A method and system for hierarchical reminding of low-speed driving on a fast lane of an expressway

Abstract

The application discloses a method and system for hierarchical reminding of low-speed driving on a fast lane of an expressway, and relates to the technical field of vehicle safe driving assistance. The method comprises the following steps: acquiring multi-dimensional sensing data in the driving process of a vehicle and performing preprocessing; determining a current weather grade based on a weather state and an external temperature, and determining a low-speed triggering threshold; judging whether the current lane position is a fast lane based on current lane information and lane speed limit information; when the current lane is a fast lane, judging whether the triggering condition of hierarchical reminding is met based on a speed time sequence and the low-speed triggering threshold; if the triggering condition of hierarchical reminding is met, determining a rear vehicle threat grade based on the state of a rear vehicle, and determining a lane-changing safety grade based on the state of a right lane; and outputting hierarchical reminding information according to a preset hierarchical decision rule based on the rear vehicle threat grade, the lane-changing safety grade and the current weather grade. The application improves the real-time performance, accuracy and driving safety of the management of low-speed driving on a fast lane.

Description

Technical Field

[0001] This invention relates to the field of vehicle safety driving assistance technology, and in particular to a graded reminder method and system for low-speed driving in the fast lane of a highway. Background Technology

[0002] With the continuous growth of my country's expressway mileage and the rapid increase in car ownership, expressway traffic safety and efficiency have received increasing attention. In actual driving, it is common for vehicles to drive at low speeds in the fast lane for extended periods. This behavior not only reduces the efficiency of the fast lane and forces vehicles behind to change lanes frequently, but also easily leads to traffic congestion and even rear-end collisions.

[0003] Currently, solutions for managing low-speed driving in express lanes mainly fall into two categories: vehicle-mounted proactive alerts and roadside enforcement. Vehicle-mounted solutions typically identify lane position and vehicle speed, issuing alerts when the speed falls below a fixed threshold. However, their judgment logic is relatively simplistic, failing to dynamically adjust the low-speed trigger threshold based on weather conditions and real-time traffic flow. This results in rigid alert strategies and insufficient accuracy in adverse weather or complex road conditions. Furthermore, existing solutions often focus on prompting drivers to accelerate or change lanes, lacking a comprehensive assessment of the surrounding safety environment. For example, they fail to consider whether the right lane is suitable for lane changing or whether there is a collision risk from vehicles behind. Directly suggesting lane changes in unsafe situations could lead to secondary accidents. Roadside solutions often employ section speed measurement or surveillance cameras for post-incident penalties. While these can have a deterrent effect, they suffer from high construction costs and delayed feedback, failing to correct driver behavior in real time during driving and hindering proactive safety intervention. Summary of the Invention

[0004] To address the problems in existing solutions, such as the simplistic logic for determining low-speed driving, the lack of dynamic threshold adjustment, and the lack of safety assessment of the surrounding environment leading to potential safety hazards in the reminder strategy, this invention proposes a graded reminder method and system for low-speed driving in the fast lane of a highway.

[0005] The present invention achieves the above objectives through the following technical solutions:

[0006] A tiered reminder method for low-speed driving in the fast lane of a highway includes:

[0007] Acquire multi-dimensional perception data during vehicle operation and preprocess it; the multi-dimensional perception data includes weather conditions, road conditions ahead, vehicle conditions behind, right lane conditions, external temperature, current vehicle speed, current lane information, and lane speed limit information.

[0008] Based on the weather conditions and external temperature, determine the current weather level, and then determine the corresponding low-speed trigger threshold based on the current weather level.

[0009] Based on the current lane information and lane speed limit information, determine whether the current lane is a fast lane;

[0010] When the current lane is a fast lane, the system determines whether the triggering conditions for a graded reminder are met based on the vehicle speed time series and the low-speed trigger threshold.

[0011] If the triggering conditions for the graded reminder are met, the threat level of the vehicle behind is determined based on the status of the vehicle behind, and the lane change safety level is determined based on the status of the right lane.

[0012] Based on the threat level of following vehicles, lane change safety level, and current weather level, graded reminder information is output according to preset graded decision rules; the graded reminder information includes level 1 information reminder, level 2 suggestion reminder, level 3 warning reminder, and level 4 emergency alarm.

[0013] As a preferred embodiment of the present invention, the preprocessing includes outlier removal, linear interpolation, and time alignment.

[0014] As a preferred embodiment of the present invention, determining the current weather level based on weather conditions and external temperature includes:

[0015] The rainfall intensity level is determined based on rain sensor signals and / or wiper mode signals.

[0016] The system acquires road images captured by the vehicle's forward-facing camera, identifies the road surface condition in the road images through image semantic segmentation or feature extraction, and determines the road surface condition level by combining the external temperature.

[0017] Based on visibility sensor data or edge sharpness parameters obtained from road image edge detection results, visibility is determined by querying a pre-calibrated visibility-sharpness mapping table, and then the visibility level is determined based on the visibility.

[0018] Obtain the confidence levels corresponding to rainfall intensity level, road surface condition level, and visibility level;

[0019] When the confidence level of any of the rainfall intensity level, road surface condition level, and visibility level is lower than the preset confidence threshold, the level is marked as invalid and removed.

[0020] Determine if any levels remain after removal;

[0021] If there are remaining levels, compare the safety restrictions corresponding to the remaining levels and determine the remaining level with the highest safety restriction as the current weather level;

[0022] If no remaining weather level exists, the default weather level will be set as the current weather level.

[0023] As a preferred embodiment of the present invention, determining the corresponding low-speed trigger threshold based on the current weather level includes:

[0024] Obtain the preset safety lower limit threshold corresponding to the current weather level; the preset safety lower limit threshold is negatively correlated with the safety restriction level of the current weather level and is not higher than the statutory minimum speed limit;

[0025] If real-time traffic flow data exists for the current road segment, dynamic calibration is performed to obtain the low-speed trigger threshold.

[0026] If there is no real-time traffic flow data for the current road segment, a preset safety lower limit threshold will be used as the low-speed trigger threshold; the low-speed trigger threshold shall not be lower than the preset absolute safety lower limit and shall not be higher than the legal maximum speed limit of the current lane.

[0027] The process of performing dynamic calibration to obtain the low-speed trigger threshold includes:

[0028] Obtain the traffic flow characteristic speed of the current lane; the traffic flow characteristic speed is one of the statistical quantile, average or median value of the vehicle speed within a preset range of the current lane;

[0029] An exponentially weighted moving average algorithm is used to smooth the traffic flow characteristic speed, resulting in a smoothed traffic flow characteristic speed; the smoothing factor of the exponentially weighted moving average is a calibrable parameter.

[0030] Based on the smoothed traffic flow characteristic velocity and the preset velocity offset, the dynamic threshold is calculated using the following formula:

[0031] ;

[0032] in, For dynamic thresholds; The characteristic velocity of the smoothed traffic flow; This is the preset speed offset.

[0033] When the dynamic threshold is greater than the preset absolute safety lower limit, the dynamic threshold is used as the low-speed trigger threshold; when the dynamic threshold is less than or equal to the preset absolute safety lower limit, the preset absolute safety lower limit is used as the low-speed trigger threshold.

[0034] As a preferred embodiment of the present invention, the step of determining whether the current lane is a fast lane based on the current lane information and the lane speed limit information includes:

[0035] Based on the current lane information, lane line recognition is performed to determine whether there is an adjacent lane to the left of the current vehicle; if there is no adjacent lane to the left, the current vehicle is determined to be in the leftmost lane.

[0036] Based on lane speed limit information, obtain the maximum speed limit of the current lane, the maximum speed limit of the adjacent right lane, and the maximum speed limit of the current road.

[0037] Starting from the current vehicle position, along the driving direction, the system calculates the rate of change of lane width based on the geometric coordinates of the lane lines ahead, obtained from road images captured by the forward-facing camera. If the lane width increases monotonically and the cumulative rate of change calculated based on the current lane width exceeds a preset diversion threshold, then it is determined that there is a diversion trend ahead of the current lane. If the lane width decreases monotonically and the cumulative rate of change calculated based on the current lane width exceeds a preset narrowing threshold, then it is determined that there is a narrowing trend ahead of the current lane.

[0038] If there is no tendency for the current lane to diverge or narrow ahead, and any of the following conditions are met, then the current lane is determined to be a fast lane:

[0039] The vehicle is currently in the leftmost lane;

[0040] The current lane's maximum speed limit is equal to the current road's maximum speed limit.

[0041] The current lane's maximum speed limit is greater than the maximum speed limit of the adjacent right lane, and the current lane's maximum speed limit is not lower than a preset minimum speed threshold for fast lanes; the preset minimum speed threshold for fast lanes is higher than the legal minimum speed limit for the current road.

[0042] As a preferred embodiment of the present invention, the step of determining whether the triggering conditions for a graded reminder are met based on the vehicle speed time series and the low-speed trigger threshold includes:

[0043] Obtain the vehicle speed within a preset sliding window and construct a vehicle speed time series;

[0044] When it is determined that the current vehicle speed is lower than the low speed trigger threshold, the statistical characteristic values ​​of the vehicle speed time series within a preset sliding window are calculated; the statistical characteristic values ​​include the coefficient of variation and the linear regression slope.

[0045] If the slope of the linear regression is greater than the preset acceleration threshold, it is determined that the vehicle is accelerating away from the low-speed state, and the triggering of the reminder is suppressed.

[0046] If the linear regression slope is less than or equal to the preset acceleration threshold, the trigger duration threshold will be dynamically adjusted based on the coefficient of variation.

[0047] If the current vehicle speed is below the low speed trigger threshold for an extended period of time, the system will determine whether there are vehicles within a preset range ahead of the vehicle, based on the road conditions ahead.

[0048] If there are no vehicles within the preset range in front of the vehicle, the triggering conditions for the tiered alert are met.

[0049] If there is a vehicle within a preset range ahead of the vehicle, the longitudinal speed of the vehicle ahead and the longitudinal distance between the vehicle ahead and the vehicle ahead are obtained based on the road conditions ahead. If the current speed of the vehicle ahead is lower than the longitudinal speed of the vehicle ahead, and the difference between the longitudinal speed of the vehicle ahead and the current speed of the vehicle ahead is greater than a preset longitudinal speed threshold, the triggering conditions for the graded reminder are still met. If the difference between the longitudinal speed of the vehicle ahead and the current speed of the vehicle ahead is less than or equal to the preset longitudinal speed threshold, and the longitudinal distance between the vehicle ahead and the vehicle ahead is less than a preset following distance threshold, the reminder is suppressed.

[0050] As a preferred embodiment of the present invention, the formula for calculating the coefficient of variation is:

[0051] ;

[0052] in, The standard deviation of the vehicle speed within the preset sliding window; The average speed of the vehicle within the preset sliding window; is the coefficient of variation.

[0053] The formula for calculating the slope of linear regression is:

[0054] ;

[0055] in, The preset number of sampling points within the sliding window; For the first Each sampling time; For the first The vehicle speed at each sampling time; The average time within the preset sliding window; The slope of the linear regression;

[0056] The dynamic adjustment of the trigger duration threshold based on the coefficient of variation includes:

[0057] When the coefficient of variation is less than the preset variation threshold, the vehicle speed is determined to be in a stable low-speed state, and the first preset duration is used as the trigger duration threshold.

[0058] When the coefficient of variation is greater than or equal to the preset variation threshold, the vehicle speed is determined to be in a fluctuating low-speed state, and the second preset duration is used as the trigger duration threshold.

[0059] As a preferred embodiment of the present invention, determining the threat level of the following vehicle based on the status of the following vehicle includes:

[0060] Based on the status of vehicles behind, obtain the target vehicles within a preset range behind the vehicle in the current lane and the relative motion state of the target vehicles; the relative motion state includes longitudinal distance and longitudinal speed.

[0061] If a target vehicle is located behind the vehicle and its longitudinal velocity is greater than zero, it is determined that the target vehicle is approaching the vehicle. Based on the longitudinal distance and longitudinal velocity, the collision time between the target vehicle and the vehicle is calculated using the following formula:

[0062] ;

[0063] in, The collision time between the target vehicle and the vehicle itself; Vertical distance; Longitudinal velocity;

[0064] The collision time is compared with a preset threat level threshold to determine the threat level of the following vehicle; the threat level of the following vehicle includes Level 1 threat, Level 2 threat and Level 3 threat; among them, Level 1 threat corresponds to the longest collision time and Level 3 threat corresponds to the shortest collision time.

[0065] If there is no target vehicle behind the vehicle, or the target vehicle is not approaching the vehicle, the threat level of the vehicle behind is marked as no threat.

[0066] The method of determining the lane change safety level based on the right lane status includes:

[0067] Based on the status of the right lane, determine whether the right lane is an emergency lane, whether the right lane line is a solid line, and whether there is a curb or guardrail on the right side of the road;

[0068] If the right lane is an emergency lane, the right lane line is a solid line, or there is a curb or guardrail on the right side of the road, then the lane change safety level is determined to be unsafe.

[0069] If the right lane is not an emergency lane, the right lane line is not a solid line, and there is no curb or guardrail on the right side of the road, then it is determined that lane changing is allowed in the right lane, and the target vehicle and its relative motion state within a preset range behind the vehicle in the right lane are obtained.

[0070] If there is no target vehicle in the right lane, or if there is a target vehicle in the right lane but the longitudinal velocity of the target vehicle in its relative motion state is less than or equal to zero, then the lane change safety level is determined to be safe.

[0071] If there is a target vehicle in the right lane and the longitudinal velocity of the target vehicle in its relative motion state is greater than zero, it is determined that the target vehicle is approaching the vehicle, and the collision time between the target vehicle and the vehicle is calculated based on the relative motion state of the target vehicle.

[0072] If the collision time is less than the preset safe lane change time threshold, the lane change safety level is determined to be risky; if the collision time is greater than or equal to the preset safe lane change time threshold, the lane change safety level is determined to be safe.

[0073] As a preferred embodiment of the present invention, the preset hierarchical decision rule includes:

[0074] When the lane change safety level is safe, a level 2 suggestion reminder is output, prompting the driver to change lanes to the right;

[0075] When the lane change safety level is unsafe or risky, and the threat level of the vehicle behind is level three, if the current weather level is normal, a level three warning will be issued; if the current weather level is any other than normal weather or icing / snow accumulation, a level four emergency alarm will be issued.

[0076] When the lane change safety level is unsafe or risky, and the threat level of the following vehicle is level two or level one, a level one information reminder is output to remind the driver to pay attention to vehicles coming from behind.

[0077] When the lane change safety level is unsafe or risky, and the threat level of the following vehicle is not a threat, a level one information reminder is output to prompt the driver to stay in the current lane.

[0078] A graded reminder system for low-speed driving in the fast lane of a highway includes:

[0079] The data acquisition and preprocessing module is used to acquire and preprocess multi-dimensional perception data during vehicle operation. The multi-dimensional perception data includes weather conditions, road conditions ahead, vehicle conditions behind, right lane conditions, external temperature, current vehicle speed, current lane information, and lane speed limit information.

[0080] The low-speed trigger threshold generation module is used to determine the current weather level based on the weather conditions and external temperature, and to determine the corresponding low-speed trigger threshold based on the current weather level.

[0081] The fast lane recognition module is used to determine whether the current lane is a fast lane based on the current lane information and the lane speed limit information.

[0082] The graded reminder trigger condition determination module is used to determine whether the trigger conditions for graded reminders are met when the current lane is a fast lane, based on the vehicle speed time series and the low speed trigger threshold.

[0083] The rear vehicle threat and lane change safety assessment module is used to determine the rear vehicle threat level based on the status of the vehicles behind and the lane change safety level based on the status of the right lane if the triggering conditions for the graded reminder are met.

[0084] The graded reminder information output module is used to output graded reminder information based on the threat level of following vehicles, the lane change safety level, and the current weather level, according to preset graded decision rules; the graded reminder information includes level 1 information reminder, level 2 suggestion reminder, level 3 warning reminder, and level 4 emergency alarm.

[0085] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention provides a comprehensive information foundation for low-speed driving judgment by acquiring multi-dimensional data including weather conditions, external temperature, and multi-directional environmental perception; on this basis, it integrates weather conditions and external temperature to dynamically determine the low-speed trigger threshold, overcoming the shortcomings of existing technologies where the threshold is fixed and cannot adapt to complex scenarios; at the same time, it introduces a trigger condition judgment mechanism based on vehicle speed time series, effectively distinguishing between real low speed and brief deceleration, reducing the false alarm rate; before triggering the reminder, it further comprehensively evaluates the lane change safety level and the threat level of the following vehicle, ensuring that only under safe conditions, it outputs hierarchical reminder information of levels one to four, realizing a closed-loop decision-making process from environmental perception to dynamic threshold to intelligent judgment to safety assessment to graded response, improving the real-time performance, accuracy, and driving safety of low-speed driving management in fast lanes. Attached Figure Description

[0086] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein: Figure 1 This is a flowchart of a graded reminder method for low-speed driving in the fast lane of a highway, as proposed in this invention. Figure 2 This is a schematic diagram of the modular structure of a graded reminder system for low-speed driving in the fast lane of a highway, as proposed in this invention. Detailed Implementation

[0087] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the described embodiments of the present invention are within the scope of protection of the present invention.

[0088] like Figure 1 As shown, this is an embodiment of the present invention, which provides a graded reminder method for low-speed driving in the fast lane of a highway, including:

[0089] S1 acquires multi-dimensional perception data during vehicle operation and performs preprocessing.

[0090] Multidimensional sensing data includes weather conditions, external temperature, road conditions ahead, vehicle conditions behind, right lane conditions, current vehicle speed, current lane information, and lane speed limit information.

[0091] Weather conditions include at least rainfall, snowfall, and visibility.

[0092] When the vehicle is equipped with a rain sensor, the domain controller reads the signal value from the rain sensor and determines the rainfall intensity level based on the signal value; when the vehicle is not equipped with a rain sensor, the body controller collects the current setting of the windshield wipers and determines the rainfall intensity level based on the current setting. In one specific embodiment, when the rain sensor signal value is less than 0.5 mm / h, or the current wiper setting is off, the rainfall intensity level is determined to be no rain; when the rain sensor signal value is between 0.5 mm / h and 5 mm / h, or the current wiper setting is intermittent, the rainfall intensity level is determined to be light rain; when the rain sensor signal value is between 5 mm / h and 15 mm / h, or the current wiper setting is low-speed continuous, the rainfall intensity level is determined to be moderate rain; when the rain sensor signal value is between 15 mm / h and 30 mm / h, or the current wiper setting is high-speed continuous, the rainfall intensity level is determined to be heavy rain; when the rain sensor signal value is greater than or equal to 30 mm / h, or the current wiper setting is high-speed continuous and the rain sensor signal value is between 15 mm / h and 30 mm / h, the rainfall intensity level is determined to be torrential rain. If the signal value of the rain sensor is inconsistent with the rainfall intensity level determined by the current setting of the wipers, the result of the rain sensor signal value shall prevail, and the inconsistency event shall be recorded.

[0093] Visibility is usually obtained directly through a visibility sensor; when the vehicle is not equipped with a visibility sensor, a forward-facing camera is used to capture road images, Sobel edge detection is performed on the road images, the average pixel gradient value within the lane line area is calculated as the edge sharpness parameter, and the visibility-sharpness mapping table stored in non-volatile memory is queried to estimate the current visibility; this visibility-sharpness mapping table is obtained through real vehicle calibration and covers a visibility range of 10m to 1000m.

[0094] The intensity level of snowfall is determined by combining the precipitation intensity assessment with the external temperature. For example, when the external temperature is below 2°C, the original assessment of light rain is revised to light snow, moderate rain to moderate snow, heavy rain to heavy snow, and torrential rain to blizzard.

[0095] External temperature is obtained through an onboard temperature sensor, which is typically installed below the exterior rearview mirror or inside the front grille.

[0096] The road conditions ahead are acquired through a fusion perception of a forward-facing camera and a forward-facing millimeter-wave radar. Specifically, radar targets and visual targets are time-stamp matched and distance correlated. After successful correlation, Kalman filtering is performed to fuse the data, outputting the fused longitudinal distance and longitudinal velocity. Radar targets refer to the reflection points of objects detected by the forward-facing millimeter-wave radar through the emission of electromagnetic waves and the reception of echoes. Each radar target includes at least longitudinal distance, longitudinal velocity, and azimuth. Visual targets refer to vehicles or obstacles identified from road images captured by the forward-facing camera. Each visual target includes at least a 2D bounding box, target type, and detection confidence. The 2D bounding box includes the upper left and lower right pixel coordinates. Target types include at least vehicles, curbs, and guardrails. The allowed time difference threshold for time-stamp matching is no more than 50ms, and the allowed longitudinal distance difference threshold for distance correlation is no more than 2.0m, and the allowed lateral distance difference threshold is no more than 1.0m. When the vehicle is not equipped with a forward-facing millimeter-wave radar, the road conditions ahead are acquired through the forward-facing camera. Although this does not include direct distance measurements from the radar, it is sufficient to determine whether there is a need for a vehicle in front of or to the right of the vehicle.

[0097] The status of vehicles behind is acquired using a rearward millimeter-wave radar, including the longitudinal distance, longitudinal speed, and azimuth angle between the rear vehicles and the vehicle. When the vehicle is not equipped with a rearward millimeter-wave radar, the estimated values ​​of the longitudinal distance and longitudinal speed between the rear vehicles and the vehicle are directly obtained using a rear-view camera. The longitudinal speed is estimated by temporally differencing the longitudinal distance across multiple consecutive frames of images and then smoothing it using a Kalman filter. The longitudinal distance is estimated using a monocular vision ranging method, specifically: based on the 2D bounding box output by YOLOv8n, the midpoint of the bottom of this bounding box is used as the contact point between the target vehicle and the ground. The pixel coordinates are converted to normalized planar coordinates using the camera intrinsic parameter matrix. Then, based on the camera mounting height and pitch angle, and assuming a flat ground surface, the longitudinal distance is calculated using a pinhole camera model.

[0098] ;

[0099] in, The pixel equivalent of the camera's focal length; The pixel ordinate of the bottom midpoint of the target; The vertical pixel coordinate of the horizon in the image; Determine the camera installation height; Vertical distance; .

[0100] The status of the right lane is obtained through a fusion perception of side radar and surround-view cameras. The side radar identifies target vehicles in the right lane and their relative motion, including longitudinal distance and speed. Simultaneously, the surround-view cameras capture images of the right lane to identify lane line attributes and road boundary features. Lane line attributes include solid and dashed lines, and road boundary features include curbs and guardrails. Furthermore, combined with high-precision map data, it identifies whether the right lane is an emergency lane. When the vehicle is not equipped with side radar, the right lane status is obtained through surround-view cameras.

[0101] The current vehicle speed is calculated from the wheel speed sensor readings from the vehicle's CAN bus, and corrected for tire slip ratio; alternatively, the absolute vehicle speed is obtained by differential calculation using vehicle positioning information received from the GPS / BeiDou positioning system, which serves as a redundancy check for the wheel speed signal. When the deviation between the two exceeds 5 km / h, the wheel speed sensor signal is taken as the standard and the anomaly is recorded.

[0102] Current lane information is based on lane line coordinates identified by the forward-facing camera, combined with the LaneNet-ED deep learning-based network to determine the vehicle's current lane position, the total number of lanes, and lateral distance. Lane speed limit information is obtained through visual recognition of roadside speed limit signs by the forward-facing camera, or by retrieving the legal maximum and minimum speed limits for the current road segment from high-precision maps / navigation electronic maps. When the speed limit information on the speed limit sign differs from the speed limit information retrieved from the high-precision map / navigation electronic map, the speed limit information on the speed limit sign shall prevail.

[0103] Preprocessing includes outlier removal, linear interpolation, and time alignment. Single frames of data from a sensor that deviate significantly from physical laws due to interference are removed. In this embodiment, the 3σ criterion is used to remove outliers for continuous data, and the majority consensus principle is used for outliers for discontinuous data. For data frame loss caused by temporary occlusion, linear interpolation is performed to complete the data based on the values ​​of the preceding and following valid frames. For example, assuming the rainfall signal value is 8.0 mm / h at 10.0 seconds and 12.0 mm / h at 12.0 seconds, and there is no valid data at the current time of 10.8 seconds, the rainfall signal value at the current time of 10.8 seconds is obtained by linear interpolation as 8.0 + (12.0 - 8.0) × (10.8 - 10.0) / (12.0 - 10.0) = 8.0 + 4.0 × 0.4 = 9.6 mm / h. Finally, the timestamps of all linearly interpolated multidimensional sensing data are aligned to a common time axis, for example, with a period of 10 milliseconds.

[0104] S2 determines the current weather level based on the weather conditions and external temperature, and then determines the corresponding low-speed trigger threshold based on the current weather level.

[0105] Based on weather conditions and external temperature, determine the current weather level, including:

[0106] The rainfall intensity level is determined based on the rain sensor signal and / or wiper setting signal.

[0107] The system acquires road images captured by the vehicle's forward-facing camera. Through image semantic segmentation or feature extraction, it identifies the road surface condition in the road images and determines the road surface condition level based on the external temperature. In one specific embodiment, when the external temperature is above 4°C, there is no precipitation, and the road image shows no abnormalities, the road surface condition level is determined to be normal; when the external temperature is above 4°C and the road image identifies areas of water accumulation on the road, the road surface condition level is determined to be water accumulation; when the external temperature is between 0°C and 4°C and there is precipitation, the road surface condition level is determined to be slippery; when the external temperature is below 0°C and there is precipitation, the road surface condition level is determined to be icy / snowy; furthermore, when the external temperature is below or equal to 2°C and there is currently no precipitation, texture analysis is performed on the road surface area in the road image to extract texture feature parameters. If the texture feature parameters are below a preset dry road surface texture threshold, it is determined that there is dark ice on the road surface, and the road surface condition level is determined to be icy / snowy. The texture feature parameters include the contrast of the gray-level co-occurrence matrix or the uniformity of the local binary mode. The preset dry road surface texture threshold is obtained by collecting image samples of dry and icy roads under different lighting and road conditions, calibrating them offline, and storing them in the non-volatile memory of the domain controller.

[0108] Visibility is determined based on visibility sensor data or edge sharpness parameters obtained statistically from road image edge detection results, and a pre-calibrated visibility-sharpness mapping table is consulted. Based on this visibility, a visibility level is then determined. Visibility levels include normal, low, low, and very low. In this embodiment, when visibility is greater than 500m, the visibility level is determined to be normal; when visibility is between 200m and 500m, the visibility level is determined to be low; when visibility is between 50m and 200m, the visibility level is determined to be low; and when visibility is less than 50m, the visibility level is determined to be very low.

[0109] Obtain the confidence levels corresponding to rainfall intensity level, road surface condition level, and visibility level.

[0110] The confidence level of rainfall intensity level is calculated based on the stability of the rain gauge signal, using an inverse variance model. The calculation formula is as follows:

[0111] ;

[0112] in, The confidence level for the rainfall intensity level; These are the normalization coefficients; The variance of the rain sensor signal values ​​within a preset time window is calculated based on the current time. The preset time window is set to 1 second. To prevent zero constant; and have Same units. In this embodiment, Take 1.0 (mm / h)², Take 0.01 (mm / h)².

[0113] Confidence level of road surface condition rating The calculation formula is:

[0114] ;

[0115] in, This is the maximum probability value of the corresponding road surface state category output after processing road images using a convolutional neural network with ResNet-18 as the backbone network. The value ranges from 0 to 1. This convolutional neural network is an existing model. It is trained offline by collecting road image samples from different lighting and weather conditions. The total number of road image samples is no less than 100,000. The road images of each category are evenly distributed. The initial learning rate is set to 0.001 and the training period is set to 50 epochs. External temperature; As the first preset temperature threshold, take ; As the second preset temperature threshold, take The coefficients 0.1, 1.2, and 0.8 mentioned above are calibrated based on the physical characteristics of the relationship between the road surface adhesion coefficient and temperature.

[0116] Confidence level of visibility rating The calculation formula is:

[0117] ;

[0118] in, As a reliability factor for visibility measurement, when a vehicle is equipped with a visibility sensor, The confidence output value of the visibility sensor itself is used, ranging from 0 to 1; when the vehicle is not equipped with a visibility sensor... The matching degree between the image edge sharpness parameter and the pre-calibrated visibility-sharpness mapping relationship is obtained. The matching degree is calculated by interpolating the sum of squared residuals of the edge sharpness parameter and each calibration sample point in the piecewise linear mapping function, and the matching degree score is obtained by interpolation through the radial basis function. The value ranges from 0 to 1. The higher the matching degree, the closer the current edge sharpness parameter is to the reliable region in the calibration sample space, and the more reliable the estimation result is. As an image quality assessment factor, it counts the proportion of pixels in a road image whose brightness value exceeds a preset overexposure threshold out of the total number of pixels. ,like If the overexposure ratio exceeds the preset overexposure threshold, then ; Statistically determine the proportion of pixels in a road image whose brightness value is less than a preset underexposure threshold out of the total number of pixels. ,like If the underexposure ratio exceeds the preset threshold, then ;like and If none of them exceed their respective proportional thresholds, then The preset overexposure threshold is 240, the preset overexposure ratio threshold is 30%, the preset underexposure threshold is 20, and the preset underexposure ratio threshold is 20%.

[0119] If the confidence level of any of the rainfall intensity level, road surface condition level, and visibility level is lower than the preset confidence threshold, that level will be marked as invalid and removed. The preset confidence threshold is set to 0.5.

[0120] Determine if any levels remain after removal.

[0121] If remaining levels exist, the safety restrictions corresponding to these remaining levels are compared, and the remaining level with the highest safety restriction is determined as the current weather level. The safety restriction level for each level is numerically calibrated: in rainfall intensity levels, no rain = 0, light rain = 1, moderate rain = 3, heavy rain / storm rain = 4; in road surface condition levels, normal = 0, slippery = 2, water accumulation = 3, icy / snow accumulation = 4; in visibility levels, normal = 0, low = 1, low = 3, very low = 4. Higher values ​​indicate higher safety restrictions. The above numerical calibration is based on mapping the increase in vehicle braking distance caused by each weather factor. Specifically, for each increase of one safety restriction level, the corresponding lower limit of safe vehicle speed is reduced by approximately 15% to 20% of the legal minimum speed limit. This mapping relationship is calibrated using actual vehicle braking distance test data. When multiple remaining levels correspond to the same safety restriction level, the system selects the level according to the preset priority. The preset priorities, from highest to lowest, are: icing / snow accumulation, extremely low visibility, heavy rain / torrential rain, water accumulation, low visibility, moderate rain, slippery road surface, relatively low visibility, light rain, and normal weather.

[0122] If no remaining weather level is available, the default weather level will be set as the current weather level. The default weather level is icy / snowy weather.

[0123] The corresponding low-speed trigger threshold is determined based on the current weather level, including:

[0124] Obtain the preset safety lower limit threshold corresponding to the current weather level. The preset safety lower limit threshold is negatively correlated with the safety restriction level of the current weather level and does not exceed the legal minimum speed limit of the current lane. In a preferred embodiment, the preset safety lower limit thresholds corresponding to each weather level are shown in Table 1.

[0125] Table 1. Preset safety thresholds for different weather levels

[0126]

[0127] If real-time traffic flow data exists for the current road segment, dynamic calibration is performed to obtain the low-speed trigger threshold.

[0128] Perform dynamic calibration to obtain the low-speed trigger threshold, including:

[0129] Obtain the traffic flow characteristic speed of the current lane. The traffic flow characteristic speed is one of the statistical quantile, average, or median values ​​of the vehicle speed within a preset range of the current lane.

[0130] An exponentially weighted moving average algorithm is used to smooth the traffic flow characteristic velocity, resulting in a smoothed traffic flow characteristic velocity. The smoothing factor in the exponentially weighted moving average is a calibrable parameter.

[0131] Based on the smoothed traffic flow characteristic velocity and the preset velocity offset, the dynamic threshold is calculated using the following formula:

[0132] ;

[0133] in, For dynamic thresholds; The characteristic velocity of the smoothed traffic flow; This is the preset speed offset.

[0134] For example, using the 15th percentile value, a sample set of instantaneous speeds of vehicles passing through the current lane in the last 5 minutes is obtained. Statistical processing is then performed on the obtained speed sample set to obtain the 15th percentile speed. And use exponentially weighted moving average to Perform smoothing:

[0135] ;

[0136] in, The smoothing factor is set to 0.8 in this embodiment; Index for the current time; After smoothing The 15th percentile velocity at time 1; After smoothing The 15th percentile velocity at time 1; for The 15th percentile velocity at time 1.

[0137] Calculate the dynamic threshold:

[0138] ;

[0139] in, For dynamic thresholds; The preset speed offset is set to 15 km / h in this embodiment. This value is based on the time margin calibration corresponding to the safe following distance on highways.

[0140] When the dynamic threshold is greater than the preset absolute safety lower limit, the dynamic threshold is used as the low-speed trigger threshold; when the dynamic threshold is less than or equal to the preset absolute safety lower limit, the preset absolute safety lower limit is used as the low-speed trigger threshold. In this embodiment, the preset absolute safety lower limit is set to 20 km / h.

[0141] S3 determines whether the current lane is a fast lane based on the current lane information and lane speed limit information.

[0142] Determining whether the current lane position is a fast lane includes:

[0143] Based on the current lane information, lane line recognition is performed to determine whether there is an adjacent lane to the left of the current vehicle; if there is no adjacent lane to the left, the current vehicle is determined to be in the leftmost lane.

[0144] Based on lane speed limit information, obtain the maximum speed limit of the current lane, the maximum speed limit of the adjacent right lane, and the maximum speed limit of the current road.

[0145] Starting from the current vehicle position and moving along the direction of travel, the rate of change of lane width is calculated based on the geometric coordinates of the lane lines ahead, obtained from road images captured by a forward-facing camera. If the lane width monotonically increases, and the cumulative rate of change calculated based on the current lane width exceeds a preset diversion threshold, then a diversion trend is determined ahead of the current lane. If the lane width monotonically decreases, and the cumulative rate of change calculated based on the current lane width exceeds a preset narrowing threshold, then a narrowing trend is determined ahead of the current lane. The preset distance is calibrated based on the vehicle speed and sensor sensing distance; in this embodiment, it is set to 200m. Lane width Represented by the difference in lateral coordinates between the left and right lane lines at the same longitudinal position. Where y is the vertical distance. and These are the lateral coordinates of the left and right lane lines at the longitudinal distance y, respectively. In this embodiment, the preset diversion judgment threshold is set to 5%, that is, the cumulative increase exceeds 5% of the current lane width; the preset narrowing threshold is set to -5%, that is, the cumulative decrease exceeds 5% of the current lane width; when the judgment conclusion of the lane width change rate is inconsistent with the judgment conclusion of the cumulative change, the judgment conclusion of the cumulative change shall prevail.

[0146] If there is no tendency for the current lane to diverge or narrow ahead, and any of the following conditions are met, then the current lane is determined to be a fast lane:

[0147] The vehicle is currently in the leftmost lane.

[0148] The current lane's maximum speed limit is equal to the current road's maximum speed limit.

[0149] The current lane's maximum speed limit is greater than the maximum speed limit of the adjacent right lane, and the current lane's maximum speed limit is not lower than a preset minimum fast lane speed threshold. The preset minimum fast lane speed threshold is used to distinguish the minimum speed requirements of fast lanes and slow lanes, and is higher than the current road's legal minimum speed limit. In this embodiment, the preset minimum fast lane speed threshold is set to 80 km / h, but the specific value may be adjusted according to the minimum speed limit regulations and actual traffic management policies of highways in different countries or regions.

[0150] If the forward-facing camera identifies a speed limit value indicated by a road speed limit sign, and the highest speed limit in that speed limit value is inconsistent with the highest speed limit of the current lane, then the highest speed limit of the current lane is updated based on the highest speed limit indicated by the road speed limit sign, and the current lane is re-determined as a fast lane based on the updated highest speed limit.

[0151] S4. When the current lane is a fast lane, determine whether the triggering conditions for a graded reminder are met based on the vehicle speed time series and the low speed trigger threshold.

[0152] Obtain the vehicle speed within a preset sliding window and construct a vehicle speed time series.

[0153] When the current vehicle speed is determined to be below the low-speed trigger threshold, the statistical characteristic values ​​of the vehicle speed time series within a preset sliding window are calculated. These statistical characteristic values ​​include the coefficient of variation and the linear regression slope. In this embodiment, the preset sliding window takes the most recent 20 sampling points, corresponding to 2 seconds.

[0154] The formula for calculating the coefficient of variation is:

[0155] ;

[0156] in, The standard deviation of the vehicle speed within the preset sliding window; The average speed of the vehicle within the preset sliding window; is the coefficient of variation.

[0157] The formula for calculating the slope of linear regression is:

[0158] ;

[0159] in, The preset number of sampling points within the sliding window; For the first Each sampling time; For the first The vehicle speed at each sampling time; The average time within the preset sliding window; The slope represents the linear regression slope; a positive value indicates an increase in vehicle speed, while a negative value indicates a decrease in vehicle speed.

[0160] If the slope of the linear regression is greater than the preset acceleration threshold, it is determined that the vehicle is accelerating away from the low-speed state, and the triggering of the reminder is suppressed; in this embodiment, the preset acceleration threshold is 0.5 km / h / s.

[0161] If the linear regression slope is less than or equal to the preset acceleration threshold, the trigger duration threshold will be dynamically adjusted based on the coefficient of variation.

[0162] The trigger duration threshold is dynamically adjusted based on the coefficient of variation, including:

[0163] When the coefficient of variation is less than the preset variation threshold, the vehicle speed is determined to be in a stable low-speed state, and a first preset duration is used as the trigger duration threshold. The preset variation threshold is 0.1, and the first preset duration is 8 seconds.

[0164] When the coefficient of variation is greater than or equal to a preset variation threshold, the vehicle speed is determined to be in a fluctuating low-speed state, and a second preset duration is used as the trigger duration threshold. The second preset duration is 25 seconds.

[0165] If the current vehicle speed is below the low-speed trigger threshold for an extended period beyond the trigger duration threshold, the system will determine whether there are vehicles within a preset range ahead of the vehicle, based on the road conditions ahead.

[0166] If there are no vehicles within the preset range in front of the vehicle, the triggering conditions for the tiered alert are met.

[0167] If a vehicle is located within a preset range ahead of the vehicle, the longitudinal speed of the vehicle ahead and the longitudinal distance between the vehicle ahead and the vehicle ahead are obtained based on the road conditions ahead. If the current speed of the vehicle ahead is lower than the longitudinal speed of the vehicle ahead, and the difference between the longitudinal speed of the vehicle ahead and the current speed of the vehicle ahead is greater than a preset longitudinal speed threshold, the triggering conditions for a tiered alert are still met. If the difference between the longitudinal speed of the vehicle ahead and the current speed of the vehicle ahead is less than or equal to the preset longitudinal speed threshold, and the longitudinal distance between the vehicle ahead and the vehicle ahead is less than a preset following distance threshold, the alert is suppressed. In this embodiment, the preset longitudinal speed threshold is set to 10 km / h.

[0168] S5, if the triggering conditions for graded reminders are met, the threat level of the following vehicle is determined based on the status of the following vehicle, and the lane change safety level is determined based on the status of the right lane.

[0169] The threat level of the vehicle behind is determined based on the status of the vehicle behind, including:

[0170] Based on the status of vehicles behind, the system acquires information about target vehicles within a preset range behind the current vehicle in the current lane, as well as their relative motion. The relative motion includes longitudinal distance and longitudinal speed.

[0171] If a target vehicle is located behind the vehicle and its longitudinal velocity is greater than zero, it is determined that the target vehicle is approaching the vehicle. Based on the longitudinal distance and longitudinal velocity, the collision time between the target vehicle and the vehicle is calculated using the following formula:

[0172] ;

[0173] in, The collision time between the target vehicle and the vehicle itself; Vertical distance; This represents the longitudinal velocity.

[0174] The collision time is compared with a preset threat level threshold to determine the threat level of the following vehicle. The threat levels of the following vehicle include Level 1 threat, Level 2 threat, and Level 3 threat; among them, Level 1 threat corresponds to the longest collision time, and Level 3 threat corresponds to the shortest collision time; the shorter the collision time, the higher the threat level of the following vehicle.

[0175] In this embodiment, the preset threat level threshold includes a Level 1 threat threshold. and Level 2 threat threshold ,in, Level 1 Threat Threshold Set to 8 seconds, Level 2 threat threshold Set to 3 seconds, Level 1 threat threshold and Level 2 threat threshold The specific values ​​are calibrated and adjusted based on the actual vehicle model's braking performance and the conservatism of the system strategy. When At that time, the threat level of the following vehicle was determined to be Level 3 threat; when At that time, the threat level of the following vehicle was determined to be Level 2 threat; when At that time, the threat level of the following vehicle was determined to be Level 1 threat.

[0176] If there is no target vehicle behind your vehicle, or if the target vehicle is not approaching your vehicle, then mark the threat level of the vehicle behind you as no threat.

[0177] Determining the lane change safety level based on the condition of the right lane includes:

[0178] Based on the status of the right lane, determine whether the right lane is an emergency lane, whether the right lane line is a solid line, and whether there is a curb or guardrail on the right side of the road;

[0179] If the right lane is an emergency lane, the right lane line is a solid line, or there is a curb or guardrail on the right side of the road, then the lane change safety level is determined to be unsafe.

[0180] If the right lane is not an emergency lane, the right lane line is not a solid line, and there is no curb or guardrail on the right side of the road, then it is determined that lane changing is allowed in the right lane, and the target vehicle and its relative motion state within a preset range behind the vehicle in the right lane are obtained.

[0181] If there is no target vehicle in the right lane, or if there is a target vehicle in the right lane but the longitudinal speed of the target vehicle in its relative motion state is less than or equal to zero, then the lane change safety level is determined to be safe.

[0182] If a target vehicle is present in the right lane and its longitudinal velocity in relative motion is greater than zero, it is determined that the target vehicle is approaching the vehicle. Based on the target vehicle's relative motion, the collision time between the target vehicle and the vehicle is calculated. This collision time is calculated in the same way as in determining the threat level of the following vehicle based on the status of the following vehicles.

[0183] If the collision time is less than the preset safe lane change time threshold, the lane change safety level is determined to be risky; if the collision time is greater than or equal to the preset safe lane change time threshold, the lane change safety level is determined to be safe. In this embodiment, the preset safe lane change time threshold is set to 6 seconds. This preset safe lane change time threshold is set based on the sum of the time required for a typical lane change operation on a highway and the safety margin, ensuring that the driver has sufficient time to complete the lane change operation without conflicting with vehicles approaching from the right rear.

[0184] S6 outputs graded alert information based on the threat level of following vehicles, lane change safety level, and current weather level, according to preset graded decision-making rules.

[0185] The tiered alert information includes Level 1 information alerts, Level 2 suggestion alerts, Level 3 warning alerts, and Level 4 emergency alarms.

[0186] Pre-defined hierarchical decision-making rules include:

[0187] When the lane change safety level is deemed safe, a Level 2 suggestion reminder is issued, prompting the driver to change lanes to the right. The Level 2 suggestion reminder is delivered via voice prompts combined with dashboard icon indicators; for example, "The road ahead is clear, the right lane is safe, it is recommended to change lanes to the slow lane."

[0188] When the lane change safety level is unsafe or risky, and the threat level from the vehicle behind is level three, a level three warning will be issued if the current weather level is normal; otherwise, a level four emergency alarm will be issued. The level three warning is delivered via voice announcement, flashing warning icons on the instrument panel, and steering wheel vibration to alert the driver that a vehicle is rapidly approaching from behind and that it is not safe to change lanes. The level four emergency alarm includes continuous audible and visual alarms, full-screen flashing prompts, and safety advice, such as: "Vehicles are rapidly approaching from behind, changing lanes is unsafe, roads are slippery in rainy weather, please accelerate immediately or prepare for evasive action!"

[0189] When the lane change safety level is unsafe or risky, and the threat level from the vehicle behind is level two or one, a level one warning message is output to alert the driver to vehicles approaching from behind. The level one warning message is displayed only as a static warning icon on the instrument panel or HUD, without any sound. For example, a yellow warning icon on the instrument panel indicates that a vehicle is approaching from behind and the driver should be aware of the approaching vehicle.

[0190] When the lane change safety level is unsafe or risky, and the threat level from following vehicles is non-threatening, a Level 1 warning is issued, prompting the driver to stay in the current lane. When the lane change is unsafe but there are no vehicles behind, the driver faces a lower level of urgency, and a Level 1 warning is issued, displayed only as an icon on the instrument panel or HUD without sound, prompting the driver to stay in the current lane.

[0191] like Figure 2 As shown, another embodiment of the present invention provides a graded reminder system for low-speed driving in the fast lane of a highway, comprising:

[0192] The data acquisition and preprocessing module is used to acquire and preprocess multi-dimensional perception data during vehicle operation. The multi-dimensional perception data includes weather conditions, road conditions ahead, vehicle conditions behind, right lane conditions, external temperature, current vehicle speed, current lane information, and lane speed limit information.

[0193] The low-speed trigger threshold generation module is used to determine the current weather level based on the weather conditions and external temperature, and then determine the corresponding low-speed trigger threshold based on the current weather level.

[0194] The fast lane recognition module is used to determine whether the current lane is a fast lane based on the current lane information and lane speed limit information.

[0195] The graded reminder trigger condition determination module is used to determine whether the trigger conditions for graded reminders are met when the current lane is a fast lane, based on the vehicle speed time series and the low speed trigger threshold.

[0196] The rear vehicle threat and lane change safety assessment module is used to determine the rear vehicle threat level based on the status of the vehicles behind and the lane change safety level based on the status of the right lane if the triggering conditions for the graded reminder are met.

[0197] The graded alert information output module is used to output graded alert information based on the threat level of following vehicles, the lane change safety level, and the current weather level, according to preset graded decision rules. The graded alert information includes Level 1 information alert, Level 2 suggestion alert, Level 3 warning alert, and Level 4 emergency alarm.

[0198] In summary, this invention provides a tiered alert method and system for low-speed driving in highway fast lanes. By acquiring multi-dimensional perception data and fusing weather conditions and real-time traffic flow information, a low-speed trigger threshold with scene adaptability is dynamically determined, solving the problems of fixed thresholds and poor adaptability in existing solutions. Furthermore, a trigger condition determination mechanism based on vehicle speed time-series statistical feature analysis is introduced. The coefficient of variation and linear regression slope dynamically distinguish between stable low speeds, fluctuating low speeds, and acceleration-disengagement states, reducing the false alarm rate of unnecessary alerts. Simultaneously, by comprehensively evaluating lane change safety levels and the threat level of following vehicles, a tiered alert system is output, including Level 1 information alerts, Level 2 suggestion alerts, Level 3 warning alerts, and Level 4 emergency alarms, ensuring safety. This overcomes the shortcomings of existing single-level alerts and lack of safety guarantees, achieving end-to-end intelligent decision-making from single threshold triggering to environmental perception, state discrimination, safety assessment, and tiered response, improving the real-time performance, accuracy, and driving safety of low-speed driving management in fast lanes.

[0199] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various variations or substitutions within the technical scope disclosed in this application, and these should all be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A graded reminder method for low-speed driving in the fast lane of a highway, characterized in that, include: Acquire multi-dimensional perception data during vehicle operation and perform preprocessing; The multi-dimensional sensing data includes weather conditions, road conditions ahead, vehicle status behind, right lane status, external temperature, current vehicle speed, current lane information, and lane speed limit information; Based on the weather conditions and external temperature, determine the current weather level, and then determine the corresponding low-speed trigger threshold based on the current weather level. Based on the current lane information and lane speed limit information, determine whether the current lane is a fast lane; When the current lane is a fast lane, the system determines whether the triggering conditions for a graded reminder are met based on the vehicle speed time series and the low-speed trigger threshold. If the triggering conditions for the graded reminder are met, the threat level of the vehicle behind is determined based on the status of the vehicle behind, and the lane change safety level is determined based on the status of the right lane. Based on the threat level of following vehicles, lane change safety level, and current weather level, graded reminder information is output according to preset graded decision rules; the graded reminder information includes level 1 information reminder, level 2 suggestion reminder, level 3 warning reminder, and level 4 emergency alarm.

2. The graded reminder method for low-speed driving in the fast lane of a highway according to claim 1, characterized in that, The preprocessing includes outlier removal, linear interpolation, and time alignment.

3. The graded reminder method for low-speed driving in the fast lane of a highway according to claim 1, characterized in that, The determination of the current weather level based on weather conditions and external temperature includes: The rainfall intensity level is determined based on rain sensor signals and / or wiper mode signals. The system acquires road images captured by the vehicle's forward-facing camera, identifies the road surface condition in the road images through image semantic segmentation or feature extraction, and determines the road surface condition level by combining the external temperature. Based on visibility sensor data or edge sharpness parameters obtained from road image edge detection results, visibility is determined by querying a pre-calibrated visibility-sharpness mapping table, and then the visibility level is determined based on the visibility. Obtain the confidence levels corresponding to rainfall intensity level, road surface condition level, and visibility level; When the confidence level of any of the rainfall intensity level, road surface condition level, and visibility level is lower than the preset confidence threshold, the level is marked as invalid and removed. Determine if any levels remain after removal; If there are remaining levels, compare the safety restrictions corresponding to the remaining levels and determine the remaining level with the highest safety restriction as the current weather level; If no remaining weather level exists, the default weather level will be set as the current weather level.

4. The graded reminder method for low-speed driving in the fast lane of a highway according to claim 3, characterized in that, The determination of the corresponding low-speed trigger threshold based on the current weather level includes: Obtain the preset safety lower limit threshold corresponding to the current weather level; the preset safety lower limit threshold is negatively correlated with the safety restriction level of the current weather level and is not higher than the statutory minimum speed limit; If real-time traffic flow data exists for the current road segment, dynamic calibration is performed to obtain the low-speed trigger threshold. If there is no real-time traffic flow data for the current road segment, a preset safety lower limit threshold will be used as the low-speed trigger threshold; the low-speed trigger threshold shall not be lower than the preset absolute safety lower limit and shall not be higher than the legal maximum speed limit of the current lane. The process of performing dynamic calibration to obtain the low-speed trigger threshold includes: Obtain the traffic flow characteristic speed of the current lane; the traffic flow characteristic speed is one of the statistical quantile, average or median value of the vehicle speed within a preset range of the current lane; An exponentially weighted moving average algorithm is used to smooth the traffic flow characteristic speed, resulting in a smoothed traffic flow characteristic speed; the smoothing factor of the exponentially weighted moving average is a calibrable parameter. Based on the smoothed traffic flow characteristic velocity and the preset velocity offset, the dynamic threshold is calculated using the following formula: ； in, For dynamic thresholds; The characteristic velocity of the smoothed traffic flow; This is the preset speed offset. When the dynamic threshold is greater than the preset absolute safety lower limit, the dynamic threshold is used as the low-speed trigger threshold; when the dynamic threshold is less than or equal to the preset absolute safety lower limit, the preset absolute safety lower limit is used as the low-speed trigger threshold.

5. The graded reminder method for low-speed driving in the fast lane of a highway according to claim 1, characterized in that, The step of determining whether the current lane is a fast lane based on the current lane information and lane speed limit information includes: Based on the current lane information, lane line recognition is performed to determine whether there is an adjacent lane to the left of the current vehicle; if there is no adjacent lane to the left, the current vehicle is determined to be in the leftmost lane. Based on lane speed limit information, obtain the maximum speed limit of the current lane, the maximum speed limit of the adjacent right lane, and the maximum speed limit of the current road. Starting from the current vehicle position, along the driving direction, the system calculates the rate of change of lane width based on the geometric coordinates of the lane lines ahead, obtained from road images captured by the forward-facing camera. If the lane width increases monotonically and the cumulative rate of change calculated based on the current lane width exceeds a preset diversion threshold, then it is determined that there is a diversion trend ahead of the current lane. If the lane width decreases monotonically and the cumulative rate of change calculated based on the current lane width exceeds a preset narrowing threshold, then it is determined that there is a narrowing trend ahead of the current lane. If there is no tendency for the current lane to diverge or narrow ahead, and any of the following conditions are met, then the current lane is determined to be a fast lane: The vehicle is currently in the leftmost lane; The current lane's maximum speed limit is equal to the current road's maximum speed limit. The current lane's maximum speed limit is greater than the maximum speed limit of the adjacent right lane, and the current lane's maximum speed limit is not lower than a preset minimum speed threshold for fast lanes; the preset minimum speed threshold for fast lanes is higher than the legal minimum speed limit for the current road.

6. The graded reminder method for low-speed driving in the fast lane of a highway according to claim 1, characterized in that, The determination of whether the triggering conditions for a graded reminder are met based on vehicle speed time series and low-speed trigger threshold includes: Obtain the vehicle speed within a preset sliding window and construct a vehicle speed time series; When it is determined that the current vehicle speed is lower than the low speed trigger threshold, the statistical characteristic values ​​of the vehicle speed time series within a preset sliding window are calculated; the statistical characteristic values ​​include the coefficient of variation and the linear regression slope. If the slope of the linear regression is greater than the preset acceleration threshold, it is determined that the vehicle is accelerating away from the low-speed state, and the triggering of the reminder is suppressed. If the linear regression slope is less than or equal to the preset acceleration threshold, the trigger duration threshold will be dynamically adjusted based on the coefficient of variation. If the current vehicle speed is below the low speed trigger threshold for an extended period of time, the system will determine whether there are vehicles within a preset range ahead of the vehicle, based on the road conditions ahead. If there are no vehicles within the preset range in front of the vehicle, the triggering conditions for the graded reminder are met; If there is a vehicle within a preset range ahead of the vehicle, the longitudinal speed of the vehicle ahead and the longitudinal distance between the vehicle ahead and the vehicle ahead are obtained based on the road conditions ahead. If the current speed of the vehicle ahead is lower than the longitudinal speed of the vehicle ahead, and the difference between the longitudinal speed of the vehicle ahead and the current speed of the vehicle ahead is greater than a preset longitudinal speed threshold, the triggering conditions for the graded reminder are still met. If the difference between the longitudinal speed of the vehicle ahead and the current speed of the vehicle ahead is less than or equal to the preset longitudinal speed threshold, and the longitudinal distance between the vehicle ahead and the vehicle ahead is less than a preset following distance threshold, the reminder is suppressed.

7. A graded reminder method for low-speed driving in the fast lane of a highway according to claim 6, characterized in that, The formula for calculating the coefficient of variation is: ； in, The standard deviation of the vehicle speed within the preset sliding window; The average speed of the vehicle within the preset sliding window; The coefficient of variation; The formula for calculating the slope of linear regression is: ； in, The preset number of sampling points within the sliding window; For the first Each sampling time; For the first The vehicle speed at each sampling time; The average time within the preset sliding window; The slope of the linear regression; The dynamic adjustment of the trigger duration threshold based on the coefficient of variation includes: When the coefficient of variation is less than the preset variation threshold, the vehicle speed is determined to be in a stable low-speed state, and the first preset duration is used as the trigger duration threshold. When the coefficient of variation is greater than or equal to the preset variation threshold, the vehicle speed is determined to be in a fluctuating low-speed state, and the second preset duration is used as the trigger duration threshold.

8. A graded reminder method for low-speed driving in the fast lane of a highway according to claim 1, characterized in that, The method of determining the threat level of a following vehicle based on the status of the following vehicle includes: Based on the status of vehicles behind, obtain the target vehicles within a preset range behind the vehicle in the current lane and the relative motion state of the target vehicles; the relative motion state includes longitudinal distance and longitudinal speed. If a target vehicle is located behind the vehicle and its longitudinal velocity is greater than zero, it is determined that the target vehicle is approaching the vehicle. Based on the longitudinal distance and longitudinal velocity, the collision time between the target vehicle and the vehicle is calculated using the following formula: ； in, The collision time between the target vehicle and the vehicle itself; Vertical distance; Longitudinal velocity; The collision time is compared with a preset threat level threshold to determine the threat level of the following vehicle; the threat level of the following vehicle includes Level 1 threat, Level 2 threat and Level 3 threat; among them, Level 1 threat corresponds to the longest collision time and Level 3 threat corresponds to the shortest collision time. If there is no target vehicle behind the vehicle, or the target vehicle is not approaching the vehicle, the threat level of the vehicle behind is marked as no threat. The method of determining the lane change safety level based on the right lane status includes: Based on the status of the right lane, determine whether the right lane is an emergency lane, whether the right lane line is a solid line, and whether there is a curb or guardrail on the right side of the road; If the right lane is an emergency lane, the right lane line is a solid line, or there is a curb or guardrail on the right side of the road, then the lane change safety level is determined to be unsafe. If the right lane is not an emergency lane, the right lane line is not a solid line, and there is no curb or guardrail on the right side of the road, then it is determined that lane changing is allowed in the right lane, and the target vehicle and its relative motion state within a preset range behind the vehicle in the right lane are obtained. If there is no target vehicle in the right lane, or if there is a target vehicle in the right lane but the longitudinal velocity of the target vehicle in its relative motion state is less than or equal to zero, then the lane change safety level is determined to be safe. If there is a target vehicle in the right lane and the longitudinal velocity of the target vehicle in its relative motion state is greater than zero, it is determined that the target vehicle is approaching the vehicle, and the collision time between the target vehicle and the vehicle is calculated based on the relative motion state of the target vehicle. If the collision time is less than the preset safe lane change time threshold, the lane change safety level is determined to be risky; if the collision time is greater than or equal to the preset safe lane change time threshold, the lane change safety level is determined to be safe.

9. A graded reminder method for low-speed driving in the fast lane of a highway according to claim 8, characterized in that, The preset hierarchical decision-making rules include: When the lane change safety level is safe, a level 2 suggestion reminder is output, prompting the driver to change lanes to the right; When the lane change safety level is unsafe or risky, and the threat level of the vehicle behind is level three, if the current weather level is normal, a level three warning will be issued; if the current weather level is any other than normal weather or icing / snow accumulation, a level four emergency alarm will be issued. When the lane change safety level is unsafe or risky, and the threat level of the following vehicle is level two or level one, a level one information reminder is output to remind the driver to pay attention to vehicles coming from behind. When the lane change safety level is unsafe or risky, and the threat level of the following vehicle is not a threat, a level one information reminder is output to prompt the driver to stay in the current lane.

10. A graded reminder system for low-speed driving in the fast lane of a highway, based on the graded reminder method for low-speed driving in the fast lane of a highway as described in any one of claims 1 to 9, characterized in that, include: The data acquisition and preprocessing module is used to acquire multi-dimensional perception data during vehicle operation and perform preprocessing. The multi-dimensional sensing data includes weather conditions, road conditions ahead, vehicle status behind, right lane status, external temperature, current vehicle speed, current lane information, and lane speed limit information; The low-speed trigger threshold generation module is used to determine the current weather level based on the weather conditions and external temperature, and to determine the corresponding low-speed trigger threshold based on the current weather level. The fast lane recognition module is used to determine whether the current lane is a fast lane based on the current lane information and the lane speed limit information. The graded reminder trigger condition determination module is used to determine whether the trigger conditions for graded reminders are met when the current lane is a fast lane, based on the vehicle speed time series and the low speed trigger threshold. The rear vehicle threat and lane change safety assessment module is used to determine the rear vehicle threat level based on the status of the vehicles behind and the lane change safety level based on the status of the right lane if the triggering conditions for the graded reminder are met. The graded reminder information output module is used to output graded reminder information based on the threat level of following vehicles, the lane change safety level, and the current weather level, according to preset graded decision rules; the graded reminder information includes level 1 information reminder, level 2 suggestion reminder, level 3 warning reminder, and level 4 emergency alarm.

Images