Intelligent prompting method for vehicle meeting in narrow space

Abstract

The invention belongs to the technical field of vehicle control, and particularly relates to a narrow space vehicle meeting intelligent prompting method which comprises the steps of narrow road space vehicle meeting scene recognition and judgment, system prompting and intervention in a vehicle meeting mode and vehicle meeting mode custom setting. According to the technical scheme, in the meeting process, the system can continuously monitor the vehicle state and predict the driving track, prompts are given out or parking is intervened in time according to different danger degrees, safety accidents such as scratching, edge jamming and landslide caused by insufficient pre-estimated safety distance of a driver are effectively avoided, and the safety of the vehicle and personnel is guaranteed. And a driver user-defined prompt distance function is provided, and individual demands of different drivers are fully considered, so that the system better fits the actual driving habits of the drivers, and the practicability and the user experience of the system are improved.

Description

Technical Field

[0001] This invention belongs to the field of vehicle control technology, specifically relating to a method for intelligent oncoming traffic alert in narrow spaces. Background Technology

[0002] In everyday driving scenarios, drivers often encounter narrow road spaces, such as those in residential areas with limited parking, roads where lanes are occupied, and rural roads, where vehicles need to pass each other. Novice drivers, lacking sufficient driving experience, often struggle to accurately estimate safe passing distances. This inaccurate estimation can easily lead to accidents such as scraping against parked vehicles or getting stuck on the roadside, causing damage to the vehicle itself and potentially threatening the safety of the driver and passengers, resulting in significant personal injury and property loss. Therefore, intelligent prompts and interventions for passing vehicles in narrow road spaces can help drivers pass each other more safely.

[0003] Patent CN202310605110.X proposes a method, device, medium, and product for generating and reminding safe passing information, mainly including:

[0004] The system acquires real-time driving information of vehicles traveling on target roads, including accident-prone sections. This real-time driving information includes the vehicle's real-time location and speed. Based on the real-time driving information generated before the vehicle reaches the accident-prone section and the information of the accident-prone section itself, the system predicts the vehicle's trajectory information while traveling on the accident-prone section. Based on the trajectory information of vehicles traveling in different directions on the accident-prone section, the system identifies target vehicles and oncoming traffic alert locations that may occur on the accident-prone section. The system sends oncoming traffic alert information, including the oncoming traffic alert location and alert content, to the target vehicle's client, so that the client can notify the target vehicle of the presence of oncoming traffic on the accident-prone section.

[0005] This solution has the following problems:

[0006] ① The focus is on accident-prone road sections. Although it can provide oncoming traffic alerts for these specific road sections, it lacks consideration for narrow road scenarios where oncoming traffic is still at risk, even though they are not accident-prone road sections.

[0007] ② This method primarily predicts the vehicle's trajectory before it reaches accident-prone sections, determines the location for oncoming traffic alerts, and sends alert messages. This approach focuses more on pre-emptive alerts and lacks effective monitoring and intervention mechanisms for real-time dynamic changes during the oncoming traffic process. For example, during oncoming traffic, vehicle and road conditions may change at any time, such as the sudden appearance of obstacles or sudden changes in the speed of oncoming vehicles; this technology may not be able to respond to these changes in a timely manner.

[0008] Patent CN202510103541.5 discloses a method, equipment, and vehicle for vehicles to pass each other, mainly including:

[0009] ① Based on the acquired vehicle perception information, when the vehicle enters a meeting scene, the obstacle status in the meeting scene is determined based on the vehicle perception information, and the obstacle meeting safety level of the vehicle is also determined.

[0010] ② Determine the oncoming vehicle safety level based on the distance between the oncoming vehicle and the right-hand traffic boundary of the oncoming vehicle.

[0011] ③ Determine the area of ​​the vehicle to be approached based on the oncoming vehicle perception information and the vehicle's perception information.

[0012] ④ Before the vehicle arrives at the waiting area, control the vehicle to drive to and stop at the position in the lane-sharing area to wait for the oncoming vehicle to pass; after the oncoming vehicle completes the passing process, control the vehicle to drive out of the lane-sharing area.

[0013] This solution has the following problems:

[0014] ① It mainly focuses on the recognition of oncoming traffic scenarios, obstacle status, oncoming traffic safety level, and automatic control during oncoming traffic. If the driver does not use this function, there is a lack of prompts and reasonable intervention based on the real-time status when the driver is oncoming traffic.

[0015] ② The vehicle meeting control mainly involves stopping the vehicle in the lane-sharing area before reaching the waiting area to wait for oncoming vehicles. The handling method is relatively simple and does not take into account the personalized needs of drivers due to factors such as driving experience and driving style. This may affect the driver's acceptance of the system and its effectiveness to some extent. Summary of the Invention

[0016] The purpose of this invention is to provide an intelligent prompting method for passing other vehicles in narrow spaces. This method can effectively identify passing scenarios in narrow road spaces and provide a safe and effective intelligent prompting and intervention method to help drivers better judge the safety conditions of passing other vehicles and complete the passing maneuver, thereby reducing the safety risks of passing other vehicles.

[0017] To achieve the above objectives, this application employs the following technical solution:

[0018] A method for intelligent vehicle passing prompts in narrow spaces includes the following steps:

[0019] S1. Data preprocessing: Obtain usable input data;

[0020] S2. Calculate the available input data obtained in step S1 and extract feature values;

[0021] S3. Algorithm Model Decision Model Reasoning: The dynamic compensation term α is obtained by linear combination of the feature values ​​calculated in step S2.

[0022] ,

[0023] Here , , , Set an initial value such as 0.1, 0.05, 0.02, 0.03, and continuously adjust and optimize it based on the training results.

[0024] S4. Perform SHAP feature contribution on the eigenvalues ​​and the inference results. calculate:

[0025] , where: N is the set of all features, S is a subset of features; v(S) is the predicted value of the feature subset S; To combine weights and ensure fairness;

[0026] S5. Dynamic Coefficient Update: The coefficients and offsets of the linear combination of the above algorithms need to be obtained in the early training and recognition of classic scenarios in narrow road spaces. Subsequently, based on the error feedback of actual data, the average value will be extracted, and a linear change will be made to the relationship between it and the actual value to adjust and optimize the dynamic compensation term. linear combination coefficients , , , ;

[0027] S6. System prompts and interventions in passing mode:

[0028] When entering the meeting point, the system first gives a voice prompt to be aware that oncoming traffic is about to pass, and then gives a voice prompt to provide driving advice based on the current speed of the two vehicles to calculate the section of road where they will meet.

[0029] During the continuous detection and alert phase, the vehicle's speed and driving direction angle are continuously monitored. Combined with a pre-set safety algorithm model, the vehicle's driving trajectory is predicted in real time to determine whether the vehicle is within a safe range and to provide alerts in the appropriate circumstances.

[0030] When it is predicted that a serious danger may occur to the vehicle within a set time, the system will automatically enter the emergency intervention and stop phase, and stop the vehicle by controlling the vehicle's braking system to avoid an accident.

[0031] When the passing process is detected to be over, the passing mode is automatically exited, the in-vehicle notification information is canceled accordingly, and the passing record is then pushed to the car owner via the APP.

[0032] Further, in step S1, data preprocessing is performed to obtain usable input data. Specifically, this involves: using multiple sensors to acquire lane width, oncoming vehicle width, and obstacle width; using the CAN bus to acquire the vehicle's speed, steering wheel angular velocity, and yaw rate; processing these data for missing values, removing invalid data, and finally obtaining usable input data: lane width. Width of oncoming vehicles obstacle width speed Steering wheel angular velocity yaw rate .

[0033] Furthermore, in step S2, the available input data obtained in step S1 is calculated to extract feature values, specifically as follows:

[0034] S21. Calculate the static clearance;

[0035] S22. Calculate the space occupied by dynamic obstacles;

[0036] S23. Calculate the effective row width;

[0037] S24. Calculate the shift demand index;

[0038] S25. Calculate the trajectory envelope width;

[0039] S26. Obtain environmental factors.

[0040] Furthermore, step S3 also includes adjusting the dynamic compensation terms. Calculate the dynamic available road width :

[0041] Where K is the safety space, which is preset to 0.5m; when the dynamic available road width Less than the effective passage width When this happens, the current road space is considered to be a narrow road.

[0042] Furthermore, it also includes calculating the probability of risk occurring under dynamic conditions. For dynamic compensation items and current effective passage width Perform linear combinations.

[0043] Furthermore, in step S6, upon entering the meeting point stage, the vehicle's infotainment system first provides a voice prompt warning of an impending meeting oncoming traffic. Subsequently, based on the current speeds of both vehicles, it calculates the meeting point and provides driving suggestions via voice prompts: if there is a safe meeting point, it provides a speed adjustment suggestion based on the vehicle's speed for safe meeting; if there is no safe meeting point, it suggests slowing down or reversing. At the same time, it displays a lane driving scene map for the driver, showing the system's predicted safe meeting route.

[0044] Furthermore, in step S6, during the continuous detection and prompting phase, the vehicle's speed and driving direction angle are continuously monitored. Combined with a pre-set safety algorithm model, the vehicle's driving trajectory is predicted in real time to determine whether the vehicle is within a safe range and to provide prompts under appropriate circumstances. Specifically, when meeting oncoming traffic at a turning point, the vehicle interface should switch to 360-degree panoramic detection and prompt that 360-degree panoramic detection is enabled; when pedestrians are detected around the vehicle, a prompt to be aware of pedestrians should be given; when it is predicted that the vehicle may be in danger within 5 seconds, a voice prompt should be given to provide adjustment suggestions and simultaneously reduce the vehicle to the set speed.

[0045] The beneficial effects of this invention are:

[0046] During oncoming traffic, the system continuously monitors vehicle status and predicts driving trajectory, issuing timely warnings or intervening to stop the vehicle based on different levels of danger. This effectively avoids safety accidents such as scrapes, getting stuck, and landslides caused by insufficient safe distance estimation by the driver, ensuring the safety of vehicles and personnel.

[0047] The system offers a customizable distance prompt function for drivers, fully considering the individual needs of different drivers, making the system more aligned with drivers' actual driving habits, and improving the system's usability and user experience. Attached Figure Description

[0048] Figure 1 This is a logic diagram of the algorithm steps of the present invention.

[0049] Figure 2 This is a block diagram of the narrow space passing system module of the present invention.

[0050] Figure 3 This is a flowchart illustrating the operation of the narrow space passing prompt system of the present invention. Detailed Implementation

[0051] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings. The following embodiments are merely exemplary and can only be used to explain and illustrate the technical solution of the present invention, and should not be construed as limiting the technical solution of the present invention.

[0052] like Figures 1 to 3As shown, this application provides a method for intelligent prompting when vehicles meet in narrow spaces, including the following steps:

[0053] 1. Recognition and determination of oncoming traffic scenarios in narrow road spaces:

[0054] S1. Data Preprocessing: Multiple sensors, including LiDAR, cameras, and positioning antennas, are used to acquire lane width, oncoming vehicle width, and the width of obstacles such as pedestrians. The vehicle's speed, steering wheel angular velocity, and yaw rate are acquired via the CAN bus. Missing values ​​are processed, removing invalid data such as lane widths smaller than the vehicle's own width, or values ​​where a sensor's estimated size of the same target differs significantly from other sensors. Finally, usable input data is obtained: lane width. Width of oncoming vehicles obstacle width speed Steering wheel angular velocity yaw rate .

[0055] S2, Feature values: For the input data, the following calculations are performed:

[0056] Calculate the static gap:

[0057] ;

[0058] Calculate the space occupied by dynamic obstacles:

[0059]

[0060] Calculate the effective passage width:

[0061] ;

[0062] Calculate the shift demand index:

[0063] ;

[0064] Calculate the trajectory envelope width:

[0065] ;

[0066] Environmental factors were identified: Considering that different weather conditions, lighting conditions, and road surface adhesion coefficients can lead to variations in data processing results, parameters were introduced for different environments. Factors and parameters used to represent the impact of weather This represents the road surface adhesion coefficient.

[0067] S3, Algorithm Model Decision Model Reasoning: Based on the calculations obtained in the previous step... , , , and The dynamic compensation term is obtained by performing linear combination calculations. ,Right now:

[0068]

[0069] Here , , , An initial value such as 0.1, 0.05, 0.02, and 0.03 will be set and continuously adjusted and optimized based on the training results.

[0070] According to dynamic compensation items Calculate the dynamic available road width :

[0071] ;

[0072] When the dynamically available road width Less than the effective passage width When this happens, the current road space is considered to be a narrow road.

[0073] In addition, it is necessary to calculate the probability of a risk occurring under dynamic conditions (i.e., whether driving in this manner would be unsafe). For dynamic compensation items and current effective passage width Perform linear combinations:

[0074] ,

[0075] .

[0076] S4. Interpretability Analysis: SHAP feature contribution calculation is performed on the eigenvalues ​​and the results of inference.

[0077] ,

[0078] in:

[0079] N represents the set of all features, and S represents a subset of features.

[0080] v(S) is the predicted value of the feature subset S.

[0081] To combine weights and ensure fairness.

[0082] Identify whether the primary influencing factor is vehicle speed or steering wheel angle, and use this information to provide adjustment suggestions: when the calculated probability... When the speed exceeds 60%, suggestions and prompts should be given; when it exceeds 80%, the speed should be reduced; and when it exceeds 90%, stopping should be considered.

[0083] S5. Dynamic Coefficient Update: The coefficients and offsets of the linear combination of the above algorithms need to be obtained in the early training and recognition of classic scenarios in narrow road spaces. Subsequently, based on the error feedback of actual data, the average value will be extracted, and a linear change will be made to the relationship between it and the actual value to adjust and optimize the dynamic compensation term. linear combination coefficients , , , .

[0084] 2. System prompts and interventions in the passing mode:

[0085] The passing mode will be divided into four stages: entering the passing stage, continuous detection and prompting stage, emergency intervention and stopping stage, and exiting the passing stage.

[0086] S6. When entering the meeting point, the system first gives a voice prompt to be aware that an oncoming vehicle is about to pass. Then, based on the current speeds of the two vehicles, it calculates the section of road where they will meet and gives a voice prompt with driving suggestions: if there is a section of road where the vehicles can pass safely, it gives a speed adjustment suggestion based on the vehicle speed to facilitate safe passing; if there is no safe driving space in the section of road where the vehicles can pass, it suggests slowing down or reversing. At the same time, it opens a real-time map of the lane for the driver and displays the safe passing route predicted by the system.

[0087] During the continuous detection and alert phase, the vehicle's speed and steering angle are continuously monitored. Combined with a pre-set safety algorithm model, the vehicle's trajectory is predicted in real time to determine if the vehicle is within a safe range and provide appropriate alerts. For example, when a vehicle is turning and meeting another vehicle, the in-vehicle interface should switch to 360-degree panoramic detection and indicate that it is enabled, allowing the driver to promptly check blind spot information. When pedestrians such as the elderly or children are detected near the vehicle, a "Caution Pedestrians" alert should be given. When a potential hazard is predicted within 5 seconds (such as scraping against a roadside vehicle or veering off the road), a voice prompt should be given suggesting adjustments and reducing the vehicle speed to 2 m / s. The volume of the alerts adjusts based on the distance between vehicles, between the vehicle and obstacles (including pedestrians), and between the vehicle and the road edge, with the urgency level of the safety alert graded according to distance; for example, the alert level increases with each 10 centimeters closer.

[0088] When it is predicted that a serious danger may occur to the vehicle within 1 second, it automatically enters the emergency intervention and stopping phase, and stops the vehicle by controlling the vehicle's braking system to avoid an accident.

[0089] Once the oncoming traffic is detected to have ended (e.g., the other vehicle has passed safely and the vehicle's driving status has returned to normal), the vehicle automatically exits the oncoming traffic mode, the in-car notification information is canceled accordingly, and then the oncoming traffic record is pushed to the driver via the APP. This record includes the oncoming traffic section, the oncoming traffic status, vehicle movement trajectory information, and a real-time image of the narrow section. The driver can view the oncoming traffic record and provide feedback on the accuracy of the service notification information and the feasibility of the planned route.

[0090] 3. Custom settings for passing mode:

[0091] To cater to the diverse driving habits and needs of different drivers, the system offers a customizable oncoming traffic mode. This includes options such as whether to use the oncoming traffic alert function and, within legally permissible safety limits, to customize the safe distance for each oncoming vehicle. Drivers can configure the distance threshold for entering oncoming traffic mode (e.g., 80 meters, 30 meters), the threshold for confined spaces (e.g., 0.5 meters, 0.3 meters), and the alert duration threshold for different hazardous situations (e.g., 4 seconds, 6 seconds) in the vehicle's settings menu. The system will then provide oncoming traffic alerts and interventions based on the driver's customized parameters.

[0092] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. A method for intelligent vehicle passing prompts in narrow spaces, characterized in that, Includes the following steps: S1. Data preprocessing: Obtain usable input data; S2. Calculate the available input data obtained in step S1 and extract feature values; S3. Algorithm Model Decision Model Reasoning: The dynamic compensation term α is obtained by linear combination of the feature values ​​calculated in step S2. ， Here , , , Set an initial value such as 0.1, 0.05, 0.02, 0.03, and continuously adjust and optimize it based on the training results. S4. Perform SHAP feature contribution on the eigenvalues ​​and the inference results. calculate: , where: N is the set of all features, S is a subset of features; v(S) is the predicted value of the feature subset S; To ensure fairness, the weights are combined. S5. Dynamic Coefficient Update: The coefficients and offsets of the linear combination of the above algorithms need to be obtained in the early training and recognition of classic scenarios in narrow road spaces. Subsequently, based on the error feedback of actual data, the average value will be extracted, and a linear change will be made to the relationship between it and the actual value to adjust and optimize the dynamic compensation term. linear combination coefficients , , , ; S6. System prompts and interventions in passing mode: When entering the meeting point, the system first gives a voice prompt to be aware that oncoming traffic is about to pass, and then gives a voice prompt to provide driving advice based on the current speed of the two vehicles to calculate the section of road where they will meet. During the continuous detection and alert phase, the vehicle's speed and driving direction angle are continuously monitored. Combined with a pre-set safety algorithm model, the vehicle's driving trajectory is predicted in real time to determine whether the vehicle is within a safe range and to provide alerts in the appropriate circumstances. When it is predicted that a serious danger may occur to the vehicle within a set time, the system will automatically enter the emergency intervention and stop phase, and stop the vehicle by controlling the vehicle's braking system to avoid an accident. When the passing process is detected to be over, the passing mode is automatically exited, the in-vehicle notification information is canceled accordingly, and the passing record is then pushed to the car owner via the APP.

2. The intelligent vehicle passing prompt method in narrow spaces according to claim 1, characterized in that, The data preprocessing in step S1, which obtains usable input data, specifically involves: using multiple sensors to obtain lane width, oncoming vehicle width, and obstacle width; and using the CAN bus to obtain the vehicle speed, steering wheel angular velocity, and yaw rate of the vehicle being driven. Missing values ​​were processed from these data, and invalid data was removed to obtain usable input data: lane width. Width of oncoming vehicles obstacle width speed Steering wheel angular velocity yaw rate .

3. The intelligent prompting method for passing vehicles in narrow spaces according to claim 1, characterized in that, In step S2, the available input data obtained in step S1 is used to calculate and extract feature values, specifically as follows: S21. Calculate the static clearance; S22. Calculate the space occupied by dynamic obstacles; S23. Calculate the effective row width; S24. Calculate the shift demand index; S25. Calculate the trajectory envelope width; S26. Obtain environmental factors.

4. The intelligent vehicle passing prompt method in narrow spaces according to claim 1, characterized in that, Step S3 also includes adjusting the dynamic compensation terms. Calculate the dynamic available road width : Where K is the safety space, which is preset to 0.5m; when the dynamic available road width Less than the effective passage width When this happens, the current road space is considered to be a narrow road.

5. The intelligent prompting method for passing vehicles in narrow spaces according to claim 4, characterized in that, It also includes calculating the probability of risk occurring under dynamic conditions. For dynamic compensation items and current effective passage width Perform linear combinations.

6. The intelligent prompting method for passing vehicles in narrow spaces according to claim 1, characterized in that, In step S6, upon entering the meeting point stage, the vehicle's infotainment system first provides a voice prompt warning of an upcoming oncoming vehicle. Then, based on the current speeds of both vehicles, it calculates the meeting point and provides driving suggestions via voice prompts: if there is a safe section of road for passing, it provides suggestions for adjusting speed to ensure safe passing; if there is no safe passage, it suggests slowing down or reversing. At the same time, it displays a lane driving scene map for the driver, showing the system's predicted safe passing route.

7. The intelligent prompting method for passing vehicles in narrow spaces according to claim 1, characterized in that, In step S6, during the continuous detection and prompting phase, the vehicle's speed and driving direction angle are continuously monitored. Combined with a pre-set safety algorithm model, the vehicle's driving trajectory is predicted in real time to determine whether the vehicle is within a safe range and to provide prompts in accordance with the circumstances. Specifically, when meeting oncoming traffic at a turning point, the vehicle interface should switch to 360-degree panoramic detection and prompt that 360-degree panoramic detection is enabled; when pedestrians are detected around the vehicle, a prompt to be aware of pedestrians should be given; when it is predicted that the vehicle may be in danger within 5 seconds, a voice prompt should be given to provide adjustment suggestions and simultaneously reduce the vehicle to the set speed.

Images