Road surface state warning system and method

Abstract

The invention provides a road surface state warning method. The method comprises the steps that target road surface information in front of a vehicle is collected; determining a target road surface danger level and a target road surface abnormal condition based on the target road surface information; according to the target road surface danger level and the target road surface abnormal condition, projection information is determined, and the projection information comprises a light blanket coverage area and a warning mark; and controlling a vehicle lamp to output a warning mark and a warning light blanket consistent with the light blanket coverage area. Through real-time fusion of the multi-source sensor data, the projection content can be accurately adjusted according to the danger level and the abnormal condition, false alarm or missing alarm is avoided, the warning information is directly presented on the road surface through light blanket projection, a driver can visually understand the danger without moving away the sight, and distraction is reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of safe driving, in particular to a road surface state warning system and method. BACKGROUND

[0002] In the process of driving, timely and accurate identification of the road surface state in front of the vehicle is crucial to driving safety. Especially in the conditions of poor light and obstructed view such as night, tunnel or bad weather, it is difficult for the driver to discover potential dangers such as water accumulation, icing, oil stain and pothole in advance by naked eye, which can easily lead to traffic accidents.

[0003] In the prior art, the advanced driver assistance system (ADAS) and related lighting technology have made some progress. For example, adaptive driving beam (ADB) and matrix headlight technology can automatically adjust the beam shape according to the oncoming vehicle situation, realizing the functions of lane lighting and anti-dazzling. In addition, some systems can identify lane lines, traffic signs and even vehicles in front through monocular or binocular cameras.

[0004] However, these existing technologies have obvious limitations, mainly in the following three aspects: 1) single perception dimension, insufficient recognition accuracy: mainstream systems rely heavily on visual sensors (cameras). Although cameras can provide rich texture information, their perception effect is greatly affected by environmental lighting and weather. In the night, rain, snow and fog weather, the image quality decreases sharply. More importantly, cameras are difficult to effectively distinguish road surface states with similar reflection characteristics but completely different physical danger levels, such as dry asphalt, wet road surface, water accumulation, thin ice and oil stain, which leads to unreliable estimation of the friction coefficient and cannot provide accurate basis for safety decision-making. 2) shallow information fusion level, poor decision robustness: although some solutions try to introduce heterogeneous sensors such as radar, they mostly stay at the level of simple data or decision fusion. There is a lack of effective fusion methods at the feature level for multi-source (radar point cloud, camera image, vehicle body sensor signal) heterogeneous data. Due to the differences in working principle, data format, sampling frequency and time and space reference of each sensor, direct fusion can easily cause data conflicts and inconsistencies, making it difficult to form a consistent and robust description of the road surface state, especially when a sensor is temporarily disabled or disturbed, the system performance will deteriorate significantly. 3) warning method is out of context, and human-computer interaction efficiency is low: traditional warning is mostly realized through instrument panel icons or auditory alarms. This method forces the driver to move his eyes away from the road to understand the abstract warning symbols, increasing the cognitive load and reaction time, which is especially dangerous in emergency situations. In addition, the existing warning strategy is usually static and fixed, and cannot be combined with dynamic context factors such as real-time vehicle speed, weather, historical accident data and individual behavior habits of the driver, which cannot provide personalized and precise risk warning and guidance, resulting in poor warning effect and low acceptance of the driver. SUMMARY

[0005] The present application aims at the defects of the prior art, and provides a road surface state warning method, comprising: collecting target road surface information in front of a vehicle; determining a target road surface danger level and a target road surface abnormal situation based on the target road surface information; determining projection information according to the target road surface danger level and the target road surface abnormal situation, the projection information comprising a light blanket coverage area and a warning sign; controlling a vehicle lamp to output the warning sign and a warning light blanket consistent with the light blanket coverage area.

[0006] Further, the specific method for determining the target road surface danger level based on the target road surface information is as follows: the target road surface information is current raw data samples from each heterogeneous sensor, the current raw data samples of each heterogeneous sensor are preprocessed, the current raw data samples after preprocessing are clustered to obtain all road surface features of the target road surface; a plurality of danger levels are preset, each danger level defines a corresponding classical domain of all road surface feature characteristic values, and the classical domain is within a section domain corresponding to the road surface feature characteristic values; according to the characteristic values of all road surface features of the target road surface, the correlation degrees of the target road surface and each danger level are calculated based on a road surface and danger level correlation degree calculation model, and the danger level with the largest correlation degree with the target road surface is taken as the target road surface danger level.

[0007] Further, the specific method for calculating the correlation degrees of the target road surface and each danger level according to the characteristic values of all road surface features of the target road surface based on the road surface and danger level correlation degree calculation model is as follows: the danger level is represented as , the first danger level, the friction coefficient classical domain, the obstacle distance classical domain, the visibility classical domain, and the road surface flatness classical domain of each danger level are not overlapped; the road surface and danger level correlation degree calculation model comprises a correlation function and a correction function; the characteristic values of all road surface features of the target road surface are compared with corresponding road surface features in each danger level; Input the correlation function to obtain the correlation degree between all road features of the target road and each hazard level. For each hazard level, the correlation degree between each road feature of the target road and the hazard level is weighted and summed to obtain the initial correlation degree between the target road and the hazard level. Input the initial correlation degree between the target road and the hazard level into the correction function to obtain the correlation degree between the target road and the hazard level. Iterate through all hazard levels to obtain the correlation degree between the target road and each hazard level.

[0008] Furthermore, the correlation function is specifically as follows: In the formula, It is the target road surface Characteristic values ​​of road surface features for Road surface features The lower limit value, for feature The upper limit, For the target road surface Features and the first kind The initial degree of correlation; In the formula, For the target road surface and the first The initial correlation of the risk levels, For the target road surface Total number of features For the target road surface Features and the first kind The association weight.

[0009] Furthermore, the correction function is as follows: In the formula, For vehicle speed influencing factors, Weather influencing factors As a historical data comparison factor, For driver behavior factors, For the target road surface and the first The degree of correlation between different risk levels; In the formula, This is the current vehicle speed. That is the base speed. It is the maximum permissible speed. It is the vehicle speed influence coefficient; In the formula, It is the first The influence coefficient of this type of weather It is the first A weather indicator function, if the current weather is the first type. This weather =1, otherwise =0; In the formula, It is the historical influence coefficient. It is the time decay coefficient. It is the base of the natural logarithm. It is the time elapsed since the historical event. It is a value indicating the severity of historical events. This is the maximum severity value; In the formula, It is the driver behavior influence coefficient. It is the probability of a driver driving safely.

[0010] Furthermore, the specific method for determining the projection information based on the target road surface hazard level and the target road surface anomaly is as follows: The abnormal conditions of the target road surface include the abnormal areas and the abnormal types of the target road surface; When the distance between the vehicle and the abnormal area of ​​the target road surface is a preset distance, a light strip is projected to surround the abnormal area of ​​the target road surface. The corresponding warning sign is selected from the warning sign library based on the type of abnormality and the hazard level of the target road surface. The corresponding warning sign is projected to the designated position according to the abnormal area of ​​the target road surface, and the coverage area of ​​the light blanket is adjusted according to the abnormal area of ​​the target road surface.

[0011] Furthermore, the specific method for projecting corresponding warning signs onto designated locations within the light carpet based on abnormal areas of the target road surface is as follows: The road surface area is divided into two sides based on the position of the road centerline; When the abnormal area of ​​the target road surface falls entirely on one side of the road surface, the corresponding warning sign will be projected onto the other side of the road surface. When an abnormal area of ​​the target road surface falls on both sides of the road, corresponding warning signs are projected onto both sides of the road surface.

[0012] When the abnormal area of ​​the target road surface is a road surface on both sides, the width of the corresponding pattern is increased by a preset multiple.

[0013] Furthermore, the specific method for adjusting the projection light blanket coverage area based on abnormal areas of the target road surface is as follows: When the entire abnormal area of ​​the target road surface falls on one side of the road surface, the light blanket coverage area shall cover at least that side of the road surface in width; When the abnormal area of ​​the target road surface overlaps with both sides of the road surface, the light blanket coverage area should cover at least both sides of the road surface in width.

[0014] A road condition warning system, comprising: The information acquisition module is used to collect information about the target road surface in front of the vehicle. The road surface perception module is used to determine the hazard level and abnormal conditions of the target road surface based on the target road surface information. The warning module determines projection information based on the target road surface hazard level and abnormal conditions. The projection information includes the light carpet coverage area and warning signs, and controls the vehicle lights to output warning signs and a warning light carpet consistent with the light carpet coverage area.

[0015] A computer program product includes a computer program / instructions that, when executed by a processor, implement the aforementioned road condition warning method.

[0016] The beneficial effects of this invention are as follows: 1. This invention employs a mean-based clustering algorithm to cluster multi-dimensional feature vectors, forming feature vector clusters representing friction coefficients, obstacle distances, visibility, road surface smoothness, etc., which are then weighted and fused using sensor weights. This method can eliminate outlier data, forming a consistent description of road surface conditions, and the system can maintain stable output even when a particular sensor is interfered with. The feature vectors obtained after clustering are used for correlation calculation, avoiding the ambiguity of hard decisions. For example, the correlation function calculates the initial correlation degree based on the classical domain, and then uses weighted summation to make risk assessment smoother and more accurate, especially suitable for complex scenarios where feature values ​​fall near boundaries. Through real-time clustering and weighting, the system can dynamically adapt to road surface changes, such as sudden obstacles or weather changes, ensuring timely and accurate warnings.

[0017] 2. Hazard levels are represented as combinations of classical domains (such as the classical domain of friction coefficient and the classical domain of obstacle distance), with no overlap between the classical domains of each level. The correlation function calculates the correlation between the target road surface characteristics and each level, taking the maximum value to determine the hazard level, thus avoiding the limitations of the traditional dichotomy method. The correction function integrates vehicle speed influence factors, weather influence factors, historical data comparison factors, and driver behavior factors, ensuring that risk assessment is based not only on real-time road data but also on the macro-environment and individual habits.

[0018] 3. Traditional warnings rely on dashboards or sounds, forcing drivers to concentrate. This invention uses a light carpet projection device (such as DLP technology) to project bright patterns (such as boundary lines and warning icons) directly onto the road surface, allowing drivers to intuitively understand the type, location, and extent of hazards without taking their eyes off the road. The system dynamically adjusts the shape of the light carpet according to the hazard level and situation. For example, when a hazard is detected on one side, the light carpet shifts towards the safe side; on curves or slopes, rotation and scaling are controlled by micromirrors. This upgrade from passive alarm to active guidance reduces the risk of secondary accidents.

[0019] 4. By fusing multi-source sensor data in real time, the system can accurately adjust the projection content according to the level of danger and abnormal conditions, avoiding false alarms or missed alarms. The light carpet projection directly presents the warning information on the road surface, allowing drivers to intuitively understand the danger without taking their eyes off the road, thus reducing distraction. Attached Figure Description

[0020] Figure 1 This is a system block diagram of the present invention.

[0021] Figure 2 This is the control logic diagram for the present invention. Detailed Implementation

[0022] To make the technical problems, technical solutions, and beneficial effects to be solved by this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and are not intended to limit the scope of this application.

[0023] Example 1 A road condition warning method, comprising: Collect information about the target road surface ahead of the vehicle; Based on the target road surface information, determine the target road surface hazard level and target road surface anomalies; Based on the target road surface hazard level and the abnormal conditions of the target road surface, projection information is determined, including the light blanket coverage area and warning signs; Control the vehicle lights to output warning signs and a warning light blanket that matches the area covered by the light blanket.

[0024] As a preferred implementation, the abnormal conditions of the target road surface are determined based on the target road surface information. These abnormal conditions include abnormal areas and abnormal types (such as water accumulation or icing). Existing road surface analysis technologies can be used to achieve this objective, such as Chinese patents CN119126141A (A method for detecting road surface and dirt conditions based on lidar) and CN118597144A (Vehicle control method, device, equipment, and storage medium), etc.

[0025] As a preferred embodiment, the specific method for determining the hazard level of the target road surface based on the target road surface information is as follows: The target road surface information is the current raw data sample from each heterogeneous sensor. The current raw data sample from each heterogeneous sensor is preprocessed, and the preprocessed current raw data sample is clustered to obtain all road surface features of the target road surface. Multiple hazard levels are preset, and each hazard level defines the corresponding classical domain of all pavement feature values. This classical domain is within the section domain of the corresponding pavement feature values. Based on the characteristic values ​​of all pavement features of the target pavement, the correlation between the target pavement and each hazard level is calculated using the pavement-hazard level correlation calculation model. The hazard level with the highest correlation with the target pavement is taken as the hazard level of the target pavement.

[0026] Heterogeneous sensors include the following types: Multispectral imaging sensor: Integrated inside the headlight, it captures images of the road ahead in different wavelengths (including visible and near-infrared). Different road surface conditions (such as dry asphalt, standing water, ice, potholes, and oil stains) exhibit significant differences in the absorption and reflection characteristics of specific wavelengths of light, providing the sensor with a basis for identification.

[0027] Short-range, high-precision millimeter-wave radar: Installed behind the front bumper, it is used to detect microscopic contour changes in the road surface ahead. By analyzing the scattering characteristics of the radar echo, it can effectively identify geometric anomalies such as potholes and uneven road surfaces, and assist in verifying the judgments of optical sensors.

[0028] Vehicle status and environment sensors: including vehicle speed sensor, windshield wiper status sensor, and ambient light sensor, provide information for judging road conditions and adjusting the brightness of the light blanket.

[0029] The specific method for preprocessing the current raw data samples of each heterogeneous sensor is as follows: A Gaussian filter is used to filter the raw data samples from each heterogeneous sensor to suppress noise. Features are then extracted from the filtered raw data samples according to the sensor category, resulting in feature vectors representing the raw data samples from each heterogeneous sensor. These feature vectors from all raw data samples constitute a multidimensional feature space with unified structure, dimensions, and spatiotemporal reference. Due to the different sources of the raw data samples, an example of the feature extraction method is provided below: 1) Feature extraction from millimeter-wave radar The raw data samples are point cloud data. Each data point may contain: distance, radial velocity, azimuth angle, signal-to-noise ratio, etc. It can be used to extract obstacle distances and traffic flow density, indirectly assisting in the judgment of road surface conditions.

[0030] For obstacle distance: Use DBSCAN (Density-Based Noise-Based Spatial Clustering) or Euclidean clustering algorithms to cluster the point cloud, grouping points belonging to the same object into one class. Track the clustered targets across multiple frames to estimate their absolute velocity and position. Extract the distance to the nearest obstacle within the lane. For example, select targets on the vehicle's trajectory that have a relatively slow speed, and calculate the average or minimum distance to them as the obstacle distance.

[0031] For traffic flow density: Count the number of all vehicles in the current lane and adjacent lanes within a certain distance (e.g., 150 meters) ahead. Divide the number of vehicles by the length of the monitoring range to obtain a line density, or use the number directly as a simplified indicator as traffic flow density.

[0032] 2) Feature extraction from the camera The raw data samples are video frames. They can be used to extract visibility, friction coefficient, obstacle distance, and traffic flow density.

[0033] Regarding visibility: Atmospheric scattering models suggest that lower visibility results in poorer image contrast and blurred details. Calculating the gray-level variance or information entropy of the entire image or sky region shows that variance and entropy decrease with low visibility. Clear images contain a large number of low-brightness pixels, while fog / haze images have higher overall dark channel values. Calculating the dark channel mean of an image can indirectly reflect visibility. This value, after calibration, can be converted into an approximate visible distance as visibility.

[0034] For friction coefficient: Images are segmented and classified using deep learning models (such as CNN convolutional neural networks). This identifies whether the road surface is wet, covered in snow, or icy. Wet roads have different reflective properties than dry roads. Snow and ice have unique textures and colors. The model outputs a classification confidence score or probability corresponding to dry, wet, snowy, or icy conditions. This category can be associated with an empirical friction coefficient range (e.g., dry: 0.8-1.0, wet: 0.5-0.7), thus obtaining an estimated friction coefficient.

[0035] For obstacle distance and traffic flow density: Algorithms such as YOLO (You Only See Once) and Faster R-CNN (a faster region-based convolutional neural network) are used to detect vehicles, pedestrians, etc. in images. If camera intrinsics and the actual size of the target (e.g., vehicle width) are known, its distance can be estimated from the pixel width of the target in the image. The distance to the nearest vehicle can be used as obstacle distance; the number of detected vehicles can be used as traffic flow density. By constructing a multi-dimensional feature space with unified structure, unified dimensions, and unified spatiotemporal benchmarks, the most fundamental challenge in heterogeneous sensor data fusion is effectively solved. It maps data from radar, cameras, and other sources with different physical properties, units, and sampling times into the same computable space, providing the possibility for subsequent clustering and weighted fusion, which is a prerequisite for the algorithm to work effectively.

[0036] 3) Feature extraction from embedded road condition sensors The raw data samples are time-series electrical signals (such as accelerometer, optical sensor, and triboelectric resistance measurements). The friction coefficient and road surface smoothness can be extracted.

[0037] For the coefficient of friction: Direct measurement: Some specialized sensors (such as optical sensors) estimate the coefficient of friction directly by emitting light of a specific wavelength and analyzing the reflection spectrum. Indirect estimation: The coefficient of friction is inferred by measuring tire-road noise or suspension vibration response and combining it with a model. The extracted feature is the direct reading after transformation by the physical model, which is used as the coefficient of friction.

[0038] For road surface smoothness: Analyze the signal from the vehicle's vertical accelerometer. Calculate the root mean square (RMS) value of the acceleration signal or the International Roughness Index (IRI). A larger RMS value indicates more severe vibration and a less smooth road surface. Map this value to a range of 1 (smooth) to 5 (severe potholes) to determine the road surface smoothness.

[0039] The specific method for clustering the preprocessed original data samples to obtain all road surface features of the target road surface is as follows: All feature vectors in the multidimensional feature space are clustered using the mean clustering algorithm to obtain feature vector clusters representing road surface features. These feature vector clusters include those representing friction coefficient, obstacle distance, visibility, road surface smoothness, and traffic flow density. The feature vectors in each feature vector cluster are weighted according to the weights of the corresponding heterogeneous sensors. These weights can be calculated by inputting the feature vectors of the original data samples from each heterogeneous sensor and the true road surface state value corresponding to the feature vector into the immune particle swarm optimization algorithm, resulting in feature vectors for friction coefficient, obstacle distance, visibility, road surface smoothness, and traffic flow density.

[0040] Using mean clustering followed by sensor weighting is an efficient feature-level fusion strategy.

[0041] Clustering algorithms can group multiple sensor readings representing the same road surface condition into one category, eliminate obvious abnormal readings, form a consistent description of the target road surface condition, and enhance the robustness of the system when some sensors are interfered with.

[0042] By explicitly weighting at the feature vector level, the optimal sensor weights obtained through optimization algorithms such as the immune particle swarm optimization can be accurately applied, maximizing the accuracy of data fusion and directly improving the reliability of subsequent hazard level assessments.

[0043] The specific method for calculating the correlation between the target pavement and each hazard level based on the characteristic values ​​of all pavement features and the pavement-hazard level correlation calculation model is as follows: The danger level is expressed as , Indicates the first Each hazard level is defined by a classical domain, which represents the range of characteristic values ​​for the corresponding road surface features within that hazard level. The classical domains for friction coefficient, obstacle distance, visibility, and road surface smoothness do not overlap for each hazard level. The calculation model for the correlation between road surface and hazard level includes a correlation function and a correction function; The characteristic values ​​of all pavement features of the target pavement are compared with the corresponding pavement features in each hazard level. Input the correlation function to obtain the correlation degree between all road features of the target road and each hazard level. For each hazard level, the correlation degree between each road feature of the target road and the hazard level is weighted and summed to obtain the initial correlation degree between the target road and the hazard level. Input the initial correlation degree between the target road and the hazard level into the correction function to obtain the correlation degree between the target road and the hazard level. Iterate through all hazard levels to obtain the correlation degree between the target road and each hazard level.

[0044] The specific association function is as follows: In the formula, It is the target road surface Characteristic values ​​of road surface features for Road surface features The lower limit value, for feature The upper limit, For the target road surface Features and the first kind The initial degree of correlation; In the formula, For the target road surface and the first The initial correlation of the risk levels, For the target road surface Total number of features For the target road surface Features and the first kind The association weight.

[0045] The correction function is as follows: In the formula, As a vehicle speed influencing factor, the impact of vehicle speed on the hazard level is considered. The higher the vehicle speed, the higher the hazard level should be for the same road conditions. Weather factors, such as rain, snow, and fog, can increase the danger level. This is a historical data comparison factor; if a dangerous incident has occurred on this road section in the past, the danger level is increased. Driver behavior factors are considered, taking into account the driver's individual driving habits and state; for example, aggressive driving or fatigued driving increases the risk. For the target road surface and the first The degree of correlation between different risk levels.

[0046] In the formula, This is the current vehicle speed. It is the base speed (e.g., road speed limit). It is the maximum permissible speed. It is the vehicle speed influence coefficient (determined based on experiments); In the formula, It is the first The influence coefficient of this type of weather It is the first A weather indicator function, if the current weather is the first type. This weather =1, otherwise =0; In the formula, It is the historical influence coefficient. It is the time decay coefficient. It is the base of the natural logarithm. It is the time elapsed since the historical event. It is a value indicating the severity of historical events. This is the maximum severity value; In the formula, It is the driver behavior influence coefficient. It is the probability of safe driving by the driver. As a preferred method, the probability of safe driving by the driver can be obtained based on a model related to driver behavior, such as patent CN116572984B: a dangerous driving control method and system based on multi-feature fusion.

[0047] The following is an example of this embodiment: Assuming the multi-source road surface sensing module has preprocessed and clustered the sensor data to obtain the feature vector of the target road surface: The coefficient of friction C1 = 0.4 (for example, a wet road surface reduces friction). The obstacle is 25 meters away from C2 (there is a vehicle ahead). Visibility C3 = 80 meters (light fog) Road surface smoothness C4=3.5 (with a few potholes) Meanwhile, assume the current context factor (used for the correction function): Vehicle speed v = 80 km / h (baseline vehicle speed v0 = 60 km / h, maximum permissible vehicle speed vmax = 120 km / h, vehicle speed influence coefficient ks = 0.5) Weather: Light rain (influence coefficient βrain = 1.2, other weather indicator functions are 0) Historical data: A minor accident occurred on this road section 1 hour ago (historical event severity Sh=0.6, maximum severity Smax=1.0, historical impact coefficient γ=0.8, time decay coefficient λ=0.1). Driver behavior: Safe driving probability Psafe=0.9 (driver behavior influence coefficient δ=0.5) The correlation between road surface and hazard level was calculated using the following model: P1 correlation ≈ 0.3 (the current feature value does not perfectly match the extreme classical domain of P1), P2 correlation ≈ 1.157 (highest), P3 correlation ≈ 0.8, P4 correlation ≈ 0.5, and P5 correlation ≈ 0.2.

[0048] Based on the principle of selecting the hazard level most relevant to the target road surface, the system determines the target road surface hazard level to be P2 (moderate hazard). This indicates that the road surface condition requires attention, such as wetness combined with obstacles and low visibility, but not to an emergency level. The warning module will then project a yellow border or exclamation mark pattern, depending on the P2 level, to remind the driver to slow down.

[0049] The correlation function can accurately measure the degree of conformity between the feature values ​​of the target pavement state and the classical domain of each hazard level. Its effect is to achieve quantitative and fine matching, rather than a simple hard decision that it belongs to a certain level if it is within a certain range. This makes the level determination results smoother and more accurate, and it can effectively handle the ambiguous cases where the feature values ​​fall near the boundaries of the classical domain.

[0050] The correction function enables dynamic and personalized hazard assessment, significantly enhancing the system's situational awareness capabilities. The vehicle speed impact factor reflects the dynamic nature of safety risks. The same road surface unevenness might only cause a bump at low speeds, but could lead to loss of control at high speeds. This correction factor links system assessment with real-time vehicle speed, making risk judgments more consistent with actual driving dynamics. The weather impact factor expands the system's environmental adaptability. A small puddle on a dry road is drastically different in danger from a small puddle on a rainy day. This factor incorporates macro-environmental conditions into micro-level assessments. The historical data comparison factor endows the system with memory and learning capabilities. By focusing on high-frequency accident sections through a time decay function, location-based warning enhancement is achieved. Even if the current sensors fail to fully capture the hazard, the system can raise the alert level in advance based on historical records, reflecting a proactive safety concept. The driver behavior factor enables personalized adaptation. For aggressive or fatigued drivers, the system adopts a more conservative warning strategy; conversely, a more lenient strategy can be used.

[0051] As a preferred implementation, the light carpet projection device specifically employs a DLP (Digital Light Processing) projection device, including a DMD (Digital Micromirror Device) chip, a light source, and an optical lens. Its high-resolution characteristics allow it to generate complex, high-precision warning symbols and seamlessly integrate them into the basic lighting light carpet.

[0052] The control logic for achieving boundary highlighting is as follows: The light carpet decision unit receives the coordinate data of the abnormal area sent by the sensor fusion unit, and generates a closed polygon vector graphic representing the boundary of the area through a contour extraction algorithm; then, the vector graphic is mapped to the pixel coordinate system of the DMD chip, and the pixels constituting the boundary line are assigned brightness parameters and dynamic flicker frequencies higher than those of the basic lighting light carpet; finally, a control command is generated to drive the corresponding micromirror array on the DMD chip to work in a pulse width modulation (PWM) mode with a higher duty cycle, thereby projecting a bright light band on the road surface that matches the contour of the abnormal area to achieve a warning effect.

[0053] The control logic for adjusting the shape and range of the light carpet is as follows: the light carpet decision unit receives lane line information, vehicle attitude information, and navigation information in real time, and calculates the target light shape parameters required for the current driving scenario accordingly; the light carpet signal generation unit converts the parameters into control signals for the micromirror group on the DMD chip, and achieves macroscopic adjustments such as translation, rotation, scaling, and local shading of the entire basic light carpet shape by controlling the opening and closing of specific micromirror arrays in a partitioned manner, so that it can dynamically fit the lane, adapt to curves and slopes, and avoid glare while ensuring optimal illumination.

[0054] As a preferred embodiment, the specific method for determining the projection information based on the target road surface hazard level and the target road surface anomaly is as follows: The abnormal conditions of the target road surface include the abnormal areas and the abnormal types of the target road surface; When the distance between the vehicle and the abnormal area of ​​the target road surface is the preset distance, a light strip is projected to surround the abnormal area of ​​the target road surface. The corresponding warning sign is selected from the warning sign library based on the type of abnormality of the target road surface (such as water accumulation, ice, etc.) and the danger level of the target road surface. The corresponding warning sign is projected to the designated position according to the abnormal area of ​​the target road surface, and the coverage area of ​​the light blanket is adjusted according to the abnormal area of ​​the target road surface.

[0055] As a preferred embodiment, the specific method for projecting corresponding warning signs onto designated locations within the light blanket based on abnormal areas of the target road surface is as follows: The road surface area is divided into two sides based on the position of the road centerline; When the abnormal area of ​​the target road surface falls entirely on one side of the road surface, the corresponding warning sign will be projected onto the other side of the road surface. When an abnormal area of ​​the target road surface falls on both sides of the road, corresponding warning signs are projected onto both sides of the road surface.

[0056] When the abnormal area of ​​the target road surface is a road surface on both sides, the width of the corresponding pattern is increased by a preset multiple.

[0057] As a preferred embodiment, the specific method for adjusting the projection light blanket coverage area according to the abnormal areas of the target road surface is as follows: When the entire abnormal area of ​​the target road surface falls on one side of the road surface, the light blanket coverage area shall cover at least that side of the road surface in width; When the abnormal area of ​​the target road surface overlaps with both sides of the road surface, the light blanket coverage area should cover at least both sides of the road surface in width.

[0058] For example, depending on the level of danger, such as Figure 2As shown, the system determines whether the road surface ahead is in an abnormal condition. For example, levels four and five are considered normal road conditions, while levels one, two, and three are considered abnormal road conditions. Therefore, there are two warning modes: Mode 1: Dynamic Warning Mode. Triggering condition: The sensor fusion unit detects an abnormal road surface ahead.

[0059] Execution process: 1) Pattern Projection: Based on the identified road surface condition type, the light carpet decision unit selects the corresponding warning icon from the built-in symbol library (e.g., water ripples represent standing water, exclamation marks represent ice or oil stains, and triangles represent potholes or road stones). The projection device is then controlled to highlight the light carpet pattern in a designated location near the abnormal area, transforming it into a bright warning icon.

[0060] 2) Highlight the boundary: At the same time, use a bright light strip to clearly mark the boundary outline of the abnormal area, so that the driver can intuitively judge the danger range.

[0061] 3) Adaptive adjustment of the light blanket: The base light blanket is adjusted synchronously according to the type and level of hazard. For example: If water accumulation / potholes are detected on one side: the light carpet will automatically shift slightly to the safe side to guide the driver to avoid the impact.

[0062] Large area of ​​ice detected ahead: While issuing a warning, the light carpet is slightly widened to provide a wider field of vision and the overall illuminance is increased to enhance road surface brightness and assist the driver in observation.

[0063] 4) Brightness adjustment: The brightness of the warning symbol can be adaptively adjusted according to the level of danger and the ambient light, ensuring that the warning effect is eye-catching but not dazzling.

[0064] Mode 2: Standard Adaptive Mode. Triggering condition: No abnormal road surface condition detected.

[0065] Execution process: The system executes the standard adaptive high beam function, adjusting the shape and range of the light carpet according to lane markings, oncoming vehicles, and other conditions to provide optimal basic lighting.

[0066] By utilizing a unified platform of high-resolution DLP hardware and different software control logics, both pixel-level precise drawing in Mode 1 (boundary highlighting, warning symbols) and macroscopic adjustment of vector light patterns in Mode 2 (lane alignment, curve rotation, etc.) were achieved.

[0067] Compared to traditional audible warnings or dashboard icons, projecting warning symbols (such as highlighted borders and exclamation mark icons) directly onto hazardous areas of the actual road surface achieves an augmented reality (AR) effect. Drivers can intuitively understand the type, location, and extent of hazards without taking their eyes off the road, significantly reducing information processing time and minimizing secondary risks caused by distraction.

[0068] By deeply integrating the warning function with the vehicle's lighting system (such as adaptive headlights), functional synergy is achieved. The light carpet provides both basic illumination and dynamically overlays warning information, saving hardware costs, eliminating the need for additional display devices inside the vehicle, and making the system more integrated.

[0069] The light carpet can dynamically adjust its shape, for example, shifting to the safe side when a danger is detected on one side. This not only warns of danger but also actively provides guidance for a safe path, achieving a functional upgrade from passive alarm to active guidance.

[0070] Example 2 A road condition warning system, such as Figure 1 As shown, it includes: The information acquisition module is used to collect information about the target road surface in front of the vehicle. The road surface perception module is used to determine the hazard level and abnormal conditions of the target road surface based on the target road surface information. The warning module determines projection information based on the target road surface hazard level and abnormal conditions. The projection information includes the light carpet coverage area and warning signs, and controls the vehicle lights to output warning signs and a warning light carpet consistent with the light carpet coverage area.

[0071] Example 3 A computer program product includes a computer program / instructions that, when executed by a processor, implement the road condition warning method in Embodiment 1.

[0072] The contents not described in detail in this specification are prior art known to those skilled in the art. Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0073] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0074] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0075] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0076] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit its scope of protection. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that after reading the present invention, they can still make various changes, modifications or equivalent substitutions to the specific implementation of the invention, but these changes, modifications or equivalent substitutions are all within the scope of protection of the pending claims of the invention.

Claims

1. A method for warning road conditions, characterized in that, include: Collect information about the target road surface ahead of the vehicle; Based on the target road surface information, determine the target road surface hazard level and target road surface anomalies; Based on the target road surface hazard level and the abnormal conditions of the target road surface, projection information is determined, including the light blanket coverage area and warning signs; Control the vehicle lights to output warning signs and a warning light blanket that matches the area covered by the light blanket.

2. The road condition warning method according to claim 1, characterized in that, The specific method for determining the hazard level of the target road surface based on the target road surface information is as follows: The target road surface information is the current raw data sample from each heterogeneous sensor. The current raw data sample from each heterogeneous sensor is preprocessed, and the preprocessed current raw data sample is clustered to obtain all road surface features of the target road surface. Multiple hazard levels are preset, and each hazard level defines the corresponding classical domain of all pavement feature values. This classical domain is within the section domain of the corresponding pavement feature values. Based on the characteristic values ​​of all pavement features of the target pavement, the correlation between the target pavement and each hazard level is calculated using the pavement-hazard level correlation calculation model. The hazard level with the highest correlation with the target pavement is taken as the hazard level of the target pavement.

3. The road condition warning method according to claim 2, characterized in that, Based on the feature values ​​of all pavement characteristics of the target pavement, the specific method for calculating the correlation between the target pavement and each hazard level based on the pavement-hazard level correlation calculation model is as follows: The danger level is expressed as , Indicates the first There are three hazard levels, and the classical domains of friction coefficient, obstacle distance, visibility, and road surface smoothness for each hazard level do not overlap. The calculation model for the correlation between road surface and hazard level includes a correlation function and a correction function; The characteristic values ​​of all pavement features of the target pavement are compared with the corresponding pavement features in each hazard level. Input the correlation function to obtain the correlation degree between all road features of the target road and each hazard level. For each hazard level, the correlation degree between each road feature of the target road and the hazard level is weighted and summed to obtain the initial correlation degree between the target road and the hazard level. Input the initial correlation degree between the target road and the hazard level into the correction function to obtain the correlation degree between the target road and the hazard level. Iterate through all hazard levels to obtain the correlation degree between the target road and each hazard level.

4. The road condition warning method according to claim 3, characterized in that, The specific association function is as follows: In the formula, It is the target road surface Characteristic values ​​of road surface features for Road surface features The lower limit value, for feature The upper limit, For the target road surface Features and the first kind The initial degree of correlation; In the formula, For the target road surface and the first The initial correlation of the risk levels, For the target road surface Total number of features For the target road surface Features and the first kind The association weight.

5. The road condition warning method according to claim 4, characterized in that, The correction function is as follows: In the formula, For vehicle speed influencing factors, Weather influencing factors As a historical data comparison factor, For driver behavior factors, For the target road surface and the first The degree of correlation between different risk levels; In the formula, This is the current vehicle speed. That is the base speed. It is the maximum permissible speed. It is the vehicle speed influence coefficient; In the formula, It is the first The influence coefficient of this type of weather It is the first A weather indicator function, if the current weather is the first type. This weather =1, otherwise =0; In the formula, It is the historical influence coefficient. It is the time decay coefficient. It is the base of the natural logarithm. It is the time elapsed since the historical event. It is a value indicating the severity of historical events. This is the maximum severity value; In the formula, It is the driver behavior influence coefficient. It is the probability of a driver driving safely.

6. The road condition warning method according to claim 1, characterized in that, The specific method for determining the projection information based on the target road surface hazard level and the target road surface anomaly is as follows: The abnormal conditions of the target road surface include the abnormal areas and the abnormal types of the target road surface; When the distance between the vehicle and the abnormal area of ​​the target road surface is a preset distance, a light strip is projected to surround the abnormal area of ​​the target road surface. The corresponding warning sign is selected from the warning sign library based on the type of abnormality and the hazard level of the target road surface. The corresponding warning sign is projected to the designated position according to the abnormal area of ​​the target road surface, and the coverage area of ​​the light blanket is adjusted according to the abnormal area of ​​the target road surface.

7. The road condition warning method according to claim 6, characterized in that, The specific method for projecting corresponding warning signs onto designated locations within the light carpet based on abnormal areas of the target road surface is as follows: The road surface area is divided into two sides based on the position of the road centerline; When the abnormal area of ​​the target road surface falls entirely on one side of the road surface, the corresponding warning sign will be projected onto the other side of the road surface. When an abnormal area of ​​the target road surface falls on both sides of the road surface, corresponding warning signs are projected on both sides of the road surface. When the abnormal area of ​​the target road surface is a road surface on both sides, the width of the corresponding pattern is increased by a preset multiple.

8. The road condition warning method according to claim 6, characterized in that, The specific method for adjusting the coverage area of ​​the projected light carpet according to the abnormal areas of the target road surface is as follows: When the entire abnormal area of ​​the target road surface falls on one side of the road surface, the light blanket coverage area shall cover at least that side of the road surface in width; When the abnormal area of ​​the target road surface overlaps with both sides of the road surface, the light blanket coverage area should cover at least both sides of the road surface in width.

9. A road condition warning system, characterized in that, include: The information acquisition module is used to collect information about the target road surface in front of the vehicle. The road surface perception module is used to determine the hazard level and abnormal conditions of the target road surface based on the target road surface information. The warning module determines projection information based on the target road surface hazard level and abnormal conditions. The projection information includes the light carpet coverage area and warning signs, and controls the vehicle lights to output warning signs and a warning light carpet consistent with the light carpet coverage area.

10. A computer program product, comprising a computer program / instructions, characterized in that, When the computer program / instruction is executed by the processor, it implements the road condition warning method according to any one of claims 1-8.

Images