Road detection method and system based on millimeter wave radar

Abstract

The invention discloses a road detection method and system based on a millimeter wave radar, and relates to the technical field of road detection methods, and the method comprises the steps: determining a plurality of road features according to the road detection of the millimeter wave radar; and the road abnormal features are determined according to the forms of the plurality of road surface features and the positions of the plurality of road surface features, so that the detection accuracy of the specific condition of the road is further ensured. Therefore, the distance of the electric vehicle relative to the road abnormal feature is determined based on the position of the electric vehicle and the position of the road abnormal feature, and the driving risk coefficient of the electric vehicle is determined according to the distance of the electric vehicle relative to the road abnormal feature, the specific condition of the road and the driving state of the electric vehicle; the safe driving logic of the electric vehicle is matched according to the driving risk coefficient of the electric vehicle and the safe driving mapping relation of the electric vehicle, emergency avoidance of the electric vehicle relative to road abnormal characteristics is achieved, and the safe driving effect of the electric vehicle in various severe environment scenes is guaranteed.

Description

[0001] The present invention relates to the technical field of road detection methods, and in particular to a road detection method and system based on millimeter wave radar. Background Art

[0002] With the development of science and technology, electric vehicles travel relative to the road under the action of electric energy. Electric vehicles perform intelligent driving on the road and implement corresponding driving logic according to different driving scenarios. In the existing technology, electric vehicles identify the road surface as flat during driving. In reality, the actual road surface may have some abnormal road features such as puddles, local depressions or gravel. Existing electric vehicles are unable to identify abnormal road features, resulting in low accuracy in detecting the specific conditions of the road. Summary of the Invention

[0003] The purpose of the present invention is to overcome the deficiencies of the prior art, and the present invention provides a road detection method and system based on millimeter-wave radar.

[0004] An embodiment of the present invention provides a road detection method based on millimeter wave radar, comprising: Determining a scene in which the electric vehicle is located based on a current position of the electric vehicle and a road image captured by the electric vehicle; If the electric vehicle is in a harsh environment, the electric vehicle's millimeter-wave radar road surface detection is triggered, and multiple road surface features are determined based on the millimeter-wave radar road surface detection; determining abnormal road features according to the shapes and positions of the plurality of road features, and marking specific conditions of the road; determining a distance of the electric vehicle from the abnormal road feature based on the location of the electric vehicle and the location of the abnormal road feature, and determining a driving risk coefficient of the electric vehicle based on the distance of the electric vehicle from the abnormal road feature, specific conditions of the road, and a driving state of the electric vehicle; The safe driving logic of the electric vehicle is matched according to the driving risk coefficient of the electric vehicle and the safe driving mapping relationship of the electric vehicle to trigger the emergency avoidance of the electric vehicle relative to abnormal road features.

[0005] An embodiment of the present invention provides a millimeter-wave radar-based road detection system, which is applied to the above-mentioned millimeter-wave radar-based road detection method. The millimeter-wave radar-based road detection system includes: A scene module, configured to determine the scene in which the electric vehicle is located based on the current position of the electric vehicle and the road image captured by the electric vehicle; A road feature module is used to trigger the road surface detection of the electric vehicle's millimeter-wave radar if the electric vehicle is in a harsh environment, and determine multiple road surface features based on the road surface detection of the millimeter-wave radar; A road abnormality feature module is used to determine the road abnormality feature according to the shape and position of multiple road surface features, and mark the specific condition of the road; a driving risk coefficient module, configured to determine the distance of the electric vehicle from the abnormal road feature based on the position of the electric vehicle and the position of the abnormal road feature, and determine the driving risk coefficient of the electric vehicle according to the distance of the electric vehicle from the abnormal road feature, the specific conditions of the road, and the driving state of the electric vehicle; The emergency avoidance module is used to match the safe driving logic of the electric vehicle according to the driving risk coefficient of the electric vehicle and the safe driving mapping relationship of the electric vehicle, so as to trigger the emergency avoidance of the electric vehicle relative to abnormal road features.

[0006] Compared with the prior art, the present invention has the following beneficial effects: In an embodiment of the present invention, through the method in the embodiment of the present invention, the scene in which the electric vehicle is located is determined based on the current position of the electric vehicle and the road surface image taken by the electric vehicle; if the scene in which the electric vehicle is located is a harsh environment scene, the road surface detection of the millimeter-wave radar of the electric vehicle is triggered, and multiple road surface features are determined based on the road surface detection of the millimeter-wave radar; abnormal road features are determined based on the morphology of multiple road surface features and the positions of multiple road surface features, and the specific conditions of the road are marked, which is compatible with the overall consideration of the morphology of multiple road surface features and the positions of multiple road surface features, ensures the detection accuracy of abnormal road features, and further ensures the detection accuracy of the specific conditions of the road.

[0007] Therefore, based on the position of the electric vehicle and the position of the abnormal road feature, the distance of the electric vehicle relative to the abnormal road feature is determined, and the driving risk coefficient of the electric vehicle is determined according to the distance of the electric vehicle relative to the abnormal road feature, the specific conditions of the road and the driving state of the electric vehicle; the safe driving logic of the electric vehicle is matched according to the mapping relationship between the driving risk coefficient of the electric vehicle and the safe driving of the electric vehicle, the driving risk coefficient of the electric vehicle is introduced, and the accuracy of the safe driving logic of the electric vehicle is guaranteed, which realizes the emergency avoidance of the electric vehicle relative to the abnormal road feature and ensures the safe driving effect of the electric vehicle in various harsh environmental scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 is a flow chart of a road detection method based on millimeter-wave radar in an embodiment of the present invention; Figure 2 1 is a flow chart of step S11 in the road detection method based on millimeter wave radar in an embodiment of the present invention; Figure 3 1 is a flow chart of step S12 in the road detection method based on millimeter wave radar in an embodiment of the present invention; Figure 41 is a flow chart of step S13 in the road detection method based on millimeter wave radar in an embodiment of the present invention; Figure 5 1 is a flow chart of step S14 in the road detection method based on millimeter wave radar in an embodiment of the present invention; Figure 6 1 is a flow chart of step S15 in the road detection method based on millimeter wave radar in an embodiment of the present invention; Figure 7 Schematic diagram of the structure of a road detection system based on millimeter-wave radar in an embodiment of the present invention. DETAILED DESCRIPTION

[0009] The technical solutions in the embodiments of the present invention will be described clearly and completely below with reference to the accompanying drawings in the embodiments of the present invention.

[0010] See also Figures 1 to 7 A road detection method based on millimeter wave radar is applied to a road condition detection scenario based on millimeter wave radar; the road detection method based on millimeter wave radar includes: Step S11: determining the scene where the electric vehicle is located according to the current position of the electric vehicle and the road image captured by the electric vehicle; Step S12: If the electric vehicle is in a harsh environment, triggering the road surface detection of the electric vehicle's millimeter-wave radar, and determining a plurality of road surface features based on the road surface detection of the millimeter-wave radar; Step S13: determining abnormal road features based on the shapes and positions of the multiple road features, and marking the specific conditions of the road; Step S14: determining the distance of the electric vehicle from the abnormal road feature based on the position of the electric vehicle and the position of the abnormal road feature, and determining the driving risk coefficient of the electric vehicle based on the distance of the electric vehicle from the abnormal road feature, the specific conditions of the road, and the driving state of the electric vehicle; Step S15: matching the safe driving logic of the electric vehicle according to the driving risk coefficient of the electric vehicle and the safe driving mapping relationship of the electric vehicle to trigger emergency avoidance of the electric vehicle relative to abnormal road features; refer to Figure 2 , in step S11, determining the scene where the electric vehicle is located according to the current position of the electric vehicle and the road image captured by the electric vehicle; In the specific implementation process of the present invention, the specific steps are: S111: collecting the current location of the electric vehicle, determining the urban area where the electric vehicle is located based on the current location of the electric vehicle and a town distribution map, and determining a first scenario coefficient based on the urban area where the electric vehicle is located and the model of the electric vehicle; S112: The electric vehicle is equipped with corresponding cameras, which are distributed around the electric vehicle and collect images of the electric vehicle's surroundings. The corresponding road area is determined based on the detection of the images of the electric vehicle's surroundings, and a second scene coefficient is determined based on the environmental characteristics of the road area. S113: collecting scene mapping relationships corresponding to the electric vehicle, and determining the scene in which the electric vehicle is located based on the first scene feature, the second scene feature, and the scene mapping relationship. The scenes in which the electric vehicle is located include standard environment scenes and harsh environment scenes.

[0011] In an embodiment of the present application, the current position of the electric vehicle is collected, the urban area where the electric vehicle is located is determined based on the current position of the electric vehicle and the urban distribution map, and the first scenario coefficient is determined based on the urban area where the electric vehicle is located and the model of the electric vehicle. This takes into account the overall consideration of the urban area where the electric vehicle is located and the model of the electric vehicle, thereby ensuring the accuracy of the first scenario coefficient.

[0012] At this time, the electric vehicle has a built-in GPS positioning system; the GPS system can obtain the latitude and longitude coordinates of the electric vehicle in real time, thereby determining its geographical location; for example, the electric vehicle is currently located at 39.9042 degrees north latitude and 116.4074 degrees east longitude, which roughly corresponds to the central area of ​​Beijing.

[0013] The town distribution map is a geographic information system (GIS) layer that contains information such as town boundaries, road networks, and population distribution. By comparing the current location of the electric vehicle with the town distribution map, it is determined whether the electric vehicle is located in a town area and which specific town area it is located in. Optionally, through comparison, it is determined that the electric vehicle is currently located in Dongcheng District or Xicheng District of Beijing (depending on the accuracy of the latitude and longitude coordinates).

[0014] The first scenario coefficient is a comprehensive indicator that reflects factors such as road conditions, traffic density, and driving speed faced by electric vehicles in different urban areas. It is dynamically adjusted according to the urban area in which the electric vehicle is located and the model of the electric vehicle (such as vehicle model, power type, configuration, etc.). Optionally, assume that the electric vehicle is a high-performance electric vehicle, model XYZ-EV, and is currently located in the central urban area of ​​Beijing. Due to the busy traffic and complex road conditions in the central urban area, and the fact that electric vehicles face more charging station requirements when driving at high speeds, a higher first scenario coefficient (such as 0.8) is set to reflect these challenges. This first scenario coefficient will serve as an important reference for evaluating the driving risks of electric vehicles and triggering millimeter-wave radar detection in subsequent steps. If the first scenario coefficient is high, it means that the current environment in which the electric vehicle is located is more complex and dangerous, so it is necessary to assess the driving risks more carefully and trigger millimeter-wave radar detection more frequently to obtain real-time road condition information.

[0015] Furthermore, the electric vehicle is equipped with corresponding cameras, which are distributed around the electric vehicle and collect images of the electric vehicle's surroundings. The corresponding road area is determined based on the detection of the images of the electric vehicle's surroundings, and the second scene coefficient is determined based on the environmental characteristics presented by the road area. This is compatible with the overall consideration of the environmental characteristics presented by the road area, ensuring the accuracy of the second scene coefficient.

[0016] At this time, when electric vehicles are designed and manufactured, multiple cameras will be installed on their sides (such as the front, rear, left, right and top). These cameras have high resolution and wide-angle field of view, and can capture image information around the electric vehicle in real time; the number and position of the cameras depend on the specific design and purpose of the electric vehicle; for example, some high-end electric vehicles will install multiple cameras on the front and rear bumpers, side mirrors, roof and other positions to achieve a 360-degree panoramic view.

[0017] When an electric vehicle is driving, the cameras on its sides will continuously capture surrounding image information, including road signs, traffic lights, pedestrians, vehicles, weather conditions, road conditions, etc.; the image information collected by the camera will be transmitted in real time to the electric vehicle's control system or central processor for subsequent processing and analysis.

[0018] The electric vehicle's control system or central processor uses image processing algorithms and deep learning technology to detect and analyze the image information collected by the camera. By identifying features such as road signs, traffic lights, and buildings in the image, the electric vehicle's current road area is determined. At this time, if a highway entrance sign appears in the image, it is judged that the electric vehicle is about to enter the highway area. If buildings and pedestrians on city streets appear in the image, it is judged that the electric vehicle is currently in a city street area.

[0019] After determining the road area where the electric vehicle is located, the control system or central processing unit will further analyze the environmental characteristics of the area, including weather conditions (such as sunny, rainy, foggy, etc.), lighting conditions (such as daytime, nighttime, dusk, etc.), road conditions (such as dry, slippery, snowy, etc.), traffic density (such as busy, sparse, etc.), etc.; based on these environmental characteristics, the control system or central processing unit will set a second scenario coefficient; the second scenario coefficient reflects the driving risks and challenges faced by the electric vehicle in the current road area; optionally, when driving in rainy or foggy days, due to reduced visibility and slippery roads, a higher second scenario coefficient needs to be set; and when it is sunny and the road is dry, a lower second scenario coefficient is set.

[0020] Specifically, suppose there is an electric car driving on a city street; the electric car is equipped with multiple cameras on the front and rear bumpers and side-view mirrors, which can capture surrounding image information in real time; when the electric car drives into a busy commercial district, the camera captures surrounding pedestrians, vehicles, traffic lights, buildings and other features; the control system or central processor uses image processing algorithms and deep learning technology to detect and analyze these image information, and determines that the road area where the electric car is currently located is an urban commercial district.

[0021] The control system or central processor further analyzes the environmental characteristics of the area. Since it is currently sunny and the road is dry, but the traffic density is high and there are many pedestrians, a medium second-scenario coefficient (such as 0.6) is set. This second-scenario coefficient reflects the driving risks and challenges faced by electric vehicles in the current road area, such as the need to pay more attention to the dynamics of pedestrians and vehicles and maintain a safe distance. This second-scenario coefficient will serve as an important reference for evaluating the driving risks of electric vehicles and triggering millimeter-wave radar detection in subsequent steps. If the second-scenario coefficient is high, it means that the current environment of the electric vehicle is more complex and dangerous, so the driving risks need to be assessed more carefully and the millimeter-wave radar detection needs to be triggered more frequently to obtain real-time road condition information.

[0022] Therefore, the scene mapping relationship corresponding to the electric vehicle is collected, and the scene in which the electric vehicle is located is determined based on the first scene characteristics, the second scene characteristics and the scene mapping relationship. The scenes in which the electric vehicle is located include standard environment scenes and harsh environment scenes, which are compatible with the overall consideration of the first scene characteristics, the second scene characteristics and the scene mapping relationship to ensure the accuracy of the scene in which the electric vehicle is located.

[0023] At this time, the scene mapping relationship corresponding to the electric vehicle is collected. The scene mapping relationship is a predefined database, which establishes the correspondence between the first scene feature (such as a coefficient based on location and model), the second scene feature (such as an environmental feature coefficient based on a camera image) and the specific scene type; optionally, this mapping relationship is constructed based on a large amount of historical data, expert experience and machine learning algorithms; it takes into account the performance of different electric vehicle models under different road and environmental conditions, as well as the impact of these conditions on driving safety; in actual applications, this mapping relationship is stored in the electric vehicle's control system or cloud server for real-time access and updating.

[0024] After obtaining the first scene features and the second scene features, the electric vehicle's control system or central processor will use the scene mapping relationship to determine the specific scene the electric vehicle is currently in. This process involves complex algorithms and calculations, including feature matching, probability assessment, decision tree analysis, etc.; ultimately, the control system will classify the electric vehicle as being in a standard environment scenario or a harsh environment scenario based on these features and mapping relationships.

[0025] Optionally, a standard environment scenario generally refers to a situation where the road conditions are good, the traffic density is moderate, and the weather conditions are normal; in this scenario, the electric vehicle is driven according to normal driving logic and speed.

[0026] Harsh environment scenarios usually refer to situations with complex road conditions, high traffic density, and severe weather conditions (such as rain, snow, fog, and haze). In such scenarios, electric vehicles need to assess driving risks more carefully and trigger additional safety mechanisms (such as millimeter-wave radar detection, emergency avoidance logic, etc.).

[0027] Specifically, assume that an electric vehicle is traveling on a city street and has obtained a first scene feature (such as a coefficient of 0.7 based on location and model) and a second scene feature (such as a coefficient of 0.6 based on the environmental feature of the camera image) through steps S111 and S112; the control system of the electric vehicle will access a predefined scene mapping relationship database, which contains performance data of various electric vehicle models under different road and environmental conditions, as well as the correspondence between these data and specific scene types.

[0028] In the database, the control system will find the entry that best matches the current first and second scene characteristics; for example, it will find an entry in which the first scene characteristic is between 0.6 and 0.8 and the second scene characteristic is between 0.5 and 0.7. Electric vehicles are classified as being in a harsh environment scenario in most cases; based on this matching result, the control system will determine that the current scene of the electric vehicle is a harsh environment scenario; therefore, it will trigger additional safety mechanisms, such as increasing the detection frequency of the millimeter-wave radar, adjusting the driving speed, enabling emergency avoidance logic, etc., to ensure that the electric vehicle can drive safely in complex and dangerous environments. This scene determination process is crucial to the driving safety of electric vehicles and the comfort of passengers; by accurately identifying the current scene type, electric vehicles make more intelligent and safe driving decisions.

[0029] In one embodiment of the present application, it is assumed that there is a matching table of scene mapping relationships, which lists scene types (St) corresponding to different combinations of first scene features (Fs) and second scene features (Sc); the matching table of scene mapping relationships is shown in Table 1: Table 1 Matching table of scene mapping relationships Fs (First Scene Features) Sc (Second Scene Features) St (scene type) 0.0-0.4 0.0-0.4 Standard environment scene 0.5-0.7 0.0-0.4 Standard environment scene 0.8-1.0 0.0-0.4 Harsh environment scenarios 0.0-0.4 0.5-0.7 Standard environment scene 0.5-0.7 0.5-0.7 Harsh environment scenarios 0.8-1.0 0.5-0.7 Harsh environment scenarios 0.0-0.4 0.8-1.0 Harsh environment scenarios 0.5-0.7 0.8-1.0 Harsh environment scenarios 0.8-1.0 0.8-1.0 Harsh environment scenarios Assume that the first scenario feature Fs of an electric vehicle is 0.6 (indicating that it is located in a medium-sized urban area and the electric vehicle model is moderate), and the second scenario feature Sc is 0.7 (indicating that the weather conditions are slightly worse, such as light rain, but the road conditions are acceptable); according to the matching table, Fs=0.6 falls within the range of 0.5-0.7, and Sc=0.7 also falls within the range of 0.5-0.7, so the corresponding scenario type St is "harsh environment scenario"; at this time, the electric vehicle's control system will prompt the driver to pay attention to weather changes and prepare to take necessary driving measures, while increasing the detection frequency of the millimeter-wave radar to monitor the road conditions ahead.

[0030] refer to Figure 3 In step S12, if the scene in which the electric vehicle is located is a harsh environment scene, the road surface detection of the millimeter-wave radar of the electric vehicle is triggered, and multiple road surface features are determined based on the road surface detection of the millimeter-wave radar; In the specific implementation process of the present invention, the specific steps are: S121: When the electric vehicle is in a harsh environment, determining corresponding harsh environmental factors based on detection of the electric vehicle, and determining a detection mode of a millimeter-wave radar of the electric vehicle according to the harsh environmental factors and a driving state of the electric vehicle; S122: The millimeter-wave radar detects the path along the detection pattern and collects point cloud data of the road surface. Based on the synthesis of the point cloud data of the road surface, a plurality of first sub-road surface features are determined, and spatial positions of the plurality of first sub-road surface features are marked. S123: The camera configured for the electric vehicle performs directionally photographing the spatial position of each first sub-road surface feature to collect the current image of the road surface, and determines the corresponding second sub-road surface feature based on the recognition of the current image of the road surface, and determines multiple road surface features based on multiple first sub-road surface features, multiple second sub-road surface features and road surface mapping relationships.

[0031] In an embodiment of the present application, when the scene in which the electric vehicle is located is a harsh environment scene, the corresponding harsh environmental factors are determined based on the detection of the scene in which the electric vehicle is located, and the detection mode of the millimeter-wave radar of the electric vehicle is determined according to the harsh environmental factors and the driving state of the electric vehicle. This takes into account the overall consideration of the harsh environmental factors and the driving state of the electric vehicle, and ensures the accuracy of the detection mode of the millimeter-wave radar of the electric vehicle.

[0032] At this time, the control system of the electric vehicle will first determine whether the current scene is a harsh environment scene, which is usually determined based on the scene mapping relationship and feature data collected in the previous step (such as S113); at the same time, once it is confirmed to be in a harsh environment scene, the control system will further analyze the specific harsh environmental factors, which include weather conditions (such as rain, snow, haze, extreme high or low temperatures), road conditions (such as slippery, icy, potholes, construction areas), traffic density (such as congestion, accident-prone areas), etc.

[0033] The control system will obtain real-time driving status information of the electric vehicle, including current speed, acceleration, steering angle, braking status, etc. This information is crucial for adjusting the detection mode of the millimeter-wave radar; based on harsh environmental factors and the driving status of the electric vehicle, the control system will dynamically adjust the detection mode of the millimeter-wave radar, which involves adjusting the radar's scanning frequency, scanning angle, detection distance, resolution and other parameters to adapt to current environmental conditions and driving requirements.

[0034] Specifically, suppose an electric car is traveling on a highway and suddenly encounters a sudden rainstorm; the electric car's control system detects that the current weather conditions are severe through cameras and sensors, and determines that it is in a severe environment based on the scene mapping relationship; the control system further analyzes that the specific severe environmental factor is rainstorm weather, which leads to risks such as reduced visibility, slippery road surface, and easy skidding of the vehicle. At the same time, the control system obtains that the electric car is currently traveling on the highway at a speed of 100km / h and is not performing any steering or braking operations.

[0035] Based on the heavy rain and the high-speed driving state of the electric vehicle, the control system decided to adjust the detection mode of the millimeter-wave radar; specifically, it will increase the scanning frequency of the radar to obtain information about the road ahead more frequently; at the same time, adjust the scanning angle to better cover the blind spot in front of the electric vehicle; and increase the detection distance to ensure that there is enough time to react in an emergency; in addition, the resolution of the radar will be improved to more accurately identify obstacles and vehicles on the road ahead; through such adjustments, the millimeter-wave radar of the electric vehicle will be able to perceive the surrounding environment more effectively in heavy rain and improve driving safety; if an emergency occurs ahead, such as a vehicle breakdown or pedestrians crossing the road, the radar will be able to detect it earlier and remind the driver to take corresponding measures.

[0036] Furthermore, the millimeter-wave radar detects the path along the detection mode and collects point cloud data of the road surface. Based on the synthesis of the point cloud data of the road surface, multiple first sub-road surface features are determined, and the spatial positions of the multiple first sub-road surface features are marked, thereby introducing the method of marking the spatial positions of multiple first sub-road surface features.

[0037] At this point, the millimeter-wave radar starts operating according to the detection mode determined in step S121, which means that the radar will detect the path ahead at a specific scanning frequency, scanning angle, detection distance, and resolution. At the same time, during operation, the millimeter-wave radar will emit millimeter-wave signals and receive reflected signals. These reflected signals contain information such as the position, shape, and speed of the road surface and objects on it. By processing these signals, the radar generates point cloud data of the road surface. Point cloud data is a collection of a large number of three-dimensional coordinate points, each of which represents the position of a reflector (such as the road surface, vehicle, obstacle, etc.) detected by the radar.

[0038] The collected point cloud data contains noise and redundant information, so it needs to be synthesized. The synthesis process includes filtering, denoising, registration (aligning point cloud data at different times or perspectives), and gridding (converting point cloud data into continuous surfaces or grids). Through these processes, more accurate and complete road surface point cloud data is obtained. Based on the synthesized road surface point cloud data, various features on the road surface are identified through algorithms. These features are called first sub-road surface features, including potholes, bumps, cracks, obstacles, lane lines, etc.

[0039] For each identified first sub-road feature, its spatial position needs to be marked, which is usually achieved by recording the three-dimensional coordinates of the feature point. These coordinate information is crucial for subsequent path planning, obstacle avoidance decisions, etc.

[0040] Specifically, assume that an electric vehicle is traveling on a city road, and the millimeter-wave radar has started operating according to the detection mode determined in step S121; the radar detects the road ahead at a higher scanning frequency and an appropriate scanning angle to ensure that detailed information on the road surface can be captured.

[0041] The radar transmits millimeter-wave signals and receives reflected signals, generating point cloud data containing a large number of three-dimensional coordinate points. This data reflects the position and shape information of the road surface and objects on it. Filtering and denoising processes remove noise and redundant information from the point cloud data. Then, through registration and gridding, more accurate and complete road surface point cloud data is obtained. In the synthesized point cloud data, the algorithm identifies several potholes, a crack, and a roadblock ahead as the first sub-road surface features.

[0042] For each identified feature, its corresponding 3D coordinates are recorded. For example, the characteristic point of a pothole is located at (x1, y1, z1), the characteristic line of a crack is composed of a series of points (x2, y2, z2), (x3, y3, z3), ..., and the location of a roadblock is represented by its center point (x4, y4, z4) and size information. Through these steps, the electric vehicle's millimeter-wave radar can accurately perceive various features on the road ahead, providing key information for subsequent path planning and obstacle avoidance decisions.

[0043] Therefore, the camera equipped with the electric vehicle takes a directional shot of the spatial position of each first sub-road feature to collect the current image of the road surface, and determines the corresponding second sub-road feature based on the recognition of the current image of the road surface. Based on multiple first sub-road features, multiple second sub-road features and road surface mapping relationships, multiple road surface features are determined, which is compatible with the overall consideration of multiple first sub-road features, multiple second sub-road features and road surface mapping relationships, and ensures the accuracy of multiple road surface features.

[0044] At this time, the camera configured on the electric vehicle performs directional shooting based on the spatial position information of the first sub-road feature determined in step S122, which means that the camera will adjust its viewing angle and focal length to ensure that the area where each feature is located can be clearly captured.

[0045] After the camera completes shooting, it will collect a series of current road surface images containing the first sub-road surface features. These images provide visual information of the road surface features, including color, texture, shape, etc.; through image processing algorithms, the system will analyze and identify the collected road surface images. This process involves edge detection, texture analysis, shape matching and other technologies to extract the second sub-road surface features in the image. These second sub-road surface features correspond to the first sub-road surface features, but provide richer visual information.

[0046] The system will associate and integrate the first sub-road surface features with the second sub-road surface features based on a predefined road surface mapping relationship; the road surface mapping relationship is a database or model that defines the association rules and priorities between different road surface features. Through this process, the system determines the final set of road surface features, which will be used for subsequent tasks such as path planning, navigation, and obstacle avoidance.

[0047] Specifically, assume that an electric vehicle is driving on the road and has determined several first sub-road features in front through step S122, including a pothole, a slippery area and a lane line; the camera on the electric vehicle adjusts its viewing angle and focal length according to the spatial position information of these features, and takes directional shots of the pothole, slippery area and lane line respectively; the camera collects a clear road image containing potholes, slippery areas and lane lines; potholes appear as dark concave areas in the image, slippery areas appear reflective and lighter in color, and lane lines are clear white lines. The system identifies the second sub-road features in the image; for potholes, the algorithm detects their shape, depth and edges; for slippery areas, the algorithm analyzes their reflective properties and water accumulation; for lane lines, the algorithm extracts their position, direction and continuity.

[0048] The system associates these second-sub-road features with the first-sub-road features based on predefined road surface mapping relationships. For example, the depth and shape of a pothole are combined with the three-dimensional coordinate information detected by the radar to form a more complete description of the pothole's characteristics. The reflective characteristics and water accumulation of the slippery area are combined with the location of the slippery area detected by the radar to confirm the existence and scope of the slippery area. The position and direction of the lane line are mutually verified with the lane line information detected by the radar, improving the accuracy of lane recognition. Through these steps, electric vehicles can more comprehensively perceive and understand the characteristic information of the road ahead, providing more reliable and rich data support for subsequent driving decisions.

[0049] In one embodiment of the present application, a road surface feature matching table is collected, and the road surface feature matching table is shown in Table 2: Table 2 Road surface feature matching table First sub-road feature ID First sub-road feature description Second sub-road feature ID Second sub-road feature description Road surface characteristics 1 potholes A Dark depression Dark sunken potholes 2 bulge B High-gloss reflective High-gloss reflective bumps 3 Lane lines C Solid white line Solid white lane markings When the electric vehicle's camera captures a road image containing a first sub-road feature (such as a pothole), the system will identify the corresponding second sub-road feature (such as a dark depression); then, by querying the road feature matching table, the system finds the comprehensive description associated with these features (such as a dark depression and pothole).

[0050] refer to Figure 4 , in step S13, determining abnormal road features according to the shapes and positions of the plurality of road features, and marking the specific conditions of the road; In the specific implementation process of the present invention, the specific steps are: S131: Real-time detection of multiple road features, and collecting the location of multiple road features, based on the location of the multiple road features and determine the regional division of the multiple road areas; S132: Performing morphological recognition on the plurality of road features, determining the morphologies of the plurality of road features based on the morphological recognition, marking the morphologies of the plurality of road features on corresponding road surface areas; and determining a first road anomaly parameter based on the morphologies of the plurality of road features and the area of ​​the road surface area. S133: Determine a second road anomaly parameter based on the morphology of the plurality of road features and the regional location of the road area, determine a road anomaly feature based on a mapping relationship between the first road anomaly parameter, the second road anomaly parameter, and the anomaly feature, and determine a specific condition of the road based on each road anomaly feature and the driving state of the electric vehicle.

[0051] In an embodiment of the present application, multiple road surface features are detected in real time, and the positions of the multiple road surface features are collected. Multiple road surface areas are determined based on the regional division of the positions of the multiple road surface features, which is compatible with the overall consideration of the regional division of the positions of the multiple road surface features and ensures the accuracy of the multiple road surface areas.

[0052] At this time, multiple road features are detected in real time. The system uses sensors such as cameras and millimeter-wave radars installed on electric vehicles to capture road information in real time. These sensors can identify a variety of road features, such as potholes, cracks, water accumulation, lane lines, traffic signs, etc.; the camera uses image processing algorithms to identify visual features on the road surface; the millimeter-wave radar uses the reflection of electromagnetic waves to detect physical changes on the road surface.

[0053] Once road features are identified, the system records their specific locations on the road surface, which typically includes information such as the feature's coordinates (such as longitude and latitude or position relative to the vehicle), direction, and size. Through sensor fusion technology, the system can accurately locate each road feature. For example, the camera uses an image matching algorithm to determine the location of the feature in the image, while the millimeter-wave radar calculates the distance to the feature by measuring the time difference of the reflected wave.

[0054] After collecting the location information of road features, the system will divide the road surface into multiple areas based on the spatial distribution and density of this information. These areas are divided based on the density, type, size or direction of the road features. At the same time, area division usually involves clustering algorithms or spatial segmentation techniques. For example, the K-means clustering algorithm is used to divide the road features into several clusters based on their location information, and each cluster represents a road surface area. Alternatively, the grid division technology is used to divide the road surface into a series of regular grid cells, each of which contains a certain number of road surface features.

[0055] Specifically, suppose an electric car is driving on the road, and the system detects the following road features in real time: there are two potholes on the left lane, located at (x1, y1) and (x2, y2) respectively; there is a crack on the central dividing strip, extending from (x3, y3) to (x4, y4); there is a water area on the right lane, with its center point at (x5, y5) and a certain area.

[0056] After collecting the location information of these features, the system divided the areas according to their spatial distribution: the two potholes on the left lane were divided into a "pothole area"; the cracks on the central dividing strip were divided into a "crack area"; and the water accumulation area on the right lane was divided into a "water accumulation area"; ultimately, the system obtained three clear road surface areas, each of which contains specific road surface features. These areas will be used for subsequent road condition assessment and analysis, such as calculating the abnormal parameters of each area and determining the abnormal road characteristics; through such area division, the system can simplify complex road surface information into a form that is easier to manage and analyze.

[0057] Furthermore, morphological recognition is performed on multiple road features, and the morphologies of the multiple road features are determined based on the morphological recognition of the multiple road features, and the morphologies of the multiple road features are marked on the corresponding road areas; based on the morphologies of the multiple road features and the area of ​​the road area, a first road anomaly parameter is determined, which is compatible with the overall consideration of the morphologies of the multiple road features and the area of ​​the road area, thereby ensuring the accuracy of the first road anomaly parameter.

[0058] At this time, the system uses image processing, machine learning or deep learning technologies to perform a detailed morphological analysis of the detected road features; morphological recognition includes information on the size (such as length, width, depth), shape (such as round, linear, irregular), texture (such as rough, smooth), color (such as dark, light), etc. of the features; optionally, for pothole features, the system calculates the actual size of the pothole by measuring the pixel size of the pothole in the image and combining the focal length and shooting angle of the camera; for crack features, the system uses edge detection algorithms to identify the boundaries of the cracks and calculate the length and width of the cracks.

[0059] Based on the results of morphological recognition, the system determines the specific morphology of each road surface feature. This usually involves matching the identified morphological information with a preset morphological template or database to determine the most suitable morphological type. In this case, the system has a database containing a variety of road surface feature morphologies, such as potholes and cracks of different shapes and sizes. By comparing the identified morphology with the morphology in the database, the system determines the morphological type of each feature.

[0060] The system will mark the identified road surface features on the previously divided road surface areas. This usually involves adding corresponding morphological information for each road surface area on the electronic map or in the data structure within the system. At the same time, the system uses different colors or icons on the electronic map to represent different types of road surface features, such as red circles for potholes and green lines for cracks.

[0061] The system combines the morphology of the road features and the area of ​​the road area to calculate the first road anomaly parameters, which include the severity of the road damage, the proportion of the damaged area, the number of specific types of damage, etc. At this time, the system calculates the total area of ​​potholes in each road area and compares it with the total area of ​​the road area to obtain the proportion of pothole areas; alternatively, the system calculates the total length of cracks to assess the impact of cracks on the integrity of the road structure.

[0062] Specifically, assume that the system detects the following road features on an urban road: there are three potholes of different sizes and depths on the left lane; there is a long crack of moderate width on the central dividing strip; and there is a large area of ​​water at the edge of the right lane.

[0063] The system performs morphological recognition on these features, determines their morphological types (such as potholes, cracks, and water accumulation), and marks these features on the corresponding road surface areas. The system then calculates the following first road anomaly parameters: The proportion of pothole area in the left lane is calculated by measuring the area of ​​each pothole, summing the area, and dividing it by the total area of ​​the left lane. Severity of median cracks: Assess the severity of cracks based on their length and width, as well as their impact on the road structure; Area of ​​waterlogged areas in the right lane: Directly measures the area of ​​waterlogged areas to assess their potential impact on traffic safety. Through these parameters, the system can more accurately understand the actual conditions of the road and provide valuable information for road maintenance and safe driving.

[0064] Therefore, the second road anomaly parameter is determined based on the morphology of multiple road features and the regional position of the road area, and the road anomaly feature is determined based on the mapping relationship between the first road anomaly parameter, the second road anomaly parameter and the anomaly feature; the specific condition of the road is determined based on each road anomaly feature and the driving state of the electric vehicle, which is compatible with the overall consideration of the first road anomaly parameter, the second road anomaly parameter and the anomaly feature mapping relationship, ensuring the accuracy of the road anomaly feature. At the same time, it is compatible with the overall consideration of the morphology of multiple road features and the position of multiple road features, ensuring the accuracy of the detection of the road anomaly feature, and further ensuring the accuracy of the detection of the specific condition of the road.

[0065] At this point, the system combines the morphology of road surface features (such as the depth of potholes and the width of cracks) and the location of road surface areas (such as lane location and near intersections) to determine the second road anomaly parameters. These parameters include the distribution of specific types of damage on lanes, the impact of damage locations on traffic flow, etc. At this point, the system analyzes the distribution of potholes on lanes to determine which lanes are more prone to pothole damage; or, the system evaluates whether cracks are located on the vehicle's driving trajectory and the impact of these cracks on vehicle handling.

[0066] The system uses the first road anomaly parameter (such as the pothole area ratio and crack severity) and the second road anomaly parameter (such as damage distribution and traffic flow impact), combined with a preset anomaly feature mapping relationship (such as severe damage if the damage area exceeds a certain threshold), to determine the specific road anomaly characteristics; at this time, the system sets a threshold, and when the pothole area ratio exceeds the threshold, the area is marked as a "severe pothole area"; at the same time, the system combines the severity and location information of the cracks to mark certain cracks as "high-risk cracks."

[0067] The system comprehensively considers abnormal road characteristics and the driving status of the electric vehicle (such as speed, acceleration, steering, etc.) to evaluate the specific impact of the road on the driving of the electric vehicle, which includes information on road safety, driving comfort, maintenance needs, etc.; at the same time, the system predicts the road surface abnormalities that the vehicle will encounter in the next few seconds based on the current speed and steering information of the electric vehicle; if it predicts that the vehicle will enter a severely pothole area, the system will issue a warning in advance and advise the driver to slow down or take a detour.

[0068] Specifically, assume that the system detects the following road features on the road: there are multiple potholes of varying depths on the left lane; there is a long crack on the central dividing strip, and it is close to a busy intersection; there is a large area of ​​water on the edge of the right lane, and this lane is the main driving lane for electric vehicles.

[0069] The system first determined the primary road anomaly parameter based on the depth and distribution of potholes (e.g., the proportion of potholes in the left lane). It then determined the secondary road anomaly parameter based on the location and severity of the cracks (e.g., a crack located near a busy intersection, significantly impacting traffic flow). The system then mapped these parameters to pre-defined anomaly characteristics to identify the following road anomaly features: The left lane is marked as a "severe pothole area"; Cracks in the median strip are marked as "high-risk cracks," especially those near intersections; The area of ​​water in the right lane is marked as a "potential safety hazard"; Finally, based on the electric vehicle's driving status (such as current speed, acceleration, steering intention, etc.), the system predicts road anomalies that the electric vehicle may encounter in the next few seconds and issues a warning to the driver in advance. For example, when the electric vehicle is about to enter a severely potholed area in the left lane, the system suggests the driver to slow down or take a detour through sound or visual prompts. Through this evaluation process, the system can provide electric vehicles with real-time road condition information, support drivers in making safe driving decisions, and also provide valuable data support for road maintenance departments.

[0070] refer to Figure 5 In step S14, the distance of the electric vehicle from the abnormal road feature is determined based on the position of the electric vehicle and the position of the abnormal road feature, and the driving risk coefficient of the electric vehicle is determined according to the distance of the electric vehicle from the abnormal road feature, the specific conditions of the road, and the driving state of the electric vehicle; In the specific implementation process of the present invention, the specific steps are: S141: Real-time monitoring of road anomaly features, and collecting the position of the electric vehicle and the position of the road anomaly features, constructing a position change map of the electric vehicle relative to the road anomaly features based on the position of the electric vehicle and the position of the road anomaly features, and marking the distance of the electric vehicle relative to the road anomaly features; S142: If the distance of the electric vehicle from the abnormal road feature is less than the estimated safe distance threshold, a risk control logic for the electric vehicle is triggered. In this risk control logic, multiple driving risk factors are determined based on the distance of the electric vehicle from the abnormal road feature and the specific conditions of the road. S143: constructing a driving risk dynamic graph of the electric vehicle according to the plurality of driving risk factors and the driving state of the electric vehicle, and determining a driving risk coefficient of the electric vehicle based on the identification of the driving risk dynamic graph of the electric vehicle.

[0071] In an embodiment of the present application, the road abnormality feature is monitored in real time, and the position of the electric vehicle and the position of the road abnormality feature are collected. A position change map of the electric vehicle relative to the road abnormality feature is constructed based on the position of the electric vehicle and the position of the road abnormality feature, and the distance of the electric vehicle relative to the road abnormality feature is marked. This introduces the method of marking the distance of the electric vehicle relative to the road abnormality feature.

[0072] At this time, the system uses sensor networks, cameras or other monitoring equipment to continuously monitor abnormal features on the road, such as potholes, cracks, water accumulation, obstacles, etc. The location, shape and status information of these abnormal features are collected and updated in real time. At this time, the system captures road images through cameras deployed on the roadside and uses image processing algorithms to identify abnormal features; or through sensors installed on electric vehicles (such as millimeter wave radars) to perceive the surrounding environment and detect abnormal features in real time.

[0073] The system needs to simultaneously collect the real-time location information of electric vehicles and the location information of various abnormal features on the road; the location information of electric vehicles is usually obtained through GPS or other positioning technologies, while the location information of abnormal features is determined through map matching, image processing or sensor data fusion methods; at the same time, the GPS module installed on the electric vehicle sends the location information to the monitoring system in real time; at the same time, the system uses high-definition maps and image processing technology to match the road images captured by the camera with the map data to determine the exact location of the abnormal features.

[0074] Based on the collected location information of electric vehicles and abnormal features, the system constructs a position change map, which shows the relative position relationship between electric vehicles and various abnormal features and is updated over time; the position change map is two-dimensional (such as the point-line relationship on a flat map) and three-dimensional (such as a stereogram with height information added); at this time, the system uses GIS (geographic information system) technology to superimpose the location information of electric vehicles and abnormal features on the map to form a dynamic position change map, which is updated in real time to reflect the relative position changes between electric vehicles and abnormal features.

[0075] On the position change map, the system marks the actual distance between the electric vehicle and each abnormal road feature. This distance information is crucial for assessing the driving risk of electric vehicles. At the same time, the system uses distance calculation algorithms (such as Euclidean distance, Manhattan distance, etc.) to calculate the straight-line distance between the electric vehicle and the abnormal feature based on their coordinates on the map. In addition, the system also considers factors such as the actual direction and curvature of the road to make more accurate corrections to the distance.

[0076] Specifically, suppose an electric vehicle is driving on a city road, and there is an obvious pothole abnormal feature on the road; the system captures the image of the pothole through a roadside camera and uses an image processing algorithm to determine its position and shape; at the same time, the GPS module on the electric vehicle sends location information to the monitoring system in real time; based on this information, the system constructs a position change map, showing the relative position relationship between the electric vehicle and the pothole; as time goes by, the electric vehicle gradually approaches the pothole, and the system updates the position change map in real time and marks the actual distance between the electric vehicle and the pothole; when the distance decreases below the preset safety threshold, the system triggers an early warning mechanism to remind the electric vehicle driver to avoid the pothole; through such real-time monitoring and position change map construction, the system can provide electric vehicle drivers with timely road anomaly information to help them make safe driving decisions.

[0077] Furthermore, if the distance of the electric vehicle relative to the abnormal road feature is less than the expected safety distance threshold, the risk management logic of the electric vehicle is triggered. In this risk management logic, multiple driving risk factors are determined based on the distance of the electric vehicle relative to the abnormal road feature and the specific conditions of the road, which is compatible with the overall consideration of the distance of the electric vehicle relative to the abnormal road feature and the specific conditions of the road, ensuring the accuracy of multiple driving risk factors.

[0078] At this time, the system continuously monitors the relative distance between the electric vehicle and the abnormal road features, and compares it with the preset safety distance threshold; the safety distance threshold is usually determined based on factors such as road type, vehicle speed, and abnormal feature type; at this time, the system uses sensor data (such as millimeter wave radar) or GPS positioning information to calculate the distance between the electric vehicle and the abnormal feature in real time; then, this distance is compared with the preset safety distance threshold to determine whether the risk management logic is triggered.

[0079] When the relative distance between an electric vehicle and an abnormal road feature is less than the safety distance threshold, the system triggers the risk management logic, which includes issuing a warning signal, adjusting the electric vehicle's driving parameters (such as deceleration, steering, etc.), and activating the emergency braking system. At this time, the system sends a warning signal to the electric vehicle driver through the on-board display screen, sound prompts, etc. At the same time, the driving parameters are adjusted by adjusting the electric vehicle's motor controller, braking system and other components. In some cases, the system also automatically activates the emergency braking system to avoid or reduce the occurrence of traffic accidents.

[0080] In the risk management logic, the system not only considers the relative distance between the electric vehicle and the abnormal features, but also combines the specific conditions of the road (such as road surface material, slope, curves, etc.) to comprehensively determine multiple driving risk factors. These factors include collision risk, rollover risk, loss of control risk, etc.; at this time, based on historical data and real-time information, the driving risk factors are predicted and evaluated; at the same time, the system also combines external factors such as road conditions and traffic flow to dynamically adjust and optimize risk factors.

[0081] Specifically, suppose an electric vehicle is driving on a road and there is an obvious abnormal curve feature in front of it; the system calculates the distance between the electric vehicle and the curve in real time through sensor data, and compares it with the preset safety distance threshold; when the distance is less than the threshold, the system triggers the risk control logic.

[0082] In the risk management logic, the system first sends a warning signal to the electric vehicle driver, reminding him that there is a curve ahead and he needs to slow down. At the same time, the system comprehensively determines the driving risk factors based on the specific conditions of the curve (such as the curve radius, road material, etc.). For example, if the curve radius is small and the road surface is slippery, the system believes that there is a high risk of rollover and loss of control. Based on these risk factors, the system further adjusts the driving parameters of the electric vehicle; for example, by reducing the motor output power and reducing the vehicle speed; or by adjusting the steering system to make the electric vehicle maintain a more stable driving posture in the curve. Through these measures, the system aims to reduce the risks of electric vehicles during driving and ensure the safety of drivers and passengers.

[0083] Therefore, a driving risk dynamic graph of an electric vehicle is constructed based on multiple driving risk factors and the driving status of the electric vehicle, and the driving risk coefficient of the electric vehicle is determined based on the identification of the driving risk dynamic graph of the electric vehicle, which is compatible with the overall consideration of the identification of the driving risk dynamic graph of the electric vehicle and ensures the accuracy of the driving risk coefficient of the electric vehicle.

[0084] At this time, the system constructs a dynamic driving risk graph based on multiple driving risk factors (such as collision risk, rollover risk, loss of control risk, etc.) and the real-time driving status of the electric vehicle (such as speed, acceleration, steering angle, etc.). This graph intuitively shows the risk change trend of the electric vehicle during driving; at this time, the system uses data visualization technology, such as line charts, bar charts, heat maps, etc., to present the driving risk factors and driving status information in a graphical manner; at the same time, the system also uses dynamic updates to reflect the risk changes during the driving of the electric vehicle in real time.

[0085] The system identifies and analyzes the constructed driving risk dynamic map to extract key risk information. This process involves advanced algorithms such as image recognition and pattern recognition. At this time, the system uses machine learning algorithms or deep learning models to automatically identify and classify the driving risk dynamic map. By training these algorithms or models, the system can identify different risk patterns and quantify and evaluate them.

[0086] Based on the recognition results of the driving risk dynamic graph, the system determines a quantitative driving risk coefficient. This coefficient is a comprehensive indicator that reflects the overall risk level of the electric vehicle during driving. At this time, the system integrates multiple driving risk factors and driving status information into a single driving risk coefficient. This coefficient is dynamically adjusted and optimized according to actual conditions to more accurately reflect the driving risk of the electric vehicle.

[0087] Specifically, suppose an electric vehicle is driving on a road and there is an emergency braking area (such as a construction area) ahead; the system constructs a dynamic driving risk map based on multiple driving risk factors (such as collision risk, loss of control risk, etc.) and the real-time driving status of the electric vehicle (such as speed, acceleration, etc.).

[0088] In the dynamic driving risk graph, the system displays the risk change trend of electric vehicles during driving in the form of a line graph; when the electric vehicle approaches the emergency braking area, the collision risk and loss of control risk gradually increase, resulting in an increase in the driving risk coefficient; the system identifies this dynamic graph and discovers the upward trend of the risk coefficient, and accordingly issues a warning signal to the electric vehicle driver; specifically, when the driving risk coefficient reaches a preset threshold, the system issues an emergency braking warning to the driver through the on-board display screen, sound prompts, etc.; at the same time, the system also automatically adjusts the driving parameters of the electric vehicle (such as reducing the motor output power, starting the emergency braking system, etc.) to reduce driving risks; in this way, step S143 provides electric vehicle drivers with an intuitive driving risk assessment and early warning mechanism, which helps them make safer driving decisions.

[0089] In one embodiment of the present application, assume that an electric vehicle is traveling on a mountain road, and the system collects the following driving risk factor data in real time: collision risk: medium (score of 6); rollover risk: high (score of 7); loss of control risk: low (score of 2); the corresponding weights are introduced to calculate the driving risk coefficient as follows: driving risk coefficient = (6 × 0.4) + (7 × 0.3) + (2 × 0.2) = 2.4 + 2.1 + 0.4 = 4.9; the system finally outputs a driving risk coefficient of 4.9, indicating that there is a certain risk in the current driving state; the driver determines whether to slow down, avoid obstacles or take other safety measures based on this coefficient.

[0090] refer to Figure 6 In step S15, the safe driving logic of the electric vehicle is matched according to the driving risk coefficient of the electric vehicle and the safe driving mapping relationship of the electric vehicle to trigger the emergency avoidance of the electric vehicle relative to the abnormal road features; In the specific implementation process of the present invention, the specific steps are: S151: collecting safe driving mapping relationships for electric vehicles, determining multiple safe driving items for the electric vehicle based on the driving risk coefficient of the electric vehicle and the safe driving mapping relationships for the electric vehicle, and constructing safe driving logic for the electric vehicle based on matching the multiple safe driving items with the driving position of the electric vehicle; S152: In the safe driving logic of the electric vehicle, the electric vehicle gradually passes through corresponding driving positions during driving, and a driving assist signal of the electric vehicle is determined according to the driving parameters of the electric vehicle and the corresponding safe driving items; S153: Determine the assisted driving content of the electric vehicle based on the analysis of the driving assist signal, and determine the emergency avoidance path of the electric vehicle relative to the road abnormality according to the assisted driving content and the corresponding road abnormality feature. The electric vehicle avoids the road abnormality along the emergency avoidance path.

[0091] In an embodiment of the present application, the safe driving mapping relationship of the electric vehicle is collected, and multiple safe driving items of the electric vehicle are determined based on the driving risk coefficient of the electric vehicle and the safe driving mapping relationship of the electric vehicle. The safe driving logic of the electric vehicle is constructed according to the matching of multiple safe driving items and the driving position of the electric vehicle, which is compatible with the overall consideration of the driving risk coefficient of the electric vehicle and the safe driving mapping relationship of the electric vehicle, and ensures the accuracy of multiple safe driving items of the electric vehicle.

[0092] At this time, the system first needs to collect safe driving data of electric vehicles under different road conditions, different driving speeds, different steering angles, etc. These data include key indicators such as the braking distance, stability, and handling of electric vehicles, as well as the correspondence between these indicators and abnormal road features (such as potholes, bends, slippery roads, etc.); at this time, the system collects data in real time through on-board sensors (such as millimeter-wave radar, cameras, acceleration sensors, etc.) and extracts key information through data analysis algorithms; at the same time, the system also uses historical data and expert experience to build a safe driving mapping relationship database for electric vehicles.

[0093] After obtaining the driving risk coefficient of the electric vehicle (such as calculated through step S143) and the safe driving mapping relationship, the system needs to determine multiple safe driving items based on this information. These items include deceleration, steering, emergency braking, lane keeping, etc., aiming to provide appropriate driving suggestions for the electric vehicle based on the current driving risk and road conditions; at this time, the system uses an algorithm to match the driving risk coefficient with the safe driving mapping relationship, and determines the corresponding safe driving items based on the matching results.

[0094] After determining multiple safe driving items, the system needs to match these items with the driving position of the electric vehicle and construct a safe driving logic diagram. This diagram shows the safe driving measures that should be taken at different driving positions and provides real-time driving guidance for the electric vehicle. At this time, the system uses map data, GPS positioning information, etc. to match the safe driving items with the current position of the electric vehicle. At the same time, the system also dynamically adjusts and optimizes the safe driving logic based on external factors such as road conditions and traffic flow.

[0095] Specifically, suppose an electric car is driving on a mountain road with an obvious curve ahead; the system first collects the electric car's safe driving data under similar curve conditions through on-board sensors, including key indicators such as braking distance and stability; then, based on the current driving risk factor (assuming it is medium risk) and the safe driving mapping relationship, the system determines the safe driving item of "slowing down and turning slightly"; finally, the system matches this item with the electric car's current position (about to enter the curve) and constructs the corresponding safe driving logic.

[0096] During driving, the system sends a "slow down and turn slightly" prompt message to the driver through the on-board display screen, and reminds the driver through sound or vibration. At the same time, the system also assists the driver in implementing this safe driving measure by adjusting the electric vehicle's motor output power, braking system and other components. In this way, the system ensures the safety and stability of electric vehicles when driving on curves.

[0097] Furthermore, in the safe driving logic of electric vehicles, electric vehicles gradually pass through corresponding driving positions during driving, and determine the driving assistance signal of the electric vehicle based on the driving parameters of the electric vehicle and the corresponding safe driving items, which is compatible with the overall consideration of the driving parameters of the electric vehicle and the corresponding safe driving items, and ensures the accuracy of the driving assistance signal of the electric vehicle.

[0098] At this time, the system first uses GPS positioning technology, on-board sensors and other means to monitor the location information of the electric vehicle in real time during driving. This information includes key parameters such as the longitude, latitude, speed, and direction of the electric vehicle, which are used to determine whether the electric vehicle is about to enter or has passed the preset driving position; at this time, the system receives satellite signals through the GPS receiver integrated in the electric vehicle and calculates the real-time position of the electric vehicle; at the same time, the system also uses millimeter-wave radar, cameras and other sensors to monitor the environment around the electric vehicle, providing additional data support for position monitoring.

[0099] After determining the real-time location of the electric vehicle, the system needs to match the electric vehicle's driving parameters (such as speed, acceleration, steering angle, etc.) with preset safe driving items. This step aims to determine appropriate safe driving recommendations for the electric vehicle based on current driving conditions and road characteristics; at this time, the system uses algorithms to compare the electric vehicle's driving parameters with the data in the safe driving item database to find the most matching safe driving item. These algorithms are based on rule matching, machine learning and other technologies, and can make intelligent decisions based on the real-time status of the electric vehicle and road conditions.

[0100] After matching the appropriate safe driving items, the system needs to determine the corresponding driving assistance signals. These signals include sound prompts, visual warnings, vibration feedback, etc., which are aimed at conveying safe driving suggestions to electric vehicle drivers and guiding them to take correct driving operations; at this time, the system issues a sound prompt through the on-board audio, or displays a visual warning message through the on-board display; at the same time, the system also uses seat vibrators, steering wheel vibrators and other devices to provide vibration feedback to enhance the effect of the driving assistance signal.

[0101] Specifically, suppose an electric vehicle is traveling on a highway, and there is an emergency braking area (such as a construction area) not far ahead; the system first monitors the position of the electric vehicle in real time through GPS positioning technology, and finds that the electric vehicle is about to enter the emergency braking area; then, the system matches the electric vehicle's real-time speed, acceleration and other driving parameters with preset safe driving items, and determines that "emergency braking" is the most appropriate safe driving recommendation at the moment; finally, the system issues an "emergency braking" sound prompt through the on-board audio, and displays the corresponding warning information through the on-board display screen, guiding the electric vehicle driver to take emergency braking measures; in such a scenario, step S152 ensures that the electric vehicle driver can receive the driving assistance signal in a timely and accurate manner, so as to make safe driving decisions.

[0102] Therefore, the assisted driving content of the electric vehicle is determined based on the analysis of the driving assistance signal, and the emergency avoidance path of the electric vehicle relative to the abnormal road characteristics is determined according to the assisted driving content and the corresponding abnormal road characteristics. The electric vehicle avoids the abnormal road characteristics along the emergency avoidance path, which is compatible with the overall consideration of the assisted driving content and the corresponding abnormal road characteristics, ensures the accuracy of the emergency avoidance path of the electric vehicle relative to the abnormal road characteristics, introduces the driving risk coefficient of the electric vehicle, and ensures the accuracy of the safe driving logic of the electric vehicle, realizes the emergency avoidance of the electric vehicle relative to the abnormal road characteristics, and ensures the safe driving effect of the electric vehicle in various harsh environment scenarios.

[0103] At this time, the system first receives and parses the driving assistance signals from step S152. These signals contain a variety of information, such as emergency braking, slight steering, deceleration, etc., which are intended to guide the electric vehicle to avoid potential dangers; at this time, the system parses the codes or protocols in the signals and converts them into specific instructions or operation suggestions. These instructions or suggestions will be directly used to guide the driving behavior of the electric vehicle.

[0104] After parsing the driving assistance signal, the system needs to further determine the specific assisted driving content, which includes the specific operations that the electric vehicle needs to perform, such as emergency braking, steering avoidance, deceleration, etc.; at this time, the system will match the parsed signal with the current state of the electric vehicle to determine the most appropriate assisted driving content.

[0105] After determining the assisted driving content, the system needs to plan an emergency avoidance path based on the specific location of the abnormal road feature and the current position of the electric vehicle. This path should be able to guide the electric vehicle to safely avoid the abnormal road feature while maintaining smooth and continuous driving. At this time, the system uses map data, GPS positioning information and real-time detection data of abnormal road features to generate an emergency avoidance path through a path planning algorithm. The emergency avoidance path takes into account multiple factors, such as road width, steering angle restrictions, and the dynamic performance of the electric vehicle.

[0106] After planning the emergency avoidance path, the system needs to control the electric vehicle to drive along this path to avoid abnormal road features. This step involves multiple subsystems of the electric vehicle, such as the power system, braking system, steering system, etc.; at this time, the system sends control instructions to the various subsystems of the electric vehicle to implement the emergency avoidance operation. These instructions include adjusting the motor output power, activating the braking system, adjusting the steering angle, etc.

[0107] Specifically, suppose an electric vehicle is driving on a road and a large pothole suddenly appears in front of it; the system first receives a driving assistance signal from step S152, prompting that emergency avoidance is required; then, the system analyzes this signal and determines "emergency steering avoidance" as the auxiliary driving content; then, the system combines the specific location of the pothole and the current position of the electric vehicle to plan an emergency avoidance path, which guides the electric vehicle to turn slightly to the left and continue driving around the pothole; finally, the system controls the electric vehicle to drive along this emergency avoidance path and successfully avoids the pothole; in such a scenario, step S153 ensures that the electric vehicle can make avoidance operations quickly and accurately, thereby avoiding potential safety risks.

[0108] In another embodiment of the present application, it is assumed that the electric vehicle receives a "steering avoidance" driving assistance signal during driving; after the system analyzes the signal, it searches the matching table and finds that the corresponding assisted driving content is "turn left", and the abnormal road feature is "potholes on the road"; then, the system plans an emergency avoidance path that turns slightly to the left based on the specific location of the pothole and the current position of the electric vehicle; the electric vehicle drives along this path and successfully avoids the pothole.

[0109] In addition, onboard millimeter-wave radars installed around the electric vehicle monitor the road around it. Multiple millimeter-wave radars are installed on electric vehicles to enable autonomous driving. Electromagnetic waves emitted by the millimeter-wave radar reflect off the ground, forming an echo signal that is received by the radar. The received signal is processed by the radar and then processed to identify one or more stationary targets. Based on this or these stationary targets, the millimeter-wave radar can detect the presence of ground ahead.

[0110] The method for millimeter wave radar to determine whether the target ahead is a stationary target is as follows: In a typical scenario, the vehicle is moving forward at a speed V. For a point on the road, it is a stationary point with an absolute speed of 0. The angle between the vehicle and the vertical axis is θ. The true relative speed Vground is:

[0111] The relative speed of all targets measured by the radar is V dop Assume that the dynamic and static separation threshold is V th ,

[0112] Obtain all stationary point information. Divide the radar field of view (FOV) into angle units with each degree as the interval, and initialize the stationary object distance of each angle unit to the radar's maximum detection distance Rmax. Traverse each angle unit in the FOV, and for the i-th angle range, select the stationary point closest to the vehicle in the current angle space as the stationary point cloud used for fitting, update the stationary object distance in the current angle unit θi to Ri, and count n plus 1. If there is no stationary point cloud that meets the conditions in the current angle space, the boundary of the stationary point in the current angle space is the measured maximum distance Rmax. For the selected stationary point, calculate:

[0113] If there are no stationary points within a certain range in front of the electric vehicle, the vehicle determines that there is no road ahead. This range can be determined based on extensive test data or empirical experience. The radar can also accumulate stationary point measurement results from multiple frames to improve judgment accuracy.

[0114] Furthermore, if the millimeter wave radar is a 4D millimeter wave radar, the radar can not only measure the distance, relative speed and horizontal azimuth of the target , and can also measure vertical pitch angle , then it is expanded to the three-dimensional coordinates of the measurement target:

[0115]

[0116] By dividing the height of the stationary object The target point is compared with the radar installation height to determine whether it is the ground, thereby improving the accuracy of ground judgment.

[0117] See also Figure 7 , Figure 7: is a schematic diagram of the structural composition of a millimeter wave radar-based road detection system in an embodiment of the present invention; the millimeter wave radar-based road detection system includes: A scene module 21 is used to determine the scene in which the electric vehicle is located based on the current position of the electric vehicle and the road image captured by the electric vehicle; A road feature module 22 is configured to trigger road surface detection by the millimeter-wave radar of the electric vehicle if the electric vehicle is in a harsh environment, and determine a plurality of road surface features based on the road surface detection by the millimeter-wave radar; A road abnormality feature module 23 is used to determine road abnormality features based on the shapes and positions of multiple road features, and mark the specific conditions of the road; a driving risk coefficient module 24 for determining a distance of the electric vehicle from the abnormal road feature based on the position of the electric vehicle and the position of the abnormal road feature, and determining a driving risk coefficient of the electric vehicle based on the distance of the electric vehicle from the abnormal road feature, the specific conditions of the road, and the driving state of the electric vehicle; The emergency avoidance module 25 is used to match the safe driving logic of the electric vehicle according to the driving risk coefficient of the electric vehicle and the safe driving mapping relationship of the electric vehicle, so as to trigger the emergency avoidance of the electric vehicle relative to abnormal road features.

[0118] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all 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.

Claims

1. A road detection method based on millimeter wave radar, characterized in that: include: Determining a scene in which the electric vehicle is located based on a current position of the electric vehicle and a road image captured by the electric vehicle; If the electric vehicle is in a harsh environment, the electric vehicle's millimeter-wave radar road surface detection is triggered, and multiple road surface features are determined based on the millimeter-wave radar road surface detection; determining abnormal road features according to the shapes and positions of the plurality of road features, and marking specific conditions of the road; determining a distance of the electric vehicle from the abnormal road feature based on the location of the electric vehicle and the location of the abnormal road feature, and determining a driving risk coefficient of the electric vehicle based on the distance of the electric vehicle from the abnormal road feature, specific conditions of the road, and a driving state of the electric vehicle; The safe driving logic of the electric vehicle is matched according to the driving risk coefficient of the electric vehicle and the safe driving mapping relationship of the electric vehicle to trigger the emergency avoidance of the electric vehicle relative to abnormal road features.

2. The road detection method based on millimeter wave radar according to claim 1, characterized in that: The determining of the scene in which the electric vehicle is located based on the current position of the electric vehicle and the road image captured by the electric vehicle includes: Collecting the current location of the electric vehicle, determining the urban area where the electric vehicle is located based on the current location of the electric vehicle and a town distribution map, and determining a first scenario coefficient based on the urban area where the electric vehicle is located and the model of the electric vehicle; The electric vehicle is equipped with corresponding cameras, which are distributed around the electric vehicle and collect images of the electric vehicle's surroundings. The corresponding road area is determined based on the detection of the images of the electric vehicle's surroundings, and the second scene coefficient is determined based on the environmental characteristics presented by the road area. The scene mapping relationship corresponding to the electric vehicle is collected, and the scene in which the electric vehicle is located is determined based on the first scene feature, the second scene feature and the scene mapping relationship. The scenes in which the electric vehicle is located include standard environment scenes and harsh environment scenes.

3. The road detection method based on millimeter wave radar according to claim 1, characterized in that: If the electric vehicle is in a harsh environment, the millimeter-wave radar of the electric vehicle is triggered to detect the road surface, and multiple road surface features are determined based on the road surface detection of the millimeter-wave radar, including: When the electric vehicle is in a harsh environment, the corresponding harsh environmental factors are determined based on the detection of the electric vehicle, and the detection mode of the millimeter-wave radar of the electric vehicle is determined according to the harsh environmental factors and the driving state of the electric vehicle; The millimeter-wave radar detects the path along the detection pattern and collects point cloud data of the road surface, determines a plurality of first sub-road surface features based on synthesis of the point cloud data of the road surface, and marks spatial positions of the plurality of first sub-road surface features; The camera equipped with the electric vehicle takes a directional shot of the spatial position of each first sub-road feature to collect the current image of the road surface, and determines the corresponding second sub-road feature based on the recognition of the current image of the road surface, and determines multiple road features based on multiple first sub-road features, multiple second sub-road features and road surface mapping relationships.

4. The road detection method based on millimeter wave radar according to claim 1, characterized in that: The determining of abnormal road features based on the shapes and positions of the plurality of road features and marking the specific conditions of the road includes: A plurality of road surface features are detected in real time, and positions of the plurality of road surface features are collected, and a plurality of road surface areas are determined based on area division of the positions of the plurality of road surface features.

5. The road detection method based on millimeter wave radar according to claim 4, characterized in that: The method of determining abnormal road features based on the shapes and positions of the plurality of road features and marking the specific conditions of the road further includes: Performing morphological recognition on a plurality of road features, determining morphologies of the plurality of road features based on the morphological recognition of the plurality of road features, and marking the morphologies of the plurality of road features on corresponding road areas; and determining a first road anomaly parameter based on the morphologies of the plurality of road features and the area of ​​the road area; A second road anomaly parameter is determined based on the morphology of multiple road features and the regional location of the road area, and a road anomaly feature is determined based on a mapping relationship between the first road anomaly parameter, the second road anomaly parameter, and the anomaly feature; and a specific condition of the road is determined based on each road anomaly feature and the driving state of the electric vehicle.

6. The road detection method based on millimeter wave radar according to claim 1, characterized in that: The method of determining the distance of the electric vehicle relative to the abnormal road feature based on the position of the electric vehicle and the position of the abnormal road feature, and determining the driving risk coefficient of the electric vehicle according to the distance of the electric vehicle relative to the abnormal road feature, the specific conditions of the road, and the driving state of the electric vehicle includes: Monitor road anomaly features in real time, collect the position of the electric vehicle and the position of the road anomaly features, build a position change map of the electric vehicle relative to the road anomaly features based on the position of the electric vehicle and the position of the road anomaly features, and mark the distance of the electric vehicle relative to the road anomaly features.

7. The road detection method based on millimeter wave radar according to claim 6, characterized in that: The method further includes determining the distance of the electric vehicle relative to the abnormal road feature based on the position of the electric vehicle and the position of the abnormal road feature, and determining the driving risk coefficient of the electric vehicle according to the distance of the electric vehicle relative to the abnormal road feature, the specific conditions of the road, and the driving state of the electric vehicle. If the distance of the electric vehicle from the road anomaly falls below the estimated safe distance threshold, the electric vehicle risk management logic is triggered. In this risk management logic, multiple driving risk factors are determined based on the distance of the electric vehicle from the road anomaly and the specific conditions of the road. A driving risk dynamic graph of the electric vehicle is constructed according to a plurality of driving risk factors and a driving state of the electric vehicle, and a driving risk coefficient of the electric vehicle is determined based on the identification of the driving risk dynamic graph of the electric vehicle.

8. The road detection method based on millimeter wave radar according to claim 1, characterized in that: The method of matching the safe driving logic of the electric vehicle according to the driving risk coefficient of the electric vehicle and the safe driving mapping relationship of the electric vehicle to trigger the emergency avoidance of the electric vehicle relative to abnormal road features includes: The safe driving mapping relationship of the electric vehicle is collected, multiple safe driving items of the electric vehicle are determined based on the driving risk coefficient of the electric vehicle and the safe driving mapping relationship of the electric vehicle, and the safe driving logic of the electric vehicle is constructed based on the matching of the multiple safe driving items and the driving position of the electric vehicle.

9. The road detection method based on millimeter wave radar according to claim 8, characterized in that: The method of matching the safe driving logic of the electric vehicle according to the driving risk coefficient of the electric vehicle and the safe driving mapping relationship of the electric vehicle to trigger the emergency avoidance of the electric vehicle relative to abnormal road features further includes: In the safe driving logic of an electric vehicle, the electric vehicle gradually passes through corresponding driving positions during driving, and the driving assistance signal of the electric vehicle is determined according to the driving parameters of the electric vehicle and the corresponding safe driving items; The assisted driving content of the electric vehicle is determined based on the analysis of the driving assist signal, and the emergency avoidance path of the electric vehicle relative to the road abnormality is determined according to the assisted driving content and the corresponding road abnormality. The electric vehicle avoids the road abnormality along the emergency avoidance path.

10. A road detection system based on millimeter wave radar, characterized in that: The millimeter-wave radar-based road detection system is applied to the millimeter-wave radar-based road detection method according to any one of claims 1 to 9, and the millimeter-wave radar-based road detection system includes: A scene module, configured to determine the scene in which the electric vehicle is located based on the current position of the electric vehicle and the road image captured by the electric vehicle; A road feature module is used to trigger the road surface detection of the electric vehicle's millimeter-wave radar if the electric vehicle is in a harsh environment, and determine multiple road surface features based on the road surface detection of the millimeter-wave radar; A road abnormality feature module is used to determine the road abnormality feature according to the shape and position of multiple road surface features, and mark the specific condition of the road; a driving risk coefficient module, configured to determine the distance of the electric vehicle from the abnormal road feature based on the position of the electric vehicle and the position of the abnormal road feature, and determine the driving risk coefficient of the electric vehicle according to the distance of the electric vehicle from the abnormal road feature, the specific conditions of the road, and the driving state of the electric vehicle; The emergency avoidance module is used to match the safe driving logic of the electric vehicle according to the driving risk coefficient of the electric vehicle and the safe driving mapping relationship of the electric vehicle, so as to trigger the emergency avoidance of the electric vehicle relative to abnormal road features.

Images