic Hierarchy Process (AHP) is used to perform weight analysis on key elements of traffic flow simulation scenarios. First, a hierarchical model is established, comprising a target layer, a criterion layer, and an indicator layer. Then, a judgment matrix is ​​constructed using expert scoring or historical data analysis. Next, the weight vector of each element is calculated using column normalization and row averaging. Finally, a consistency check is performed to ensure the rationality of the judgment logic. The obtained weight results are used to guide the selection of high-value test scenarios and can serve as a priori basis for state probabilities in information entropy calculation, improving the rationality of complexity assessment. S202: For any of the aforementioned reference simulation scenarios, the weight values ​​of each scenario element in the reference simulation scenario are used as prior knowledge to assist the information entropy theory in calculating the state probability distribution of each scenario element. The state probability distributions are then summarized and processed to determine the scenario complexity of the reference simulation scenario.

[0059] In this step, for any reference simulation scenario, the weight values ​​of each scenario element in the reference simulation scenario are used as prior knowledge to assist the information entropy theory in calculating the state probability distribution of each scenario element. The state probability distributions are then summarized to determine the scenario complexity of the reference simulation scenario.

[0060] Here, the information entropy theory is used to quantify the complexity of a scenario, and its calculation formula is as follows:

[0061] in, ) is the first The state probability distribution of each scene element is obtained from historical statistical data or expert experience; the higher the entropy value, the greater the uncertainty of the scene and the stronger the challenge.

[0062] S104: Based on the complexity of the scenario, select high-complexity simulation scenarios and output a set of target simulation test scenarios for testing autonomous driving systems.

[0063] In this step, high-complexity simulation scenarios are selected based on scenario complexity, and a set of target simulation test scenarios for testing autonomous driving systems is output.

[0064] In this application, an improvement scheme is proposed to address the following technical problems existing in traffic flow simulation scene extraction methods: 1. The problem of narrow coverage and poor representativeness of test scenes due to reliance on a single data source; 2. The problem of spatiotemporal inconsistency and semantic inconsistency among multiple data sources, making it difficult to effectively integrate them; 3. The problem of low efficiency in identifying high-risk scenes due to the lack of quantitative basis for assessing the importance of scene elements; 4. The problem of the difficulty in objectively representing the complexity of simulation scenes, affecting the controllable adjustment of the test challenge.

[0065] This application provides a method for extracting traffic flow simulation scenarios. The method includes: acquiring multiple types of data sources and preprocessing the data from these data sources; extracting static and dynamic scene elements from the traffic flow simulation scenario in the preprocessed data sources; fusing the static and dynamic scene elements from different data sources to generate multiple structured reference simulation scenarios; using the analytic hierarchy process (AHP) and information entropy theory to perform weight analysis and state probability distribution processing on each scene element in each reference simulation scenario to determine the scenario complexity of each reference simulation scenario; and filtering out high-complexity simulation scenarios based on the scenario complexity to output a target simulation test scenario set for testing autonomous driving systems. This method fully utilizes information from multiple data sources, improving the realism and complexity of the simulation scenario. Furthermore, by employing the AHP and information entropy theory, it can more accurately reflect the characteristics of the actual traffic environment, providing more effective support for the testing and verification of autonomous vehicles.

[0066] Please see Figure 3 , Figure 3 This is a schematic diagram of a traffic flow simulation scene extraction device provided in an embodiment of this application. Figure 3 As shown, the extraction device 300 includes: Data source processing module 310 is used to acquire multiple types of data sources, perform data preprocessing on multiple data sources, and extract static scene elements and dynamic scene elements in the traffic flow simulation scenario from the data sources after data preprocessing. The fusion module 320 is used to perform data fusion on the static scene elements and the dynamic scene elements from different data sources to generate multiple structured reference simulation scenes. Analysis module 330 is used to perform weight analysis and state probability distribution processing on each scene element in each of the reference simulation scenarios using the analytic hierarchy process and information entropy theory, so as to determine the scene complexity of each of the reference simulation scenarios. The filtering module 340 is used to filter out high-complexity simulation scenarios based on the scenario complexity and output a set of target simulation test scenarios for testing autonomous driving systems.

[0067] Furthermore, the data source processing module 310 is used to extract static scene elements and dynamic scene elements from the traffic flow reference simulation scenario in the data source after data preprocessing: The road feature analysis, natural environment condition analysis, and static obstacle analysis are performed on multiple traffic flow reference simulation scenarios in the data source after data preprocessing to determine the static scene elements. Collision scene identification processing, traffic flow feature analysis processing, traffic participant behavior analysis processing, and intersection infrastructure analysis processing are performed on multiple traffic flow reference simulation scenarios to determine the dynamic scene elements.

[0068] Furthermore, the fusion module 320 is used to perform data fusion on the static scene elements and the dynamic scene elements from different data sources to generate multiple structured reference simulation scenes: Spatiotemporal alignment and semantic unification processing are performed on the static scene elements and dynamic scene elements from different data sources to determine the processed static scene elements and the processed dynamic scene elements. Based on a neural network model, the processed static scene elements and the processed dynamic scene elements are mapped to a unified feature space to generate a joint feature vector. The joint feature vector is then subjected to feature fusion processing to generate multiple structured reference simulation scenes.

[0069] Furthermore, the analysis module 330 is used to perform weight analysis and state probability distribution processing on each scene element in each of the reference simulation scenarios using the analytic hierarchy process and information entropy theory, to determine the scene complexity of each of the reference simulation scenarios: The analytic hierarchy process is used to perform weight analysis on each scene element in each of the reference simulation scenarios to determine the weight value of each scene element in each of the reference simulation scenarios. For any of the aforementioned reference simulation scenarios, the weight values ​​of each scenario element in the reference simulation scenario are used as prior knowledge to assist the information entropy theory in calculating the state probability distribution of each scenario element. The state probability distributions are then summarized and processed to determine the scenario complexity of the reference simulation scenario.

[0070] Furthermore, the analysis module 330 is used to perform weight analysis on each scene element in each of the reference simulation scenarios using the analytic hierarchy process (AHP) to determine the weight value of each scene element in each of the reference simulation scenarios: A hierarchical model is constructed, comprising a target layer, a criterion layer, and an indicator layer; wherein, the target layer is the decision objective of the reference simulation scenario, the criterion layer includes various scenario elements, and the indicator layer consists of scenario variables for each scenario element; A judgment matrix is ​​constructed by comparing the importance of multiple scene elements in each level; The judgment matrix is ​​column normalized, and the arithmetic mean of each row of elements in the normalized judgment matrix is ​​calculated to determine the weight value of each scene element. The judgment matrix is ​​subjected to a consistency check. If the judgment matrix passes the consistency check, the judgment matrix is ​​not adjusted. If the judgment matrix fails the consistency test, the judgment matrix is ​​adjusted and the weight values ​​of each scene element in the judgment matrix are redefined.

[0071] Furthermore, the analysis module 330 is used to perform a consistency check on the judgment matrix: Based on the maximum weight value and the matrix order in the judgment matrix, the target consistency index value of the judgment matrix is ​​determined. The consistency ratio is determined based on the ratio between the target consistency index value and the random consistency index value corresponding to the matrix order. If the consistency ratio is less than or equal to a preset threshold, the judgment matrix passes the consistency test; if the consistency ratio is greater than a preset age threshold, the judgment matrix fails the consistency test.

[0072] This application provides a traffic flow simulation scenario extraction device, comprising: a data source processing module for acquiring multiple types of data sources and preprocessing the data from these data sources to extract static and dynamic scene elements from the traffic flow simulation scenario; a fusion module for fusing the static and dynamic scene elements from different data sources to generate multiple structured reference simulation scenarios; an analysis module for using the Analytic Hierarchy Process (AHP) and information entropy theory to perform weight analysis and state probability distribution processing on each scene element in each reference simulation scenario to determine the scenario complexity of each reference simulation scenario; and a filtering module for filtering high-complexity simulation scenarios based on the scenario complexity and outputting a target simulation test scenario set for autonomous driving system testing. This device fully utilizes information from multiple data sources, improving the realism and complexity of the simulation scenario. Furthermore, by employing the AHP and information entropy theory, it can more accurately reflect the characteristics of the actual traffic environment, providing more effective support for the testing and verification of autonomous vehicles.

[0073] Please see Figure 4 , Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 4 As shown, the electronic device 400 includes a processor 410, a memory 420, and a bus 430.

[0074] The memory 420 stores machine-readable instructions executable by the processor 410. When the electronic device 400 is running, the processor 410 communicates with the memory 420 via the bus 430. When the machine-readable instructions are executed by the processor 410, they can perform the operations described above. Figure 1 as well as Figure 2 The steps of the traffic flow simulation scenario extraction method in the illustrated method embodiment can be found in the method embodiment for specific implementation, and will not be repeated here.

[0075] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, can perform the above-described actions. Figure 1 as well as Figure 2 The steps of the traffic flow simulation scenario extraction method in the illustrated method embodiment can be found in the method embodiment for specific implementation, and will not be repeated here.

[0076] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0077] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the shown or discussed mutual couplings, direct couplings, or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0078] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0079] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0080] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a processor-executable, non-volatile, computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0081] Finally, it should be noted that the above-described embodiments are merely specific implementations of this application, used to illustrate the technical solutions of this application, and not to limit them. The scope of protection of this application is not limited thereto. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features, within the scope of the technology disclosed in this application. Such modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for extracting traffic flow simulation scenarios, characterized in that, The extraction method includes: Multiple types of data sources are acquired and preprocessed to extract static and dynamic scene elements from the traffic flow simulation scenario from the preprocessed data sources. Data fusion is performed on the static scene elements and the dynamic scene elements from different data sources to generate multiple structured reference simulation scenes; The analytic hierarchy process (AHP) and information entropy theory are used to perform weight analysis and state probability distribution processing on each scene element in each of the reference simulation scenarios to determine the scene complexity of each of the reference simulation scenarios. Based on the complexity of the scenarios, high-complexity simulation scenarios are selected, and a set of target simulation test scenarios for testing autonomous driving systems is output.

2. The extraction method according to claim 1, characterized in that, The process of extracting static and dynamic scene elements from the traffic flow simulation scenario from the data source after data preprocessing includes: The road feature analysis, natural environment condition analysis, and static obstacle analysis are performed on multiple traffic flow simulation scenarios in the data source after data preprocessing to determine the static scene elements. Collision scene identification processing, traffic flow feature analysis processing, traffic participant behavior analysis processing, and intersection infrastructure analysis processing are performed on multiple traffic flow reference simulation scenarios to determine the dynamic scene elements.

3. The extraction method according to claim 1, characterized in that, The process of fusing static and dynamic scene elements from different data sources to generate multiple structured reference simulation scenes includes: Spatiotemporal alignment and semantic unification processing are performed on the static scene elements and dynamic scene elements from different data sources to determine the processed static scene elements and the processed dynamic scene elements. Based on a neural network model, the processed static scene elements and the processed dynamic scene elements are mapped to a unified feature space to generate a joint feature vector. The joint feature vector is then subjected to feature fusion processing to generate multiple structured reference simulation scenes.

4. The extraction method according to claim 1, characterized in that, The method employs the analytic hierarchy process (AHP) and information entropy theory to perform weight analysis and state probability distribution processing on each scene element in each reference simulation scene, thereby determining the scene complexity of each reference simulation scene, including: The analytic hierarchy process is used to perform weight analysis on each scene element in each of the reference simulation scenarios to determine the weight value of each scene element in each of the reference simulation scenarios. For any of the aforementioned reference simulation scenarios, the weight values ​​of each scenario element in the reference simulation scenario are used as prior knowledge to assist the information entropy theory in calculating the state probability distribution of each scenario element. The state probability distributions are then summarized and processed to determine the scenario complexity of the reference simulation scenario.

5. The extraction method according to claim 4, characterized in that, For any of the aforementioned reference simulation scenarios, the step of using the analytic hierarchy process (AHP) to perform weight analysis on each scene element in each reference simulation scenario to determine the weight value of each scene element in each reference simulation scenario includes: A hierarchical model is constructed, comprising a target layer, a criterion layer, and an indicator layer; wherein, the target layer is the decision objective of the reference simulation scenario, the criterion layer includes various scenario elements, and the indicator layer consists of scenario variables for each scenario element; A judgment matrix is ​​constructed by comparing the importance of multiple scene elements in each level; The judgment matrix is ​​column normalized, and the arithmetic mean of each row of elements in the normalized judgment matrix is ​​calculated to determine the weight value of each scene element. The judgment matrix is ​​subjected to a consistency check. If the judgment matrix passes the consistency check, the judgment matrix is ​​not adjusted. If the judgment matrix fails the consistency test, the judgment matrix is ​​adjusted and the weight values ​​of each scene element in the judgment matrix are redefined.

6. The extraction method according to claim 5, characterized in that, The consistency check of the judgment matrix includes: Based on the maximum weight value and the matrix order in the judgment matrix, the target consistency index value of the judgment matrix is ​​determined. The consistency ratio is determined based on the ratio between the target consistency index value and the random consistency index value corresponding to the matrix order. If the consistency ratio is less than or equal to a preset threshold, the judgment matrix passes the consistency test; if the consistency ratio is greater than a preset age threshold, the judgment matrix fails the consistency test.

7. A device for extracting traffic flow simulation scenarios, characterized in that, The extraction device includes: The data source processing module is used to acquire multiple types of data sources, perform data preprocessing on multiple data sources, and extract static scene elements and dynamic scene elements in the traffic flow simulation scenario from the data sources after data preprocessing. The fusion module is used to fuse the static scene elements and the dynamic scene elements from different data sources to generate multiple structured reference simulation scenes. The analysis module is used to perform weight analysis and state probability distribution processing on each scene element in each of the reference simulation scenarios using the analytic hierarchy process and information entropy theory, so as to determine the scene complexity of each of the reference simulation scenarios. The filtering module is used to filter out high-complexity simulation scenarios based on the scenario complexity and output a set of target simulation test scenarios for testing autonomous driving systems.

8. The extraction apparatus according to claim 7, characterized in that, The data source processing module is used to extract static and dynamic scene elements from the traffic flow reference simulation scenario in the data source after data preprocessing. The road feature analysis, natural environment condition analysis, and static obstacle analysis are performed on multiple traffic flow reference simulation scenarios in the data source after data preprocessing to determine the static scene elements. Collision scene identification processing, traffic flow feature analysis processing, traffic participant behavior analysis processing, and intersection infrastructure analysis processing are performed on multiple traffic flow reference simulation scenarios to determine the dynamic scene elements.

9. An electronic device, characterized in that, include: The device includes a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the electronic device is running, the processor communicates with the memory via the bus. The machine-readable instructions are executed by the processor to perform the steps of the traffic flow reference simulation scenario extraction method as described in any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, performs the steps of the traffic flow reference simulation scenario extraction method as described in any one of claims 1 to 6.

Images