Vehicle control method, system and medium based on V2X scenario training

Abstract

The present invention relates to a vehicle control method, system and medium based on V2X scenario training, and relates to the field of intelligent driving technology. The method comprises acquiring vehicle data, wherein the vehicle data at least comprises basic vehicle model data; acquiring and training a preset scenario model based on the vehicle data to obtain a scenario generation model; extracting a user portrait based on the scenario generation model, converting the user portrait into a V2X scenario model, and sending the V2X scenario model to the vehicle; and controlling the vehicle based on the V2X scenario model. The present invention is used to perform personalized customization according to user driving habits, and improve user experience and safety.

Description

Technical Field

[0001] The present invention relates to the field of intelligent driving technology, and in particular to a vehicle control method, system and medium based on V2X scenario training. Background Art

[0002] At present, there are two main types of common vehicle-side V2X scenario applications. First, based on the V2X protocol stack, in accordance with the provisions of the first phase of the national standard, detailed product requirements are formulated to complete the seventeen basic scenarios specified in the first phase of the national standard. The advantage of this application method is that it can unify standards in the early stages of industry development, which is conducive to standardized development and promotes the popularization of functions. Secondly, this method refines typical scenarios and is applicable to most situations. However, its disadvantage is that its universality is relatively low, because it lacks specificity and is sometimes not applicable to actual road conditions. In addition, it does not distinguish between different user groups, usually only provides prompts, does not really implement the functions, and some scene functions may overlap, causing trouble to users.

[0003] The second application scenario uses the communication and networking capabilities provided by V2X to realize new applications of single-vehicle intelligence + vehicle-road collaboration. The advantage of this method is that it fully utilizes the beyond-line-of-sight, low-latency, and ignoring occlusion characteristics of V2X communication capabilities, as well as networking capabilities, effectively making up for the shortcomings of traditional single-vehicle intelligent sensors such as cameras and radars, and truly realizing the effect of vehicle-road collaboration. When traditional autonomous driving components are added (such as driving intentions, historical trajectories, etc.), the scope of the V2X scenario can be expanded, the accuracy of warnings can be improved, and latency can be reduced. However, this method also has disadvantages, including the current high cost of autonomous driving equipment, which makes the cost of a single vehicle too high and difficult to apply on a large scale. Summary of the invention

[0004] In view of the shortcomings of the prior art mentioned above, the present invention provides a vehicle control method, system and medium based on V2X scenario training, which are used to perform personalized customization according to user driving habits to improve user experience and safety.

[0005] To achieve the above purpose, the present invention adopts the following technical solution.

[0006] In a first aspect, the present invention provides a vehicle control method based on V2X scenario training, which adopts the following technical solution:

[0007] A vehicle control method based on V2X scenario training, comprising:

[0008] Acquiring vehicle data, wherein the vehicle data at least includes basic vehicle model data;

[0009] Based on the vehicle data, a preset scene model is acquired and trained to obtain a scene generation model;

[0010] Based on the scenario generation model, extract a user profile, convert the user profile into a V2X scenario model, and send the V2X scenario model to the vehicle;

[0011] And, controlling the vehicle based on the V2X scenario model.

[0012] Furthermore, in the above-mentioned vehicle control method based on V2X scenario training, the basic vehicle model data includes at least one of vehicle body CAN / LIN information, positioning information and V2X data information.

[0013] Furthermore, in the above-mentioned vehicle control method based on V2X scenario training, the step of acquiring and training a preset scenario model based on the vehicle data to obtain a scenario generation model includes:

[0014] Based on the basic vehicle model data, the preset map-free model is obtained and trained to obtain the map-free scene generation model.

[0015] Furthermore, in the above-mentioned vehicle control method based on V2X scenario training, the vehicle data also includes intelligent driving model data, and the intelligent driving model data includes at least one of high-precision positioning and navigation data.

[0016] Furthermore, in the above-mentioned vehicle control method based on V2X scenario training, the step of acquiring and training a preset scenario model based on the vehicle data to obtain a scenario generation model includes:

[0017] Based on the basic vehicle model data and intelligent driving vehicle model data, the preset conventional road model is acquired and trained to obtain a conventional road scene generation model.

[0018] Furthermore, in the above-mentioned vehicle control method based on V2X scenario training, the vehicle data also includes smart driving model data and smart cabin model data, and the smart cabin model data includes ADAS data.

[0019] Furthermore, in the above-mentioned vehicle control method based on V2X scenario training, the step of acquiring and training a preset scenario model based on the vehicle data to obtain a scenario generation model includes:

[0020] Based on the basic vehicle model data, the smart driving vehicle model data and the smart cabin vehicle model data, a preset intersection road model is acquired and trained to obtain an intersection road scene generation model.

[0021] Furthermore, in the above-mentioned vehicle control method based on V2X scenario training, extracting a user portrait based on the scenario generation model includes:

[0022] Based on the scene generation model, key frames determined based on user habits are extracted to depict user portraits.

[0023] Furthermore, in the above-mentioned vehicle control method based on V2X scenario training, the conversion into a V2X scenario model based on the user portrait includes:

[0024] Based on the user portrait, risk information in the user portrait is evaluated, and a scenario corresponding to the risk information is converted into a V2X scenario model.

[0025] Furthermore, in the above-mentioned vehicle control method based on V2X scenario training, sending the V2X scenario model to the vehicle includes:

[0026] Convert the extracted V2X scenario model into coded text;

[0027] In the coded text, key evaluation criteria in the V2X scenario model are extracted and stored in the coded text according to the agreed key.

[0028] In a second aspect, the present invention provides a vehicle control system based on V2X scenario training, which adopts the following technical solution:

[0029] A vehicle control system based on V2X scenario training, comprising:

[0030] A data acquisition module, at least used to acquire vehicle data, wherein the vehicle data at least includes basic vehicle model data;

[0031] A cloud training module, at least used to acquire and train a preset scene model based on the vehicle data to obtain a scene generation model;

[0032] A cloud management module, at least used to extract a user profile based on the scenario generation model, convert the user profile into a V2X scenario model, and send the V2X scenario model to the vehicle;

[0033] The vehicle-side scenario application module is at least used to control the vehicle based on the V2X scenario model.

[0034] In a third aspect, the present invention provides a readable storage medium, which adopts the following technical solution:

[0035] A readable storage medium stores computer instructions, and when the computer instructions are executed by a processor, a vehicle control method as described in any one of the first aspects above is implemented.

[0036] In summary, compared with the prior art, the present invention includes at least one of the following beneficial technical effects:

[0037] 1. The present invention solves the problem that conventional V2X application scenarios cannot be customized according to users under existing methods. The present invention uses the cloud to send scenarios to achieve different faces for different users;

[0038] 2. The present invention uses big data for training, quickly implements model training, can reuse user portraits, accurately meet user needs, and improve user experience;

[0039] 3. The present invention can use basic vehicle model data as the data source for big data training, and can realize the function at a low cost. By adding sensory data such as cameras and radars, the scene can be quickly upgraded, and the application range is wide. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0041] Figure 1 It is a flowchart of a specific embodiment of a vehicle control method based on V2X scenario training of the present invention.

[0042] Figure 2 It is a structural schematic diagram of a specific embodiment of a vehicle control system based on V2X scenario training of the present invention.

[0043] Figure 3 It is a structural schematic diagram of another specific embodiment of a vehicle control system based on V2X scenario training of the present invention.

[0044] Figure 4 It is a structural schematic diagram of another specific embodiment of a vehicle control system based on V2X scenario training of the present invention.

[0045] Figure 5 It is a structural schematic diagram of another specific embodiment of a vehicle control system based on V2X scenario training of the present invention.

[0046] Figure 6 It is a structural schematic diagram of another specific embodiment of a vehicle control system based on V2X scenario training of the present invention.

[0047] Figure 7 It is a topological diagram of a specific embodiment of a vehicle control system based on V2X scenario training of the present invention. DETAILED DESCRIPTION

[0048] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present application. In addition, it should be understood that the specific implementation methods described herein are only used to illustrate and explain the present application, and are not used to limit the present application.

[0049] It should be noted that the description order of the following embodiments is not intended to limit the preferred order of the embodiments of the present application. In addition, in the following embodiments, the description of each embodiment has its own emphasis, and for parts not described in detail in one embodiment, reference can be made to the relevant description of other embodiments.

[0050] The method steps described in the embodiments of the present invention may be executed in the order described in the specific implementation manner, or the execution order of each step may be adjusted according to actual needs and on the premise that the technical problem can be solved, which will not be listed here one by one.

[0051] The following is combined with Figure 1-7 The present invention is described in further detail.

[0052] Reference Figure 1 , a vehicle control method based on V2X scenario training, comprising:

[0053] S1, obtaining vehicle data, wherein the vehicle data at least includes basic vehicle model data;

[0054] S2, based on the vehicle data, acquiring and training a preset scene model to obtain a scene generation model;

[0055] S3, extracting a user profile based on the scenario generation model, converting the user profile into a V2X scenario model, and sending the V2X scenario model to the vehicle;

[0056] S4: Control the vehicle based on the V2X scenario model.

[0057] Specifically, the specific implementation method of the vehicle control method based on V2X scenario training is as follows: First, in the S1 stage, vehicle data is obtained through the vehicle system, which at least includes basic vehicle model data, such as CAN / LIN signals, vehicle speed, vehicle light status, pedal information, etc. In the S2 stage, the acquired vehicle data is used to perform the training process of the preset scenario model. This includes using machine learning or deep learning algorithms to generate a scenario generation model through analysis and learning of vehicle data, which can predict the behavior and reaction of the vehicle in different scenarios.

[0058] In the S3 stage, the user profile is refined based on the trained scenario generation model. This involves extracting key user behavior features, reaction time, habits, etc. from the scenario generation model to form a user profile. Subsequently, the user profile is converted into a V2X scenario model, which describes the user's expected behavior and decision-making in a specific traffic scenario. Finally, in the S4 stage, the generated V2X scenario model is sent to the vehicle, and the vehicle is controlled in real time through the V2X scenario module. This includes intelligent vehicle control based on the current traffic situation using predicted user behavior, such as early deceleration, lane change, and maintaining a safe distance, to improve the driving safety and efficiency of the vehicle in complex traffic environments. The entire method realizes the entire process of intelligent vehicle control by integrating vehicle data, scenario model training, and V2X scenario control.

[0059] The vehicle control method of the present invention obtains a scenario generation model that can accurately predict vehicle behavior by training a scenario model based on vehicle data based on vehicle data. By refining user portraits into V2X scenario models, an accurate description of user behavior and decisions is achieved. This process enables the vehicle to make corresponding decisions intelligently based on user portraits in real-time traffic scenarios, thereby achieving safer and more efficient vehicle driving under the control of the V2X scenario module. Advantages include strong adaptability to diverse traffic conditions, and personalized user portraits are reflected in the V2X scenario model, making vehicle control more in line with actual driving scenarios and improving driving safety and comfort.

[0060] Furthermore, as an embodiment of the present invention, the basic vehicle model data includes at least one of vehicle body CAN / LIN information, positioning information and V2X data information.

[0061] Specifically, the basic vehicle model data includes: CAN / LIN information such as gear position, vehicle speed, lights, pedals, as well as positioning information (latitude and longitude, heading, etc.), V2X data (BSM, SPAT, RSI, MAP, RSM). The basic vehicle model data refers to multiple elements: First, it covers CAN / LIN information, including key vehicle status information such as gear position, vehicle speed, lights, pedals, etc. Secondly, it includes positioning information, such as longitude and latitude and heading, to provide the accurate position and orientation of the vehicle in geographic space. In addition, it also includes V2X data, covering BSM (basic safety message), SPAT (signal light phase and timing message), RSI (road condition information), MAP (map information) and RSM (road section information), to comprehensively obtain the vehicle body and the surrounding environment of the vehicle, etc., providing a rich data foundation for subsequent scene model training. This meticulous and comprehensive collection of basic vehicle model data helps to build a more accurate and comprehensive scene generation model and enhance the vehicle's intelligent driving capabilities in various traffic scenarios.

[0062] Further, as an implementation mode of the present invention, step S2, based on the vehicle data, obtains and trains a preset scene model to obtain a scene generation model, including: based on the basic vehicle model data, obtains and trains a preset map-free model to obtain a map-free scene generation model.

[0063] Specifically, based on the vehicle data, especially the basic vehicle model data, the preset scenario model is acquired and trained. Especially for the map-free situation, the map-free model is acquired and trained to obtain the map-free scenario generation model. During the implementation process, machine learning or deep learning algorithms can be used to analyze and learn the basic vehicle model data to obtain the typical behavior and reaction mode of the vehicle in a map-free environment. This involves modeling short-distance collisions, vehicle acceleration and deceleration conditions, adjacent vehicle status, etc. to form a map-free scenario generation model. Through this specific implementation method, the system can more accurately predict the behavior of the vehicle without map information support based on the basic vehicle model data obtained from the lower-configuration vehicle model, and provide an accurate reference for subsequent vehicle control.

[0064] Furthermore, as an embodiment of the present invention, the vehicle data also includes intelligent driving model data, and the intelligent driving model data includes at least one of high-precision positioning and navigation data.

[0065] Specifically, when acquiring data, when data collection is performed based on a medium-configuration vehicle model, the vehicle data includes not only basic vehicle model data, but also intelligent driving model data, including at least one of high-precision positioning and navigation data. In terms of high-precision positioning, the system collects accurate latitude and longitude, heading and other location information of the vehicle to provide more precise geographic positioning. In terms of navigation data, the system obtains information related to vehicle driving, such as driving intentions, destinations, etc., making the system more intelligent and able to better understand the vehicle's driving purpose and intentions. By combining basic vehicle model data and intelligent driving model data, the system can provide more comprehensive and advanced information, providing a more accurate basis for scene model training and vehicle control.

[0066] Furthermore, as an embodiment of the present invention, based on the vehicle data, a preset scene model is acquired and trained to obtain a scene generation model, including: based on the basic vehicle model data and the intelligent driving model data, a preset conventional road model is acquired and trained to obtain a conventional road scene generation model.

[0067] Specifically, the system acquires and trains scene models for vehicle data collected from medium-configuration models. This includes presetting basic vehicle model data and intelligent driving model data, obtaining a regular road model through training, and forming a regular road scene generation model. During the training process, the system analyzes and learns basic vehicle model data (such as CAN / LIN information, vehicle speed, headlight status, etc.) and intelligent driving model data (such as high-precision positioning and navigation data) through machine learning or deep learning algorithms.

[0068] In particular, when constructing a conventional road model, the behavior characteristics of the vehicle on conventional roads are taken into account, and the high-precision positioning and navigation information of the intelligent driving model data are used to comprehensively consider factors such as the vehicle's position, driving intention, and destination to obtain more comprehensive scene information. This enables the conventional road scene generation model to more accurately reflect the vehicle's driving behavior and decision-making in a standard road environment, providing more reliable scene predictions for subsequent vehicle control. The advantage of this implementation method is its high adaptability to conventional roads, and it can more accurately simulate the behavior of vehicles in various common driving scenarios.

[0069] Furthermore, as an embodiment of the present invention, the vehicle data also includes smart driving model data and smart cabin model data, and the smart cabin model data includes ADAS data.

[0070] Specifically, when collecting data, based on high-end configuration models, the vehicle data obtained not only includes basic model data, but also covers smart driving model data and smart cabin model data, among which smart cabin model data includes ADAS data, including high-precision maps, lidar data, camera data, etc. In actual implementation, through the high-end configuration of the vehicle system, the system can obtain richer and more detailed vehicle information. Basic model data includes CAN / LIN information, vehicle speed, headlight status, etc., smart driving model data includes high-precision positioning and navigation data, and smart cabin model data provides more advanced perception information.

[0071] In particular, in the smart cabin vehicle data, the collection of ADAS data involves the acquisition of high-precision maps, the recording of lidar data, and the collection of camera data. Through high-precision maps, the system can obtain more accurate geographic information, lidar provides high-precision distance and obstacle information about the vehicle's surroundings, and camera data provides more details for visual perception. This detailed data collection method provides a more comprehensive and advanced information foundation for subsequent scene model training and intelligent driving decisions. By comprehensively utilizing this data, the system can more accurately simulate the vehicle's ability to perceive and understand the surrounding environment, improving the safety and efficiency of the vehicle in complex driving scenarios.

[0072] Furthermore, as an implementation mode of the present invention, step S2, based on the vehicle data, obtains and trains a preset scene model to obtain a scene generation model, including: based on the basic vehicle model data, the smart driving model data and the smart cabin vehicle model data, obtains and trains a preset intersection road model to obtain an intersection road scene generation model.

[0073] Specifically, for vehicle data collected from high-end configuration models, the corresponding steps involve the acquisition and training of scene models based on vehicle data. This process includes the preset based on basic vehicle model data, smart driving vehicle model data, and smart cabin vehicle model data, obtaining the intersection road model through training, and generating the intersection road scene generation model. In the specific steps, using machine learning or deep learning algorithms, the system analyzes and learns basic vehicle model data (such as CAN / LIN information, vehicle speed, headlight status, etc.), integrates smart driving vehicle model data (high-precision positioning and navigation information) and smart cabin model data (including ADAS data, such as high-precision maps, lidar data, camera data).

[0074] In particular, when constructing the intersection road model, the system comprehensively considers the behavioral characteristics of vehicles at intersections. The intelligent driving model data provides information about the vehicle's driving intentions, destinations, etc., while the intelligent cabin model data provides more advanced perception data, including detailed knowledge of obstacles, traffic signs, and signal lights near the intersection. In this way, the intersection road scene generation model can more accurately predict the driving behavior of vehicles at intersections, providing more accurate scene simulation for the intelligent driving system to achieve safe and efficient vehicle control.

[0075] Further, as an embodiment of the present invention, the scene model can be divided into general categories such as conventional road model, intersection model, highway model, parking lot model, etc. Under the general categories, subcategories can be subdivided according to common traffic accident situations, such as rear-end collision, blind spot and other collision warnings and signal lights, signs and other message warnings and other subsystem models. Conventional road models require V2XMAP data / high-precision maps, positioning / high-precision positioning to determine the specific location of the vehicle in the road network. If there is V2X SPAT data / intelligent traffic light information, you can choose a signal light related model; if there is V2X BSM data / perception data (radar, camera), you can choose a collision model; if there is navigation information (intention, destination), you can choose a lane selection / keeping model, and so on.

[0076] The intersection road model requires V2X MAP data / high-precision map and positioning / high-precision positioning to determine the vehicle's position at the intersection in the road network. Compared with conventional roads, intersections pay more attention to collision information. The model should include vehicle deceleration information, steering wheel speed, steering wheel angle, accelerator pedal usage, etc.

[0077] V2X has no map model, only V2X BSM data, and the short-range collision subsystem model can be selected.

[0078] Furthermore, as an implementation mode of the present invention, extracting a user portrait based on the scene generation model includes: based on the scene generation model, extracting key frames determined based on user habits to depict the user portrait.

[0079] Specifically, through the implementation of the scenario generation model, the system realizes the process of refining the user portrait. This step includes using the scenario generation model, especially the scenario information based on the user's habits, to extract key frames to depict the user portrait. By analyzing the user's reactions, decisions and behaviors in different traffic scenarios, the system can extract key moments in the user's habits from the model, which may include the user's reaction speed, the frequency of specific behaviors, driving preferences in specific environments, etc.

[0080] These key frames form a comprehensive user portrait that reflects the user's personalized characteristics in driving. In this way, the system can better understand and predict the behavior patterns of specific users in different driving scenarios, providing a basis for the subsequent customization of V2X scenario models. This specific implementation enables the system to provide a more personalized intelligent driving experience based on the driver's personality and habits, and enhances the system's accurate understanding and feedback of the user's driving behavior.

[0081] Specifically, for car owners, when facing the countdown of approaching traffic lights, the distance to the stop line when changing lanes, the reaction time after the front car slows down, etc., we can mark the basic user portrait of the car owner based on reaction speed, reaction attitude, reaction plan, etc., such as aggressive car owners with fast reactions and conservative car owners with average reactions. Based on the reactions of car owners to various models in the model library, we can portray accurate portraits based on statistics.

[0082] Furthermore, as an embodiment of the present invention, the conversion into a V2X scenario model based on the user portrait includes: based on the user portrait, evaluating risk information in the user portrait, and converting into a V2X scenario model based on the scenario corresponding to the risk information.

[0083] Specifically, the process of evaluating user portraits and converting them into V2X scenario models includes the following steps: First, evaluate potential driving risks based on key information in the user portraits, taking into account user preferences and behavior patterns in different scenarios. For example, when evaluating user habits, the system may focus on the user's reactions in scenarios such as conventional forward collisions and left turn assists. Subsequently, the system adjusts the parameters of the V2X scenario model in a targeted manner based on the evaluation results. Taking the user's U-turn behavior as an example, if the portrait shows that the user usually has a large turning radius when turning, the system can add a U-turn judgment scenario and adjust key parameters in it to reflect this feature. This ensures that the V2X scenario model is more in line with the user's driving habits and improves the accuracy of scenario simulation.

[0084] In addition, when a user comes to a new city, the system will extract the scene model of similar people based on the previous city portraits, and replace the key information of the user with this information to initialize the scene model. Through this specific implementation method, the system realizes personalized analysis of user portraits and fine-tuning of V2X scene models to better adapt to the user's driving style and environment, and improve the system's intelligent driving performance in different cities and traffic scenarios.

[0085] Furthermore, as an embodiment of the present invention, the V2X scenario model is sent to a vehicle, including: converting the extracted V2X scenario model into a coded text; in the coded text, extracting key evaluation criteria in the V2X scenario model and storing them in the coded text according to an agreed key.

[0086] Specifically, the specific implementation steps of sending the V2X scenario model to the vehicle include two main stages: First, the extracted V2X scenario model is converted into coded text, which may be in a variety of formats such as JSON, ARXML, DOC, etc., for effective data transmission. In this process, the system integrates various information of the V2X scenario model into a codable text format to facilitate subsequent transmission and parsing.

[0087] Secondly, in the encoded text, the system will extract the key evaluation criteria in the V2X scenario model and store them in the text according to the pre-agreed key. This includes various key parameters required in the scenario model, such as speed, deceleration, distance, tti, etc. By storing them according to the specified key value, the system can ensure that the V2X application scenario parser on the vehicle side can accurately interpret and apply these key evaluation criteria to achieve accurate simulation and intelligent decision-making of actual driving scenarios. This specific implementation method ensures the effective distribution of the V2X scenario model and the correct parsing on the vehicle side, thereby providing accurate scenario information and real-time decision support for the intelligent driving system.

[0088] Further, step S4, controlling the vehicle based on the V2X scenario model, includes:

[0089] First, on the vehicle side, the encoded text corresponding to the V2X scenario module sent from the cloud is received and parsed, which contains the extracted key V2X scenario model. Through parsing, the vehicle can understand the key evaluation criteria and parameters in the model;

[0090] Next, the scene is updated in real time based on the scene model information obtained through analysis. This includes comparing the actual environment perceived by the vehicle and the status of other vehicles with the scene model sent from the cloud to ensure that the real-time information is consistent with the model;

[0091] Finally, based on the analysis results and real-time environmental perception, the vehicle's behavior is controlled, including issuing voice or text prompts, implementing braking measures, or performing other corresponding driving assistance operations to adapt to current road conditions and predict future traffic scenarios. Through this specific implementation method, real-time control of the vehicle is achieved, combining the cloud-based intelligent scene model with the actual driving conditions of the vehicle, improving the intelligence and safety of driving, and ensuring that the vehicle responds appropriately in different scenarios.

[0092] An embodiment of the present invention also discloses a vehicle control system based on V2X scenario training.

[0093] Reference Figure 2 , a vehicle control system based on V2X scenario training, comprising:

[0094] A data acquisition module, at least used to acquire vehicle data, wherein the vehicle data at least includes basic vehicle model data;

[0095] A cloud training module, at least used to acquire and train a preset scene model based on the vehicle data to obtain a scene generation model;

[0096] A cloud management module, at least used to extract a user profile based on the scenario generation model, convert the user profile into a V2X scenario model, and send the V2X scenario model to the vehicle;

[0097] The vehicle-side scenario application module is at least used to control the vehicle based on the V2X scenario model.

[0098] Specifically, refer to Figure 3The data collection module includes a vehicle-side collection trigger module, a cloud-side scene management module, and a vehicle-side data collection module. Among them, the vehicle-side collection trigger module is used to determine whether the V2X environment collection conditions are met and actively report to the cloud. The cloud-side scene management module is used to select a targeted collection script based on the upload parameters after receiving the message from the vehicle-side trigger module and issue the collection task. The vehicle-side data collection module is used to continuously collect and upload data after receiving the collection task until the collection task is completed.

[0099] Reference Figure 4 The cloud training module includes a data receiving / preprocessing module, an algorithm selection / model selection module, and a model training / result evaluation module. Among them, the data receiving / preprocessing module is used to receive the data collected by the vehicle and perform some preprocessing. Determine whether the vehicle is under V2X signal coverage, whether the data is continuous and complete, etc., and filter out data that does not meet the conditions. The algorithm selection / model selection module is used to analyze whether the vehicle is at an intersection or on the road, etc., and analyze whether the data source has traffic light information, Map information, radar information, camera information, etc., and select preset models and algorithms according to different situations. The model training / result evaluation module is used to use the preprocessed data for training, and continuously adjusts in big data training until the accuracy, recall rate and other requirements of the evaluation module are met.

[0100] Reference Figure 5 The cloud management module includes a training result conversion module, a scene evaluation module, and a scene delivery module. Among them, the training result conversion module is used to extract user portraits from the training results. When facing other vehicles, traffic, and intersections, key frames such as user reaction time, user behavior, and user common habits are extracted to describe user portraits and summarize a group of people. The scene evaluation module is used to evaluate the risk information in the user's portrait and convert it into a V2X scene model in a targeted manner. For example, for conventional forward collisions and left turn assistance, the key parameters of conventional scenes can be adjusted according to the habits of the owner. For example, the turning radius of users who usually turn around is relatively large, and a new U-turn judgment scene can be added. If you come to a new city, based on the portrait of the previous city, the scene model of the same group of people is extracted, the key information of the user is replaced, and the scene model is initialized. The scene delivery module is used to convert the extracted v2x scene model into coded text (including but not limited to json, arxml, doc), and send it to the vehicle end for that module. In the coded text, the key evaluation criteria in the scene model, such as speed, deceleration, distance, tti, etc., are extracted and stored in the coded text according to the agreed key.

[0101] Reference Figure 6The vehicle-side scene application module includes a vehicle-side scene receiving module, a vehicle-side scene parsing module and a vehicle-side scene loading module. The vehicle-side scene receiving module is used to receive the V2X scene files sent from the cloud, manage them by account, and update the scene files under the current account. The vehicle-side scene parsing module is used to parse the scene files and store them in the database. The vehicle-side scene loading module is used to reload the updated scene files and update the v2x scene.

[0102] Reference Figure 7 In a vehicle control system based on V2X scenario training described in an embodiment of the present invention, the vehicle-side acquisition module sends a collection task request to the cloud management module when the conditions are met. After receiving the request, the cloud management module performs task evaluation, determines the data to be collected, and generates a collection task through a script. Subsequently, the collection task is sent to the vehicle-side data acquisition module through the cloud management module.

[0103] After receiving the task, the vehicle-side data acquisition module collects data according to the instructions of the acquisition script and uploads the collected data to the cloud data receiving module. In the cloud, the received data is first preprocessed, including judging whether it meets the collection conditions, data continuity and other filtering conditions.

[0104] Subsequently, the cloud selects specific algorithm libraries and model libraries for model training, and uses the collected data to train the corresponding model. After the training is completed, the results are evaluated to ensure the accuracy and applicability of the model. If the evaluation results meet the requirements, the results are converted and the scenario is evaluated.

[0105] During the scenario evaluation process, application scenarios that meet the conditions are updated to the cloud application scenario library. Finally, the cloud sends the updated application scenarios to the vehicle side, which receives, parses, and loads these scenarios to make them V2X scenarios that can remind users. This system process realizes data interaction, model training, and application scenario updates between the vehicle side and the cloud side, providing an effective mechanism for real-time optimization of V2X scenarios and vehicle-side reminders.

[0106] The embodiment of the present invention also discloses a readable storage medium.

[0107] A readable storage medium stores a computer program, which, when executed by a processor, implements the steps of a vehicle control method based on V2X scenario training described in any one of the above embodiments. The computer-readable storage medium may include: any entity or device capable of carrying a computer program, a recording medium, a USB flash drive, a mobile hard disk, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), and a software distribution medium, etc. The computer program includes a computer program code. The computer program code may be in source code form, object code form, an executable file, or some intermediate form, etc. The computer-readable storage medium may include: any entity or device capable of carrying a computer program code, a recording medium, a USB flash drive, a mobile hard disk, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), and a software distribution medium, etc.

[0108] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, segment or portion of code that includes one or more executable instructions for implementing the steps of a specific logical function or process, and the scope of the preferred embodiments of the present invention includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by those skilled in the art to which the embodiments of the present invention belong.

[0109] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, apparatus or device (such as a computer-based system, a system including a processing module, or other system that can fetch instructions from an instruction execution system, apparatus or device and execute instructions), or used in combination with these instruction execution systems, apparatuses or devices.

[0110] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A vehicle control method based on V2X scenario training, characterized in that: include: Acquiring vehicle data, wherein the vehicle data at least includes basic vehicle model data; Based on the vehicle data, a preset scene model is acquired and trained to obtain a scene generation model; Based on the scenario generation model, extract a user profile, convert the user profile into a V2X scenario model, and send the V2X scenario model to the vehicle; and, controlling the vehicle based on the V2X scenario model; in, The step of generating a user portrait based on the scenario generation model includes: Based on the scene generation model, extract key frames determined based on user habits to depict user portraits; The converting into a V2X scenario model based on the user portrait includes: Based on the user portrait, risk information in the user portrait is evaluated, and the scenario corresponding to the risk information is converted into a V2X scenario model.

2. The vehicle control method based on V2X scenario training according to claim 1, characterized in that: The basic vehicle model data includes at least one of vehicle body CAN / LIN information, positioning information and V2X data information.

3. The vehicle control method based on V2X scenario training according to claim 2 is characterized in that: The step of acquiring and training a preset scene model based on the vehicle data to obtain a scene generation model includes: Based on the basic vehicle model data, the preset map-free model is obtained and trained to obtain the map-free scene generation model.

4. The vehicle control method based on V2X scenario training according to claim 1, characterized in that: The vehicle data also includes intelligent driving model data, and the intelligent driving model data includes at least one of high-precision positioning and navigation data.

5. The vehicle control method based on V2X scenario training according to claim 4 is characterized in that: The step of acquiring and training a preset scene model based on the vehicle data to obtain a scene generation model includes: Based on the basic vehicle model data and intelligent driving vehicle model data, the preset conventional road model is acquired and trained to obtain a conventional road scene generation model.

6. The vehicle control method based on V2X scenario training according to claim 1, characterized in that: The vehicle data also includes smart driving model data and smart cabin model data, and the smart cabin model data includes ADAS data.

7. The vehicle control method based on V2X scenario training according to claim 6, characterized in that: The step of acquiring and training a preset scene model based on the vehicle data to obtain a scene generation model includes: Based on the basic vehicle model data, the smart driving vehicle model data and the smart cabin vehicle model data, a preset intersection road model is acquired and trained to obtain an intersection road scene generation model.

8. The vehicle control method based on V2X scenario training according to claim 1, characterized in that: Sending the V2X scenario model to the vehicle includes: Convert the extracted V2X scenario model into coded text; In the coded text, key evaluation criteria in the V2X scenario model are extracted and stored in the coded text according to the agreed key.

9. A vehicle control system based on V2X scenario training, applied to the vehicle control method according to any one of claims 1 to 8, characterized in that: The system comprises: A data acquisition module, at least used to acquire vehicle data, wherein the vehicle data at least includes basic vehicle model data; A cloud training module, at least used to acquire and train a preset scene model based on the vehicle data to obtain a scene generation model; A cloud management module, at least used to extract a user profile based on the scenario generation model, convert the user profile into a V2X scenario model, and send the V2X scenario model to the vehicle; The vehicle-side scenario application module is at least used to control the vehicle based on the V2X scenario model.

10. A readable storage medium, characterized in that: The readable storage medium stores computer instructions, and when the computer instructions are executed by a processor, the vehicle control method according to any one of claims 1 to 8 is implemented.

Images