Scene library updating method and device, vehicle-mounted controller, medium and program product
By filtering and introducing scene disturbance information from the original vehicle traffic data stream to generate a risk scene data stream, the problem of insufficient coverage of risk scenes in the existing technology is solved, and the scene library is updated efficiently and standardized, meeting the needs of autonomous driving simulation testing.
Patent Information
- Application Number
- CN202611480587.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-22
- Publication Date
- 2026-11-10
AI Technical Summary
In existing technologies, the SOTIF scenario library relies on manual analysis for updates, resulting in insufficient coverage of risk scenarios, difficulty in covering unknown risk scenarios and accumulated redundant data, and difficulty in meeting the needs of autonomous driving simulation testing and expected functional safety.
Based on the original scenario library, the system filters out the first traffic data stream with potential risks from the original vehicle traffic data stream, introduces scenario disturbance information to generate the second traffic data stream corresponding to the risk scenario, performs scenario verification, and updates the original scenario library.
It enriches the risk scenarios in the scenario library, improves the coverage and diversity of risk scenarios, reduces manual workload, and enhances the standardization and adaptability of the scenario library, thus meeting the requirements of autonomous driving simulation testing and expected functional safety.
Smart Images

Figure CN123072209A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of database update technology, and in particular to a method, apparatus, vehicle controller, medium, and program product for updating a scene database. Background Technology
[0002] With the continuous development of autonomous driving technology, conducting vehicle function verification based on the Safety of the Intended Functionality (SOTIF) standard has become an important industry requirement. The SOTIF scenario library is the core foundation for conducting expected functional safety analysis and risk verification.
[0003] In related technologies, the updating of the SOTIF scenario library mainly relies on manual analysis and screening of the original vehicle traffic data stream to determine the vehicle operation scenario, and then updating the scenario library based on the data stream under the vehicle operation scenario.
[0004] However, the raw traffic data stream of vehicles generally cannot cover risk scenarios, resulting in insufficient coverage of risk scenarios in the scenario library. Summary of the Invention
[0005] Therefore, it is necessary to provide a scenario library update method, device, vehicle controller, medium, and program product to address the above-mentioned technical problems, which can enrich the risk scenarios in the scenario library.
[0006] Firstly, this application provides a method for updating a scene library, including:
[0007] Based on the original scenarios in the original scenario library, the first traffic data stream with potential risks is selected from the original vehicle traffic data stream;
[0008] Based on the first traffic data stream and scene disturbance information, a second traffic data stream corresponding to the risk scenario is generated;
[0009] The second traffic data stream is validated for a specific scenario, and the original scenario database is updated using the validated second traffic data stream.
[0010] In one embodiment, a second traffic data stream corresponding to the risk scenario is generated based on the first traffic data stream and scenario disturbance information, including:
[0011] Add scenario disturbance information corresponding to any risk scenario to the first traffic data stream to obtain the traffic data stream corresponding to any risk scenario;
[0012] Based on the traffic data stream corresponding to any risk scenario, a second traffic data stream corresponding to the risk scenario is generated.
[0013] In one embodiment, any risk scenario is a potential risk scenario, and the scenario disturbance information includes the potential disturbance information corresponding to the potential risk scenario in the risk scenario;
[0014] Add scenario disturbance information corresponding to any risk scenario to the first traffic data stream to obtain a traffic data stream corresponding to any risk scenario, including:
[0015] Add the lowest level of potential disturbance information to the first traffic data stream to obtain the low-level traffic data stream corresponding to the potential risk scenario;
[0016] If the low-level traffic data stream passes the verification, higher-level potential disturbance information is added to the low-level traffic data stream in order of the disturbance level of the potential disturbance information. All traffic data streams that pass the verification are used as the traffic data streams corresponding to the potential risk scenarios.
[0017] In one embodiment, the method further includes:
[0018] If the low-level traffic data stream fails the verification, the next level of potential disturbance information is added to the first traffic data stream according to the disturbance level of the potential disturbance information to obtain the next level traffic data stream.
[0019] If the next level of traffic data stream fails the verification, new potential disturbance information of the next level will continue to be added to the first traffic data stream until the new next level of traffic data stream passes the verification.
[0020] In one embodiment, the scene disturbance information also includes other disturbance information corresponding to other risk scenarios besides potential risk scenarios in the risk scenario;
[0021] Based on the traffic data stream corresponding to any risk scenario, generate a second traffic data stream corresponding to the risk scenario, including:
[0022] By sequentially adding other disturbance information to the traffic data stream corresponding to the potential risk scenario, the traffic data stream corresponding to other risk scenarios is obtained.
[0023] Traffic data streams corresponding to potential risk scenarios and traffic data streams corresponding to other risk scenarios are used as the second traffic data streams corresponding to the risk scenarios.
[0024] In one embodiment, other disturbance information is sequentially added to the traffic data stream corresponding to the potential risk scenario to obtain the traffic data stream corresponding to other risk scenarios, including:
[0025] Obtain the risk scenario similarity between other risk scenarios and potential risk scenarios;
[0026] Based on the similarity of risk scenarios, different levels of disturbance information from other risk scenarios are added sequentially to the traffic data streams corresponding to potential risk scenarios to obtain traffic data streams corresponding to other risk scenarios.
[0027] In one embodiment, a second traffic data stream corresponding to any given risk scenario is generated, including:
[0028] The traffic data stream corresponding to any risk scenario is used as the second traffic data stream corresponding to the risk scenario.
[0029] In one embodiment, scenario verification is performed on the second traffic data stream, including:
[0030] Based on the original scenarios in the original scenario library, calculate the matching degree between the second traffic data stream corresponding to the risk scenario and the original scenario library;
[0031] Based on the matching degree of the original scenario library, the verification result of the second traffic data stream is determined.
[0032] In one embodiment, based on the original scenarios in the original scenario library, the matching degree between the second traffic data stream corresponding to the risk scenario and the original scenario library is calculated, including:
[0033] Calculate the fit between the second traffic data stream corresponding to the risk scenario and the scenario library; calculate the overlap between the second traffic data stream corresponding to the risk scenario and the original scenario in the original scenario library; and quantitatively evaluate the second traffic data stream corresponding to the risk scenario according to the compliance degree quantification value to obtain the compliance fit.
[0034] Adaptability, overlap, and compliance adaptability are all considered as the matching degree between the second traffic data stream and the original scenario library.
[0035] In one embodiment, based on the original scenarios in the original scenario library, a first traffic data stream with potential risks is selected from the original vehicle traffic data stream, including:
[0036] The raw vehicle traffic data stream is divided into multiple segments of raw traffic data stream;
[0037] Based on the original scenarios in the original scenario library, potential risks are predicted for each segment of original traffic data stream to obtain the potential risk quantification value corresponding to each segment of original traffic data stream.
[0038] The original traffic data stream with a potential risk quantification value greater than the preset risk threshold is regarded as the first traffic data stream with potential risk.
[0039] In one embodiment, potential risks are predicted for each segment of original traffic data stream based on the original scenarios in the original scenario library, resulting in a quantified potential risk value for each segment of original traffic data stream, including:
[0040] For any segment of original traffic data stream, calculate the probability that the original traffic data stream belongs to the original scene in the original scene database; and identify the scene danger level of the original traffic data stream to obtain the scene danger level corresponding to the original traffic data stream.
[0041] By integrating various probabilities and scenario hazard levels, a quantitative value of the potential risk corresponding to the original traffic data stream is obtained.
[0042] In one embodiment, before generating the second traffic data stream corresponding to the risk scenario based on the first traffic data stream and scenario disturbance information, the method further includes:
[0043] The static environmental features and dynamic interaction timing features in the first traffic data stream are standardized based on standard test scenario data.
[0044] Secondly, this application also provides a scene library updating device, including:
[0045] The filtering module is used to filter out the first traffic data stream with potential risks from the original traffic data stream of vehicles based on the original scenarios in the original scenario library;
[0046] The generation module is used to generate a second traffic data stream corresponding to the risk scenario based on the first traffic data stream and scenario disturbance information;
[0047] The verification module is used to perform scene verification on the second traffic data stream and update the original scene library with the verified second traffic data stream.
[0048] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the content of any of the scenario library update methods in the first aspect above.
[0049] Fourthly, this application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the content of any of the scenario library update methods in the first aspect described above.
[0050] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, implements the content of any of the scenario library update methods in the first aspect described above.
[0051] The aforementioned scenario library update method, device, vehicle controller, medium, and program product, based on the original scenarios in the original scenario library, filters out the first traffic data stream with potential risks from the original vehicle traffic data stream; generates a second traffic data stream corresponding to the risk scenario based on the first traffic data stream and scenario disturbance information; performs scenario verification on the second traffic data stream, and updates the original scenario library using the verified second traffic data stream. This method first filters the original vehicle traffic data based on the original scenarios in the original scenario library to extract the first traffic data stream with potential risks. Introducing scenario disturbance information based on the first traffic data stream can generate risk scenarios that are difficult to reproduce and collect. Then, by filtering out illogical second traffic data streams through scenario verification, the rationality of updating the second traffic data stream in the scenario library can be ensured. This approach can continuously expand the types of risk scenarios in the scenario library, comprehensively improving the coverage and diversity of risk scenarios in the original scenario library. Attached Figure Description
[0052] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.
[0053] Figure 1 This is an application environment diagram of the scene library update method in one embodiment;
[0054] Figure 2 This is a schematic diagram of the first process of a scene library update method in one embodiment;
[0055] Figure 3 This is a schematic diagram of the second process of the scene library update method in one embodiment;
[0056] Figure 4 This is a schematic diagram of the third process of the scene library update method in one embodiment;
[0057] Figure 5 This is a schematic diagram of the fourth process of the scene library update method in one embodiment;
[0058] Figure 6 This is a schematic diagram of the fifth process of the scene library update method in one embodiment;
[0059] Figure 7 This is a schematic diagram of the sixth process of the scene library update method in one embodiment;
[0060] Figure 8 This is a schematic diagram of the seventh process of the scene library update method in one embodiment;
[0061] Figure 9 This is a schematic diagram of the eighth process of the scene library update method in one embodiment;
[0062] Figure 10 This is a schematic diagram of the ninth process of the scene library update method in one embodiment;
[0063] Figure 11 This is a schematic diagram of the tenth process of the scene library update method in one embodiment;
[0064] Figure 12 This is a schematic diagram of the eleventh step of the scene library update method in one embodiment;
[0065] Figure 13 This is a structural block diagram of a scene library update device in one embodiment. Detailed Implementation
[0066] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.
[0067] Before providing a detailed description of the technical solution of this application, a brief explanation of the background technology of this application will be given first.
[0068] Currently, updates to the SOTIF scenario library largely rely on manual analysis and screening of raw vehicle traffic data streams. This manual approach has several limitations: firstly, it is difficult to extract rare risk scenarios from massive amounts of raw vehicle traffic data streams, and it cannot automatically generate data on unknown risk scenarios that vehicles have not actually collected data from; secondly, the standardization of manual screening and updating methods is low, leading to a large accumulation of redundant data in the scenario library, making maintenance difficult and failing to meet the expected functional safety standards' verification requirements for unknown risks.
[0069] To address the aforementioned issues, this application provides a scenario library update method, apparatus, vehicle controller, medium, and program product, which can effectively increase the number of high-risk unknown scenarios, enrich the risk scenarios in the scenario library, reduce manual workload, improve the standardization of the scenario library, and better adapt to autonomous driving simulation testing and expected functional safety compliance review. The technical solution of this application will be described in detail below.
[0070] The scene library update method provided in this application embodiment can be applied to, for example... Figure 1The application environment is shown. This application scenario includes an on-board controller 101 and a traffic data stream acquisition device 102, which communicate with each other. The traffic data stream acquisition device 102 is used to collect raw vehicle traffic data streams and send the collected raw vehicle traffic data streams to the on-board controller 101. After receiving the raw vehicle traffic data streams, the on-board controller 101, based on the original scenarios in the original scenario library, filters out a first traffic data stream with potential risks from the raw vehicle traffic data streams; generates a second traffic data stream corresponding to the risk scenario based on the first traffic data stream and scenario disturbance information; performs scenario verification on the second traffic data stream, and updates the original scenario library with the verified second traffic data stream.
[0071] The traffic data acquisition equipment 102 may include vehicle-mounted acquisition equipment, roadside sensing equipment, and simulation test platform data acquisition equipment. Vehicle-mounted acquisition equipment is mounted on various test vehicles and can collect real-time traffic data such as vehicle status data, surrounding vehicles, pedestrians, and road conditions during vehicle operation, covering various real-world road driving scenarios. Roadside sensing equipment is deployed at actual road condition locations such as urban roads, highway intersections, and park roads, enabling large-scale and continuous collection of dynamic traffic flow data across the entire area, compensating for the limitations of the vehicle-mounted acquisition equipment's perspective. The simulation test platform data acquisition equipment can collect standardized traffic data streams under virtual simulation test scenarios.
[0072] In one exemplary embodiment, such as Figure 2 As shown, a scene library update method is provided, which is applied to... Figure 1 Taking the vehicle-mounted controller as an example, the explanation includes the following steps S101 to S103. Wherein:
[0073] S101, based on the original scenarios in the original scenario library, filters out the first traffic data stream with potential risks from the original vehicle traffic data stream.
[0074] The original scene library refers to a pre-built database that stores original scenes; this scene library is a SOTIF scene library. Original scenes refer to scene samples already saved within the original scene library. Each original scene contains a set of scene elements, such as road type, traffic participants, weather, vehicle speed, lane status, and obstacle information.
[0075] In this embodiment, the vehicle controller can send a data acquisition command to the traffic data stream acquisition device according to the update requirements of the original scene library. After receiving the data acquisition command, the traffic data stream acquisition device can send the collected original vehicle traffic data stream to the vehicle controller. Alternatively, the original vehicle traffic data stream collected by the traffic data stream acquisition device can be stored in a database, and the vehicle controller can search for the original vehicle traffic data stream corresponding to the update requirement from the database according to the update requirement.
[0076] After acquiring the original vehicle traffic data stream, the onboard controller can divide the original traffic data stream into multiple segments using a time sliding window. It then calculates the similarity between each segment and the data in the original scene database, filtering out segments with similarity values exceeding a preset threshold. The remaining segments undergo potential risk assessment, and those with potential risk assessment values exceeding a preset threshold are designated as the first traffic data stream with potential risk.
[0077] Alternatively, the vehicle controller can slide through the original traffic data stream according to preset potential risk characteristics, monitoring the changes in each scene element. If a potential risk characteristic exists, and the scene corresponding to that segment of the original traffic data stream has not been updated in the original scene library, that segment of the original traffic data stream is designated as the first traffic data stream with potential risk. Potential risk characteristics could include the sudden appearance or disappearance of traffic participants, drastic changes in lighting conditions, and loss of road sign information.
[0078] Alternatively, the vehicle controller can divide the raw traffic data stream into multiple segments using a time sliding window, and input each segment into a preset potential risk detection model. The potential risk detection model extracts features from each segment and analyzes whether any potential risks exist within it. If the determination result indicates the presence of a potential risk, and the scenario corresponding to that segment is not updated in the original scenario library, then that segment is considered the first traffic data stream with potential risk. This application does not limit the specific method used to select the first traffic data stream with potential risk from the vehicle's raw traffic data stream.
[0079] S102, based on the first traffic data stream and scene disturbance information, generate the second traffic data stream corresponding to the risk scenario.
[0080] Scene disturbance information refers to pre-configured operational condition change information used to modify the original traffic scene, such as changing vehicle speed, adjusting obstacle distance, changing lighting, and modifying participant behavior, in order to construct derivative risk conditions. Risk scenarios can include environmental risk scenarios and traffic risk event scenarios. Environmental risk scenarios include at least lighting intensity interference risk, rain and fog interference risk, or lane line visibility interference risk; traffic risk event scenarios include at least pedestrian crossing risk and vehicle cutting in line risk.
[0081] In this embodiment, the vehicle controller can modify environmental elements in the first traffic data stream according to scene disturbance information, such as changing light intensity, increasing rain and fog interference, or adjusting lane line visibility, and re-render and output the complete time sequence segment to obtain the second traffic data stream. It is understood that one type of environmental element can be added at a time, or multiple types of environmental elements can be added at a time.
[0082] Alternatively, the vehicle controller can inject new traffic events into the first traffic data stream based on scene disturbance information, such as adding pedestrian crossings or vehicles cutting in line, while retaining most of the background information of the original scene, and generating a second traffic data stream with the newly added risk events.
[0083] Alternatively, the vehicle controller can add environmental elements and new traffic events to the first traffic data stream to obtain a second traffic data stream corresponding to multiple risk scenarios.
[0084] S103, perform scenario verification on the second traffic data stream, and update the original scenario library using the verified second traffic data stream.
[0085] In this embodiment, the vehicle controller can pre-set multiple evaluation dimensions for scenario verification, including adaptability, overlap, and compliance adaptability. The second traffic data stream is then scored item by item, and verification is deemed successful if the overall score reaches a preset verification threshold.
[0086] Alternatively, the vehicle controller can input the second traffic data stream into a preset scenario verification model. The scenario verification model analyzes the completeness of scenario elements, the rationality of timing interaction logic, the effectiveness of risk triggering, and the executability of simulation of the second traffic data stream to determine whether the second traffic data stream passes the verification.
[0087] Understandably, after the second traffic data stream that has passed verification updates the original scenario library, the update logs, safety test reports, and other contents of the original scenario library can be output and directly connected to the autonomous driving simulation platform and the vehicle safety review system to complete the closed-loop archiving.
[0088] In the aforementioned scenario library update method, based on the original scenarios in the original scenario library, a first traffic data stream with potential risks is selected from the original vehicle traffic data stream. Based on the first traffic data stream and scenario disturbance information, a second traffic data stream corresponding to the risk scenario is generated. The second traffic data stream undergoes scenario verification, and the original scenario library is updated using the verified second traffic data stream. This method first filters the original vehicle traffic data based on the original scenarios in the original scenario library to extract the first traffic data stream with potential risks. Introducing scenario disturbance information based on the first traffic data stream can generate risk scenarios that are difficult to reproduce and collect. Then, by filtering out illogical second traffic data streams through scenario verification, the rationality of updating the second traffic data stream in the scenario library can be ensured. This approach can continuously expand the types of risk scenarios in the scenario library, comprehensively improving the coverage and diversity of risk scenarios in the original scenario library.
[0089] The key to updating the scenario library lies in generating traffic data streams corresponding to risk scenarios. Therefore, in one embodiment, such as... Figure 3 As shown, the specific content of the second traffic data stream corresponding to the risk scenario generated based on the first traffic data stream and scenario disturbance information includes:
[0090] S201, add scenario disturbance information corresponding to any risk scenario to the first traffic data stream to obtain the traffic data stream corresponding to any risk scenario.
[0091] In this embodiment of the application, each risk scenario corresponds to a scenario disturbance information. The vehicle controller can select a potential risk scenario as a derived scenario from multiple risk scenarios and add the scenario disturbance information corresponding to the potential risk scenario to the first traffic data stream. The occurrence process of the potential risk scenario is simulated in the first traffic data stream to obtain the traffic data stream corresponding to the potential risk scenario.
[0092] Alternatively, the vehicle controller can also determine the collection season, daytime collection time, and rain conditions during the collection process based on the information collected from the first traffic data stream. Based on this information, it selects the risk scenario that best matches the collected data from multiple risk scenarios and adds the corresponding scenario disturbance information to the first traffic data stream, thus obtaining the traffic data stream corresponding to the risk scenario. For example, if the first traffic data stream is collected during the rainy season, during the day, and during a period of no rainfall, then the best-matching risk scenario is a rainy scenario.
[0093] S202, based on the traffic data stream corresponding to any risk scenario, generate a second traffic data stream corresponding to the risk scenario.
[0094] In this embodiment of the application, when a traffic data stream corresponding to any risk scenario is obtained, and the traffic data stream corresponding to the risk scenario is not updated in the original scenario library, the vehicle controller can perform format and timing alignment processing on the traffic data stream corresponding to the risk scenario to generate a second traffic data stream corresponding to the risk scenario.
[0095] Alternatively, the vehicle controller can add new scene disturbance information corresponding to other risk scenarios sequentially to the traffic data stream corresponding to each risk scenario, based on scene disturbance information corresponding to other risk scenarios. After each addition of new scene disturbance information, it is necessary to determine whether the traffic data stream corresponding to the new risk scenario has been updated in the original scenario database. If none of the traffic data streams corresponding to any of the risk scenarios have been updated in the original scenario database, then each traffic data stream after adding a risk scenario is treated as the second traffic data stream corresponding to that risk scenario.
[0096] In the aforementioned scenario library update method, scenario disturbance information corresponding to any risk scenario is added to the first traffic data stream to obtain a traffic data stream corresponding to any risk scenario; based on the traffic data stream corresponding to any risk scenario, a second traffic data stream corresponding to the risk scenario is generated. This method uses the selected first traffic data stream as a base, adds any type of scenario disturbance, and independently generates various types of risk condition data. It can accurately and purposefully derive risk scenario data of different dimensions, enriching the types of risk scenarios.
[0097] Assuming any risk scenario is a potential risk scenario, the scenario disturbance information includes the potential disturbance information corresponding to the potential risk scenarios within the risk scenarios. Then, in one embodiment, as... Figure 4 As shown, the detailed content of the traffic data stream corresponding to any risk scenario obtained by adding scenario disturbance information corresponding to any risk scenario to the first traffic data stream includes:
[0098] S301, add the lowest level of potential disturbance information to the first traffic data stream to obtain the low-level traffic data stream corresponding to the potential risk scenario.
[0099] Among them, the lowest level of potential disturbance information refers to weak disturbance changes with controllable intensity.
[0100] In this embodiment, if the potential disturbance information is an environmental disturbance, the lowest level of environmental disturbance information can be added to the overall segment of the first traffic data stream to obtain a low-level traffic data stream corresponding to the potential risk scenario. For example, if the first traffic data stream is a time-series segment of an urban intersection collected from real vehicles, and the original scene has weak lighting, rainfall disturbance information can be added to the first traffic data stream.
[0101] If the potential disturbance information is a traffic event disturbance, a minimum-level traffic event disturbance can be added at a certain location in the first traffic data stream to obtain a low-level traffic data stream corresponding to the potential risk scenario. For example, if the first traffic data stream is a time-series segment of an urban intersection collected from real vehicles, and there are no pedestrians crossing the road in the original scenario, then a pedestrian crossing disturbance can be added to the first traffic data stream.
[0102] S302: If the low-level traffic data stream passes the verification, add higher-level potential disturbance information to the low-level traffic data stream in sequence according to the disturbance level of the potential disturbance information, and take all the verified traffic data streams as the traffic data streams corresponding to the potential risk scenarios.
[0103] In this embodiment, provided that the low-level traffic data stream passes verification, the vehicle controller can superimpose higher-level potential disturbance information onto the verified low-level traffic data stream in ascending order of the preset gradient levels of potential disturbance information. Each level of potential disturbance information follows a gradient progression principle, with the disturbance intensity and scenario risk induction effect increasing progressively. Each level of disturbance is adjusted incrementally based on the previous level of qualified data stream.
[0104] In the aforementioned scenario library update method, the lowest-level potential disturbance information is added to the first traffic data stream to obtain a low-level traffic data stream corresponding to the potential risk scenario. If the low-level traffic data stream passes verification, higher-level potential disturbance information is added sequentially to the low-level traffic data stream according to the disturbance level order. All verified traffic data streams of different levels are then used as the traffic data streams corresponding to the potential risk scenario. This method sequentially superimposes different levels of potential disturbance information into the first traffic data stream according to their level. Through a gradient-progressive disturbance generation method, it can derive multi-gradient potential risk data streams with different risk levels based on the same baseline scenario, ensuring the continuity and correlation of risk scenarios at each level and enriching the data types of potential risk scenarios.
[0105] In one embodiment, the relevant content regarding the failure of low-level traffic data stream verification is described, such as... Figure 5 As shown, the relevant content includes:
[0106] S401: If the low-level traffic data stream fails the verification, add the next level of potential disturbance information to the first traffic data stream according to the disturbance level of the potential disturbance information to obtain the next level traffic data stream.
[0107] In this embodiment of the application, when the low-level traffic data stream fails the verification, the disturbance is no longer superimposed on the unqualified low-level traffic data stream. Instead, it returns to the first traffic data stream as the disturbance base and selects the next level of potential disturbance information to be applied to the first traffic data stream according to the pre-set disturbance level order of potential disturbance information from low to high, so as to obtain the next level of traffic data stream.
[0108] S402 If the next-level traffic data stream fails the verification, continue to add new potential disturbance information of the next level to the first traffic data stream until the new next-level traffic data stream passes the verification.
[0109] In this embodiment, if the generated next-level traffic data stream still fails verification, the vehicle controller discards the currently unqualified next-level traffic data stream without accumulating disturbances based on the unqualified data stream, and continues to use the original first traffic data stream as the unchanged base. New potential disturbance information for the next level is selected according to the disturbance level order and superimposed onto the first traffic data stream to obtain a new next-level traffic data stream, which is then verified. This disturbance superposition and verification step is repeated until the generated new next-level traffic data stream passes verification.
[0110] In the aforementioned scenario database update method, if the low-level traffic data stream fails verification, the next level of potential disturbance information is added to the first traffic data stream according to the disturbance level order, resulting in the next-level traffic data stream. If the next-level traffic data stream fails verification, new next-level potential disturbance information is added to the first traffic data stream until the new next-level traffic data stream passes verification. This method avoids further adding disturbances to the failed low-level data when the lowest-level disturbance fails to meet verification requirements. Instead, it reverts to the first traffic data stream to re-add the next-level disturbance, preventing unqualified disturbances from accumulating and amplifying, which could cause the scenario to deviate significantly from real traffic conditions.
[0111] Scene disturbance information also includes other disturbance information corresponding to other risk scenarios besides the potential risk scenario. Therefore, after obtaining the traffic data stream corresponding to the potential risk scenario, it is necessary to add other disturbance information to the traffic data stream corresponding to the potential risk scenario. In one embodiment, for example... Figure 6 As shown, the detailed content of generating a second traffic data stream corresponding to a risk scenario based on the traffic data stream corresponding to any risk scenario is introduced. This detailed content includes:
[0112] S501, other disturbance information is added sequentially to the traffic data stream corresponding to the potential risk scenario to obtain the traffic data stream corresponding to other risk scenarios.
[0113] In this embodiment, the vehicle controller can use the original traffic data stream corresponding to the potential risk scenario as a unified base. Each time a set of other disturbance information is selected, it is superimposed on the original base, without using the previous disturbance output as the input for the next. In this way, multiple sets of other disturbance information are added sequentially, and traffic data streams corresponding to other risk scenarios for each set of disturbances are obtained.
[0114] Alternatively, the vehicle controller can use the traffic data stream corresponding to the potential risk scenario as the initial base, and then sequentially superimpose each group of other disturbance information onto the data stream output by the previous disturbance. The disturbance effect accumulates gradually, and after each superposition, a traffic data stream corresponding to another risk scenario is output.
[0115] S502 defines the traffic data streams corresponding to potential risk scenarios and the traffic data streams corresponding to other risk scenarios as the second traffic data streams corresponding to the risk scenarios.
[0116] In this embodiment of the application, when traffic data streams corresponding to other risk scenarios are obtained, since risk scenarios include potential risk scenarios and other risk scenarios, the vehicle controller can use the traffic data streams corresponding to both potential risk scenarios and other risk scenarios as the second traffic data streams corresponding to the risk scenarios.
[0117] In the aforementioned scenario library update method, other perturbation information is sequentially added to the traffic data stream corresponding to potential risk scenarios to obtain traffic data streams corresponding to other risk scenarios. These traffic data streams corresponding to potential risk scenarios and other risk scenarios are then used as a second traffic data stream corresponding to the risk scenario. This method sequentially superimposes other perturbation information onto the traffic data stream of potential risk scenarios, preserving a traffic data stream close to the real scenario while obtaining multiple derived scenario data under the same scenario through the superposition of perturbations, thus enriching the types of risk scenarios.
[0118] In one embodiment, such as Figure 7 As shown, the detailed content of the traffic data streams corresponding to other risk scenarios obtained by sequentially adding other disturbance information to the traffic data streams corresponding to potential risk scenarios includes:
[0119] S601, obtain the risk scenario similarity between other risk scenarios and potential risk scenarios.
[0120] In this application embodiment, other risk scenarios may include multiple risk scenarios. Therefore, for any other risk scenario, the vehicle controller can extract the static environment, dynamic interaction, and risk trigger feature vectors of the potential risk scenario and any other risk scenario, and calculate the risk scenario similarity between any other risk scenario and the potential risk scenario based on these feature vectors.
[0121] S602, based on the similarity of risk scenarios, add different levels of other risk scenario disturbance information to the traffic data streams corresponding to potential risk scenarios in turn to obtain traffic data streams corresponding to other risk scenarios.
[0122] In this embodiment, the vehicle controller can sort other risk scenarios from high to low based on the already acquired risk scenario similarity, obtaining a sorting result. The lowest-level disturbance information corresponding to the other risk scenario with the highest similarity is added to the traffic data stream corresponding to the potential risk scenario, and it is determined whether the obtained traffic data stream passes verification. If it passes, the next level of disturbance information is added to the lowest-level traffic data stream until all levels of traffic data streams are obtained.
[0123] Then, add disturbance information for other risk scenarios in the sorting process, and use all traffic data streams that pass the verification as the traffic data streams corresponding to other risk scenarios.
[0124] For example, the potential risk scenario is a nighttime scenario. The similarity to the potential risk scenario from high to low is: lane line visibility interference risk, rain and fog interference risk, and vehicle cutting in risk. Light intensity interference risk is not introduced.
[0125] In the aforementioned scenario library update method, the similarity between other risk scenarios and potential risk scenarios is obtained. Based on the risk scenario similarity, different levels of disturbance information from other risk scenarios are sequentially added to the traffic data streams corresponding to the potential risk scenarios to obtain the traffic data streams corresponding to the other risk scenarios. This method combines scenario similarity to determine the order of adding disturbances, prioritizing risk scenarios with higher similarity for disturbance addition. This ensures that the newly added disturbance content highly matches the characteristics of the potential risk scenarios, guaranteeing the coherence and rationality of scenario derivation.
[0126] The above embodiments are solutions provided under the assumption that any risk scenario is a potential risk scenario. The following embodiments do not limit the scope of any risk scenario, and will describe in detail the process of generating a second traffic data stream corresponding to a risk scenario based on the traffic data stream corresponding to any risk scenario. This detail includes:
[0127] The traffic data stream corresponding to any risk scenario is used as the second traffic data stream corresponding to the risk scenario.
[0128] In this embodiment, when a traffic data stream corresponding to any risk scenario is obtained, the vehicle controller first queries the original scenario database to determine whether the traffic data stream corresponding to the risk scenario has been updated in the original scenario database. If the traffic data stream corresponding to the risk scenario does not exist in the original scenario database, the vehicle controller can perform data format conversion and standardization processing on the traffic data stream corresponding to the risk scenario, and take the processed traffic data stream corresponding to the risk scenario as the final result, i.e., the second traffic data stream corresponding to the risk scenario.
[0129] In the aforementioned scenario database update method, the traffic data stream corresponding to any risk scenario is used as the second traffic data stream corresponding to that risk scenario. This method directly uses the traffic data stream corresponding to the risk scenario itself as the second traffic data stream for that risk scenario, enabling the second traffic data stream to better retain the original data characteristics of the traffic scenario data and ensuring the authenticity of the risk scenario data in the updated scenario database.
[0130] After obtaining the second traffic data stream corresponding to the risk scenario, scenario verification is still required. The following example illustrates the scenario verification process. Figure 8 As shown, the detailed content of the above-mentioned scenario verification for the second traffic data stream includes:
[0131] S701, based on the original scenarios in the original scenario library, calculates the matching degree between the second traffic data stream corresponding to the risk scenario and the original scenario library.
[0132] In this embodiment, the vehicle controller can extract the scene feature vector of the second traffic data stream and the feature vector of each original scene corresponding to the data in the original scene library. Then, it calculates the similarity between the second traffic data stream and each original scene, and takes the maximum similarity as the matching degree of the second traffic data stream relative to the original scene library.
[0133] Alternatively, the vehicle controller can evaluate the second traffic data stream corresponding to the risk scenario based on the original scenario in the original scenario library, in terms of adaptability, overlap, and compliance adaptability, and use the assessed adaptability, overlap, and compliance adaptability as the matching degree between the second traffic data stream and the original scenario library.
[0134] S702, based on the matching degree of the original scene library, determine the verification result of the second traffic data stream.
[0135] In this embodiment, the vehicle controller can compare the fit, overlap, and compliance fit in the matching degree with preset thresholds. If at least one of the fit, overlap, and compliance fit is greater than or equal to the preset threshold, the second traffic data stream verification is determined to have failed. If the fit, overlap, and compliance fit are all less than the preset thresholds, the second traffic data stream verification is determined to have passed.
[0136] It should be noted that the matching degree includes adaptability, overlap, and compliance adaptability, each corresponding to a preset threshold. The matching degree threshold corresponding to adaptability can be calculated using the following formula:
[0137]
[0138] in, E(S) represents the historical compliance adaptation balance coefficient; E(S) represents the average scarcity entropy value of the scenarios in the original scenario library.
[0139] The threshold corresponding to redundancy overlap can be calculated using the following formula:
[0140]
[0141] in, This represents the redundancy suppression adjustment coefficient; This represents the average fusion validity value of the scenes in the original scene library.
[0142] The threshold corresponding to compliance and adaptability can be calculated using the following formula:
[0143]
[0144] in, This indicates the compliance benchmarking correction factor; This represents the global matching quantization score.
[0145] In the aforementioned scenario library update method, based on the original scenarios in the original scenario library, the matching degree between the second traffic data stream corresponding to the risk scenario and the original scenario library is calculated; based on the matching degree of the original scenario library, the verification result of the second traffic data stream is determined. This method utilizes the existing original scenarios in the original scenario library to calculate the matching degree between the second traffic data stream corresponding to the risk scenario and the original scenario, and then completes scenario verification based on the matching degree. This can accurately identify invalid and misjudged second traffic data streams, ensuring that the second traffic data streams used to update the scenario library are valuable new risk scenarios, avoiding the accumulation of a large amount of redundant data in the scenario library, and reducing the maintenance difficulty of the scenario library.
[0146] In one embodiment, such as Figure 9 As shown, the detailed content of the matching degree between the second traffic data stream corresponding to the risk scenario calculated based on the original scenario library and the original scenario library includes:
[0147] S801, calculate the fit between the second traffic data stream corresponding to the risk scenario and the scenario library; calculate the overlap between the second traffic data stream corresponding to the risk scenario and the original scenario in the original scenario library; and, quantitatively evaluate the second traffic data stream corresponding to the risk scenario according to the compliance degree quantification value to obtain the compliance fit.
[0148] In this embodiment, the vehicle controller can verify the data format, key fields, and timing information of the second traffic data stream according to the entry specifications of the scenario library, calculate the proportion of items conforming to the specifications, and obtain the fit degree. Simultaneously, it extracts the scenario feature vector of the second traffic data stream and calculates its similarity with all original scenarios in the original scenario library, selecting the highest similarity as the overlap degree. Furthermore, the vehicle controller can refer to the compliance degree quantification value to score the scenario traffic logic and simulation usability item by item, obtaining the compliance fit degree.
[0149] S802 uses adaptability, overlap, and compliance adaptability as criteria for matching the second traffic data stream with the original scenario library.
[0150] In this embodiment of the application, after obtaining the adaptability, overlap and compliance adaptability of the second traffic data stream, the vehicle controller can directly use these three results as the matching degree between the second traffic data stream and the original scenario library.
[0151] The aforementioned scenario library update method calculates the fit between the second traffic data stream corresponding to the risk scenario and the scenario library; calculates the overlap between the second traffic data stream corresponding to the risk scenario and the original scenario in the original scenario library; and quantifies the second traffic data stream corresponding to the risk scenario according to the compliance degree quantification value to obtain the compliance fit; the fit, overlap, and compliance fit are all used as the matching degree between the second traffic data stream and the original scenario library. This method quantifies the second traffic data stream from three dimensions: scenario library fit, scenario overlap, and compliance fit. By comprehensively verifying the results of these three quantifications, it can more accurately filter out second traffic data streams that are weakly fitted to the scenario library, highly repetitive, or non-compliant, ensuring the effectiveness of the scenario library update samples.
[0152] In one embodiment, such as Figure 10 As shown, the detailed content of filtering out the first traffic data stream with potential risks from the original traffic data stream based on the original scenarios in the original scenario library is introduced. This detailed content includes:
[0153] S901 divides the original vehicle traffic data stream into multiple segments of the original traffic data stream.
[0154] In this embodiment, the vehicle controller can segment the original vehicle traffic data stream according to a preset fixed-length time window, dividing the continuous long data stream into multiple segments of original traffic data stream with consistent duration. For example, each segment can be set to a duration of 5 seconds, and data from consecutive 5-second intervals can be extracted as a segment to achieve uniform segmentation.
[0155] Alternatively, the vehicle controller can detect the degree of change of scene elements within the vehicle's original traffic data stream. When a significant change is detected in the road type, lighting conditions, or participant composition, segmentation is performed. If there is no significant change in the scene, segmentation is not performed, generating multiple segments of original traffic data stream with varying lengths.
[0156] S902, based on the original scenarios in the original scenario library, performs potential risk prediction on each segment of the original traffic data stream to obtain the potential risk quantification value corresponding to each segment of the original traffic data stream.
[0157] In this embodiment, for any segment of original traffic data stream, the vehicle controller can first determine whether the original traffic data stream is duplicated with the standard data stream corresponding to the original scenario in the original scenario library. If it is duplicated, the potential risk quantification value of the original traffic data stream is determined to be less than a preset threshold. If it is not duplicated, the original traffic data stream is scored according to the preset potential risk scoring rules to obtain the potential risk quantification value corresponding to the segment of original traffic data stream.
[0158] Alternatively, the vehicle controller can construct a normal operating condition distribution benchmark based on a large number of normal original scenarios in the original scenario library, and then calculate the degree of deviation of each segment of original traffic data stream relative to the benchmark distribution, converting each outlier degree into a potential risk quantification value corresponding to each segment of original traffic data stream.
[0159] S903 defines the original traffic data stream with a potential risk quantification value greater than a preset risk threshold as the first traffic data stream with potential risk.
[0160] In this embodiment, for any segment of raw traffic data stream, the vehicle controller can compare the potential risk quantification value corresponding to that segment with a preset risk threshold. If the comparison result shows that the potential risk quantification value is greater than the preset risk threshold, then the vehicle controller determines that the segment of raw traffic data stream has a high potential risk. After all potential risk quantification values have been compared, the vehicle controller can identify the raw traffic data streams with potential risk quantification values greater than the preset risk threshold as the first traffic data streams with potential risks.
[0161] In the aforementioned database update method, the original vehicle traffic data stream is divided into multiple segments. Based on the original scenarios in the original scenario library, potential risks are predicted for each segment, resulting in a quantified potential risk value for each segment. The original traffic data stream with a quantified potential risk value greater than a preset risk threshold is identified as the first traffic data stream with potential risk. This method, by segmenting the original vehicle traffic data stream and relying on the original scenarios in the original scenario library to predict risks and generate corresponding quantified risk values for each segment, and using a preset risk threshold as a filtering condition, can accurately locate potential risk segments in the original vehicle traffic data stream.
[0162] The following example illustrates the potential risk prediction process mentioned in the previous embodiments. Figure 11 As shown, the detailed content of the above-mentioned potential risk prediction for each segment of original traffic data stream based on the original scenarios in the original scenario library, and the resulting potential risk quantification value for each segment of original traffic data stream, includes:
[0163] S1001, for any segment of original traffic data stream, calculate the probability that the original traffic data stream belongs to the original scene in the original scene database; and identify the scene danger level of the original traffic data stream to obtain the scene danger level corresponding to the original traffic data stream.
[0164] In this embodiment, the vehicle controller can extract the scene feature vector of any segment of original traffic data stream, calculate the similarity with the feature vectors of each original scene in the original scene library, and then map the similarity to obtain the probability that the segment of original traffic data stream belongs to each original scene. Simultaneously, based on preset hazard judgment rules and combined with indicators such as relative distance, relative speed, and conflict trend of participants, the controller can score the segment of original traffic data stream, and determine the scene hazard level corresponding to the segment of original traffic data stream based on the mapping relationship between the scoring results and the hazard level.
[0165] S1002, by integrating various probabilities and scenario hazard levels, obtains the potential risk quantification value corresponding to the original traffic data stream.
[0166] In this embodiment, the vehicle controller can configure preset weighting coefficients for the attribution probability and the scene danger level of each original scene, perform a weighted summation of the attribution probabilities, and then perform a weighted calculation in combination with the level score corresponding to the danger level. The weighted calculation result is used as the potential risk quantification value of the original traffic data stream of that segment.
[0167] In another embodiment, the above-mentioned probability calculation, hazard level identification, and fusion process can all be completed by calculating the scene scarcity entropy weight screening formula, which can be expressed as:
[0168]
[0169] Where E(S) is the quantified value of the potential risk corresponding to the original traffic data stream; ω i The weights are assigned to four indicators: illumination, traffic interaction, shading, and vehicle dynamics; p i is the probability that the original traffic data stream belongs to the original scene in the original scene library; n is the number of evaluation dimensions; R(S) represents the safety risk correction coefficient; R(S) represents the scenario hazard level corresponding to the original traffic data stream, with a higher value indicating a higher degree of scenario hazard.
[0170] In the aforementioned scenario database update method, for any given segment of original traffic data stream, the probability that the original traffic data stream belongs to an existing scenario in the original scenario database is calculated; and the scenario hazard level of the original traffic data stream is identified to obtain the corresponding scenario hazard level; the probabilities and scenario hazard levels are then fused to obtain the potential risk quantification value corresponding to the original traffic data stream. This method, by considering both the matching probability of the traffic data stream with various existing scenarios in the original scenario database and simultaneously identifying the scenario hazard level of the traffic data stream itself when processing each segment of original traffic data stream, and by fusing the scenario matching probability and hazard level information, can accurately calculate the potential risk quantification value corresponding to the original traffic data stream.
[0171] After identifying the first traffic data stream with potential risks from the raw vehicle traffic data stream, this first traffic data stream with potential risks can be standardized. For example, standardization can involve completing missing elements or unifying the time sequence and format. In one embodiment, the standardization process is described in detail, including:
[0172] The static environmental features and dynamic interaction timing features in the first traffic data stream are standardized based on standard test scenario data.
[0173] Standard test scenario data refers to a pre-defined benchmark scenario dataset that specifies common environmental parameters and interaction behavior paradigms for traffic scenarios, serving as a reference for normalization processing. Static environmental characteristics refer to elements in the first traffic data stream that do not change over time, such as road alignment, number of lanes, road signs, surrounding buildings, and fixed obstacles. Dynamic interaction temporal characteristics refer to temporally sequenced interaction information that changes over time, such as vehicle driving status, the movement trajectories of surrounding traffic participants, and temporal interaction behaviors between participants.
[0174] In this embodiment, the vehicle controller can substitute the first traffic data stream into a semantic fusion formula that includes standard test scenario data. Through the calculation of the semantic fusion formula, the standardized first traffic data stream is obtained. The mathematical expression of the semantic fusion formula is as follows:
[0175]
[0176] in, The degree of matching between potential risk scenarios and standard test scenarios; These are the static environmental characteristics in the first traffic data stream; The dynamic interaction time-series features in the first traffic data stream are represented by CosSim; similarity calculation is performed using CosSim; and temporal correlation is represented by Corr. This is the deviation suppression coefficient; This is to address deviations in compliance elements for specific scenarios and ensure that the generated scenarios are standardized and consistent.
[0177] In the aforementioned database update method, the static environmental features and dynamic interaction time-series features in the first traffic data stream are standardized based on standard test scenario data. This method uses standard test scenario data as a benchmark to standardize the static environmental features and dynamic interaction time-series features of the first traffic data stream, effectively solving the problems of inconsistent feature dimensions and formats between different data sets, and achieving unified and standardized data features.
[0178] In a detailed embodiment, such as Figure 12 As shown, the above-mentioned scene library update method includes the following:
[0179] S1101 divides the original vehicle traffic data stream into multiple segments of the original traffic data stream;
[0180] S1102, For any segment of original traffic data stream, calculate the probability that the original traffic data stream belongs to the original scene in the original scene database; and identify the scene danger level of the original traffic data stream to obtain the scene danger level corresponding to the original traffic data stream.
[0181] S1103, integrates various probabilities and scenario hazard levels to obtain the potential risk quantification value corresponding to the original traffic data stream;
[0182] S1104, the original traffic data stream with a potential risk quantification value greater than the preset risk threshold is regarded as the first traffic data stream with potential risk;
[0183] S1105, add the lowest level of potential disturbance information to the first traffic data stream to obtain the low-level traffic data stream corresponding to the potential risk scenario;
[0184] S1106, If the low-level traffic data stream passes the verification, according to the order of the disturbance level of the potential disturbance information, add higher-level potential disturbance information to the low-level traffic data stream in sequence, and take all the verified traffic data streams as the traffic data streams corresponding to the potential risk scenarios.
[0185] S1107, Other disturbance information is added sequentially to the traffic data stream corresponding to the potential risk scenario to obtain the traffic data stream corresponding to other risk scenarios;
[0186] S1108, the traffic data stream corresponding to the potential risk scenario and the traffic data stream corresponding to other risk scenarios are used as the second traffic data stream corresponding to the risk scenario.
[0187] S1109, Based on the original scenarios in the original scenario library, calculate the matching degree between the second traffic data stream corresponding to the risk scenario and the original scenario library;
[0188] S1110: Based on the matching degree of the original scenario library, determine the verification result of the second traffic data stream, and update the original scenario library with the verified second traffic data stream.
[0189] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.
[0190] Based on the same inventive concept, this application also provides a scene library updating apparatus for implementing the scene library updating method described above. The solution provided by this apparatus is similar to the implementation described in the above method; therefore, the specific limitations in one or more scene library updating apparatus embodiments provided below can be found in the limitations of the scene library updating method described above, and will not be repeated here.
[0191] In one exemplary embodiment, such as Figure 13 As shown, a scene library update device is provided, including: a filtering module 11, a generation module 12, and a verification module 13, wherein:
[0192] The filtering module 11 is used to filter out the first traffic data stream with potential risks from the original traffic data stream of vehicles based on the original scenarios in the original scenario library;
[0193] The generation module 12 is used to generate a second traffic data stream corresponding to the risk scenario based on the first traffic data stream and the scenario disturbance information;
[0194] The verification module 13 is used to perform scene verification on the second traffic data stream and update the original scene library through the verified second traffic data stream.
[0195] In an exemplary embodiment, the above-described generation module includes an adding unit and a generation unit, wherein:
[0196] An add unit is used to add scenario disturbance information corresponding to any risk scenario to the first traffic data stream to obtain a traffic data stream corresponding to any risk scenario.
[0197] The generation unit is used to generate a second traffic data stream corresponding to any given risk scenario, based on the traffic data stream corresponding to that risk scenario.
[0198] In an exemplary embodiment, the above-mentioned adding unit is further configured to add the lowest level of potential disturbance information to the first traffic data stream to obtain a low-level traffic data stream corresponding to the potential risk scenario; if the low-level traffic data stream passes the verification, higher level potential disturbance information is added to the low-level traffic data stream in sequence according to the disturbance level of the potential disturbance information, and all levels of traffic data streams that pass the verification are used as the traffic data streams corresponding to the potential risk scenario.
[0199] In an exemplary embodiment, the above-mentioned adding unit is further configured to, if the low-level traffic data stream fails the verification, add the next level of potential disturbance information to the first traffic data stream according to the disturbance level order of the potential disturbance information to obtain the next level traffic data stream; if the next level traffic data stream fails the verification, then continue to add new next level potential disturbance information to the first traffic data stream until the new next level traffic data stream passes the verification.
[0200] In an exemplary embodiment, the generation unit is further configured to sequentially add other disturbance information to the traffic data stream corresponding to the potential risk scenario to obtain the traffic data stream corresponding to other risk scenarios; and use the traffic data stream corresponding to the potential risk scenario and the traffic data stream corresponding to other risk scenarios as the second traffic data stream corresponding to the risk scenario.
[0201] In an exemplary embodiment, the generation unit is further configured to obtain the risk scenario similarity between other risk scenarios and potential risk scenarios; according to the risk scenario similarity, to sequentially add different levels of other risk scenario disturbance information to the traffic data stream corresponding to the potential risk scenario, thereby obtaining the traffic data stream corresponding to other risk scenarios.
[0202] In an exemplary embodiment, the generation unit is further configured to use the traffic data stream corresponding to any one of the risk scenarios as the second traffic data stream corresponding to the risk scenario.
[0203] In an exemplary embodiment, the above-mentioned verification module includes: a calculation unit and a result determination unit, wherein:
[0204] The calculation unit is used to calculate the matching degree between the second traffic data stream corresponding to the risk scenario and the original scenario library, based on the original scenarios in the original scenario library.
[0205] The result determination unit is used to determine the verification result of the second traffic data stream based on the matching degree of the original scenario library.
[0206] In an exemplary embodiment, the above-mentioned calculation unit is further configured to calculate the adaptability of the second traffic data stream corresponding to the risk scenario with the scenario library; calculate the overlap between the second traffic data stream corresponding to the risk scenario and the original scenario in the original scenario library; and quantify and evaluate the second traffic data stream corresponding to the risk scenario according to the compliance degree quantification value to obtain the compliance adaptability; and use the adaptability, overlap and compliance adaptability as the matching degree between the second traffic data stream and the original scenario library.
[0207] In an exemplary embodiment, the above-mentioned filtering module includes: a partitioning unit, a prediction unit, and a data flow determination unit, wherein:
[0208] A segmentation unit is used to divide the raw vehicle traffic data stream into multiple segments of the raw traffic data stream.
[0209] The prediction unit is used to predict the potential risks of each segment of the original traffic data stream based on the original scenarios in the original scenario library, and to obtain the potential risk quantification value corresponding to each segment of the original traffic data stream.
[0210] The data flow determination unit is used to identify the original traffic data flow with a potential risk quantification value greater than a preset risk threshold as the first traffic data flow with potential risk.
[0211] In an exemplary embodiment, the prediction unit is further configured to calculate the probability that any segment of original traffic data stream belongs to an original scenario in the original scenario library; and to identify the scenario hazard level of the original traffic data stream to obtain the scenario hazard level corresponding to the original traffic data stream; and to fuse the probabilities and scenario hazard levels to obtain the potential risk quantification value corresponding to the original traffic data stream.
[0212] In an exemplary embodiment, the above-mentioned scene library update device further includes: a standardization processing module, wherein:
[0213] The standardization processing module is used to standardize the static environmental features and dynamic interaction timing features in the first traffic data stream based on standard test scenario data.
[0214] Each module in the aforementioned scene library update device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the operations corresponding to each module.
[0215] In one exemplary embodiment, an in-vehicle controller is provided, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the content of any embodiment of the scene library update method described above.
[0216] In one embodiment, a computer-readable storage medium is provided that stores a computer program thereon, which, when executed by a processor, implements the content of any embodiment of the above-described scene library update method.
[0217] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the content of any embodiment of the scene library update method described above.
[0218] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.
[0219] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, database, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.
[0220] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.
[0221] The above embodiments are merely illustrative of several implementation methods of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.
Claims
1. A method for updating a scene library, characterized in that, The method includes: Based on the original scenarios in the original scenario library, the first traffic data stream with potential risks is selected from the original vehicle traffic data stream; Based on the first traffic data stream and the scene disturbance information, a second traffic data stream corresponding to the risk scenario is generated; The second traffic data stream is validated for a scenario, and the original scenario library is updated using the validated second traffic data stream.
2. The method according to claim 1, characterized in that, The step of generating a second traffic data stream corresponding to the risk scenario based on the first traffic data stream and scenario disturbance information includes: Add scenario disturbance information corresponding to any risk scenario to the first traffic data stream to obtain the traffic data stream corresponding to the arbitrary risk scenario; Based on the traffic data stream corresponding to any one of the risk scenarios, a second traffic data stream corresponding to the risk scenario is generated.
3. The method according to claim 2, characterized in that, Any one of the risk scenarios is a potential risk scenario, and the scenario disturbance information includes the potential disturbance information corresponding to the potential risk scenario in the risk scenario; The step of adding scenario disturbance information corresponding to any risk scenario to the first traffic data stream to obtain the traffic data stream corresponding to the arbitrary risk scenario includes: Add the lowest level of potential disturbance information to the first traffic data stream to obtain the low-level traffic data stream corresponding to the potential risk scenario; If the low-level traffic data stream passes the verification, higher-level potential disturbance information is added to the low-level traffic data stream in sequence according to the disturbance level of the potential disturbance information. All traffic data streams that pass the verification are used as the traffic data streams corresponding to the potential risk scenario.
4. The method according to claim 3, characterized in that, The method further includes: If the verification of the low-level traffic data stream fails, the next level of potential disturbance information is added to the first traffic data stream according to the disturbance level order of the potential disturbance information to obtain the next level traffic data stream; If the next-level traffic data stream fails the verification, new potential disturbance information of the next level will continue to be added to the first traffic data stream until the new next-level traffic data stream passes the verification.
5. The method according to claim 3, characterized in that, The scenario disturbance information also includes other disturbance information corresponding to other risk scenarios besides the potential risk scenario in the risk scenario; The step of generating a second traffic data stream corresponding to any one of the risk scenarios based on the traffic data stream corresponding to the risk scenario includes: The other disturbance information is sequentially added to the traffic data stream corresponding to the potential risk scenario to obtain the traffic data stream corresponding to the other risk scenario; The traffic data streams corresponding to the potential risk scenarios and the traffic data streams corresponding to other risk scenarios are used as the second traffic data streams corresponding to the risk scenarios.
6. The method according to claim 5, characterized in that, The step of sequentially adding the other disturbance information to the traffic data stream corresponding to the potential risk scenario to obtain the traffic data stream corresponding to the other risk scenario includes: Obtain the risk scenario similarity between the other risk scenarios and the potential risk scenarios; Based on the similarity of the risk scenarios, different levels of other risk scenario disturbance information are added to the traffic data streams corresponding to the potential risk scenarios in sequence to obtain the traffic data streams corresponding to the other risk scenarios.
7. The method according to any one of claims 2-5, characterized in that, The step of generating a second traffic data stream corresponding to any one of the risk scenarios based on the traffic data stream corresponding to the risk scenario includes: The traffic data stream corresponding to any one of the risk scenarios is taken as the second traffic data stream corresponding to the risk scenario.
8. The method according to any one of claims 1-5, characterized in that, The scenario verification of the second traffic data stream includes: Based on the original scenarios in the original scenario library, calculate the matching degree between the second traffic data stream corresponding to the risk scenario and the original scenario library; Based on the matching degree of the original scene library, the verification result of the second traffic data stream is determined.
9. The method according to claim 8, characterized in that, The step of calculating the matching degree between the second traffic data stream corresponding to the risk scenario and the original scenario library based on the original scenarios in the original scenario library includes: Calculate the fit between the second traffic data stream corresponding to the risk scenario and the scenario library; calculate the overlap between the second traffic data stream corresponding to the risk scenario and the original scenario in the original scenario library; and quantitatively evaluate the second traffic data stream corresponding to the risk scenario according to the compliance degree quantification value to obtain the compliance fit. The adaptability, overlap, and compliance adaptability are all used as the matching degree between the second traffic data stream and the original scenario library.
10. The method according to any one of claims 1-5, characterized in that, The process of filtering out the first traffic data stream with potential risks from the original traffic data stream based on the original scenarios in the original scenario library includes: The original vehicle traffic data stream is divided into multiple segments of the original traffic data stream; Based on the original scenarios in the original scenario library, potential risks are predicted for each segment of original traffic data stream to obtain the potential risk quantification value corresponding to each segment of original traffic data stream. The original traffic data stream with a potential risk quantification value greater than a preset risk threshold is considered as the first traffic data stream with potential risk.
11. The method according to claim 10, characterized in that, The process involves predicting potential risks for each segment of the original traffic data stream based on the original scenarios in the original scenario library, resulting in a quantified value of the potential risk for each segment of the original traffic data stream, including: For any segment of original traffic data stream, calculate the probability that the original traffic data stream belongs to the original scene in the original scene database; and identify the scene danger level of the original traffic data stream to obtain the scene danger level corresponding to the original traffic data stream. By integrating the probabilities and the scenario hazard levels, a potential risk quantification value corresponding to the original traffic data stream is obtained.
12. The method according to any one of claims 1-5, characterized in that, Before generating the second traffic data stream corresponding to the risk scenario based on the first traffic data stream and the scenario disturbance information, the method further includes: The static environmental features and dynamic interaction timing features in the first traffic data stream are standardized based on standard test scenario data.
13. A scene library update device, characterized in that, The device includes: The filtering module is used to filter out the first traffic data stream with potential risks from the original traffic data stream of vehicles based on the original scenarios in the original scenario library; The generation module is used to generate a second traffic data stream corresponding to the risk scenario based on the first traffic data stream and the scenario disturbance information; The verification module is used to perform scene verification on the second traffic data stream and update the original scene library with the verified second traffic data stream.
14. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 12.
15. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 12.
16. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 12.
