Intelligent connected vehicle test scenario feature collection, construction and application method

Abstract

The present invention relates to the technical field of autonomous driving test scenarios and discloses a method for collecting, constructing, and applying features for intelligent connected vehicle test scenarios. The method uses a vehicle's own onboard unit to collect road signs and classify them according to a set classification pattern. The method then identifies the sign information according to the classification and extracts key information from the sign according to a set recognition method to obtain corresponding road parameter information and form road features. The method reduces the amount of data collected, simplifies data complexity, and enables rapid construction of road scenarios. This method meets the needs of scenario construction and research requiring large sample sizes and low precision, thereby reducing construction difficulty, shortening the construction cycle, reducing construction costs, and improving road scenario utilization.

Description

Technical Field

[0001] The present invention relates to the technical field of autonomous driving test scenarios, and in particular to methods for collecting, constructing, and applying test scenario features of intelligent connected vehicles. Background Art

[0002] With the continuous advancement of intelligent driving technology, the safety requirements for intelligent driving systems are becoming increasingly stringent. To improve intelligent driving technology, intelligent driving systems must be tested during development, and driving performance must be continuously improved and enhanced based on the test results. To meet these testing requirements and provide an adequate testing environment, test scenarios are currently constructed. To improve test accuracy, existing test scenarios are constructed based on existing road scene features to closely replicate the real driving environment. This acquisition and construction method generally requires the installation of various sensors on the vehicle to comprehensively collect road feature information to ensure accurate reproduction. While this construction method offers high accuracy, it also comes at a high cost, and the construction cycle for each test scenario is typically very long. Scenario construction involving large sample sizes is particularly costly and time-consuming. The high cost and time consumption of scenario construction makes it difficult to meet the diverse needs of scenario construction.

[0003] In addition, this type of collection equipment is generally exposed outside the vehicle, and has certain requirements for the equipment's installation, durability, and waterproofness. In addition, different equipment manufacturers have different recognition rates, which makes it difficult to unify the collected feature data and difficult to synchronize and balance in actual test applications, making it difficult to ensure test accuracy. Summary of the Invention

[0004] The present invention aims to provide a method for collecting, constructing and applying test scenario features of intelligent connected vehicles to solve the problems of high cost and long time consumption in the prior art of test scenario construction.

[0005] To achieve the above-mentioned objectives, the present invention adopts the following technical solutions: a method for collecting test scene features of intelligent connected vehicles, which is used to quickly construct my country's road feature information, reduce the cost and market of data feature collection, and provide support for the construction of an autonomous driving scene library. The method includes using the vehicle's own on-board unit to collect road signs, and identifying and classifying the sign signs according to a set classification mode; identifying the sign sign information according to the classification, and extracting the key information from the sign sign according to the set identification method, obtaining the corresponding road parameter information, and forming road features.

[0006] The principles and advantages of this solution are:

[0007] There is a stereotype about the construction of intelligent connected vehicle test scenarios. It is believed that the scene simulation used for unmanned driving tests needs to restore the road environment as much as possible, and make the unmanned driving test more accurate based on the real road environment, thereby improving the unmanned driving performance. Therefore, there are higher requirements for accuracy when constructing the road model. Therefore, there will be high requirements for the accuracy of the collected road size, digital features, environment, etc., so a variety of sensors will be installed in different positions of the vehicle to meet the collection of data with different accuracies.

[0008] Although the ultimate goal of scene construction is to meet the needs of driving tests, the purpose of scene construction is also to simulate the road characteristics of my country. For scenes with a large sample size and a lot of road information, if the data volume is accurately restored, the information is large and difficult to process, and it is time-consuming and costly. It is very troublesome to quickly obtain the road characteristics of a certain area, and it is difficult to meet the needs of rapid construction and analysis of road characteristics.

[0009] Therefore, this solution has developed a method for rapidly constructing road scenarios and effectively researching and analyzing them. This method doesn't rely on the need to replicate roads by accurately controlling and realistically reproducing the road environment. Instead, it focuses on meeting the analysis needs of road characteristics. By rationally selecting objects and effectively constructing road information based on object identification and information extraction, it deemphasizes the road itself and, therefore, doesn't rely on parameters such as size and numerical features. This reduces data collection and simplifies data complexity. Based on a different R&D approach, this solution breaks away from construction pitfalls and rapidly constructs road scenarios, meeting the needs of scenario construction and research with large sample sizes and low precision requirements. This reduces construction difficulty, shortens the construction cycle, reduces construction costs, and improves road scenario utilization.

[0010] The present invention also provides a method for constructing test scene features of intelligent connected vehicles, which is applied to the above-mentioned method for collecting test scene features of intelligent connected vehicles. It also includes collecting the corresponding geographical locations of road signs, and establishing a corresponding relationship between the geographical location coordinates and road parameter information, so that the formed road scene features correspond to the geographical location coordinates and are stored in a data acquisition system.

[0011] This invention also provides a method for applying test scenario features for intelligent connected vehicles, providing a driving scenario library for rapidly building road information to form test scenarios. This rapid construction of road information in my country supports the autonomous driving scenario library, thereby reducing the development and labor costs of test road scenario construction, expanding the application scope of scenario features, and improving the flexibility and utilization of scenario construction. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 Schematic diagram of the process of the collection method of embodiment 1 of the present invention;

[0013] Figure 2This is a flow chart of the construction method of the second embodiment of the present invention. DETAILED DESCRIPTION

[0014] The following is further described in detail through specific implementation methods:

[0015] Example 1

[0016] The intelligent connected vehicle test scene feature collection method in this embodiment is used to solve the problem that the existing test scene feature collection is time-consuming and costly, thereby improving the test scene collection efficiency while meeting the test scene requirements.

[0017] Specifically, as attached Figure 1 As shown, the signboards on the road are collected by the vehicle's own on-board unit, and the signboards are identified and classified according to the set classification mode; the signboard information is identified according to the classification, and the key information in the signboard is extracted according to the set identification method to obtain the corresponding road parameter information and form road features.

[0018] Specifically, the front camera in the vehicle-mounted unit is first used to collect road sign information on the road, and the collected sign image information is transmitted to the DSP (digital signal processing) processing system in the vehicle-mounted unit for preprocessing of the image information.

[0019] Preprocessing involves classifying the captured images according to a classification model and identifying the sign information based on the classification. In this embodiment, the sign is classified by color. In my country, signage is mainly classified into red, yellow, blue, and green. By identifying the main color of the sign, the type of sign can be quickly determined, thereby classifying the sign into an indication type and a rule type.

[0020] Directional signs, including blue and green signs, primarily serve as road signs, providing information such as driving direction and distance. Regulation signs, including red and yellow signs, primarily serve as warning and prohibition signs, providing information on road conditions, precautions, restrictions, and other driving information. Color can be used to quickly identify potential road characteristics, enabling more accurate assessment of current road parameters and supporting the rapid construction of scene features, thereby reducing the need for large amounts of data collection and processing.

[0021] According to the classification mode, the key fields related to the classification information are determined, and the key information related to the road parameter information is extracted from the signboard according to the key fields. The key information is associated with the road characteristics, and different degrees of association between the key information and the road characteristics are established.

[0022] Specifically, based on the classification of instruction signs and rule signs, in this embodiment, the key fields that can be set for instruction signs include, but are not limited to: direction signs, including left turn, right turn, straight ahead, keep right, meeting, and U-turn; road signs, including dedicated lanes, exits, downhill, intersections, interchanges, forks, and distances. Key fields set for rule signs include, but are not limited to: warning signs, including signal lights, tunnels, road surfaces, attention, construction, detours, and forks; and prohibition signs, including speed limits, no passage, height, quality, and yield. In this embodiment, key fields are set based on the specified sign content and in combination with features that reflect current road parameter information, thereby extracting key information that reflects the current road characteristics.

[0023] In this embodiment, the key information in the signboard content can be extracted by using the TSR (Traffic Sign Recognition) system included in the DSP processing system, which is a system that efficiently recognizes the content of traffic signs in signboards, thereby accurately extracting effective key information according to the set key fields and forming corresponding road parameter information.

[0024] After obtaining the key information from the signboards, the key information is filtered and sorted. The sorting method associates the key information with road features, thereby establishing different levels of correlation between the associated information and road features. The correlation levels are divided into three levels: primary, secondary, and tertiary, depending on whether the key information directly reflects the current road features. The primary correlation level is the most directly reflective of road features. For example, when extracting key information such as intersections, right turns, parking lots, pedestrians, and predicted distances, based on their correlation with road features, if intersections, right turns, and pedestrians directly reflect the road features of the retaining wall, right turn lanes, and crosswalks, then the current key information is classified as primary. However, parking lots and predicted distances cannot directly reflect the current road parameters and require further analysis and determination, thus being classified as tertiary.

[0025] According to the divided correlation levels, the current main road scene features are first formed through the key information of the first level of correlation, and the key information of other levels is stored as items to be supplemented, so as to quickly construct an effective road scene, effectively save the time of data collection and data analysis, shorten the time of road feature construction, save a lot of development time and labor costs, and meet the construction needs of road features.

[0026] In this embodiment, by collecting only the content of signboards and classifying them by color, key information from the signboards can be quickly extracted according to the identification method. Without requiring high precision, a large amount of valid road parameter information can be quickly obtained to form scene features. Constructing the basic framework of a road scene solely by acquiring information from signboards significantly reduces the amount of data to be collected and the high-precision requirements. This allows data collection and preprocessing analysis to be performed directly using the vehicle's own onboard unit, eliminating the need for various cumbersome sensor devices. This reduces development costs, simplifies data collection workload, effectively improves data uniformity, and reduces the intermediate processing of large amounts of data. This allows for the rapid collection of large amounts of sample data, forming valid road feature information, and rapidly constructing my country's road feature information, providing effective support for the construction of an autonomous driving scenario library.

[0027] At the same time, in this embodiment, the content in the signboard is quickly acquired by only recording the line speed, which reduces the amount of collected data and is more convenient for storage. Through rapid acquisition, 500 kilometers of data can be acquired per day, and the construction of a city information can be quickly completed. Compared with the traditional urban scene acquisition and construction, the acquisition speed is increased by at least 50%-250%, which has great advantages for rapid construction and application in research fields, thereby shortening the acquisition and construction cycle and saving a lot of scene construction time. The application framework constituted by this solution can also be directly used as a framework model, which is more convenient and flexible to be applied to the rapid construction of various scenarios.

[0028] Example 2

[0029] In this embodiment, a method for constructing test scenario features of an intelligent connected vehicle is provided. Figure 2 As shown, the test scene feature collection method used in Example 1 also includes collecting the corresponding geographic location of the road sign when collecting the road sign. Specifically, the GPS system built into the vehicle-mounted unit can be used to collect the corresponding geographic location coordinates of the sign, and the acquired key information of the sign is associated with the geographic location coordinates. The road features formed by the key information are restored using the geographic location coordinates to form road scene features with geographic identification. The generated data is stored in the data collection system for real-time retrieval and flexible construction.

[0030] As signage is collected, the GPS system is used to locate the signage, generating corresponding geographic coordinates. When constructing test road scenarios, these coordinates can be used to directly assign specific road features to specific areas, quickly recreating my country's road characteristics for application in intelligent connected vehicle testing. Furthermore, based on test requirements, the system can flexibly and quickly build the required test scenarios for testing or research and analysis through direct retrieval, improving the efficiency of test scenario construction.

[0031] Example 3

[0032] This embodiment also provides a method for applying test scenario features for intelligent connected vehicles. This method utilizes the acquisition method described in Example 1 and the construction method described in Example 2 to build a driving scenario library, facilitating the rapid retrieval of road feature scenarios from different regions as a test scenario application framework. This reduces the time required to repeatedly build different road scenarios, lowers development and labor costs, and improves the versatility and applicability of road scenarios.

[0033] At the same time, in this embodiment, in order to meet the testing needs and research needs of different requirements, road features with different correlations can be added to the called road scene to form the road scene required for testing or the road scene required for research. The road features with different correlations are stored in the data acquisition system and can be called in real time according to the needs. The application is more flexible to build, and it also saves the data collection process, improves the construction efficiency, and can meet the application of various testing needs.

[0034] Example 4

[0035] Unlike the first embodiment, this embodiment also includes extended road information. This extended road information is extended road parameter information generated based on the established correlation degree for the current road scene features. Based on the established correlation degree of key information, the system can analyze other key information related to the current key information based on the established correlation network. This key information may not be indicated by the current road sign or may not have a sign. In this case, based on the correlation network, the system can generate extended road parameter information for backup. When constructing the road feature scene, the system can select the extended road parameters for the current road features based on actual conditions and needs.

[0036] At the same time, in this embodiment, based on the formed extended road parameter information, when constructing the test road scene, different road parameters can be independently constructed according to the test requirements and the current road characteristics to achieve different test effects, thereby improving the comprehensive testing of intelligent connected vehicles, improving the flexibility and variability of the test scene characteristics, and at the same time reducing the amount of data for scene construction, achieving the effect of flexible operation.

[0037] The above is only an embodiment of the present invention, and the common knowledge such as the specific technical solutions and / or characteristics in the solution are not described in detail here. It should be pointed out that for those skilled in the art, without departing from the technical solution of the present invention, several variations and improvements can be made, which should also be regarded as the scope of protection of the present invention, and these will not affect the effect of the implementation of the present invention and the practicality of the patent. The scope of protection required by this application shall be based on the content of its claims, and the specific implementation methods and other records in the description can be used to interpret the content of the claims.

Claims

1. A method for collecting characteristics of intelligent connected vehicle test scenarios, characterized by: The vehicle's own onboard unit is used to collect road signs and identify and classify the sign signs according to a set classification mode; the sign sign information is identified according to the classification, and key information in the sign signs is extracted according to a set identification method to obtain corresponding road parameter information and form road features; the identification method is to set corresponding valid key fields according to the classification, extract key information of the sign signs according to the key fields, associate the key information with the road features, and establish different correlation degrees between the key information and the road features; obtain corresponding road parameter information according to the correlation degree, and form the current road scene features; It also includes extended road information, which is extended road parameter information formed based on the established correlation degree for the current road scene features.

2. The method for collecting test scenario features of an intelligent connected vehicle according to claim 1, characterized in that: The vehicle-mounted unit includes a front camera and a DSP processing system of the vehicle; the front camera collects road signs, and the DSP processing system recognizes the signs and extracts key information.

3. The method for collecting characteristics of intelligent connected vehicle test scenarios according to claim 1, characterized in that: The classification mode is to classify the collected signboards according to their colors, divide the signboards into indication categories and rule categories by color, and determine key fields according to the classification, thereby obtaining key information.

4. The method for collecting test scenario features of an intelligent connected vehicle according to claim 3, characterized in that: The indicator signs include blue and green signs, which are road signposts; the rule signs include red and yellow signs, which are warning and prohibition signs.

5. A method for constructing test scenario features for intelligent connected vehicles, characterized by: It includes the intelligent connected vehicle test scene feature collection method described in one of claims 1 to 4, and also includes collecting the corresponding geographical location of road signs, and establishing a corresponding relationship between the geographical location coordinates and road parameter information, so that the formed road scene features correspond to the geographical location coordinates and are stored in a data collection system.

6. The method for constructing test scenario features for an intelligent connected vehicle according to claim 5, characterized in that: The DSP system in the vehicle-mounted unit is used to collect the geographical location of the road sign and form the corresponding geographical location coordinates, which correspond to the road parameter information.

7. A method for applying characteristics of intelligent connected vehicle test scenarios, characterized by: It includes the intelligent connected vehicle test scene feature collection method described in one of claims 1 to 4, which includes providing a driving scene library for quickly building road information to form a test scene.

Images