Automatic driving scene retrieval method and device, electronic equipment and storage medium

Abstract

Embodiments of the invention disclose an automatic driving scene retrieval method and apparatus, an electronic device and a storage medium. The method comprises the steps of obtaining scene case files corresponding to a plurality of scene test cases; for each scene test case, performing analysis and key element extraction processing on the scene case file to obtain static scene content and dynamic scene content corresponding to each scene test case; constructing a scene shelf list based on the scene identification information, the static scene content and the dynamic scene content of each scene test case; in response to the scene retrieval request, the target scene test case is retrieved from the scene shelf list based on the target retrieval condition carried in the scene retrieval request, so that comprehensive coverage of scene elements is realized, and a composite scene containing complex interaction is accurately positioned; and a more complete and more accurate scene retrieval capability is provided for the test of the automatic driving system.

Description

Technical Field

[0001] The present invention relates to the field of autonomous driving technology, and in particular to an autonomous driving scene retrieval method, device, electronic device, and storage medium. Background Art

[0002] Autonomous driving is a growing trend in automotive development. Before self-driving cars can hit the road, they must undergo rigorous functional safety testing. Test scenarios are a core element of autonomous driving testing. Compared to the standard operating conditions used for traditional vehicle testing and verification, autonomous driving functions require a vast array of test scenarios. To fully verify autonomous driving functions, a scenario library is essential. As autonomous driving functions increase and scenarios accumulate, the number and complexity of scenarios within the library continue to grow. Faced with this vast number of scenarios, achieving accurate retrieval and classification is a pressing issue.

[0003] Current scene retrieval methods rely on manually defined scene names and tags for retrieval and location. However, with the increasing complexity of scenes and the emergence of combined scenes, relying solely on name distinctions and tags can no longer cover all scene elements. This leads to technical issues such as incomplete tag element coverage, inability to locate precise test scenes, and unsatisfactory retrieval results. Summary of the Invention

[0004] The present invention provides an autonomous driving scene retrieval method, device, electronic device, and storage medium to achieve comprehensive coverage of scene elements and precise positioning of composite scenes containing complex interactions, providing a more complete and accurate scene retrieval capability for the testing of autonomous driving systems.

[0005] In a first aspect, an embodiment of the present invention provides an autonomous driving scene retrieval method, the method comprising:

[0006] Get scenario test case files corresponding to multiple scenario test cases;

[0007] For each of the scenario test cases, the scenario case file is parsed and key elements are extracted to obtain static scene content and dynamic scene content corresponding to each of the scenario test cases; wherein the static scene content includes at least one static scene element and a first parameter value under each of the static scene elements; and the dynamic scene content includes at least one dynamic scene element and a second parameter value under each of the dynamic scene elements;

[0008] Building a scene shelf list based on the scene identification information of each of the scene test cases, the static scene content and the dynamic scene content;

[0009] In response to a scenario retrieval request, a target scenario test case is retrieved from the scenario shelf list based on a target retrieval condition carried in the scenario retrieval request.

[0010] In a second aspect, an embodiment of the present invention further provides an autonomous driving scene retrieval device, the device comprising:

[0011] A scenario file acquisition module is used to obtain scenario case files corresponding to multiple scenario test cases;

[0012] A scenario content extraction module is used to parse and extract key elements from each scenario test case and the scenario case file to obtain static scenario content and dynamic scenario content corresponding to each scenario test case; wherein the static scene content includes at least one static scene element and a first parameter value under each static scene element; and the dynamic scene content includes at least one dynamic scene element and a second parameter value under each dynamic scene element;

[0013] A scenario list construction module, configured to construct a scenario shelf list based on the scenario identification information of each scenario test case, the static scenario content, and the dynamic scenario content;

[0014] The scenario retrieval module is used to respond to a scenario retrieval request and retrieve a target scenario test case from the scenario shelf list based on a target retrieval condition carried in the scenario retrieval request.

[0015] In a third aspect, an embodiment of the present invention further provides an electronic device, the electronic device comprising:

[0016] one or more processors;

[0017] a storage device for storing one or more programs,

[0018] When one or more programs are executed by one or more processors, the one or more processors implement an autonomous driving scene retrieval method as described in any of the embodiments of the present invention.

[0019] In a fourth aspect, an embodiment of the present invention further provides a storage medium comprising computer-executable instructions, which, when executed by a computer processor, are used to execute an autonomous driving scene retrieval method as described in any one of the embodiments of the present invention.

[0020] The technical solution of the embodiment of the present invention obtains scenario use case files corresponding to multiple scenario test cases, and then parses and extracts key elements from the scenario use case files for each scenario test case to obtain the static scene content and dynamic scene content corresponding to each scenario test case. Thus, based on the scene identification information, static scene content, and dynamic scene content of each scenario test case, a scenario shelf list is constructed. On this basis, in response to a scenario search request, the target self-driving scenario is retrieved from the scenario shelf list based on the target search conditions carried in the scenario search request. The technical solution of this embodiment constructs a structured scenario shelf list by parsing the scenario use case files and extracting static and dynamic key elements, effectively solving the three major pain points of traditional manual labeling methods: first, through automated element extraction, comprehensive coverage of scene elements is achieved, avoiding omissions when manually defining labels; second, by decomposing the scene into a combined representation of static environmental features and dynamic behavioral features, it can accurately locate complex scenes containing complex interactions; finally, structured retrieval based on multi-dimensional features significantly improves query accuracy compared to single label matching, allowing testers to quickly locate target scenes through combined conditions, providing more complete and accurate scenario retrieval capabilities for testing autonomous driving systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] To more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings introduced here only illustrate some of the embodiments to be described by the present invention, and are not exhaustive. A person skilled in the art can derive other drawings based on these drawings without inventive effort.

[0022] Figure 1 A schematic diagram of a flow chart of an autonomous driving scene retrieval method provided by an embodiment of the present invention;

[0023] Figure 2 A schematic diagram of the structure of an autonomous driving scene retrieval device provided by an embodiment of the present invention;

[0024] Figure 3 The present invention provides a schematic structural diagram of an electronic device. DETAILED DESCRIPTION

[0025] The present invention will be further described in detail below with reference to the accompanying drawings and examples. It will be understood that the specific embodiments described herein are intended only to illustrate the present invention and are not intended to limit the present invention. It should also be noted that, for ease of description, the accompanying drawings only illustrate portions relevant to the present invention, not all structures.

[0026] It should be noted that similar reference numerals and letters represent similar items in the following figures. Therefore, once an item is defined in one figure, it does not need to be further defined or explained in subsequent figures. Furthermore, in the description of the present invention, the terms "first," "second," etc. are used only to distinguish descriptions and should not be understood to indicate or imply relative importance. The acquisition, storage, use, and processing of data in the technical solution of this application comply with the relevant provisions of national laws and regulations.

[0027] It should be noted that in the embodiments of the present application, certain software, components, models and other existing solutions in the industry may be mentioned. They should be regarded as exemplary. Their purpose is only to illustrate the feasibility of implementing the technical solution of the present application, but it does not mean that the applicant has or will necessarily use the solution.

[0028] The acquisition, storage, use, and processing of data in this application's technical solution comply with relevant national laws and regulations.

[0029] Below, the present application is further described in conjunction with the accompanying drawings and specific implementation methods. It should be noted that, under the premise of no conflict, the various embodiments or technical features described below can be arbitrarily combined to form new embodiments.

[0030] Before introducing this technical solution, an exemplary application scenario can be first described. This technical solution can be applied in a scenario where a target test scene needs to be retrieved from a large number of autonomous driving test scenes.

[0031] With the development of vehicle intelligence, autonomous driving has become a general trend. Autonomous driving vehicles must undergo rigorous functional safety testing before they can be put on the road in order to minimize the accident rate. Test scenarios are the core elements of autonomous driving functional safety testing. In order to fully verify the autonomous driving function, it is necessary to establish a scenario library containing a large number of scenarios. How to quickly and accurately retrieve the scenarios that need to be tested in the scenario library is an urgent problem to be solved. The current mainstream method is to retrieve and locate by manually defining scene names and labeling. There are problems such as incomplete coverage of label elements, inability to locate values, and low efficiency of manual labeling. The present invention proposes a scene retrieval method based on scene file parsing. By extracting relevant information elements and their corresponding values ​​in the scene use case file, the extracted information is formed into a multi-data dimension scene shelf list, thereby realizing scene retrieval operations based on the scene shelf list, and realizing rapid retrieval, classification and positioning of scenes.

[0032] Figure 1This is a flow chart of an autonomous driving scene retrieval method provided in an embodiment of the present invention. This embodiment is applicable to scenarios where it is necessary to retrieve a target test scene from a large number of autonomous driving test scenes. The method can be executed by an autonomous driving scene retrieval device, which can be implemented in the form of software and / or hardware. The hardware can be an electronic device, such as a mobile terminal, PC, or server.

[0033] like Figure 1 As shown, the autonomous driving scene retrieval method includes:

[0034] S110: Obtain scenario use case files corresponding to multiple scenario test cases.

[0035] A scenario test case is a standardized description of a specific driving scenario designed during the development and testing of autonomous driving technology. It uses a scenario test case file to define in detail the environmental elements and event sequences within that scenario. Each scenario test case can include a unique identifier, static scenario content, and dynamic scenario content. For example, static scenario content can include fixed environmental elements and their parameters, such as road structure, traffic signs, and weather conditions; dynamic scenario content can include the motion states of vehicles, pedestrians, and obstacles, as well as their changing parameters. This structured data together constitutes a reusable test unit, used to simulate and verify autonomous vehicle behavior in real-world or edge scenarios. For example, a typical scenario such as "turning left at night at a city intersection and encountering a pedestrian running a red light" can be encapsulated as a complete scenario test case. A scenario test case file is a file that stores the structured data for a scenario test case. It can use a standardized format and fully records the static environmental configuration and dynamic event sequence for a specific driving scenario. This scenario test case file encapsulates all key features of the test scenario in a machine-readable format, enabling the autonomous driving system or testing tool to accurately interpret and reproduce the scenario.

[0036] In this embodiment, first, scenario use case files corresponding to multiple scenario test cases can be imported in batches from the autonomous driving test database, simulation platform, or the enterprise's internal scenario management system; second, scenario description files in open source data sets can be captured through automated scripts or API interfaces, and files from different sources can be formatted and standardized, and finally integrated into a uniformly stored scenario file collection to ensure that each file contains complete static environment configuration and dynamic event sequence information.

[0037] S120 , for each scenario test case, the scenario test case file is parsed and key elements are extracted to obtain the static scenario content and dynamic scenario content corresponding to each scenario test case.

[0038] Static scene content refers to the fixed physical environment elements and their parameters within the autonomous driving test scenario. Dynamic scene content describes the behavioral characteristics and time-varying parameters of movable objects within the scene. Static scene content includes at least one static scene element and a first parameter value for each static scene element; dynamic scene content includes at least one dynamic scene element and a second parameter value for each dynamic scene element. Static scene elements are the basic elements that constitute the fixed environment framework within the autonomous driving test scenario; dynamic scene elements refer to all object entities within the scene that exhibit time-varying behavior.

[0039] In this embodiment, the static scene content can be composed of at least one static scene element of road topology information, lane line information, traffic light information, road attachment information and road sign information, and the first parameter values ​​corresponding to these elements, such as a road width of 5 meters, a number of lane lines of 2, etc., which together constitute the reference framework of the scene. Among them, the road topology information describes the basic physical structural characteristics of the road, including the number of lanes (such as 6 lanes in both directions), lane size (width of 3.5 meters), curvature (turning radius of 200 meters), material (asphalt pavement) and intersection type (cross intersection) and other core parameters that determine the geometry of the road; the lane line information specifically defines the lane division characteristics, including the lane line type (solid line / dashed line), acceptance results (in accordance with GB5768 standards) and size (line width of 15 cm) and other key factors that affect vehicle positioning; the traffic light information standardizes signal control The properties of the control device include timing control parameters such as type (arrow light / disc light), quantity (4 groups), lighting rules (red light-green light-yellow light cycle) and lighting duration (red light 30 seconds); road accessory information includes fixed objects around the road such as guardrails (corrugated beam guardrails), buildings (shops 10 meters away from the roadside) and obstacles (construction cones); road sign information includes visual guidance elements such as traffic signs (speed limit 60 signs) and ground signs (pedestrian crossing markings). These static elements together constitute the basic framework of environmental perception of autonomous vehicles.

[0040] The dynamic scene content can be composed of the vehicle information to be tested, the vehicle interference information of at least one interfering vehicle, and at least one dynamic scene element in the dynamic trigger event, as well as the second parameter values ​​corresponding to these elements, such as the target vehicle acceleration of 2m / s 2, pedestrian appearance time t = 3s, etc., together constitute the dynamic evolution logic of the scene. Among them, the vehicle information under test refers to the basic attributes and initial state parameters of the autonomous driving vehicle under test, including vehicle type (such as SUV), size (length 4.8 meters / width 1.9 meters), initial position (center line of lane 2) and speed (60km / h) and other factors that define the main vehicle characteristics; vehicle interference information describes the dynamic attributes of other traffic participants in the scene, including the number of interfering vehicles (3), type (truck / sedan), size (length 12 meters / width 2.5 meters), initial position (adjacent lane) and speed (50-70km / h range), etc., which constitute the variables of complex traffic flow; dynamic trigger events refer to preset time sequence interaction rules, including key parameters such as event type (cut-in / sudden braking), trigger conditions (distance <5m), trigger position (curve entrance) and trigger time (t = 15s). The three together constitute the core logic chain of vehicle behavior interaction in dynamic scenarios, which are used to simulate various challenging working conditions in real road environments.

[0041] Specifically, for these scenario test cases, the parsing and key element extraction processes of each scenario test case are consistent. In order to clearly introduce this technical solution, one of the scenario test cases can be used as an example. First, the format parser can be used to deconstruct the scenario case file hierarchy and extract the original data; then, the semantic analysis engine can be used to classify the invariant elements such as road topology and traffic facilities as static scene content based on predefined scene element classification rules and extract the first parameter value (such as lane width value), while identifying the time-varying elements such as vehicle trajectory and pedestrian behavior as dynamic scene content and extracting the second parameter value (such as acceleration time series); finally, the extracted parameters are unified and verified for validity through the data verification module, and finally structured static scene content and dynamic scene content with timestamps are generated to provide standardized data input for scene shelf construction.

[0042] Optionally, the scenario use case file includes a road description file and a scenario description file. The road description file refers to a standardized file that specifically records static environmental elements in the autonomous driving test scenario, such as an OpenDRIVE format file. It mainly contains fixed scene parameters such as road geometry, infrastructure, and environmental properties; the scenario description file is a supporting file that defines a dynamic event sequence, such as an OpenSCENARIO format file, which focuses on describing time-varying elements such as traffic participant behavior, trigger conditions, and object interaction logic. The two form a complete scene description through a file association mechanism, and together constitute the scenario use case file of the autonomous driving test case. On this basis, the scenario description file is parsed and key elements are extracted to obtain the static scene content and dynamic scene content corresponding to each scenario test case. The specific implementation steps may include:

[0043] S1. Parse the road description file and the scene description file based on a preset parsing standard to obtain a road parsing file and a scene parsing file.

[0044] The preset parsing standard refers to a predefined set of automated parsing rules that guide the structured processing of files. The road parsing file is an intermediate data representation generated by parsing the road description file, fully preserving the hierarchical relationships and parameter attributes of static elements such as the original road topology and traffic facilities. The scene parsing file is the structured output of the scene description file after parsing and conversion, systematically organizing the behavioral logic and spatiotemporal parameters of dynamic elements. These two parsing files, as standardized intermediate products, provide a unified data interface and operational foundation for subsequent key element extraction.

[0045] In this embodiment, a road description file that complies with the ASAM OpenDRIVE standard and a scene description file in the OpenSCENARIO format can be loaded, and the road description file and the scene description file can be converted into a structured object model in memory through a preset parsing standard; then, according to the preset mapping rules, for example, <road>Labels are mapped to road objects, <scenario>The event sequence is converted into a timeline data structure, and the geometric elements in the road description file and the entity behaviors in the scene description file are extracted respectively, and finally a road parsing file containing complete topological information and a scene parsing file with a time event chain are generated.

[0046] S2. Extract a first parameter value from at least one static scene element in the road parsing file based on the target field extraction model to obtain static scene content.

[0047] The target field extraction model is an intelligent processing module used to automatically identify and extract key parameters from parsed road analysis files. Built based on predefined scene element classification rules or machine learning algorithms, this model accurately locates target data fields within the file, such as lane width values ​​and vehicle acceleration curves. Through data conversion rules and validity checks, it outputs first parameter values ​​that meet specifications, thereby converting the original parsed file into a standardized scene element dataset that can be directly used by the scene shelf.

[0048] In this embodiment, the structured road parsing file can be loaded and input into the target field extraction model. The target field extraction model locates the target data field in the road parsing file through pre-configured static element feature recognition rules, that is, the field content corresponding to at least one static scene element; then, the specific parameter value under each target data field is extracted, that is, the first parameter value in each static scene element. Finally, these static scene elements and the corresponding first parameter values ​​are regularly integrated to obtain the static scene content.

[0049] Optionally, after extracting the specific parameter values ​​under each target data field, a multi-level verification mechanism can be used to process the parameters, including format conversion, such as converting the string "3.5m" to the floating-point number 3.5; unit standardization, such as uniformly converting to metric units; spatial coordinate conversion, such as converting the local coordinate system to the world coordinate system; finally, the parameter association engine is used to reorganize the extracted discrete parameters into a complete static scene content object according to the scene logic, and the original field position of the data source is marked for traceability, and a static scene data structure that meets industry standards is output.

[0050] S3. Extract a second parameter value from at least one dynamic scene element in the scene parsing file based on the dynamic content extraction model to obtain dynamic scene content.

[0051] Among them, the dynamic content extraction model refers to an intelligent processing system specifically used to identify and extract time-varying parameters from scene parsing files. By integrating temporal pattern recognition algorithms, spatial relationship reasoning engines and multi-entity interaction analysis modules, the model can accurately extract the spatiotemporal behavior parameters of dynamic elements from complex scene data, and convert the original data into standardized second parameter values ​​through spatiotemporal alignment and physical rule verification, ultimately generating dynamic scene content with strict time synchronization and causal logic.

[0052] In this embodiment, the structured scene parsing file can be loaded and input into the dynamic content extraction model. The dynamic event stream in the scene parsing file is parsed by the time series analysis engine in the dynamic content extraction model, and the key dynamic element nodes are located using the spatiotemporal pattern recognition algorithm, such as vehicle lane change events and pedestrian crossing actions. Then, multi-dimensional parameter extraction technology is used to extract the dynamic element nodes from the event triggering conditions (ifDistance<5m), spatiotemporal constraints (t=10s@x=150m,y=3.2m) and behavioral parameters (acceleration 2.5m / s 2 ) three dimensions to extract the normalized second parameter value; then, through the dynamic scene builder, these parameters are temporally and spatially associated with the corresponding entities to generate a scene event sequence with a timestamp; finally, the time-space-state three-dimensional dynamic scene description data that conforms to the OSI standard is output, which is the dynamic scene content.

[0053] S130. Build a scene shelf list based on the scene identification information, static scene content, and dynamic scene content of each scene test case.

[0054] Scenario identification information refers to metadata used to uniquely identify and categorize autonomous driving test cases. The scenario shelf list is a scenario repository organized in a database or index structure. It uses multi-dimensional feature vectors to structure the parsed scenario content. By establishing a bidirectional mapping relationship between scenario identification information and scene content, it forms an efficient retrieval system that supports multi-condition combination queries. Essentially, it is a standardized autonomous driving scenario feature database.

[0055] Specifically, the specific implementation method of constructing the scene shelf list is: create a unique index entry for each scene test case, use the scene identification information as the primary key, and establish an association mapping with the static scene content and the dynamic scene content to obtain the scene shelf list.

[0056] Optionally, based on the scene identification information, static scene content and dynamic scene content of each scene test case, the specific implementation method of constructing a scene shelf list may include: generating a structured list of the static scene content and the dynamic scene content according to the scene identification information to obtain the scene shelf list.

[0057] Among them, the scene shelf list has scene identification information as rows, at least one static scene element and at least one dynamic scene element as columns, and the first parameter value and the second parameter value as cell content.

[0058] In this embodiment, a two-dimensional relational data structure is first established, with scenario identification information as the primary key. Each row corresponds to an independent test case, and the columns are divided into two categories: static scenario elements and dynamic scenario elements. A hierarchical storage strategy is then employed to structure the storage of parameter values ​​within the cells. Specifically, static first parameter values ​​are stored as key-value pairs, while dynamic second parameter values ​​are recorded using time series arrays. This ultimately creates a hybrid scenario database that maintains the integrity of the original data while supporting efficient queries.

[0059] S140 : In response to the scenario retrieval request, retrieve a target scenario test case from the scenario shelf list based on the target retrieval condition carried in the scenario retrieval request.

[0060] A scenario retrieval request is a query command initiated by a user or system, searching the scenario library for matching scenarios by specifying search criteria. The target retrieval criteria are the specific query criteria carried in the retrieval request. These criteria include a combination of at least one static scene element, at least one dynamic scene element, and specific parameter values, which define the retrieval matching rules. The target scenario test case is the final retrieved scenario that meets all the criteria and can be directly used for testing and verification of the autonomous driving system or algorithm training.

[0061] In this embodiment, the user can edit or circle the target search condition in the search condition editing area. Once the user completes editing the target search condition and triggers the confirmation control, a scenario search request is generated. At this point, the target search condition in the scenario search request is converted into a standardized query condition. A multi-level index matching engine then combines the standardized query condition with the search scenario shelf list to obtain one or more target scenario test cases that meet the target search condition.

[0062] Exemplary target retrieval conditions may include static element conditions, such as "road type = highway"; dynamic element conditions, such as "there is vehicle cutting-in behavior"; and parameter range conditions, such as "the initial speed range of the vehicle to be tested is [60,80] km / h." During the specific retrieval process, the scenario shelf list can be searched in parallel through a multi-level index matching engine. The inverted index is first used to quickly filter the subset of scenarios that meet the static conditions. Dynamic behavior patterns are then matched within the candidate set using a time series analysis algorithm. Finally, the parameter verification module is used to accurately match specific parameter values. A weighted scoring strategy is used to calculate the scenario matching degree for complex query conditions, returning a result list sorted by relevance. The result is then linked to the complete scenario test case data through scenario identification information, ultimately outputting a target scenario set that meets all retrieval conditions.

[0063] The technical solution of the embodiment of the present invention obtains scenario use case files corresponding to multiple scenario test cases, and then parses and extracts key elements from the scenario use case files for each scenario test case to obtain the static scene content and dynamic scene content corresponding to each scenario test case. Based on the scene identification information, static scene content, and dynamic scene content of each scenario test case, a scenario shelf list is constructed. On this basis, in response to a scenario search request, the target self-driving scenario is retrieved from the scenario shelf list based on the target search conditions carried in the scenario search request. The technical solution of this embodiment constructs a structured scenario shelf list by parsing the scenario use case files and extracting static and dynamic key elements, effectively solving the three major pain points of traditional manual labeling methods: first, through automated element extraction, comprehensive coverage of scene elements is achieved, avoiding omissions when manually defining labels; second, by decomposing the scene into a combined representation of static environmental features and dynamic behavioral features, it can accurately locate complex scenes containing complex interactions; finally, structured retrieval based on multi-dimensional features significantly improves query accuracy compared to single label matching, allowing testers to quickly locate target scenes through combined conditions, providing more complete and accurate scenario retrieval capabilities for testing autonomous driving systems.

[0064] Based on the above embodiment, optionally, the autonomous driving scenario retrieval method also includes: when a scenario update event of a scenario test case is detected, updating the scenario shelf list based on the updated scenario case file to retrieve the target scenario test case based on the updated scenario shelf list.

[0065] Scenario update events refer to any operational behavior in the autonomous driving scenario library that causes a change in scenario content, including but not limited to file system changes such as the creation, modification, and deletion of scenario use case files. These events are captured in real time through file monitoring services or message middleware. Each event contains at least one of the following metadata: scenario identifier, change type, and version stamp. These events are used to trigger the update process of the scenario shelf list, ensuring the timeliness and consistency of the test scenario library.

[0066] In this embodiment, the update process may include: deleting the line corresponding to the removed scene description file, and updating the scene element information and parameter value information corresponding to the modified scene description file.

[0067] In the specific application process, the file monitoring service can be used to monitor the change events of the scenario use case files in real time. When a file update is detected, that is, when a scenario update event is detected, the update processing flow can be triggered. Specifically: for deletion events, the corresponding row records in the shelf list are quickly located based on the scenario identification information, and the associated static feature index and dynamic event index are cascaded and deleted at the same time; for modification events, a difference comparison algorithm is used to identify the changed static / dynamic elements, and the corresponding parameter values ​​in the shelf list are synchronously modified through atomic update operations, and the relevant inverted index is rebuilt; during the update process, a double buffering mechanism is used to maintain the availability of the query service, and after the data is fully synchronized, the version number is switched to achieve imperceptible update, ultimately ensuring that the scenario shelf list is consistent with the latest test case file in real time, while maintaining the retrieval performance unaffected.

[0068] Example 2

[0069] Figure 2 This is a structural diagram of an autonomous driving scene retrieval device provided by an embodiment of the present invention, which includes: a scene file acquisition module 210, a scene content extraction module 220, a scene list construction module 230 and a scene retrieval module 240.

[0070] The scenario file acquisition module 210 is used to obtain scenario use case files corresponding to multiple scenario test cases;

[0071] The scenario content extraction module 220 is configured to parse and extract key elements from the scenario test case file for each scenario test case to obtain static scenario content and dynamic scenario content corresponding to each scenario test case; wherein the static scenario content includes at least one static scenario element and a first parameter value for each static scenario element; and the dynamic scenario content includes at least one dynamic scenario element and a second parameter value for each dynamic scenario element.

[0072] A scenario list construction module 230 is configured to construct a scenario shelf list based on the scenario identification information of each scenario test case, the static scenario content, and the dynamic scenario content;

[0073] The scenario retrieval module 240 is configured to respond to a scenario retrieval request and retrieve a target scenario test case from the scenario shelf list based on a target retrieval condition carried in the scenario retrieval request.

[0074] The technical solution of the embodiment of the present invention obtains scenario use case files corresponding to multiple scenario test cases, and then parses and extracts key elements from the scenario use case files for each scenario test case to obtain the static scene content and dynamic scene content corresponding to each scenario test case. Based on the scene identification information, static scene content, and dynamic scene content of each scenario test case, a scenario shelf list is constructed. On this basis, in response to a scenario search request, the target self-driving scenario is retrieved from the scenario shelf list based on the target search conditions carried in the scenario search request. The technical solution of this embodiment constructs a structured scenario shelf list by parsing the scenario use case files and extracting static and dynamic key elements, effectively solving the three major pain points of traditional manual labeling methods: first, through automated element extraction, comprehensive coverage of scene elements is achieved, avoiding omissions when manually defining labels; second, by decomposing the scene into a combined representation of static environmental features and dynamic behavioral features, it can accurately locate complex scenes containing complex interactions; finally, structured retrieval based on multi-dimensional features significantly improves query accuracy compared to single label matching, allowing testers to quickly locate target scenes through combined conditions, providing more complete and accurate scenario retrieval capabilities for testing autonomous driving systems.

[0075] Based on the above device, optionally, the scenario use case file includes a road description file and a scenario description file, and the scenario content extraction module 220 includes:

[0076] A file parsing unit, configured to parse the road description file and the scene description file based on a preset parsing standard to obtain a road parsing file and a scene parsing file;

[0077] a static content determination unit, configured to extract a first parameter value of at least one static scene element in the road parsing file based on a target field extraction model to obtain static scene content;

[0078] The dynamic content determination unit is configured to extract a second parameter value from at least one dynamic scene element in the scene parsing file based on a dynamic content extraction model to obtain dynamic scene content.

[0079] Based on the above-mentioned device, optionally, the static scene elements include at least one of road topology information, lane line information, traffic light information, road attachment information and road sign information; the road topology information includes at least one of the number of lanes, lane size, curvature, material and intersection type; the lane line information includes at least one of lane line type, acceptance result and lane line size; the traffic light information includes at least one of traffic light type, number, lighting rules and lighting duration; the road attachment information includes at least one of guardrails, buildings and obstacles, and the road sign information includes at least one of traffic sign information and ground sign information.

[0080] Based on the above-mentioned device, optionally, the dynamic scene elements include: vehicle information to be tested, vehicle interference information of at least one interfering vehicle, and at least one of dynamic triggering events; the vehicle information to be tested includes at least one of vehicle type, size, initial position, and vehicle speed; the vehicle interference information includes at least one of the number, type, size, initial position, and speed of interfering vehicles; the dynamic triggering event includes at least one of event type, triggering condition, triggering position, and triggering time.

[0081] Based on the above-mentioned device, optionally, a scene content extraction module 230 is specifically used to generate a structured list of the static scene content and the dynamic scene content according to the scene identification information to obtain a scene shelf list; wherein, the scene shelf list has scene identification information as rows, at least one static scene element and at least one dynamic scene element as columns, and the first parameter value and the second parameter value as cell content.

[0082] Based on the above device, optionally, the autonomous driving scene retrieval device further includes: a list updating module;

[0083] The list update module is used to update the scenario shelf list based on the updated scenario case file when a scenario update event of the scenario test case is detected, so as to retrieve the target scenario test case based on the updated scenario shelf list.

[0084] Based on the above device, optionally, the update processing includes: deleting the line corresponding to the removed scene description file, and updating the scene element information and parameter value information corresponding to the modified scene description file.

[0085] The autonomous driving scene retrieval device provided in an embodiment of the present invention can execute the autonomous driving scene retrieval method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0086] It is worth noting that the various units and modules included in the above system are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the embodiments of the present invention.

[0087] Example 3

[0088] Figure 3 The present invention provides a schematic structural diagram of an electronic device. Figure 3 A block diagram of an exemplary electronic device 30 suitable for implementing exemplary embodiments of the present invention is shown. Figure 3 The electronic device 30 shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0089] like Figure 3 As shown, electronic device 30 is a general-purpose computing device. Components of electronic device 30 may include, but are not limited to, one or more processors or processing units 301, system memory 302, and a bus 303 connecting various system components (including system memory 302 and processing unit 301).

[0090] Bus 303 represents one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor, or a local bus using any of a variety of bus architectures. Examples of these architectures include, but are not limited to, the Industry Standard Architecture (ISA) bus, the Micro Channel Architecture (MAC) bus, the Enhanced ISA bus, the Video Electronics Standards Association (VESA) local bus, and the Peripheral Component Interconnect (PCI) bus.

[0091] The electronic device 30 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by the electronic device 30, including volatile and non-volatile media, removable and non-removable media.

[0092] System memory 302 may include computer system readable media in the form of volatile memory, such as random access memory (RAM) 304 and / or cache memory 305. Electronic device 30 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, storage system 306 may be used to read and write non-removable, non-volatile magnetic media ( Figure 3 Not shown, usually called a "hard drive"). Although Figure 3 Not shown, a disk drive for reading and writing to a removable non-volatile disk (e.g., a "floppy disk"), and an optical disk drive for reading and writing to a removable non-volatile optical disk (e.g., a CD-ROM, DVD-ROM, or other optical media) may be provided. In these cases, each drive may be connected to bus 303 via one or more data medium interfaces. Memory 302 may include at least one program product having a set (e.g., at least one) of program modules configured to perform the functions of various embodiments of the present invention.

[0093] A program / utility 308 having a set (at least one) of program modules 307 may be stored, for example, in memory 302. Such program modules 307 include, but are not limited to, an operating system, one or more application programs, other program modules, and program data, each of which, or some combination thereof, may include an implementation of a network environment. Program modules 307 generally implement the functions and / or methods of the embodiments described herein.

[0094] The electronic device 30 may also communicate with one or more external devices 309 (e.g., keyboard, pointing device, display 810, etc.), and may also communicate with one or more devices that enable a user to interact with the electronic device 30, and / or any device that enables the electronic device 30 to communicate with one or more other computing devices (e.g., network card, modem, etc.). Such communication may be performed through an input / output (I / O) interface 311. Furthermore, the electronic device 30 may also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) through a network adapter 312. As shown, the network adapter 312 communicates with other modules of the electronic device 30 via the bus 303. It should be understood that although Figure 3 Not shown, other hardware and / or software modules may be used in conjunction with the electronic device 30, including but not limited to microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0095] The processing unit 301 executes various functional applications and page processing by running the programs stored in the system memory 302, such as implementing the autonomous driving scene retrieval method provided in an embodiment of the present invention.

[0096] Example 4

[0097] An embodiment of the present invention further provides a storage medium containing computer-executable instructions. When the computer-executable instructions are executed by a computer processor, the computer-executable instructions are used to perform an autonomous driving scene retrieval method, the method comprising:

[0098] Get scenario test case files corresponding to multiple scenario test cases;

[0099] For each of the scenario test cases, the scenario case file is parsed and key elements are extracted to obtain static scene content and dynamic scene content corresponding to each of the scenario test cases; wherein the static scene content includes at least one static scene element and a first parameter value under each of the static scene elements; and the dynamic scene content includes at least one dynamic scene element and a second parameter value under each of the dynamic scene elements;

[0100] Building a scene shelf list based on the scene identification information of each of the scene test cases, the static scene content and the dynamic scene content;

[0101] In response to a scenario retrieval request, a target scenario test case is retrieved from the scenario shelf list based on a target retrieval condition carried in the scenario retrieval request.

[0102] The computer storage medium of the embodiment of the present invention may adopt any combination of one or more computer-readable media. The computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or component, or any combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device.

[0103] A computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0104] Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0105] Computer program code for performing the operations of embodiments of the present invention can be written in one or more programming languages, or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0106] Note that the above are only preferred embodiments of the present invention and the technical principles employed. Those skilled in the art will appreciate that the present invention is not limited to the specific embodiments herein, and that various obvious changes, readjustments, and substitutions are possible for those skilled in the art without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in detail through the above embodiments, the present invention is not limited to the above embodiments and may include many other equivalent embodiments without departing from the scope of the present invention. The scope of the present invention is determined by the scope of the appended claims.< / scenario> < / road>

Claims

1. A method for retrieving an autonomous driving scene, characterized in that: include: Get scenario test case files corresponding to multiple scenario test cases; For each of the scenario test cases, the scenario case file is parsed and key elements are extracted to obtain static scene content and dynamic scene content corresponding to each of the scenario test cases; wherein the static scene content includes at least one static scene element and a first parameter value under each of the static scene elements; and the dynamic scene content includes at least one dynamic scene element and a second parameter value under each of the dynamic scene elements; Building a scene shelf list based on the scene identification information of each of the scene test cases, the static scene content and the dynamic scene content; In response to a scenario retrieval request, a target scenario test case is retrieved from the scenario shelf list based on a target retrieval condition carried in the scenario retrieval request.

2. The method according to claim 1, characterized in that The scenario test case file includes a road description file and a scenario description file. The scenario description file is parsed and key elements are extracted to obtain static scenario content and dynamic scenario content corresponding to each scenario test case, including: Parsing the road description file and the scene description file based on a preset parsing standard to obtain a road parsing file and a scene parsing file; extracting a first parameter value of at least one static scene element in the road parsing file based on a target field extraction model to obtain static scene content; A second parameter value in at least one dynamic scene element in the scene parsing file is extracted based on the dynamic content extraction model to obtain dynamic scene content.

3. The method according to claim 2, characterized in that The static scene elements include at least one of road topology information, lane line information, traffic light information, road attachment information and road sign information; the road topology information includes at least one of the number of lanes, lane size, curvature, material and intersection type; the lane line information includes at least one of lane line type, acceptance result and lane line size; the traffic light information includes at least one of traffic light type, number, lighting rules and lighting duration; the road attachment information includes at least one of guardrails, buildings and obstacles, and the road sign information includes at least one of traffic sign information and ground sign information.

4. The method according to claim 2, characterized in that The dynamic scene elements include: vehicle information to be tested, vehicle interference information of at least one interfering vehicle, and at least one of dynamic triggering events; the vehicle information to be tested includes at least one of vehicle type, size, initial position, and vehicle speed; the vehicle interference information includes at least one of the number, type, size, initial position, and speed of interfering vehicles; the dynamic triggering event includes at least one of event type, triggering condition, triggering position, and triggering time.

5. The method according to claim 1, characterized in that The step of constructing a scene shelf list based on the scene identification information of each scene test case, the static scene content, and the dynamic scene content includes: Generate a structured list of the static scene content and the dynamic scene content according to the scene identification information to obtain a scene shelf list; wherein, the scene shelf list has scene identification information as rows, at least one static scene element and at least one dynamic scene element as columns, and the first parameter value and the second parameter value as cell content.

6. The method according to claim 1, wherein The method further comprises: When a scenario update event of a scenario test case is detected, the scenario shelf list is updated based on the updated scenario case file, so as to retrieve the target scenario test case based on the updated scenario shelf list.

7. The method according to claim 6, characterized in that The update process includes: deleting the row corresponding to the removed scene description file, and updating the scene element information and parameter value information corresponding to the modified scene description file.

8. An autonomous driving scene retrieval device, characterized in that: include: A scenario file acquisition module is used to obtain scenario case files corresponding to multiple scenario test cases; A scenario content extraction module is used to parse and extract key elements from each scenario test case and the scenario case file to obtain static scenario content and dynamic scenario content corresponding to each scenario test case; wherein the static scene content includes at least one static scene element and a first parameter value under each static scene element; and the dynamic scene content includes at least one dynamic scene element and a second parameter value under each dynamic scene element; A scenario list construction module, configured to construct a scenario shelf list based on the scenario identification information of each scenario test case, the static scenario content, and the dynamic scenario content; The scenario retrieval module is used to respond to a scenario retrieval request and retrieve a target scenario test case from the scenario shelf list based on a target retrieval condition carried in the scenario retrieval request.

9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the autonomous driving scene retrieval method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the autonomous driving scene retrieval method according to any one of claims 1 to 7 when executed.

Images