Safety warning methods and systems for vehicle driving

Abstract

This application relates to the technical field of vehicle early warning, and in particular to a method and system for vehicle driving safety early warning. The method involves acquiring vehicle driving monitoring data, predicting the vehicle's future driving trajectory based on the data, obtaining a predicted driving trajectory, acquiring environmental monitoring data, analyzing obstacles and road conditions in the vehicle's environment based on the environmental monitoring data, obtaining environmental perception information, and analyzing the predicted driving trajectory and environmental perception information using locally deployed vehicle perception AI to obtain vehicle driving status information. This driving status information is then uploaded to a cloud-based positioning map for information interaction and analysis with other vehicles in the vicinity, resulting in vehicle driving safety early warning information. This invention executes the early warning function through locally deployed vehicle equipment and a cloud server, eliminating the need for the infrastructure support required by traditional V2X early warning technologies, and thus having a wider range of application scenarios.

Description

Technical Field

[0001] This application relates to the technical field of vehicle warning, and in particular to a method and system for providing safety warnings for vehicle operation. Background Technology

[0002] Current vehicle early warning systems rely on simple integrated processing of data from vehicle-to-vehicle (V2V), vehicle-to-infrastructure (V2I), and other V2X communications. However, data transmission between V2X systems is subject to delays and false alarms. If facilities are damaged, data loss can occur, leading to serious deviations in the integrated data calculation. The current infrastructure supporting V2X is inadequate, and remote and underdeveloped areas lack such infrastructure altogether, making it impossible to support existing V2X early warning technologies. Summary of the Invention

[0003] Therefore, it is necessary to provide a vehicle driving safety warning method and system to address the above-mentioned technical problems, which can realize vehicle-to-vehicle warning without the need for infrastructure support.

[0004] In a first aspect, this application provides a vehicle driving safety warning method, the method comprising: Acquire vehicle driving monitoring data, and predict the vehicle's future driving trajectory based on the driving monitoring data to obtain the predicted driving trajectory; The system acquires environmental monitoring data of the vehicle and analyzes the obstacles and road conditions of the vehicle's environment based on the environmental monitoring data to obtain environmental perception information. The predicted driving trajectory and the environmental perception information are analyzed by locally deployed vehicle perception AI to obtain the vehicle's driving status information. The driving status information is uploaded to a cloud-based positioning map to interact and analyze with other vehicles in the vicinity, thereby obtaining driving safety warning information for the vehicle.

[0005] Secondly, this application also provides a vehicle driving safety warning system for implementing the vehicle driving safety warning method described in any one of the first aspects, comprising: The driving prediction module is used to acquire vehicle driving monitoring data and predict the future driving trajectory of the vehicle based on the driving monitoring data to obtain the predicted driving trajectory. The environmental perception module is used to acquire environmental monitoring data of the vehicle and analyze the obstacles and road conditions of the vehicle's environment based on the environmental monitoring data to obtain environmental perception information. The condition analysis module is used to analyze the predicted driving trajectory and the environmental perception information through locally deployed vehicle perception AI to obtain the vehicle's driving condition information. The safety warning module is used to upload the driving status information to the cloud positioning map for information interaction and analysis with other vehicles in the nearby area to obtain vehicle driving safety warning information.

[0006] The aforementioned vehicle driving safety warning method predicts driving trajectories, enabling advance planning of vehicle routes. It analyzes environmental information to accurately grasp the surrounding conditions, uses local AI to analyze driving conditions, and combines trajectory and environmental information to improve judgment accuracy. By uploading information to the cloud and interacting with nearby vehicles for analysis, it can obtain the overall traffic situation, detect potential collision risks in advance, and issue timely warnings. This helps drivers or autonomous driving systems take countermeasures to avoid accidents, improve vehicle driving safety and traffic efficiency, and can achieve vehicle-to-vehicle warning functions even without infrastructure support. Attached Figure Description

[0007] Figure 1 This is a schematic diagram of the steps of a vehicle driving safety warning method in one embodiment; Figure 2 This is a schematic diagram of the structure of a vehicle driving safety warning system in one embodiment. Detailed Implementation

[0008] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0009] The vehicle driving safety warning method provided in this application embodiment can be applied to, for example... Figure 1 In the application environment shown: S1: Acquire vehicle driving monitoring data, and predict the future driving trajectory of the vehicle based on the driving monitoring data to obtain the predicted driving trajectory; S2: Acquire the vehicle's environmental monitoring data, and analyze the obstacles and road conditions of the vehicle's environment based on the environmental monitoring data to obtain environmental perception information; S3: Analyze the predicted driving trajectory and the environmental perception information using locally deployed vehicle perception AI to obtain vehicle driving status information; S4: Upload the driving status information to the cloud positioning map to interact and analyze information with other vehicles in the nearby area to obtain vehicle driving safety warning information.

[0010] Specifically, in step S1 of the embodiment provided in this application, the vehicle's speed can be obtained through a speed sensor installed on the vehicle. This sensor generally works based on principles such as electromagnetic induction or Hall effect, and can measure the vehicle's speed in real time and accurately. The driving direction can be collected using a Global Positioning System (GPS) or an Inertial Measurement Unit (IMU). GPS determines the vehicle's position by receiving satellite signals and then calculates the driving direction based on position changes. The IMU measures the vehicle's acceleration and angular velocity using accelerometers and gyroscopes to deduce the driving direction. The driving speed and driving direction data collected at different times are combined into a continuous reference data stream in chronological order for subsequent analysis. Driving speed and driving direction are two key parameters describing the vehicle's driving state, and they directly affect the vehicle's future driving trajectory. By collecting these data and combining them into a reference data stream, basic data can be provided for subsequent driving style recognition and trajectory prediction.

[0011] More specifically, features that reflect driving style are extracted from the reference data stream, such as the rate of change of acceleration and speed fluctuations. Machine learning or deep learning algorithms, such as decision trees and neural networks, are used to train and analyze the extracted features. Models for different driving styles are trained using a large amount of historical data. Then, the features of the current vehicle are input into the model to identify the vehicle's driving style and generate a driving style profile.

[0012] More specifically, different driving styles lead to different vehicle behaviors during driving. For example, aggressive drivers will frequently accelerate, decelerate, and change lanes, while conservative drivers will drive more smoothly. By identifying a vehicle's driving style, one can more accurately predict the vehicle's future trajectory, because driving style affects the driver's decisions and actions in different situations.

[0013] More specifically, corresponding mathematical or physical models are established based on different driving styles. For example, for aggressive driving styles, more frequent acceleration and lane changes can be considered; for conservative driving styles, models can be built based on relatively stable speed and direction changes. The driving monitoring data (driving speed and driving direction) at the current moment is input into the corresponding prediction model, and combined with the driving style profile, the driving trajectory of the vehicle in the future is predicted.

[0014] More specifically, driving style profiles reflect the driver's behavioral habits and decision-making patterns. Combined with current driving monitoring data, they can more accurately predict the vehicle's future trajectory. Considering driving style can make the prediction results more consistent with reality, improve the accuracy of the prediction, and thus provide a more reliable basis for vehicle safety warnings.

[0015] Specifically, in step S2 of the embodiments provided in this application, multiple cameras are usually installed on the vehicle, such as front-view cameras, rear-view cameras, and surround-view cameras. These cameras capture images of the vehicle's surrounding environment at a certain frame rate (e.g., 30 frames per second). The cameras convert light signals into electrical signals, which are then converted from analog to digital and stored as digital image data. This data is environmental monitoring data. The images can intuitively reflect various information about the vehicle's environment, including the location of obstacles, road conditions, traffic signs, etc. By collecting images through cameras, rich environmental information can be obtained, providing basic data for subsequent environmental analysis.

[0016] More specifically, lighting conditions can be determined by analyzing features such as brightness, contrast, and color distribution in an image. For example, histograms can be used to statistically analyze the brightness distribution of an image. If the brightness values ​​are concentrated in the higher range, the lighting is strong; if the brightness values ​​are concentrated in the lower range, the lighting is weak. The presence of shadows in the image can also help determine the direction and intensity of the light. Image processing and machine learning algorithms can be used to identify and classify features such as road textures, boundaries, and lane lines in an image. For example, by identifying the number, color, and shape of lane lines, it can be determined whether it is a highway, an urban road, or a rural road.

[0017] More specifically, lighting conditions affect image quality and clarity. Under different lighting conditions, obstacles and road features in the image will appear differently. Understanding the characteristics of lighting conditions can provide a reference for subsequent data processing and analysis. For example, in the case of insufficient light, methods such as enhancing image contrast can be used to improve image recognizability. Different road types have different driving rules and potential risks for vehicles. Clarifying the characteristics of road types helps to more accurately assess the safety of the vehicle's environment.

[0018] More specifically, different data filtering modes are pre-set based on the characteristics of lighting conditions and road types. For example, under strong light conditions, a filtering mode is used to remove overexposed areas in the image; for urban roads, key information such as lane lines and traffic signs is extracted. By applying the corresponding data filtering modes to the environmental monitoring data, noise and irrelevant information are removed, and key data related to obstacles and road conditions are extracted.

[0019] More specifically, environmental monitoring data contains a large amount of redundant information and noise, which increases the complexity and computational load of subsequent analysis. By selecting an appropriate data filtering mode, key data can be extracted in a targeted manner, improving the efficiency and accuracy of data processing and reducing unnecessary consumption of computing resources.

[0020] More specifically, target detection algorithms (such as YOLO, Faster R-CNN, etc.) are used to identify and locate obstacles in key monitoring data, determine the type (such as pedestrians, vehicles, obstacles, etc.) and location of obstacles, and at the same time, image processing and analysis techniques are used to evaluate the road's smoothness, slope, curves and other conditions. The results of identification and evaluation are integrated to obtain environmental perception information of the vehicle's environment.

[0021] More specifically, environmental perception information is an important basis for safe vehicle driving. By identifying obstacles and road conditions through key monitoring data, vehicles can understand the surrounding environment, detect potential dangers in a timely manner, and provide support for subsequent driving decisions, such as avoiding collisions with obstacles and adjusting driving speed to adapt to road conditions.

[0022] Specifically, in step S3 of the embodiment provided in this application, the vehicle perception AI will fuse the image data, radar data, etc. of the obstacle in the environmental perception information. For example, the camera will be used to obtain the visual features of the obstacle, and the radar will obtain the distance and speed information of the obstacle. The data of the two will be combined to determine the position of the obstacle more accurately. The position information of the obstacle will be converted into a coordinate system with the vehicle as the origin, and the relative positioning data such as the distance and direction of the obstacle relative to the vehicle will be calculated. The position of the obstacle in the actual space can be deduced by mathematical methods such as trigonometric functions based on the pixel position of the obstacle in the image and the parameters of the camera.

[0023] More specifically, accurate relative positioning data is the foundation for subsequent collision risk analysis. Only by clarifying the relative positional relationship between the vehicle and each obstacle can we determine whether the vehicle will collide with the obstacle during driving, thereby ensuring the safe driving of the vehicle.

[0024] More specifically, vehicle perception AI simulates the vehicle's driving path and the obstacle's trajectory over a future period based on predicted driving trajectory and relative positioning data of obstacles. It considers factors such as the obstacle's speed and acceleration to predict its position at future moments. By comparing the positions of the vehicle and the obstacle at different future moments, it determines whether there is a possibility of collision. A certain safe distance threshold can be set. When the distance between the vehicle and the obstacle is less than the threshold, it is considered that there is a risk of collision. At the same time, collision risk features are generated based on factors such as the probability of collision and the time of collision.

[0025] More specifically, collision risk analysis is a key aspect of ensuring vehicle driving safety. By analyzing predicted driving trajectories and relative positioning data of obstacles, potential collision hazards can be identified in advance, allowing for timely implementation of appropriate measures such as deceleration and avoidance to prevent accidents from occurring.

[0026] More specifically, the collision risk characteristics of a vehicle at different times (such as the probability of collision and the time of collision) are recorded in chronological order to form a driving risk sequence. The changing trend of the driving risk sequence is analyzed to determine the safety of the vehicle during driving. For example, if the collision risk characteristics continue to increase over a period of time, it means that the risk faced by the vehicle during driving is gradually increasing; if the risk characteristics remain at a low level, it means that the vehicle is relatively safe to drive. By comprehensively considering the overall situation of the risk sequence, the driving status information of the vehicle is obtained, such as safe, some risk, or high risk.

[0027] More specifically, continuous recording and analysis of vehicle driving risks can provide a comprehensive understanding of the vehicle's safety status during operation. This driving status information can provide decision-making support for drivers or autonomous driving systems, helping them to adjust driving strategies in a timely manner and ensure safe driving. At the same time, the driving risk sequence can also serve as an important reference for subsequent accident analysis and improvement of safety measures.

[0028] Specifically, in step S4 of the embodiment provided in this application, the vehicle sends driving status information (including the vehicle's position, speed, driving direction, collision risk, etc.) to the cloud positioning map server through the vehicle communication module (such as 4G, 5G and other wireless communication technologies). After receiving the vehicle's driving status information, the cloud positioning map server marks the vehicle at the corresponding location on the map. The marking content may include information such as the vehicle's identification, driving status, and risk level, so as to intuitively display the relevant situation of the vehicle.

[0029] More specifically, uploading driving status information to a cloud-based location map enables centralized management and sharing of information. By marking the location on the map, vehicles and other relevant parties (such as traffic management departments and other vehicles) can intuitively understand the vehicle's location and status. This helps improve traffic transparency and manageability, and provides a foundation for subsequent information exchange and analysis.

[0030] More specifically, taking the current vehicle's labeling information as the center, a specified range of neighboring areas is delineated on the cloud positioning map. This range can be set according to actual needs, such as a circular area with a radius of several kilometers. The cloud positioning map server filters the labeling information of other vehicles in the neighboring area from the database, including their position, speed, driving direction, collision risk, etc. This information is integrated to obtain the positional relationship information (such as the distance between vehicles, relative position, etc.) and driving status information of each vehicle in the neighboring area.

[0031] More specifically, understanding the location and driving conditions of other vehicles in the vicinity is crucial for assessing vehicle driving safety. By collecting information on neighboring vehicles, it is possible to analyze the mutual influence between vehicles and potential collision risks. For example, if the driving direction and speed of an adjacent vehicle conflict with those of the current vehicle, there may be a risk of collision.

[0032] More specifically, the cloud-based positioning map server uses a pre-established collision risk prediction model, combined with the positional relationship information and driving status information of each vehicle, to predict the collision risk between vehicles. The model can consider factors such as vehicle speed, driving direction, acceleration, and distance. Through mathematical calculations and algorithm analysis, it assesses the possibility of a collision between vehicles. Based on the collision risk prediction results, it generates corresponding driving safety warning information. The warning information can include the risk level (such as low risk, medium risk, high risk), the time and location where a collision may occur, etc. This warning information is fed back to the relevant vehicles to remind the drivers to take appropriate measures.

[0033] More specifically, by predicting the collision risk between vehicles in the vicinity, potential dangers can be detected in advance, providing timely safety warnings for vehicles. This helps drivers or autonomous driving systems take effective measures, such as slowing down or avoiding collisions, to prevent traffic accidents and improve road traffic safety. At the same time, this method of information interaction and analysis can achieve collaborative safety between vehicles, making the entire transportation system more efficient and safer.

[0034] In one embodiment, the steps of acquiring vehicle driving monitoring data and predicting the future driving trajectory of the vehicle based on the driving monitoring data to obtain the predicted driving trajectory include: S11: Collect vehicle speed and direction of travel to obtain vehicle driving monitoring data, and combine the driving monitoring data at each time moment into a reference data stream; S12: Based on the reference data stream, identify the vehicle's driving style to obtain a vehicle driving style profile; S13: Based on the driving style profile, predict the future driving trajectory of the current driving monitoring data to obtain the predicted driving trajectory.

[0035] Specifically, vehicles are typically equipped with speed sensors, such as electromagnetic induction sensors or Hall effect sensors. Electromagnetic induction sensors determine the wheel rotation speed by measuring the induced electromotive force generated when the wheel rotates, and then calculate the vehicle speed. Hall effect sensors use Hall elements to detect changes in magnetic fields, thereby sensing the wheel rotation speed and measuring the vehicle speed. The sensor collects speed data at certain time intervals (e.g., once per second) and converts it into digital signals that are transmitted to the vehicle's electronic control unit (ECU).

[0036] More specifically, the vehicle's direction of travel can be obtained using the Global Positioning System (GPS). The GPS module receives signals from multiple satellites and determines the vehicle's direction of travel by calculating the changes in the vehicle's position at different times. An Inertial Measurement Unit (IMU) can also be used to assist in measuring the direction of travel. The IMU contains an accelerometer and a gyroscope. The accelerometer can measure the vehicle's acceleration in various directions, while the gyroscope can measure the vehicle's rotational angular velocity. By processing and analyzing this data, the vehicle's direction of travel can be calculated.

[0037] More specifically, the ECU arranges the collected driving speed and driving direction data at different times in chronological order to form a continuous data stream, namely the reference data stream. This data stream can be stored in the vehicle's memory for subsequent analysis.

[0038] More specifically, driving speed and driving direction are basic parameters that describe the vehicle's motion state. They directly affect the vehicle's driving trajectory. By collecting these data and combining them into a reference data stream, basic information can be provided for subsequent driving style recognition and trajectory prediction. Only by accurately grasping the vehicle's current driving state can its future driving trajectory be predicted more accurately.

[0039] More specifically, features related to driving style are extracted from the reference data stream, such as the rate of change of acceleration, speed fluctuations, and the frequency of rapid acceleration and deceleration. For example, frequent rapid acceleration and deceleration indicate that the driver's driving style is more aggressive, while relatively stable speed changes indicate a more conservative driving style.

[0040] More specifically, machine learning or deep learning algorithms are used to analyze the extracted features. Common algorithms include decision trees, support vector machines, and neural networks. A large amount of historical data on different driving styles is collected as a training set to train the model. During training, the model learns the feature patterns corresponding to different driving styles. Features extracted from the current vehicle's reference data stream are input into the trained model, and the model identifies the vehicle's driving style based on the learned patterns, generating a driving style profile, such as aggressive, conservative, or economical.

[0041] More specifically, different driving styles lead to different behaviors of a vehicle during driving, which in turn affects the vehicle's future trajectory. For example, drivers with an aggressive driving style tend to overtake quickly and change lanes frequently, while drivers with a conservative driving style will maintain a relatively stable driving speed and lane. By identifying a vehicle's driving style, it is possible to more accurately predict the vehicle's future trajectory, because driving style reflects the driver's behavioral habits and decision-making patterns.

[0042] More specifically, corresponding prediction models are established based on different driving styles. For example, for aggressive driving styles, more frequent acceleration, deceleration, and lane changes can be considered; for conservative driving styles, models are built based on relatively stable speed and direction changes. Prediction models can be constructed based on physical principles and mathematical algorithms, combined with the vehicle's dynamic characteristics and driving environment factors (such as road slope, curvature, etc.).

[0043] More specifically, the current driving monitoring data (driving speed and driving direction) is input into the corresponding prediction model. Combined with the driving style profile, the model will calculate the possible driving trajectory of the vehicle in the future (e.g., the next 10 seconds, 30 seconds, etc.) according to the preset rules and algorithms. The prediction results can be represented in the form of coordinate points, which constitute the predicted driving trajectory of the vehicle.

[0044] More specifically, driving style profiles reflect the behavioral tendencies of vehicle drivers. Combined with current driving monitoring data, they can more accurately predict the future driving trajectory of vehicles. Considering driving style can make the prediction results more consistent with the actual situation and improve the accuracy of the prediction. This is of great significance for applications such as vehicle safety warnings and autonomous driving, and can help vehicles take countermeasures in advance to avoid potential dangers.

[0045] In one embodiment, the steps of acquiring vehicle environmental monitoring data and analyzing the obstacle and road conditions of the vehicle's environment based on the environmental monitoring data to obtain environmental perception information include: S21: The camera module collects images of the vehicle's surroundings to obtain environmental monitoring data; S22: Analyze the lighting conditions and road types of the environmental monitoring data to obtain the lighting condition characteristics and road type characteristics of the environment in which the vehicle is located; S23: Select a specified data filtering mode based on the lighting condition characteristics and the road type characteristics to extract key data from the environmental monitoring data and obtain key monitoring data; S24: Based on the key monitoring data, identify obstacles and road conditions in the vehicle's environment to obtain environmental perception information.

[0046] Specifically, multiple cameras are reasonably installed on the vehicle, such as front-view cameras, rear-view cameras, and surround-view cameras. The front-view camera is generally installed at the front of the vehicle to capture environmental information in the direction the vehicle is traveling; the rear-view camera is installed at the rear of the vehicle to assist in operations such as reversing; and the surround-view cameras are distributed around the vehicle to provide a 360-degree environmental view. The cameras capture images of the environment around the vehicle at a certain frame rate (such as 30 frames per second). The image sensors inside the cameras convert light signals into electrical signals, and after analog-to-digital conversion and image processing, they are stored as digital image data. This data constitutes environmental monitoring data.

[0047] More specifically, images can intuitively reflect various information about the vehicle's environment, including the location of obstacles, road conditions, traffic signs, etc. By collecting images through cameras, rich environmental information can be obtained, providing basic data for subsequent environmental analysis. The arrangement of multiple cameras can provide a more comprehensive environmental perspective, reduce blind spots, and improve the completeness and accuracy of environmental information.

[0048] More specifically, the average brightness value of the image is calculated. A higher brightness value indicates stronger lighting, while a lower brightness value indicates weaker lighting. The brightness distribution can also be observed by analyzing the image's histogram to determine if there is overexposure or underexposure. Different lighting conditions affect the color performance of the image. For example, under sunlight on a sunny day, the colors of the image are more vibrant, while on a cloudy day or at night, the colors become dull. By analyzing the color characteristics of the image, the lighting conditions can be further determined.

[0049] More specifically, image processing algorithms are used to extract features such as road texture, boundaries, and lane lines from images. For example, lane lines on highways are usually more regular and clear, while lane lines on rural roads may be less obvious. Machine learning or deep learning models are used to classify and identify the extracted features to determine whether the road type is a highway, urban road, or rural road. Common models include convolutional neural networks (CNNs). By training on a large number of images of different road types, the model can learn the feature patterns of different road types, thereby achieving accurate classification.

[0050] More specifically, lighting conditions affect image quality and clarity. Under different lighting conditions, obstacles and road features in images will appear differently. Understanding the characteristics of lighting conditions can provide a reference for subsequent data processing and analysis. For example, in insufficient lighting conditions, methods such as enhancing image contrast can be used to improve image recognizability. Different road types have different driving rules and potential risks for vehicles. Clearly defining the characteristics of road types helps to more accurately assess the safety of the vehicle's environment and provides a basis for subsequent decision-making.

[0051] More specifically, corresponding data filtering modes are pre-set for different lighting conditions and road types. For example, under strong light conditions, a filtering mode is set to remove overexposed areas in the image; for urban roads, a filtering mode is set to extract key information such as lane lines and traffic signs. Based on the characteristics of lighting conditions and road types, appropriate data filtering modes are selected to process environmental monitoring data. By filtering out noise and irrelevant information, key data related to obstacles and road conditions, such as the outline of obstacles and the boundaries of roads, are extracted.

[0052] More specifically, environmental monitoring data contains a large amount of redundant information and noise, which increases the complexity and computational load of subsequent analysis. By selecting an appropriate data filtering mode, key data can be extracted in a targeted manner, improving the efficiency and accuracy of data processing and reducing unnecessary consumption of computing resources. At the same time, key data can better reflect the actual situation of the vehicle's environment, which is conducive to the accurate identification of obstacles and road conditions in the future.

[0053] More specifically, object detection algorithms (such as YOLO, Faster R-CNN, etc.) are used to identify and locate obstacles in key monitoring data. These algorithms can identify different types of obstacles, such as pedestrians, vehicles, and other obstacles, by extracting and classifying features from images, and determine their positions and boundaries. By combining data from other sensors (such as radar, lidar, etc.), the accuracy of obstacle identification can be improved. For example, radar can provide distance and speed information of obstacles, which, when fused with image data from cameras, can provide a more comprehensive understanding of the obstacle situation.

[0054] More specifically, by analyzing road images in key monitoring data, the smoothness, slope, curves, and other conditions of the road can be identified. For example, by analyzing the texture and color changes of the road in the image, it can be determined whether there are potholes, water accumulation, etc. The extracted road features are matched with pre-stored road models to determine the specific type and condition of the road.

[0055] More specifically, environmental perception information is an important basis for safe vehicle driving. By identifying obstacles and road conditions through key monitoring data, vehicles can understand the surrounding environment, detect potential dangers in a timely manner, and provide support for subsequent driving decisions, such as avoiding collisions with obstacles and adjusting driving speed to adapt to road conditions. Accurate environmental perception can improve vehicle safety and reliability.

[0056] In one embodiment, the step of analyzing the predicted driving trajectory and the environmental perception information using locally deployed vehicle perception AI to obtain vehicle driving status information includes: S31: The vehicle perception AI deployed locally performs relative positioning analysis on the environmental perception information to obtain the relative positioning data of the vehicle relative to each obstacle. S32: Using locally deployed vehicle perception AI, collision risk analysis is performed on the predicted driving trajectory based on the relative positioning data of each obstacle to obtain the collision risk characteristics of the vehicle relative to each obstacle at future moments. S33: Record the collision risk characteristics of the vehicle at each moment to obtain the vehicle's driving risk sequence, and perform a safety analysis on the vehicle's driving process based on the driving risk sequence to obtain the vehicle's driving status information.

[0057] Specifically, vehicle perception AI integrates environmental perception information from different sensors, such as image data acquired by cameras and distance and speed data provided by radar or lidar. For example, cameras can identify the appearance and approximate location of obstacles, while radar can accurately measure the distance and relative speed between obstacles and vehicles. Through data fusion algorithms, these multi-source data are processed to obtain more accurate obstacle information.

[0058] More specifically, the location information of obstacles is uniformly transformed into a coordinate system with the vehicle as the origin. Typically, the vehicle itself establishes a local coordinate system. By calculating the coordinates of the obstacle in this coordinate system, its position relative to the vehicle is determined. This coordinate system transformation can be completed using trigonometric functions and geometric relationships, combined with the installation position and angle information of the sensors.

[0059] More specifically, under a unified coordinate system, relative positioning data such as distance and angle between the vehicle and each obstacle are calculated. For example, the straight-line distance between the obstacle and the vehicle is obtained by calculating the Euclidean distance between two points, and the orientation of the obstacle relative to the vehicle is determined by angle calculation.

[0060] More specifically, accurate relative positioning data is the foundation for subsequent collision risk analysis. Only by clearly defining the relative positional relationships between the vehicle and various obstacles can we determine whether the vehicle will collide with an obstacle during operation. Different types of sensors each have their advantages and disadvantages; data fusion can fully leverage their strengths, improve the accuracy of obstacle positioning, and provide reliable information support for safe vehicle operation.

[0061] More specifically, vehicle perception AI simulates the movement trajectories of vehicles and obstacles over a future period of time based on predicted driving trajectories and relative positioning data of obstacles. It takes into account the vehicle's speed and acceleration as well as the motion state of the obstacles (if the obstacles are moving), and uses physical models and kinematic principles to predict the trajectory.

[0062] More specifically, based on the simulated trajectory, to determine whether the positions of the vehicle and the obstacle will overlap at different times in the future, a safe distance threshold can be set. When the distance between the vehicle and the obstacle is less than this threshold, a collision risk is considered to exist. At the same time, collision risk features are generated based on factors such as the probability of a collision and the collision time. For example, if it is expected that the distance between the vehicle and the obstacle will be less than the safe distance within the next 5 seconds, a high collision risk is considered to exist.

[0063] More specifically, in order to more accurately assess collision risk, the risk can be quantified. For example, based on factors such as the probability of a collision and the severity of the collision, a numerical value can be assigned to the collision risk, such as a risk score of 0-100, where 0 represents no risk and 100 represents a collision that will definitely occur.

[0064] More specifically, collision risk analysis is a key aspect of ensuring vehicle driving safety. By analyzing predicted driving trajectories and relative obstacle positioning data, potential collision hazards can be identified in advance, allowing for timely implementation of appropriate measures such as deceleration and avoidance to prevent accidents. Quantifying collision risk characteristics enables vehicle systems to more intuitively understand the degree of risk, thereby facilitating more rational decision-making.

[0065] More specifically, collision risk characteristics of a vehicle at different times (such as collision probability, collision time, risk score, etc.) are recorded chronologically to form a driving risk sequence. This data can be stored in the vehicle's local memory or a data recording system. The driving risk sequence can then be analyzed to observe trends in risk over time. For example, statistical measures such as the average and standard deviation of the risk can be calculated to assess its stability. Furthermore, fluctuations in risk can be analyzed to identify periods of increased or decreased risk.

[0066] More specifically, based on the analysis results of the driving risk sequence, the safety of the vehicle's driving process is assessed. If the risk values ​​in the risk sequence are generally low and stable, it indicates that the vehicle is relatively safe to drive. If the risk values ​​fluctuate frequently and show an upward trend, it indicates that there are certain safety hazards during the vehicle's driving process. Taking into account various factors, the vehicle's driving status information is obtained, such as safe driving, certain risks, or high-risk driving.

[0067] More specifically, continuous recording and analysis of vehicle driving risks can provide a comprehensive understanding of the vehicle's safety status during operation. This driving status information can provide decision-making support for drivers or autonomous driving systems, helping them to adjust driving strategies in a timely manner and ensure safe driving. At the same time, the driving risk sequence can also serve as an important reference for subsequent accident analysis and improvement of safety measures, so as to continuously optimize the vehicle's safety performance.

[0068] In one embodiment, the method further includes: acquiring weather information of the area where the vehicle is located, generating corresponding driving impact labels for the vehicle based on the weather information, and adjusting the parameters of the vehicle perception AI based on the driving impact labels.

[0069] Specifically, vehicles connect to meteorological data service platforms via onboard communication modules (such as 4G and 5G networks). These platforms typically provide detailed weather information, including temperature, humidity, rainfall, wind speed, and visibility. Vehicles can request real-time weather data for their location from the platform based on their GPS positioning information. Vehicles can also be equipped with local weather sensors, such as rain sensors and light sensors. Rain sensors can detect whether there is rainfall and the intensity of rainfall in real time, while light sensors can sense the intensity of light to help determine weather conditions (such as sunny or cloudy). Combining local sensor data with weather information queried online improves the accuracy of weather information acquisition.

[0070] More specifically, a series of rules related to weather conditions are predefined, and corresponding driving impact labels are generated based on different weather parameters. For example, when the rainfall is heavy, a wet road surface label is generated, indicating that the slippery road surface will affect the vehicle's braking performance and handling. When the visibility is below a certain threshold (such as 500 meters), a low visibility label is generated, prompting the driver or vehicle system to reduce driving speed and increase observation of the road conditions ahead. When the temperature is below freezing, a low temperature icy road surface label is generated, reminding drivers that the road surface may be icy and that they should be extra careful when driving. The obtained weather information is matched with the predefined rules to generate corresponding driving impact labels for the vehicle. If multiple rules are met at the same time, multiple labels are generated.

[0071] More specifically, corresponding parameter adjustment strategies are defined for each driving impact label. For example, for the wet and slippery road surface label, the safe distance threshold for collision risk assessment in the vehicle perception AI is increased because the vehicle's braking distance increases on wet and slippery surfaces. At the same time, the parameters of the target detection algorithm are adjusted to improve the sensitivity of identifying obstacles such as puddles and water accumulation on the road surface. For the low visibility label, the time span for predicting the driving trajectory is reduced because it is difficult to accurately predict road conditions over long distances in low visibility conditions. The detection frequency of obstacles ahead is increased to improve the response speed to potential hazards. For the low temperature and icy road surface label, the vehicle's dynamic model parameters are adjusted to more accurately simulate the vehicle's driving state, taking into account the reduced friction on icy roads. The vehicle perception AI automatically adjusts the corresponding parameters based on the generated driving impact labels. These parameter adjustments can take effect in real time, ensuring that the vehicle can make accurate judgments and decisions under different weather conditions.

[0072] More specifically, different weather conditions can significantly affect vehicle operation. For example, rain makes roads slippery, reducing tire-road friction and increasing braking distance; low visibility affects the driver's vision, increasing the risk of collision. By acquiring weather information and generating driving impact labels, the vehicle system can be promptly alerted to these potential dangers. Adjusting the parameters of the vehicle's perception AI can enable the vehicle to perceive the environment and assess risks more accurately under different weather conditions, thereby taking appropriate safety measures and reducing the probability of accidents.

[0073] More specifically, changes in weather conditions can affect sensor performance and the accuracy of environmental perception. For example, in heavy rain, camera visibility can be obstructed by raindrops, and radar signals may also be affected. By adjusting the parameters of the vehicle's perception AI based on weather information, these performance degradations can be compensated for, ensuring that the vehicle's perception system functions properly under various weather conditions and improving the overall performance and reliability of the vehicle.

[0074] More specifically, different weather conditions require different driving strategies. By dynamically adjusting the parameters of the vehicle's perception AI, the vehicle can make more appropriate decisions based on real-time weather conditions. For example, it can automatically reduce its speed and enhance its monitoring of the surrounding environment in severe weather, making the vehicle's driving more adaptable to actual weather conditions and improving driving safety and comfort.

[0075] In one embodiment, the method further includes analyzing the potential driving risks of the vehicle at the current moment and the corresponding countermeasures based on the driving condition information, so as to generate corresponding risk avoidance suggestions.

[0076] Specifically, the driving status information is deeply mined, and key indicators such as vehicle speed, acceleration, distance to obstacles, and collision risk characteristics are analyzed. Through trend analysis and correlation analysis of this data, potential driving risks at the current moment can be predicted. For example, if the vehicle speed is too high and the distance to the obstacle in front is rapidly shortened, there is a risk of collision; if the vehicle is traveling at too high a speed while driving on a curve, there is a risk of rollover.

[0077] More specifically, risk prediction models are built using machine learning or deep learning algorithms. A large amount of historical driving data and accident data are collected to train the model so that it can learn the patterns and rules of potential risks under different driving conditions. When the current driving condition information is input into the trained model, the model will output the type and probability of potential risks.

[0078] More specifically, a rule base is established that includes various potential risks and their corresponding countermeasures. When a potential risk is predicted, the corresponding countermeasure is searched in the rule base. For example, if a collision is predicted ahead, the corresponding countermeasures in the rule base are immediate braking and swerving. For some complex potential risk situations, scenario simulation technology is used to simulate the vehicle's driving state and results under different countermeasures, evaluate the effectiveness of each countermeasure, and select the optimal countermeasure. For example, when encountering a pedestrian suddenly crossing the road ahead, the system simulates whether the vehicle will avoid a collision under different measures such as braking and swerving, as well as the impact on other road users, thereby determining the best countermeasure.

[0079] More specifically, the predicted potential driving risks and the analyzed countermeasures are integrated to generate risk avoidance recommendations in a clear and easy-to-understand manner. These recommendations can include the risk type, risk level, and specific countermeasures. For example, if there is a collision risk ahead (high risk), brake immediately and maintain a safe distance. These risk avoidance recommendations are visualized through the vehicle's dashboard, head-up display (HUD), or in-vehicle infotainment system. Additionally, sound and vibration can be used to alert the driver to the risk and prompt them to take appropriate action.

[0080] More specifically, by predicting and analyzing potential driving risks, it is possible to identify dangerous situations in advance and take timely countermeasures. This helps to avoid or reduce traffic accidents and protect the lives of drivers, passengers and other road users. For example, by issuing risk avoidance advice in time before a collision is about to occur, drivers have enough time to react and take measures such as braking or swerving, thereby avoiding an accident.

[0081] More specifically, risk avoidance advice provides drivers with clear guidance to help them make correct decisions in complex driving environments. Especially in emergency situations, drivers may make incorrect judgments due to tension, and risk avoidance advice can provide objective and scientific solutions to improve driving safety. Timely and accurate risk avoidance advice can make drivers feel the intelligence and care of the vehicle, increasing user satisfaction and trust in the vehicle. At the same time, reducing the occurrence of accidents can also reduce vehicle maintenance costs and insurance premiums, bringing real benefits to users.

[0082] In one embodiment, the step of uploading the driving status information to a cloud-based positioning map for information interaction and analysis with other vehicles in the vicinity to obtain vehicle driving safety warning information includes: S31: Upload the driving status information to the cloud positioning map, and mark the vehicle information in the cloud positioning map to obtain the vehicle's marking information; S32: Based on the annotation information, select a nearby area within a specified range on the cloud positioning map, and collect the annotation information of the remaining vehicles in the nearby area to obtain the positional relationship information and driving status information of each vehicle in the nearby area. S33: Based on the positional relationship information of each vehicle, the collision risk is predicted based on the driving status information of each vehicle, and driving safety warning information of the vehicle in the vicinity is obtained.

[0083] Specifically, the vehicle establishes a connection with the cloud-based positioning map server through an onboard communication module (such as 4G, 5G, and other wireless communication technologies). The vehicle sends the previously analyzed driving status information, including the vehicle's position, speed, driving direction, collision risk characteristics, and other data, to the cloud server in accordance with the prescribed data format and communication protocol. After receiving the vehicle's driving status information, the cloud-based positioning map server marks the actual location of the vehicle on the map. In addition to basic location information, the markings can also include visual indicators such as the vehicle's status (such as normal driving, braking, turning, etc.) and risk level (low, medium, high) for easy viewing and analysis later.

[0084] More specifically, uploading vehicle driving status information to a cloud-based location map allows for centralized storage and management of information from multiple vehicles, facilitating information exchange and sharing between different vehicles. After marking information on the map, both the vehicles themselves and relevant parties such as traffic management departments can intuitively understand the location and status of each vehicle from a global perspective, which helps to better grasp the traffic situation.

[0085] More specifically, centered on the current vehicle's location, a specified neighboring area is defined on the cloud-based positioning map according to actual needs. This area can be circular, rectangular, or other shapes, and the radius or side length can be adjusted according to different application scenarios. For example, in urban roads, it can be set to several hundred meters to several kilometers. Based on the defined neighboring area, the cloud server filters out the location information of other vehicles in that area. By querying the database, it obtains the driving status information of these vehicles, such as their location, speed, driving direction, and risk level, and organizes and analyzes the information to obtain the positional relationship information between the vehicles (such as distance and relative orientation).

[0086] More specifically, understanding the location and driving conditions of other vehicles in the vicinity allows vehicles to have a clearer understanding of the surrounding traffic environment, promptly detect potential dangers and traffic congestion, and collect information on neighboring vehicles is the basis for collision risk prediction. Only by mastering the positional relationships and driving conditions of each vehicle can the possibility of a collision between vehicles be accurately assessed.

[0087] More specifically, by using mathematical models and algorithms, and combining vehicle positional information (such as distance, relative speed, and direction of travel) and driving condition information (such as acceleration and braking status), a collision risk prediction model is established. Common models can be based on physical kinematics principles, taking into account vehicle dynamics and traffic rules. Relevant information of each vehicle is input into the collision risk model to calculate the probability of a collision between any two vehicles in the future (such as in the next 10 seconds or 30 seconds). Based on the calculation results, the collision risk is assessed and classified into risk levels (such as low risk, medium risk, and high risk). Based on the collision risk assessment results, driving safety warning information for vehicles in the vicinity is generated. The warning information may include the risk level, the identification of vehicles that may collide, the expected time and location of the collision, and other information is fed back to the relevant vehicles.

[0088] More specifically, by predicting and warning of collision risks between vehicles, drivers or autonomous driving systems can be informed of potential dangers in advance, giving them sufficient time to take appropriate measures, such as slowing down or avoiding collisions, thereby preventing or reducing traffic accidents. This cloud-based information interaction and collision risk prediction mechanism helps achieve collaborative safety between vehicles and improves the safety and efficiency of the entire transportation system. For example, when a vehicle detects a high risk of collision with another vehicle, it can promptly send warning information to surrounding vehicles, reminding them to take joint measures to avoid an accident.

[0089] In one embodiment, such as Figure 2 As shown, a vehicle driving safety warning system is provided to implement the vehicle driving safety warning method described in any one of the first aspects, comprising: The driving prediction module is used to acquire vehicle driving monitoring data and predict the future driving trajectory of the vehicle based on the driving monitoring data to obtain the predicted driving trajectory. The environmental perception module is used to acquire environmental monitoring data of the vehicle and analyze the obstacles and road conditions of the vehicle's environment based on the environmental monitoring data to obtain environmental perception information. The condition analysis module is used to analyze the predicted driving trajectory and the environmental perception information through locally deployed vehicle perception AI to obtain the vehicle's driving condition information. The safety warning module is used to upload the driving status information to the cloud positioning map for information interaction and analysis with other vehicles in the nearby area to obtain vehicle driving safety warning information.

[0090] In this embodiment, the specific implementation of each module in the above system embodiment is described in the above method embodiment, and will not be repeated here.

[0091] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data shall comply with the relevant laws, regulations and standards of the relevant countries and regions.

[0092] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0093] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0094] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A safety warning method for a vehicle traveling, characterized by, include: Acquire vehicle driving monitoring data, and predict the vehicle's future driving trajectory based on the driving monitoring data to obtain the predicted driving trajectory; The system acquires environmental monitoring data of the vehicle and analyzes the obstacles and road conditions of the vehicle's environment based on the environmental monitoring data to obtain environmental perception information. The predicted driving trajectory and the environmental perception information are analyzed by locally deployed vehicle perception AI to obtain the vehicle's driving status information. The driving status information is uploaded to a cloud-based positioning map to interact and analyze with other vehicles in the vicinity, thereby obtaining driving safety warning information for the vehicle.

2. The vehicle driving safety warning method as described in claim 1, characterized in that, The steps of acquiring vehicle driving monitoring data and predicting the vehicle's future driving trajectory based on the driving monitoring data to obtain the predicted driving trajectory include: The vehicle's speed and direction of travel are collected to obtain vehicle driving monitoring data, and the driving monitoring data at each time point are combined into a reference data stream; The vehicle's driving style is identified based on the reference data stream to obtain a vehicle driving style profile. Based on the driving style profile, the driving monitoring data at the current moment is used to predict the future driving trajectory, thus obtaining the predicted driving trajectory.

3. The vehicle driving safety warning method as described in claim 1, characterized in that, The steps of acquiring vehicle environmental monitoring data and analyzing the obstacle and road conditions of the vehicle's environment based on the environmental monitoring data to obtain environmental perception information include: The camera module captures images of the vehicle's surroundings to obtain environmental monitoring data. The environmental monitoring data is analyzed for lighting conditions and road type to obtain the lighting condition characteristics and road type characteristics of the environment in which the vehicle is located. Based on the lighting conditions and road type characteristics, a specified data filtering mode is selected to extract key data from the environmental monitoring data and obtain key monitoring data. Based on the key monitoring data, obstacles and road conditions in the vehicle's environment are identified to obtain environmental perception information.

4. The vehicle driving safety warning method as described in claim 1, characterized in that, The steps of analyzing the predicted driving trajectory and the environmental perception information using locally deployed vehicle perception AI to obtain vehicle driving status information include: The vehicle perception AI deployed locally performs relative positioning analysis on the environmental perception information to obtain the relative positioning data of the vehicle relative to each obstacle. The vehicle perception AI deployed locally performs collision risk analysis on the predicted driving trajectory based on the relative positioning data of each obstacle, and obtains the collision risk characteristics of the vehicle relative to each obstacle at future moments. The collision risk characteristics of the vehicle at each moment are recorded to obtain the vehicle's driving risk sequence. Based on the driving risk sequence, the safety analysis of the vehicle's driving process is performed to obtain the vehicle's driving status information.

5. The vehicle driving safety warning method as described in claim 4, characterized in that, Also includes: The system acquires weather information about the area where the vehicle is located, generates corresponding driving impact labels for the vehicle based on the weather information, and adjusts the parameters of the vehicle perception AI based on the driving impact labels.

6. The vehicle driving safety warning method as described in claim 4, characterized in that, It also includes analyzing the potential driving risks of the vehicle at the current moment based on the driving status information and the corresponding countermeasures, so as to generate corresponding risk avoidance suggestions.

7. The vehicle driving safety warning method as described in claim 1, characterized in that, The steps of uploading the driving status information to a cloud-based positioning map for information interaction and analysis with other vehicles in the vicinity to obtain vehicle driving safety warning information include: The driving status information is uploaded to a cloud-based positioning map, and the vehicle information is marked on the cloud-based positioning map to obtain the vehicle's marking information; Based on the annotation information, a specified range of neighboring areas is selected on the cloud positioning map, and the annotation information of other vehicles in the neighboring area is collected to obtain the positional relationship information and driving status information of each vehicle in the neighboring area. Based on the positional relationship information of each vehicle, the collision risk is predicted according to the driving status information of each vehicle, and the driving safety warning information of the vehicle in the vicinity is obtained.

8. A vehicle driving safety warning system, characterized in that, A method for implementing a vehicle driving safety warning as described in any one of claims 1-7 includes: The driving prediction module is used to acquire vehicle driving monitoring data and predict the future driving trajectory of the vehicle based on the driving monitoring data to obtain the predicted driving trajectory. The environmental perception module is used to acquire environmental monitoring data of the vehicle and analyze the obstacles and road conditions of the vehicle's environment based on the environmental monitoring data to obtain environmental perception information. The condition analysis module is used to analyze the predicted driving trajectory and the environmental perception information through locally deployed vehicle perception AI to obtain the vehicle's driving condition information. The safety warning module is used to upload the driving status information to the cloud positioning map for information interaction and analysis with other vehicles in the nearby area to obtain vehicle driving safety warning information.

Images