Driving risk dynamic early warning method and system based on human factor and road condition fusion

Abstract

The invention provides a driving risk dynamic early warning method and system based on human factor and road condition fusion, and relates to the technical field of driving safety, first, dynamic state data of a driver and an escort and road real-time state data in the driving process are collected, and a human factor state dynamic evolution sequence and a road condition real-time change sequence are generated; analyzing the interactive influence relationship between the two to establish a human factor road condition two-way correlation model; based on the human factor road condition bidirectional correlation model, risk triggering conditions are extracted to identify a driving risk conduction path set; and performing feature extraction on the driving risk conduction path set to determine a risk influence range, thereby generating a driving risk dynamic early warning instruction and sending the instruction to the driving control terminal to trigger an early warning prompt, thereby effectively and accurately early warning the driving risk, and improving the driving safety.

Description

Technical Field

[0001] This invention relates to the field of driving safety technology, and more specifically, to a dynamic early warning method and system for driving risks based on the fusion of human factors and road conditions. Background Technology

[0002] In the field of driving safety, ensuring the safety of vehicles during operation is paramount. Currently, most existing driving risk warning methods focus solely on monitoring and analyzing road conditions, such as using sensors to obtain information like traffic flow and road surface smoothness to determine the presence of potential risks. However, these warning methods, which rely solely on road conditions, neglect the significant impact of factors such as drivers and security personnel on driving safety. Driver fatigue, concentration, emotional state, and the cooperation of security personnel can all lead to different driving risks under varying road conditions, but current technology fails to incorporate this crucial human factor information into its warning system.

[0003] On the other hand, while some methods consider human factors, they often analyze the driver's or escort's condition in isolation, failing to organically combine human factors with road conditions. Human condition and road conditions are interconnected and mutually influential; different human conditions will produce different risk outcomes under different road conditions. However, existing technologies lack effective analysis and modeling of these complex interactions, making it impossible to accurately identify the transmission path and scope of driving risks. Consequently, it is difficult to provide accurate and effective driving risk warnings, posing potential hazards to driving safety. Summary of the Invention

[0004] In view of the aforementioned problems, and in conjunction with the first aspect of the present invention, embodiments of the present invention provide a method for dynamic early warning of driving risks based on the fusion of human factors and road conditions, the method comprising: The system collects dynamic status data of drivers, escorts, and real-time road status data during the driving process. Based on the collected dynamic status data of drivers and escorts, it generates a sequence of dynamic evolution of human factors and a sequence of real-time changes in road conditions. The interaction between each state node in the dynamic evolution sequence of human factors and each road condition node in the real-time change sequence of road conditions is analyzed, and a two-way correlation model between human factors and road conditions is established. Based on the aforementioned bidirectional correlation model between human factors and road conditions, risk triggering conditions under different combinations of state nodes and road condition nodes are extracted, and a set of driving risk transmission paths is identified. For each driving risk transmission path in the set of driving risk transmission paths, path features are extracted, and the risk impact range corresponding to each driving risk transmission path is determined based on the extracted path features; Based on the risk impact range corresponding to each driving risk transmission path, a dynamic driving risk warning instruction is generated, which includes a description of the risk transmission direction and the risk impact range. The dynamic driving risk warning instruction is then sent to the driving control terminal, triggering the driving control terminal to perform a warning prompt operation.

[0005] Furthermore, embodiments of the present invention also provide a dynamic early warning system for driving risks based on the fusion of human factors and road conditions, characterized in that it includes: A processor; a machine-readable storage medium for storing machine-executable instructions of the processor; wherein the processor is configured to execute the above-described dynamic warning method for driving risks based on the fusion of human factors and road conditions by executing the machine-executable instructions.

[0006] In another aspect, embodiments of the present invention also provide a computer program product, the computer program product including machine-executable instructions, the machine-executable instructions being stored in a computer-readable storage medium, the processor of the dynamic warning system for driving risks based on the fusion of human factors and road conditions reading the machine-executable instructions from the computer-readable storage medium, the processor executing the machine-executable instructions, causing the dynamic warning system for driving risks based on the fusion of human factors and road conditions to execute the aforementioned dynamic warning method for driving risks based on the fusion of human factors and road conditions.

[0007] Based on the above, by collecting dynamic status data of drivers and escorts during driving, as well as real-time road status data, dynamic evolution sequences of human factors and real-time change sequences of road conditions are generated, presenting the changing trends of human factors and road conditions over time. By analyzing the interaction between the two and establishing a two-way correlation model between human factors and road conditions, risk triggering conditions are extracted and a set of driving risk transmission paths is identified. Path features of driving risk transmission paths are extracted, and the scope of risk impact is determined, making risk assessment more detailed and comprehensive. The final generated dynamic driving risk warning command, containing descriptions of the risk transmission direction and scope of impact, can send warning information to the driving control terminal in a timely and accurate manner, triggering warning prompts and providing drivers with effective risk warnings, thereby effectively reducing driving risks. Attached Figure Description

[0008] Figure 1 This is a schematic diagram of the execution flow of the dynamic early warning method for driving risks based on the fusion of human factors and road conditions provided in an embodiment of the present invention.

[0009] Figure 2 This is a schematic diagram of exemplary hardware and software components of a dynamic early warning system for driving risks based on the fusion of human factors and road conditions provided in an embodiment of the present invention. Detailed Implementation

[0010] The present invention will now be described in detail with reference to the accompanying drawings. Figure 1 This is a flowchart illustrating a dynamic early warning method for driving risks based on the fusion of human factors and road conditions, provided in one embodiment of the present invention. The following is a detailed description of this dynamic early warning method for driving risks based on the fusion of human factors and road conditions.

[0011] Step S110: Collect dynamic status data of the driver, the escort, and the real-time status data of the road during the driving process. Generate a dynamic evolution sequence of human factors based on the collected dynamic status data of the driver and the escort, and generate a real-time change sequence of road conditions based on the collected real-time status data of the road.

[0012] This embodiment uses a long-haul freight vehicle traveling on a highway at night as an example. In this scenario, the vehicle needs to maintain high speed for an extended period, which can easily lead to driver fatigue and other changes in condition. The interaction between the escort and the driver, as well as real-time changes in road conditions, can all affect driving safety. Therefore, it is necessary to collect relevant data for risk warning.

[0013] Step S111: Using an image acquisition device installed in the driver's seat, collect dynamic facial image data of the driver at preset time intervals. Extract the driver's eye state data and facial expression data from the collected dynamic facial image data. The eye state data records the driver's eyelid closure frequency and eyeball rotation trajectory, while the facial expression data records the driver's mouth corner change range and facial muscle contraction state.

[0014] For example, a high-definition image acquisition device is installed above the dashboard in front of the driver's seat in a long-haul freight vehicle. This device uses a wide-angle lens to capture the driver's entire face. A frame is captured every 0.5 seconds to ensure subtle facial dynamics are captured. During nighttime driving, the image acquisition device automatically activates infrared illumination to ensure clear facial images even in low-light conditions. After acquiring the facial dynamic image data, image preprocessing techniques are used to denoise, convert to grayscale, and locate the face region. For eye state data extraction, a feature point detection method is used to further determine the position of the eyes within the located face region. Eyelid closure frequency is calculated by continuously analyzing the degree of eyelid opening and closing in multiple frames. When the degree of eyelid opening and closing is below a preset threshold, it is considered a single closure. The number of closures per unit time is the eyelid closure frequency. The extraction of the eyeball rotation trajectory is achieved by tracking the positional changes of the pupil center in the image. The coordinates of the pupil center in each frame are connected sequentially over time to form the eyeball rotation trajectory. For facial expression data, feature point detection is also used to extract feature points of the corners of the mouth and key facial muscle groups. The amplitude of changes in the corners of the mouth is determined by calculating the displacement of the feature points in the vertical and horizontal directions. The state of facial muscle contraction is determined by analyzing the texture changes and grayscale distribution of the facial muscle areas. For example, the grayscale value of the glabella muscle area changes when frowning. The state of facial muscle contraction is recorded by detecting these changes.

[0015] Step S112: By using a physiological sensing device worn by the driver, the driver's physiological dynamic data is collected synchronously. The physiological dynamic data records the driver's heart rate change data, respiratory rate data, and skin conductance response data. The collected driver's eye state data, facial expression data, and physiological dynamic data are integrated to obtain the driver's dynamic state data.

[0016] The driver's physiological sensing devices include a heart rate sensor integrated into the steering wheel grip, a respiratory rate sensor mounted on the seat back, and a skin conductance sensor worn on the wrist. These sensors are wirelessly connected to the vehicle's central data processing unit via Bluetooth to achieve synchronized data transmission. The sampling frequency is consistent with the image acquisition equipment, once every 0.5 seconds, to ensure data time synchronization. The heart rate sensor uses photoelectric sensing technology to acquire heart rate data by detecting changes in the absorption of light by hemoglobin in the blood. The heart rate change data records the interval between each heartbeat and the trend of changes in the number of heartbeats per unit time. The respiratory rate sensor detects the driver's breathing movements by sensing changes in the pressure of the seat back. When the driver inhales, the body leans back slightly, increasing the pressure on the seat back; when exhaling, the pressure decreases. The respiratory rate is calculated by capturing the period of these pressure changes, and changes in breathing depth are also recorded; for example, a larger pressure change indicates a deeper breath. The skin conductance sensor reflects a driver's emotional changes by measuring changes in the skin's surface resistance. When a driver is in a state of tension or anxiety, sweat gland secretion increases, leading to a decrease in skin resistance. The sensor converts this resistance change into an electrical signal for recording. During the data integration phase, eye state data, facial expression data, and physiological dynamic data are aligned according to timestamps to ensure that data from the same point in time correspond. Then, feature extraction and standardization are performed on various data types. Eyelid closure frequency, eye movement trajectory characteristics, mouth corner changes, facial muscle contraction characteristics, heart rate changes, respiratory rate characteristics, and skin conductance characteristics are combined to form the driver's dynamic state data, with each dimension corresponding to a specific feature parameter.

[0017] Step S113: Collect the escort's voice interaction data and action posture data using the audio acquisition equipment and image acquisition equipment installed in the carriage. The voice interaction data records the content of the escort's voice communication with the driver and the changes in voice tone. The action posture data records the escort's body movement range and direction. Integrate the collected escort's voice interaction data and action posture data to obtain the escort's dynamic state data.

[0018] Inside the long-haul freight vehicle, a 360-degree rotating image acquisition device is installed on the ceiling near the escort's seat. This device covers the entire area where the escort is moving. Simultaneously, two high-sensitivity audio acquisition devices are installed in the compartment, one near the driver's seat and the other near the escort's seat, to clearly capture the voice interaction between the two. The audio acquisition devices use noise-canceling microphones to effectively filter out environmental noise such as engine noise and wind noise during vehicle operation. Voice interaction data is continuously acquired at a sampling frequency of 16kHz and a quantization bit depth of 16 bits to ensure the quality of the voice signal. For recording the content of the voice communication, speech recognition technology converts the acquired voice signal into text information, while also recording information such as the duration of the voice signal and pause intervals. Recording changes in voice intonation is achieved by analyzing features such as the frequency, amplitude, and speech rate of the voice signal. For example, when the escort is emotionally agitated, the frequency and amplitude of the voice increase, and the speech rate accelerates. The acquisition of motion and posture data is achieved through image acquisition equipment, employing a feature point detection method similar to that used for driver facial image acquisition to extract key feature points of the escort's limbs, such as the head, shoulders, elbows, wrists, hips, knees, and ankles. The amplitude of limb movements is determined by calculating the distance change between limb feature points in adjacent frames; for example, when the arm swings, the greater the displacement of the elbow feature point, the greater the amplitude of the limb movement. The direction of the movement is determined by calculating the angle between the vector formed by the limb feature points and the vehicle coordinate system; for example, when the escort extends their arm pointing forward, the angle between the vector formed by the arm feature points and the vehicle's direction of travel is smaller. During data integration, the voice interaction data and motion and posture data are aligned according to timestamps. Keywords and intonation features are extracted from the voice interaction data, and motion amplitude and direction features are extracted from the motion and posture data. These features are then combined to form the escort's dynamic state data, including feature parameters from both voice and movement aspects.

[0019] Step S114: The collected dynamic status data of the driver is segmented according to the time sequence. Each segment of data corresponds to a time window. The core features of the dynamic status data of the driver within each time window are extracted to form a driver status node. The collected dynamic status data of the escort is segmented according to the same time window. The core features of the dynamic status data of the escort within each time window are extracted to form an escort status node.

[0020] Step S1141: Determine the duration of the time window. The duration of the time window is determined based on the vehicle's speed. The faster the speed, the shorter the duration of the time window; the slower the speed, the longer the duration of the time window.

[0021] The vehicle's speed is acquired via the vehicle's CAN bus and transmitted in real time to the central data processing unit. The time window duration is determined using a dynamic adjustment mechanism. A base time window duration of 5 minutes is set. When the vehicle speed exceeds 100 km / h, the time window duration is shortened to 3 minutes; when the speed is between 60-100 km / h, the base duration of 5 minutes is maintained; and when the speed is below 60 km / h, the time window duration is extended to 8 minutes. This is because at high speeds, road conditions change rapidly, requiring more frequent analysis of driver status changes, while at low speeds, status changes are relatively slow, allowing for a longer time window and reduced data processing load. For example, in the highway nighttime driving scenario of this embodiment, the vehicle speed is typically maintained between 90-110 km / h, therefore the time window duration will be dynamically adjusted between 3-5 minutes.

[0022] Step S1142: Starting from the start time of data collection, the collected driver dynamic status data is segmented according to the determined time window duration to obtain multiple segments of driver dynamic status data. Each segment of driver dynamic status data corresponds to a time window identifier.

[0023] Starting from the data collection start time, time periods are sequentially divided according to a defined time window duration. Each time window corresponds to a unique time window identifier, which includes start and end time information. For example, the start time of the first time window is the time when data collection begins after the vehicle starts, and the end time is the start time plus the time window duration. The start time of the second time window is the end time of the first time window, and so on. During data segmentation, the central data processing unit allocates the driver's dynamic status data to the corresponding time windows based on timestamps, ensuring that the data within each time window is continuous and complete. For data at the boundaries of time windows, if the timestamp of a data point happens to fall at the intersection of two time windows, it is assigned to the latter time window.

[0024] Step S1143: Perform statistical analysis on the eye state data in each segment of driver dynamic state data, calculate the core features of the driver's dynamic state data within the time window, and form driver state nodes.

[0025] Step S11431: Perform statistical analysis on the eye state data in each segment of driver dynamic state data, calculate the average value of eyelid closure frequency and the coverage of eyeball rotation trajectory within the time window, and use the average value and coverage as the core features of eye state.

[0026] Within each time window, the frequency of eyelid closure is calculated by summing all the eyelid closure counts within that window and then dividing by the window duration to obtain the average eyelid closure frequency. For example, if 15 eyelid closures are detected within a 5-minute time window, the average eyelid closure frequency is 3 times per minute. The coverage area of ​​the eye movement trajectory is determined by calculating the maximum displacement of the eye movement trajectory in the horizontal and vertical directions. Multiplying the maximum horizontal and vertical displacements yields an area value, which represents the coverage area of ​​the eye movement trajectory. A larger area value indicates a wider range of eye movement.

[0027] Step S11432: Analyze the facial expression data in the segmented data of the driver's dynamic state, extract the maximum and minimum values ​​of the change in the corner of the mouth and the duration of the facial muscle contraction state, and use the maximum, minimum and duration as the core features of facial expression.

[0028] For the amplitude of mouth corner changes, the displacement of the mouth corner feature points is calculated frame by frame within the time window, and then the maximum and minimum values ​​are found. The maximum amplitude of mouth corner change represents the maximum degree of upward or downward movement of the mouth corner within the time window, and the minimum value represents the maximum degree of movement in the opposite direction. The duration of facial muscle contraction is calculated by detecting the time interval from the occurrence to the end of the facial muscle contraction. When the facial muscle contraction feature exceeds a preset threshold, timing begins and stops until the feature value returns to below the threshold. This time is the duration of the facial muscle contraction. If multiple muscle contractions occur within the time window, the durations of each contraction are added together to obtain the total duration.

[0029] Step S11433: Statistically analyze the physiological dynamic data in the segmented data of the driver's dynamic state, calculate the fluctuation amplitude of heart rate change data, the average value of respiratory rate data, and the peak value of skin conductance response data, and use the fluctuation amplitude, average value and peak value as the core features of physiological dynamics.

[0030] The fluctuation range of heart rate data is calculated by measuring the standard deviation of the heart rate data. The larger the standard deviation, the greater the heart rate fluctuation, and the more unstable the driver's physiological state may be. The average value of respiratory rate data is obtained by adding up all respiratory rate values ​​within the time window and then dividing by the number of data points. The peak value of skin conductance response data is obtained by finding the maximum amplitude of the skin conductance response signal within the time window. This peak value reflects the maximum level of emotional stimulation experienced by the driver within that time window.

[0031] Step S11434: Standardize the core features of eye state, facial expression, and physiological dynamics respectively. Integrate the standardized core features of eye state, facial expression, and physiological dynamics into a feature vector to form the driver state node corresponding to the time window. The driver state node includes the time window identifier and the corresponding feature vector.

[0032] The standardization process employs the Z-score standardization method. For each core feature, its value is subtracted from the historical average and then divided by the standard deviation, ensuring that the standardized feature values ​​follow a normal distribution with a mean of 0 and a standard deviation of 1. For example, for the average eyelid closure frequency, assuming a historical average of 2 times per minute and a standard deviation of 1, and an average of 3 times per minute within the current time window, the standardized value is (3-2) / 1 = 1. After standardizing all core features, the core features of eye state (including the average eyelid closure frequency and the coverage of eye movement trajectory), facial expression (including the maximum and minimum values ​​of mouth corner changes and the duration of facial muscle contraction), and physiological dynamics (including the fluctuation range of heart rate data, the average respiratory rate data, and the peak value of skin conductance response data) are arranged sequentially to form an 8-dimensional feature vector. This feature vector is then associated with the corresponding time window identifier to form the driver state node.

[0033] Step S1144: Using the same time window duration, the collected dynamic status data of the escort personnel is segmented to obtain multiple segments of escort personnel dynamic status data. Each segment of escort personnel dynamic status data corresponds to the same time window identifier as the driver status node.

[0034] Similar to the segmentation method for driver dynamic status data, the security guard's dynamic status data is segmented according to a defined time window duration, starting from the data collection start time. Since the security guard's dynamic status data is collected synchronously with the driver's, the same time window duration and time window identifier are used to ensure accurate association between driver and security guard status nodes within the same time window. For example, a driver status node with a time window identifier of 19:00-19:05 corresponds to a security guard dynamic status segment with the same time window identifier.

[0035] Step S1145: Analyze the voice interaction data in each segment of the escort's dynamic status data, extract the number of keywords and the number of tone changes in the voice communication content, and use the number of keywords and the number of tone changes as the core features of voice interaction.

[0036] In the analysis of voice interaction data, the first step is to perform speech recognition on the voice signal and convert it into text information. Then, keyword matching technology is used to search for keywords related to driving safety in the text information, such as "caution," "slow down," "obstacles," and "fatigue." The total number of these keywords is counted as a keyword quantity feature. The number of changes in voice intonation is determined by analyzing the frequency and amplitude changes of the voice signal. When the change in frequency or amplitude exceeds a preset threshold, it is considered a tone change, and the total number of tone changes within a time window is counted. For example, in the sentence "There is construction ahead, please slow down," the keywords "caution" and "slow down" will be counted in the keyword count. Additionally, because the intonation of this sentence may be higher than in normal communication, it will be considered a tone change.

[0037] Step S1146: Analyze the action posture data in the segmented dynamic state data of the escort personnel, calculate the average value of the limb movement amplitude and the angle between the direction of the movement and the direction of vehicle travel, and use the average value and the angle as the core features of the action posture.

[0038] The amplitude of limb movements is determined by calculating the average displacement of the guard's limb feature points within a time window. The displacement of each limb feature point in each frame is summed, then divided by the total number of frames to obtain the average displacement of that feature point. The maximum average displacement of all limb feature points is then taken as the average amplitude of the limb movements. The direction of the movement is determined by comparing the direction vector of the limb movement with the vehicle's driving direction vector. The vehicle's driving direction vector is a preset reference direction (e.g., 0 degrees directly in front of the vehicle). The angle between the direction of the limb movement and the vehicle's driving direction is determined by calculating the angle between the limb movement direction vector and the reference direction. For example, when the guard extends his arm pointing to the right side of the vehicle, the angle between the direction of the movement and the vehicle's driving direction is 90 degrees.

[0039] Step S1147: Standardize the core features of voice interaction and the core features of action posture respectively, and integrate the standardized core features of voice interaction and the core features of action posture into a feature vector to form the escort status node corresponding to the time window. The escort status node contains the same time window identifier and corresponding feature vector as the driver status node.

[0040] Similarly, the Z-score normalization method is used to normalize the core features of voice interaction (number of keywords and number of tone changes) and the core features of movement posture (average amplitude of limb movements and angle between the direction of movement and the vehicle's direction of travel). After normalization, these four features are arranged in order to form a 4-dimensional feature vector. This feature vector is associated with the corresponding time window identifier to form the escort officer's status node, and this time window identifier is completely consistent with the time window identifier of the driver's status node within the same time window.

[0041] Step S115: Associate and bind the driver status node and the escort status node corresponding to the same time window to form a human factor status node. Connect all human factor status nodes in sequence according to the time window to generate a dynamic evolution sequence of human factor status.

[0042] Within each time window, driver and escort status nodes with the same time window identifier are associated and bound. This association involves concatenating the feature vectors of the two nodes to form a new 12-dimensional feature vector (8-dimensional driver status feature vector + 4-dimensional escort status feature vector). This new feature vector represents the core data of the human factors status node corresponding to that time window. Simultaneously, the human factors status node also contains the time window identifier and the original feature information of the driver and escort status nodes, allowing for tracing the specific source of the human factors data during subsequent analysis. All human factors status nodes are arranged sequentially according to the time window order. Each node establishes a temporal connection with the preceding and following nodes through its time window identifier, thereby generating a dynamic evolution sequence of human factors status. This dynamic evolution sequence reflects the changes in the driver and escort status over time throughout the entire driving process. For example, in the nighttime driving scenario of a long-haul freight vehicle in this embodiment, from the start of the vehicle, the first time window of the human factor state node reflects the driver's awake state and the escort's preparation state in the initial stage. As the driving time increases, subsequent human factor state nodes will gradually reflect the driver's possible signs of fatigue and the escort's corresponding reminder actions and other state changes.

[0043] Step S116: Collect dynamic image data of the road surface using a road image acquisition device installed outside the vehicle. Extract road surface cover status data and road surface damage status data from the dynamic image data of the road surface. The cover status data records the type and coverage of the road surface cover, and the road surface damage status data records the distribution of road surface cracks and the location of potholes.

[0044] Road image acquisition devices are installed at the front bumper and rearview mirrors of the vehicle. The front device primarily captures images of the road surface directly in front of the vehicle, while the rearview mirror device captures images of the road surface on both sides of the vehicle. Working together, these three devices comprehensively capture the road surface conditions along the vehicle's path. The image acquisition frequency is 10 frames per second, with a resolution of 1920×1080 pixels to ensure clear capture of detailed road surface features. The acquired dynamic road surface image data is first stitched together, fusing the images from different devices into a complete panoramic image of the road surface. Then, image segmentation technology is used to separate the road surface area from the background. For the extraction of road surface cover status data, a deep learning-based image recognition model is used. This model, trained on a large number of images of different types of cover, can identify common road surface cover types such as water accumulation, snow accumulation, oil stains, and fallen leaves. After identifying the cover type, the coverage area is determined by calculating the pixel ratio of the cover in the road surface area; a larger pixel ratio indicates a wider coverage area. For the data on road surface damage, a deep learning image recognition model was also used. This model was specifically trained for road damage types such as cracks and potholes. The distribution of cracks was represented by identifying the coordinate positions of crack pixels and connecting them into lines, while also recording the length and width of the cracks. The location of potholes was determined by locating the center coordinates of the pothole area and recording the size of the pothole area.

[0045] Step S117: Collect dynamic environmental data around the road using environmental sensing devices installed on the vehicle. The dynamic environmental data around the road includes the movement trajectory of obstacles around the road, changes in light intensity around the road, and precipitation around the road.

[0046] The environmental sensing equipment installed on the vehicle includes lidar, millimeter-wave radar, light sensors, and precipitation sensors. Lidar and millimeter-wave radar are installed at the front and four corners of the vehicle to detect obstacles around the road. Lidar provides high-precision 3D point cloud data; by analyzing continuous frames of point cloud data, it can track changes in obstacle position and thus obtain obstacle movement trajectories. Millimeter-wave radar has strong anti-interference capabilities and can effectively detect obstacles even in adverse weather conditions. Data fusion of the two improves the accuracy and reliability of obstacle detection. Light changes around the road are collected by a light sensor installed on the vehicle roof. This sensor measures ambient light intensity in real time, sampling once per minute, and records trends in light intensity changes, such as gradual increase, gradual decrease, or sudden changes. Precipitation is collected by a precipitation sensor installed on the vehicle's windshield. This sensor determines the presence and intensity of precipitation (e.g., light rain, moderate rain, heavy rain) by detecting the scattering of infrared light by raindrops, and records the start and end times of precipitation.

[0047] Step S118: Collect road traffic dynamic data through vehicle positioning equipment and traffic information receiving equipment. The road traffic dynamic data records the lane occupancy, the speed of adjacent vehicles, and changes in road traffic signals.

[0048] The vehicle positioning equipment uses a Global Navigation Satellite System (GNSS) receiver to acquire the vehicle's real-time location coordinates. Combined with electronic map data, it can determine the vehicle's current road and lane information. By analyzing the position and speed information of vehicles in front and behind within the same lane, lane occupancy can be calculated, such as the current lane's vehicle density and vehicle spacing. The speeds of adjacent vehicles are acquired through vehicle-to-vehicle communication technology. Vehicles periodically exchange information such as their speed, position, and direction of travel. The central data processing unit filters the received information and extracts the speed data of vehicles in adjacent lanes (left, right, and within a certain distance in front and behind). Changes in road traffic signals are received from real-time traffic signal information released by traffic management departments through traffic information receiving equipment. This equipment can receive the status information of traffic lights along the road, including the switching time of red, green, and yellow lights, as well as countdown information. Simultaneously, when a vehicle approaches an intersection, the positioning equipment combines the intersection's location information from the electronic map to anticipate changes in the intersection's traffic signals.

[0049] Step S119: The collected road surface cover status data, road surface damage status data, road surrounding environment dynamic data, and road traffic dynamic data are segmented in chronological order. The segmented time window is consistent with the time window for generating human condition status nodes. The core features of the above data within each time window are extracted to form road condition nodes.

[0050] Similar to the time window settings for generating human-related state nodes, various road-related dynamic data are segmented and processed chronologically to ensure that the time windows of road condition nodes are completely consistent with the time windows of their corresponding human-related state nodes, facilitating subsequent analysis of their interactive influence. Within each time window, the average value of cover type (e.g., water accumulation, snow accumulation) and coverage area are extracted as core features from road surface cover status data; the average length of cracks and the average area of ​​potholes are extracted as core features from road surface damage status data; the maximum moving speed of obstacles, the average rate of change of light intensity, and the precipitation intensity level are extracted as core features from road surrounding environmental dynamic data; and lane occupancy rate (a quantitative indicator of lane occupancy), the average speed of adjacent vehicles, and the frequency of traffic signal changes are extracted as core features from road traffic dynamic data. After standardizing these core features, they are integrated into a multi-dimensional feature vector, which, together with the corresponding time window identifier, constitutes the road condition node.

[0051] Step S120: Analyze the interaction between each state node in the dynamic evolution sequence of human factors and each road condition node in the real-time change sequence of road conditions, and establish a two-way correlation model between human factors and road conditions.

[0052] Step S121: Extract all human state nodes from the dynamic evolution sequence of human state, each human state node includes a driver state sub-node and an escort state sub-node. Extract all road condition nodes from the real-time road condition change sequence, each road condition node includes a road surface state sub-node, a road surrounding environment state sub-node, and a road traffic state sub-node.

[0053] The dynamic evolution sequence of human factors' states is stored in the form of a linked list. By traversing this linked list, all human factors' state nodes can be extracted sequentially. For each extracted human factors' state node, based on the driver's state feature vector and the escort's state feature vector stored within it, driver state sub-nodes and escort state sub-nodes are separated. The driver state sub-nodes contain core features such as the driver's eye movements, facial expressions, and physiological dynamics, while the escort state sub-nodes contain core features such as the escort's voice interaction and gestures. Similarly, the real-time change sequence of road conditions is also stored in the form of a linked list. By traversing this linked list, all road condition nodes are extracted. For each road condition node, based on the composition of its core feature vectors, road surface state sub-nodes (including cover and damage features), road surrounding environment state sub-nodes (including obstacle, lighting, and precipitation features), and road traffic state sub-nodes (including lane occupancy, adjacent vehicle speed, and traffic signal features) are separated.

[0054] Step S122: Pair each human factor status node with the road condition node corresponding to the same time window to form multiple pairs of human factor road condition nodes. Each pair of human factor road condition nodes includes a human factor status node and a corresponding road condition node.

[0055] Since both human-related status nodes and road condition nodes contain time window identifiers, these nodes are paired by matching the time window identifiers. For example, a human-related status node with a time window identifier of 19:00-19:05 will be paired with a road condition node with the same identifier, forming a human-related / road condition node pair. During the pairing process, it is necessary to check for mismatched time window identifiers. If a time window contains only human-related status nodes or only road condition nodes, that node is marked as an abnormal node and will not participate in subsequent interaction analysis until later data verification. After pairing is complete, all human-related / road condition node pairs are arranged in chronological order, forming a new sequence.

[0056] Step S123: Decompose the features of the human state node and the road condition node in each pair of human-factor road condition nodes, extract the driver operation tendency feature, driver fatigue feature and escort prompt feature from the human state node, and extract the road resistance feature, road vision interference feature and road traffic conflict feature from the road condition node.

[0057] For the feature decomposition of the human state node, features related to operational tendencies are extracted from the feature vector of the driver state sub-node, such as the frequency of steering operations, braking operations, and acceleration operations. These features reflect the driver's driving habits and tendencies. Driver fatigue features are extracted from core features of eye state (such as the average frequency of eyelid closure) and core features of physiological dynamics (such as the fluctuation range of heart rate data and the average value of respiratory rate data). These features reflect the degree of driver fatigue. Escort officer prompting features are extracted from the core features of voice interaction (such as the number of keywords and the number of changes in voice tone) and core features of movement posture (such as the average amplitude of limb movements and the angle between the direction of movement and the vehicle's direction of travel) of the escort officer state sub-node. These features reflect the escort officer's prompting behavior and its effectiveness towards the driver. For the feature decomposition of the road condition node, road resistance features are extracted from the road surface state sub-node's data on the state of the road cover (such as the influence of the type of cover on the friction coefficient) and the data on the state of damage (such as potholes increasing driving resistance). These features reflect the magnitude of the road surface's resistance to vehicle movement. Road visibility interference features are extracted from illumination change data (e.g., strong light or backlight can affect visibility), precipitation data (e.g., heavy rain can blur visibility), and obstacle occlusion data of the road surrounding environment sub-nodes, reflecting the degree of interference with the driver's vision. Road traffic conflict features are extracted from lane occupancy data (e.g., high lane occupancy rates easily lead to conflicts), the speed of adjacent vehicles (e.g., large speed differences easily lead to rear-end collisions), and traffic signal changes (e.g., sudden signal changes easily lead to running red lights) of the road traffic status sub-nodes, reflecting the potential conflict risks during road traffic.

[0058] Step S124: Calculate the degree of interaction between the driver's operational tendency characteristics and road resistance characteristics for the same person at road condition nodes, calculate the degree of interaction between the driver's fatigue characteristics and road vision interference characteristics, and calculate the degree of interaction between the escort's prompting characteristics and road traffic conflict characteristics.

[0059] Step S1241: Extract driver operation tendency features from the human state nodes of the same human-factor road condition node pair. The driver operation tendency features include steering operation frequency, braking operation frequency, and acceleration operation frequency. Extract road resistance features from the road condition nodes of the same node pair. The road resistance features include road gradient, road surface friction coefficient, and road smoothness.

[0060] Within the same driver-road condition node pair, driver operation tendency characteristics are extracted from specific dimensions of features in the driver state sub-node. Steering frequency is determined by analyzing the number of steering wheel angle changes within a time window; braking frequency is calculated by the number of brake pedal presses; and acceleration frequency is calculated by the number of accelerator pedal presses. These operation frequency data are recorded during driver dynamic state data acquisition and, after feature extraction and standardization, are stored in the feature vector of the driver state sub-node. Road resistance characteristics are extracted by measuring road slope using a vehicle tilt sensor mounted on the chassis. This sensor detects the vehicle's longitudinal tilt angle in real time and converts it into road slope. The road surface friction coefficient is estimated based on the type and condition of the road surface cover. For example, dry asphalt pavement has a higher friction coefficient, while water or snow-covered pavement has a lower coefficient. Cracks and potholes also reduce the friction coefficient. Road smoothness is measured by vibration sensors in the vehicle's suspension system. Greater vibration indicates poorer road smoothness and greater driving resistance. These road resistance characteristics are also standardized and stored in the feature vector of the road condition node.

[0061] Step S1242: Establish an interaction matrix between driver operation tendency characteristics and road resistance characteristics. The rows of the interaction matrix represent the parameters of driver operation tendency characteristics, and the columns represent the parameters of road resistance characteristics. Each element in the matrix is ​​used to store the interaction influence value of the corresponding parameter pair.

[0062] The interaction matrix is ​​a 3x3 matrix. The rows correspond to the frequency of steering, braking, and acceleration operations, respectively, and the columns correspond to the road gradient, road surface friction coefficient, and road smoothness, respectively. Each element in the matrix, such as the element in the i-th row and j-th column, represents the interaction value between the i-th parameter of the driver's operating tendency characteristics and the j-th parameter of the road resistance characteristics.

[0063] Step S1243: Based on historical driving data, calculate the correlation coefficient between the standardized steering operation frequency and the standardized road slope as the interaction influence value; calculate the correlation coefficient between the standardized braking operation frequency and the standardized road surface friction coefficient as the interaction influence value; calculate the correlation coefficient between the standardized acceleration operation frequency and the standardized road smoothness as the interaction influence value.

[0064] By analyzing a large amount of historical driving data, a statistical relationship was established between driver operation tendency parameters and road resistance parameters. Regarding steering frequency and road gradient, historical data shows that as road gradient increases, drivers typically increase steering operations to keep the vehicle within the lane, thus a positive correlation exists between the two. The correlation coefficient is obtained by calculating the covariance of the standardized steering frequency and the standardized road gradient, and then dividing by the product of their standard deviations. This correlation coefficient is the element value corresponding to steering frequency and road gradient in the interaction matrix. Similarly, regarding braking frequency and road surface friction coefficient, a negative correlation is typically observed when the road surface friction coefficient decreases, requiring more frequent braking to maintain a safe following distance. The correlation coefficient is calculated as the corresponding element value. Regarding acceleration frequency and road smoothness, a positive correlation is observed when poorer road smoothness leads to fewer acceleration operations. The correlation coefficient is calculated as the corresponding element value.

[0065] Step S1244: Normalize all interaction influence values ​​in the interaction matrix so that all values ​​are in the same numerical range, and then calculate the trace of the interaction matrix. Use the trace value as the degree of interaction influence between driver operation tendency characteristics and road resistance characteristics.

[0066] The normalization process employs a min-max normalization method, linearly mapping each interaction influence value in the interaction matrix to a value between 0 and 1. A larger normalized interaction influence value indicates a stronger interaction between the corresponding parameter pairs. The trace of the interaction matrix is ​​the sum of the elements on the main diagonal. By calculating the trace of the normalized interaction matrix, a comprehensive value is obtained, representing the degree of interaction between driver operating tendency characteristics and road resistance characteristics. A larger value indicates a stronger overall interaction between the two types of characteristics.

[0067] Step S1245: Extract driver fatigue features from the human state nodes of the same human-cause road condition node pair. Driver fatigue features include the proportion of eyelid closure time, heart rate variability coefficient, and reaction delay time. Extract road visibility interference features from the road condition nodes of the same node pair. Road visibility interference features include the amplitude of light intensity change, visibility caused by precipitation, and the range of obstruction by surrounding objects.

[0068] The extraction of driver fatigue characteristics includes: Eyelid closure duration percentage (the ratio of total eyelid closure time to total time within a time window), calculated from core eye condition features; Heart rate variability (the ratio of standard deviation to mean of heart rate variation data), reflecting heart rate fluctuations, extracted from core physiological dynamic features; Reaction delay time, determined by analyzing the driver's response time to specific stimuli (such as traffic signal changes or obstacle appearances), extracted from driver operational reaction data; Road visibility interference features, specifically the amplitude of light intensity variation (the difference between the maximum and minimum light intensity within a time window), extracted from light variation data of road surrounding environment status sub-nodes; Visibility due to precipitation, estimated based on precipitation intensity and type (e.g., lower visibility during heavy rain and higher visibility during light rain), extracted from precipitation data; and Obstruction range of surrounding objects, determined by calculating the angle occupied by roadside obstacles in the driver's field of vision, extracted from obstacle detection data.

[0069] Step S1246: Construct a correlation function between driver fatigue characteristics and road vision interference characteristics. The input of the correlation function is the parameters of driver fatigue characteristics and the parameters of road vision interference characteristics, and the output is the interaction influence value.

[0070] The correlation function is a multi-input, single-output nonlinear function, trained using a large amount of sample data. The sample data includes labels representing the actual interaction impact under different combinations of driver fatigue characteristic parameters and road visibility interference characteristic parameters. Machine learning algorithms, such as support vector machines and neural networks, are used to train the correlation function on the sample data.

[0071] Step S1247: Input the standardized eyelid closure duration ratio and the standardized light intensity variation amplitude into the first sub-module of the correlation function to calculate their combined effect value. The combined effect value is the result of a function calculation based on historical data training and outputting a dimensionless evaluation value. Input the standardized heart rate variability coefficient and the standardized visibility caused by precipitation into the second sub-module of the correlation function to calculate their coupled effect value. The coupled effect value is the result of a function calculation based on historical data training and outputting a dimensionless evaluation value. Input the standardized reaction delay time and the standardized occlusion range of surrounding objects into the third sub-module of the correlation function to calculate their superimposed effect value. The superimposed effect value is the result of a function calculation based on historical data training and outputting a dimensionless evaluation value.

[0072] The first submodule of the correlation function specifically handles the interaction between the proportion of eyelid closure time and the amplitude of light intensity variation. When the driver's eyelid closure time is high (indicating fatigue) and the amplitude of light intensity variation is large (such as the alternation of strong light when meeting oncoming traffic at night), the synergistic effect of the two will seriously affect the driver's vision and reaction ability. The first submodule calculates the above synergistic effect value using a trained function. The second submodule handles the interaction between the coefficient of variation of heart rate and visibility caused by precipitation. A low coefficient of variation of heart rate indicates a decline in the driver's autonomic nervous system regulation ability (fatigue), and low visibility increases driving difficulty. The coupling of the two will further increase the risk. The second submodule calculates the coupling effect value. The third submodule handles the interaction between reaction time and the occlusion range of surrounding objects. When the reaction time is long and the occlusion range is large, it is difficult for the driver to detect and respond to danger in time. The third submodule calculates the superimposed effect value. The functions of the above submodules are all trained using historical data. The input is standardized feature parameters, and the output is a dimensionless evaluation value. The larger the evaluation value, the more significant the interaction effect.

[0073] Step S1248: The output values ​​of the first submodule, the second submodule, and the third submodule are weighted and summed to obtain the degree of interaction between driver fatigue characteristics and road vision interference characteristics.

[0074] Based on the importance of each submodule's output value in the overall interaction, different weights are assigned to each submodule. The weights are determined using the analytic hierarchy process (AHP) or statistical analysis based on sample data. For example, if historical data indicates that the synergistic effect of the proportion of eyelid closure duration and the magnitude of light intensity changes has the greatest contribution to the overall risk, then the first submodule has the highest weight. The output values ​​of each submodule are multiplied by their corresponding weights and then summed to obtain the degree of interaction between driver fatigue characteristics and road visibility interference characteristics.

[0075] Step S1249: Extract the escort officer prompting features from the human state nodes of the same human-cause road condition node pair. The escort officer prompting features include prompting frequency, prompting accuracy, and prompting response time. Extract the road traffic conflict features from the road condition nodes of the same node pair. The road traffic conflict features include the approach speed of adjacent vehicles, the number of lane occupancy conflicts, and the frequency of traffic signal changes.

[0076] The extraction of escort officer prompt features includes: prompt frequency, which is the number of times the escort officer issues prompts per unit time, calculated by combining the number of keywords and tone changes in the core features of voice interaction; prompt accuracy, which is evaluated by comparing the escort officer's prompt content with the actual road conditions (e.g., a prompt "There is an obstacle ahead" indicates high accuracy if an obstacle actually exists, otherwise low accuracy), extracted from the comparison results of voice interaction data and road condition data; prompt response time, which is the time interval between the occurrence of a road condition anomaly and the escort officer's prompt, calculated from timestamp data; and road traffic conflict features, including: adjacent vehicle approach speed, which is the relative speed of adjacent vehicles relative to the escort officer, calculated from adjacent vehicle speed data; lane occupancy conflict count, which is the number of lane usage conflicts detected between vehicles and other vehicles or obstacles within a time window, extracted from lane occupancy data; and traffic signal change frequency, which is the number of traffic signal changes per unit time, extracted from traffic signal change data.

[0077] Step S1250: Establish an impact assessment model for the characteristics of escort personnel prompts and road traffic conflicts. The impact assessment model includes an input layer, a processing layer, and an output layer. The input layer receives various parameters of the two types of features, the processing layer performs interactive calculations on the parameters, and the output layer outputs the degree of interactive impact.

[0078] The impact assessment model employs a neural network structure. The input layer contains six neurons, corresponding to six parameters related to the escort personnel's prompts: prompt frequency, prompt accuracy, prompt response time, and road traffic conflict characteristics: adjacent vehicle approach speed, lane occupancy conflict frequency, and traffic signal change frequency. The processing layer contains two hidden layers: the first hidden layer has 12 neurons, and the second hidden layer has 8 neurons, using the ReLU activation function. The output layer contains one neuron, outputting the interaction impact level value. The model is trained using backpropagation, employing historical data on escort personnel's prompts and road traffic conflict characteristics, along with their corresponding interaction impact level labels, as training data. Network weights and biases are adjusted to minimize the model's output error.

[0079] Step S1251: The input layer inputs the standardized warning frequency and the standardized approach speed of adjacent vehicles into the first processing unit of the processing layer to calculate the mitigation coefficient of the warning frequency on the risk of approaching adjacent vehicles. The mitigation coefficient is the result of a function calculation based on historical data training and outputting dimensionless coefficients.

[0080] The first computational unit is a submodule within the processing layer, employing a logistic regression function as its computational function. The function's inputs are the standardized alert frequency and the approach speed of adjacent vehicles, with parameters obtained through training on historical data. A higher alert frequency and lower approach speed result in a larger mitigation coefficient, indicating a stronger mitigation effect of the alert frequency on the risk of adjacent vehicle approach; conversely, a lower alert frequency results in a smaller mitigation coefficient.

[0081] Step S1252: The input layer inputs the standardized prompt accuracy and the standardized lane occupancy conflict count into the second processing unit of the processing layer to calculate the lane occupancy conflict avoidance efficiency of the prompt accuracy. The avoidance efficiency is the result of a function calculation based on historical data training and outputting dimensionless coefficients.

[0082] The second computational unit also uses a logistic regression function, with standardized prompt accuracy and lane occupancy conflict counts as inputs. Higher prompt accuracy results in fewer lane occupancy conflicts and higher avoidance efficiency; conversely, lower accuracy leads to lower avoidance efficiency. This function's parameters are also obtained through training on historical data.

[0083] Step S1253: The input layer inputs the standardized prompt response time and the standardized traffic signal change frequency into the third processing unit of the processing layer to calculate the response time's effectiveness in responding to traffic signal change risks. The effectiveness is the result of a function calculation based on historical data training, outputting dimensionless coefficients.

[0084] The third computational unit employs an exponential function, with the standardized prompt response time and traffic signal change frequency as inputs. A shorter prompt response time and a lower traffic signal change frequency result in better response and a larger dimensionless coefficient in the output; conversely, a longer response time leads to poorer response and a smaller coefficient. The function parameters are determined through training with historical data.

[0085] Step S1254: The processing layer comprehensively evaluates the output results of the first, second and third operation units to determine the proportion of positive and negative influences. The output layer calculates the degree of interaction between the escort officer prompt feature and the road traffic conflict feature based on the proportion of positive and negative influences.

[0086] The processing layer performs a weighted average of the outputs from the three computational units (mitigation coefficient, avoidance efficiency, and response effect). The weights are determined based on the importance of each output in the comprehensive evaluation, and these weights are obtained through statistical analysis of historical data. The comprehensive evaluation yields the sum of positive and negative impacts. The percentage of positive impact is calculated by dividing the sum of positive and negative impacts by the sum of positive and negative impacts, while the percentage of negative impact is calculated by subtracting the percentage of positive impact from 1. The output layer uses the percentage of positive impact as the degree of interaction between the escort officer's prompting feature and the road traffic conflict feature. The closer this value is to 1, the stronger the positive interaction, meaning the more significant the positive impact of the escort officer's prompting feature on the road traffic conflict feature; the closer it is to 0, the stronger the negative interaction.

[0087] Step S125: Standardize the calculated three interaction influence levels respectively, and then weight and integrate the standardized three interaction influence levels to obtain the overall correlation strength value of each pair of human-road condition node pairs. In the weighted integration process, assign corresponding weight coefficients according to the importance of each interaction influence level under different driving scenarios.

[0088] The interaction effects of driver operational tendencies and road resistance characteristics, driver fatigue characteristics and road visibility interference characteristics, and escort personnel's prompting characteristics and road traffic conflict characteristics were standardized separately using the Z-score standardization method, ensuring that each interaction effect value follows a normal distribution with a mean of 0 and a standard deviation of 1. During weighted integration, the weight coefficients were dynamically adjusted according to different driving scenarios. For example, in highway driving scenarios, road traffic conflict characteristics are more important, so the weight coefficient for the interaction effect between escort personnel's prompting characteristics and road traffic conflict characteristics is set relatively high; in mountainous road driving scenarios, road resistance characteristics and road visibility interference characteristics are more important, and the corresponding weight coefficients for their interaction effects are set relatively high. The determination of the weight coefficients was based on a combination of expert experience and historical accident data analysis to ensure that the weight allocation reflects the importance ranking of actual driving risk factors. Multiply the three standardized interaction influences by their respective weight coefficients, and then add them together. The sum is the overall correlation strength value of each pair of human-factor road condition node pairs. This value reflects the overall degree of correlation between human-factor state and road condition within this time window.

[0089] Step S126: According to the order of time windows, compare the overall correlation strength values ​​of human factors and road conditions node pairs in adjacent time windows, analyze the changing trend of the overall correlation strength values, identify the time windows in which the overall correlation strength values ​​change abruptly, and mark the human factors and road conditions node pairs corresponding to the abrupt change time windows as key node pairs.

[0090] The difference in overall correlation strength values ​​between adjacent time windows is calculated sequentially. When the absolute value of this difference exceeds a preset abrupt change threshold, a sudden change in the overall correlation strength value is determined, and this time window is designated as the abrupt change time window. The abrupt change threshold is set based on the normal fluctuation range of the overall correlation strength value in historical data; for example, the upper limit of the normal fluctuation range can be used as the abrupt change threshold. The human-factor road condition node pairs corresponding to the abrupt change time windows are marked as key node pairs. These node pairs reflect the moments when the correlation between human factors and road conditions changes significantly and are the focus of subsequent analysis of the two-way influence relationship.

[0091] Step S127: Extract the changes in human state characteristics and road condition characteristics from all key node pairs, analyze the triggering effect of changes in human state characteristics on changes in road condition characteristics, and analyze the feedback effect of changes in road condition characteristics on changes in human state characteristics to form a description of the two-way influence relationship.

[0092] For each key node pair, compare it with the human factor state node and road condition node of the previous time window, and calculate the change in each dimension of the human factor state feature vector (the feature value of the current node minus the feature value of the previous node) to obtain the change in human factor state features. Similarly, calculate the change in each dimension of the road condition feature vector to obtain the change in road condition features. Analyze the triggering effect of the change in human factor state features on the change in road condition features. For example, an increase in the change in driver fatigue features (increased fatigue level) may lead to driver operational errors, which in turn may lead to an increase in the change in road traffic conflict features (such as an increase in the number of lane occupancy conflicts). Analyze the feedback effect of the change in road condition features on the change in human factor state features. For example, an increase in the change in road visibility interference features (reduced visibility) may lead to driver tension, which in turn may lead to an increase in the change in physiological dynamic features (such as increased heart rate). Through a detailed analysis of the above triggering and feedback effects, describe the two-way influence relationship between humans and human factor state and road conditions in key node pairs using natural language, forming a description of the two-way influence relationship.

[0093] Step S128: Based on the bidirectional influence relationship description of all key node pairs and the overall association strength value of non-key node pairs, construct a bidirectional association model of human factors and road conditions that includes node association rules, feature interaction rules, and time evolution rules. The node association rules define the association conditions between different types of human factor state nodes and road condition nodes. The feature interaction rules define the interaction mode between human factor features and road condition features. The time evolution rules define the change pattern of the association relationship over time.

[0094] Node association rules are constructed through statistical analysis of human-related state node types and road condition node types for both critical and non-critical node pairs. For example, when the human-related state node type is "fatigue state," the corresponding road condition node type may often be "low line-of-sight interference road condition," thus defining the association condition between the "fatigue state" human-related node and the "low line-of-sight interference road condition" node. Feature interaction rules are based on the description of the bidirectional influence relationship of critical node pairs, summarizing common interaction patterns between human-related features and road condition features, such as the rule that "an increase in driver fatigue characteristics will lead to an aggravation of the impact of road line-of-sight interference characteristics on driving risks." Temporal evolution rules are constructed by analyzing the changing trend of the overall association strength value over time. For example, if the overall association strength value gradually increases over multiple consecutive time windows, it indicates that the association between human-related state and road condition is gradually strengthening, and this gradually strengthening change pattern is defined as a temporal evolution rule; if the overall association strength value shows periodic fluctuations, it is defined as a temporal evolution rule of periodic change patterns. By integrating the above node association rules, feature interaction rules, and time evolution rules, a two-way association model between human factors and road conditions is formed. This model can be represented in the form of production rules, which is convenient for computers to store and reason about.

[0095] Step S130: Based on the bidirectional correlation model of human factors and road conditions, extract the risk triggering conditions under different combinations of state nodes and road condition nodes, and identify the set of driving risk transmission paths.

[0096] Step S131: Extract all node association rules from the bidirectional human-factor road condition association model. Each node association rule corresponds to a combination of human-factor state node type and road condition node type.

[0097] The node association rules in the human-factor and road condition bidirectional correlation model are stored in a rule base. Each rule contains two parts: a precondition and a conclusion. The preconditions are the human-factor state node type and the road condition node type, and the conclusion is the correlation between the two. By traversing the rule base, the preconditions of all rules are extracted, thus obtaining all combinations of human-factor state node types and road condition node types. For example, the rule "If the human-factor state node type is 'fatigue state' and the road condition node type is 'low friction coefficient road condition,' then the correlation strength between the two is high" corresponds to the combination of the human-factor state node type "fatigue state" and the road condition node type "low friction coefficient road condition."

[0098] Step S132: For each combination of human factor state node type and road condition node type, analyze the possible abnormal human factor characteristics and abnormal road condition characteristics under this combination. Abnormal human factor characteristics include delayed driver operation response, driver distraction, and untimely prompts from escort personnel. Abnormal road condition characteristics include reduced road surface friction coefficient, sudden appearance of obstacles around the road, and reduction of road traffic lanes.

[0099] For each combination, based on the characteristics of the human factor state node type and the road condition node type, and in conjunction with historical accident cases and risk analysis reports, potential abnormal human factor characteristics and road condition characteristics are analyzed. For example, in the combination of the human factor state node type "fatigue state" and the road condition node type "low friction coefficient road condition," potential abnormal human factor characteristics include delayed driver reaction (due to fatigue leading to slower reaction time), and abnormal road condition characteristics include a decrease in the road surface friction coefficient (such as water or snow accumulation on the road surface). In the combination of the human factor state node type "distracted state" and the road condition node type "high traffic signal change frequency road condition," potential abnormal human factor characteristics include driver inattention, and abnormal road condition characteristics include a reduction in the number of road lanes (such as road construction ahead causing lane narrowing).

[0100] Step S133: Combine each type of human factor characteristic anomaly with the corresponding road condition characteristic anomaly to form multiple sets of anomaly combination scenarios. Each set of anomaly combination scenarios includes one type of human factor characteristic anomaly and one corresponding road condition characteristic anomaly.

[0101] After analyzing the abnormal human-cause characteristics and road-cause characteristics under each combination of human-cause state node type and road-cause node type, the two are matched one-to-one. For example, in the combination of "fatigue state" and "low friction coefficient road condition" mentioned above, the abnormal human-cause characteristic "delayed driver operation reaction" is combined with the abnormal road-cause characteristic "reduced road surface friction coefficient" to form the abnormal combination scenario of "delayed driver operation reaction and reduced road surface friction coefficient". For each combination of human-cause state node type and road-cause node type, multiple abnormal human-cause characteristics and multiple abnormal road-cause characteristics may occur, thus forming multiple abnormal combination scenarios.

[0102] Step S134: Analyze the triggering conditions for each abnormal combination scenario, determine the initial triggering factors that lead to the risk under the abnormal combination scenario, and determine the order in which the initial triggering factors cause subsequent feature changes.

[0103] The trigger condition analysis employs fault tree analysis, treating the abnormal combination scenario as the top event. Through layer-by-layer decomposition, the initial triggering factors leading to the top event are identified. For example, for the abnormal combination scenario of "delayed driver reaction and reduced road surface friction coefficient," the top event is the occurrence of this scenario. Analysis reveals possible initial triggering factors as excessively long driving time (leading to fatigue and consequently delayed reaction) and a sudden increase in precipitation (leading to a decrease in the road surface friction coefficient). After identifying the initial triggering factors, event tree analysis is used to analyze the sequence of possible changes in human factors and road conditions following their occurrence. For instance, a sudden increase in precipitation (the initial triggering factor) first leads to a decrease in the road surface friction coefficient (a change in road conditions). The driver, due to excessively long driving time (the initial triggering factor), is fatigued. When encountering low-friction road conditions, the delayed reaction (a change in human factors) prevents timely braking, thus increasing the vehicle's braking distance (subsequent changes in road conditions or vehicle driving parameters).

[0104] Step S135: Based on the initial triggering factors and the sequence of feature changes, construct the risk transmission chain corresponding to each abnormal combination scenario. The risk transmission chain starts from the initial triggering factor, connects sequentially to the triggered human factor feature changes or road condition feature changes, and ends at the possible driving risk outcome.

[0105] For example, step S1351: Determine the type of initial triggering factor in each abnormal combination scenario. If the initial triggering factor is a human factor abnormality, then the human factor abnormality is taken as the starting node of the risk transmission chain; if the initial triggering factor is a road condition abnormality, then the road condition abnormality is taken as the starting node of the risk transmission chain.

[0106] In each abnormal combination scenario, the initial triggering factors identified in the triggering condition analysis phase are categorized to determine whether they belong to abnormal human factors or abnormal road conditions. For example, in the abnormal combination scenario of "driver inattention and reduced road lanes," if the initial triggering factor is driver inattention (abnormal human factors), then "driver inattention" is taken as the starting node of this risk transmission chain; if the initial triggering factor is reduced road lanes (abnormal road conditions), then "reduced road lanes" is taken as the starting node.

[0107] Step S1352: Analyze the next feature change that may be directly triggered by the feature anomaly corresponding to the starting node. If the starting node is a human factor feature anomaly, analyze the type of road condition feature change caused by the human factor feature anomaly, or other related human factor feature change types. If the starting node is a road condition feature anomaly, analyze the type of human factor feature change caused by the road condition feature anomaly, or other related road condition feature change types.

[0108] For scenarios where the starting point is an abnormal human factor characteristic, such as "delayed driver reaction," analyze the subsequent changes that this abnormality may cause. If the driver's reaction is delayed, it may lead to insufficient safe distance between the vehicle and the vehicle in front (a change in vehicle driving parameters, which can be indirectly related to changes in road traffic conditions), or further cause driver anxiety leading to an increased heart rate (other related types of human factor characteristic changes). For scenarios where the starting point is an abnormal road condition characteristic, such as "reduced road surface friction coefficient," analyze the possible human factor characteristic changes it may cause, such as the driver needing to adjust the steering wheel more frequently to maintain vehicle stability (increased operation frequency in the human factor characteristic change type), or an increased vehicle braking distance (change in road traffic resistance in other related types of road condition characteristic changes).

[0109] Step S1353: Take the directly triggered feature change as the second node of the risk transmission chain, and record the triggering relationship between the starting node and the second node. The triggering relationship includes the triggering delay time and the degree to which the triggering conditions are met.

[0110] After identifying the directly triggered characteristic change, it is added as the second node in the risk transmission chain. For example, if the starting node is "driver inattention" and the directly triggered characteristic change is "no traffic signal change observed," then the second node is "no traffic signal change observed." Simultaneously, the triggering relationship between the two nodes is recorded. The trigger delay time refers to the time interval between the occurrence of the abnormal characteristic at the starting node and the appearance of the characteristic change at the second node. The trigger condition satisfaction level refers to the level of abnormality at the starting node that triggers the change at the second node.

[0111] Step S1354: Starting from the second node, repeat the above analysis process to determine the next feature change that the second node may directly trigger, which will be the third node in the risk transmission chain. Record the triggering relationship between the second and third nodes.

[0112] Starting with the second node "no traffic signal change observed", analyze the next characteristic change that it may trigger, such as "the vehicle does not decelerate as indicated by the traffic signal", and treat it as the third node. Record the triggering relationship between the two, such as "when no traffic signal change is observed for a certain period of time and the vehicle speed is higher than a certain percentage of the speed limit of the road segment, the degree to which the condition for triggering the vehicle not to decelerate is met", and record the trigger delay time range.

[0113] Step S1355: Continue to repeat steps S1352 to S1354 until the changes in the analyzed features directly lead to the driving risk result, and take the driving risk result as the terminal node of the risk transmission chain.

[0114] The process of continuously repeating feature change analysis and node addition is as follows: For example, the third node, "the vehicle failed to slow down according to traffic signal instructions," may trigger "increased collision risk with vehicles traveling laterally at the intersection" (the fourth node). This node further triggers "a side collision occurs" (driving risk outcome), at which point "a side collision occurs" is taken as the termination node. If multiple possible branches appear during the propagation process, such as the fourth node potentially triggering both "emergency braking leading to skidding" and "increased collision risk," then two branch chains are constructed respectively, both terminating at the corresponding driving risk outcome.

[0115] Step S1356: During the construction process, if a certain feature change node may trigger multiple subsequent feature change nodes at the same time, then for each possible subsequent node, a branch transmission chain is constructed to form a multi-branch risk transmission chain structure.

[0116] When a node has multiple trigger directions, such as the abnormal road condition node "sudden appearance of obstacles around the road," which may simultaneously trigger two nodes due to feature changes: "driver's emergency steering" and "driver's emergency braking," then this node is used as the parent node, and two branch chains are constructed respectively: one is "sudden appearance of obstacles around the road → driver's emergency steering → excessive steering angle leading to vehicle rollover," and the other is "sudden appearance of obstacles around the road → driver's emergency braking → insufficient braking distance leading to rear-end collision." Each branch chain independently records the trigger relationship and node information.

[0117] Step S1357: For each node in the constructed risk transmission chain, supplement the corresponding feature change parameters. The feature change parameters include the change magnitude, change duration, and the changed feature value.

[0118] For each feature change node in the chain, specific parameters are added. For example, for the node "increased frequency of driver eyelid closure", the change magnitude is "increased by a certain percentage compared to the previous time window", the change duration is "continuously occurring over multiple time windows", and the changed feature value is "eyelid closure frequency reaches a certain level". For the road condition feature node "increased number of lane occupancy conflicts", the change magnitude is "increased by a certain number of times compared to the average over a previous period", the change duration is "persistent within the current time window", and the changed feature value is "the number of lane occupancy conflicts reaches a certain quantity".

[0119] Step S1358: For each triggering relationship in the risk transmission chain, supplement the triggering probability information. The triggering probability information is obtained based on the frequency of the triggering relationship in historical driving data. The higher the frequency value, the higher the triggering probability value.

[0120] Based on historical data, a trigger probability is added to each trigger relationship. For example, for the trigger relationship "increased driver fatigue characteristics → failure to detect road visibility interference in time," if this combination occurs multiple times in historical data and the actual failure to detect it in time accounts for a certain proportion, then the trigger probability is that proportion. For the trigger relationship "decreased road surface friction coefficient → increased vehicle braking distance," if this combination occurs multiple times in historical data and the increased braking distance occurs in a high proportion, then the trigger probability is that proportion. The trigger probability is recorded as a percentage in the trigger relationship and used for weighting calculations in subsequent risk assessments.

[0121] Step S1359: Connect the starting node, intermediate feature change nodes, and ending node, along with their corresponding parameter information and probability information, according to the node order and triggering relationship to form a risk transmission chain. Each risk transmission chain is represented by a structured node sequence, which includes node type, node parameters, triggering relationship, and triggering probability.

[0122] All nodes in the chain are arranged in the transmission order to form a structured node sequence. Taking the "human-triggered" chain as an example, its node sequence is represented as follows: [Starting node: Type = Driver fatigue characteristic (increased eyelid closure frequency), Parameter = Variation amplitude, duration, feature value; Triggering relationship: to intermediate node 1, delay time, condition satisfaction level, trigger probability; Intermediate node 1: Type = Human operation characteristic (delayed braking operation), Parameter = Variation amplitude, duration, feature value; Triggering relationship: to intermediate node 2, delay time, condition satisfaction level, trigger probability; Intermediate node 2: Type = Road condition characteristics (extended braking distance), Parameter = Variation amplitude, duration, feature value; Triggering relationship: to the ending node, delay time, condition satisfaction level, trigger probability; Ending node: Type = Driving risk outcome (rear-end collision)]. The information of each node and triggering relationship is included in this structured representation, which facilitates subsequent model matching and verification.

[0123] Step S136: Match all the constructed risk transmission chains with the time evolution rules in the bidirectional correlation model of human factors and road conditions, verify whether the time interval of each link in the risk transmission chain meets the time threshold requirements in the time evolution rules, and eliminate risk transmission chains that do not meet the time threshold requirements.

[0124] The time evolution rules in the human-road-condition bidirectional correlation model define the changing pattern of the correlation over time, including time interval threshold requirements between different characteristic change stages. For example, the time evolution rule might stipulate that "after the road surface friction coefficient decreases, the driver should react by braking within 2 seconds, otherwise the risk will increase," where 2 seconds is the time interval threshold. The time interval between two adjacent stages in the risk transmission chain is compared with the corresponding time threshold in the time evolution rule. If the time interval is less than or equal to the time threshold, the chain passes verification; if the time interval is greater than the time threshold, the chain does not meet the time requirement and is discarded. For example, if the time interval between "decrease in road surface friction coefficient" and "delay in driver braking operation" in a certain risk transmission chain is 3 seconds, while the time threshold in the time evolution rule is 2 seconds, then the chain fails verification and is discarded.

[0125] Step S137: Supplement the missing feature change links in the verified risk transmission chain to make the risk transmission chain fully reflect the transmission process from the initial triggering factor to the driving risk result.

[0126] When constructing a risk transmission chain, incomplete initial analysis may lead to missing characteristic change links in the chain. For example, a risk transmission chain might be "driver inattention → failure to notice traffic signal changes → running a red light → vehicle collision," but in reality, there is a characteristic change link between "failure to notice traffic signal changes" and "running a red light," namely, "failure to slow down." By comparing with time evolution rules and actual accident cases, these missing links can be identified and added to the risk transmission chain, making the chain more complete and accurate. The supplemented chain becomes: "driver inattention → failure to notice traffic signal changes → failure to slow down → running a red light → vehicle collision."

[0127] Step S138: Classify and organize all the completed risk transmission chains, and divide them into three groups according to the type of risk transmission initiation factors: human-induced triggering, road condition-induced triggering, and human-induced and road condition-induced triggering. The risk transmission chains in each group constitute the driving risk transmission subset under that type.

[0128] Human-triggered risk transmission chains refer to chains where the initial triggering factor is an abnormal condition caused by human factors, such as "driver fatigue → delayed reaction → loss of vehicle control." Road condition-triggered risk transmission chains refer to chains where the initial triggering factor is an abnormal condition caused by road conditions, such as "large potholes on the road surface → vehicle bumps → loss of steering control → vehicle deviates from its lane." Human- and road condition-triggered risk transmission chains refer to chains where the initial triggering factors simultaneously include abnormal human factors and abnormal road conditions, such as "driver inattention and sudden appearance of an obstacle on the road → failure to avoid it in time → vehicle collides with the obstacle." All completed risk transmission chains are categorized according to the type of their initial triggering factor and assigned to their respective driving risk transmission subsets.

[0129] Step S139: Integrate all driving risk transmission subsets to form a driving risk transmission path set containing different types of risk transmission paths. Each driving risk transmission path corresponds to a unique path identifier in the set. The path identifier includes information on the type of initiating factor, the number of transmission links, and the type of risk outcome.

[0130] All risk transmission chains in the three subsets of driving risk transmission—human-triggered, road-condition-triggered, and human- and road-condition-triggered—are integrated to form a set of driving risk transmission paths. Each driving risk transmission path is assigned a unique path identifier, which is a string-encoded identifier containing the following: the type of initiating factor (e.g., "R" for human-triggered, "L" for road-condition-triggered, and "RL" for human- and road-condition-triggered), the number of transmission links (represented by a number, e.g., "5" for 5 transmission links), and the type of risk outcome (e.g., "CZ" for vehicle collision and "PL" for lane departure). For example, a human-triggered risk transmission path with 4 transmission links and a risk outcome of vehicle collision might have the path identifier "R-4-CZ".

[0131] Step S140: Extract path features for each driving risk transmission path in the set of driving risk transmission paths, and determine the risk impact range corresponding to each driving risk transmission path based on the extracted path features.

[0132] Step S141: Select a driving risk transmission path from the set of driving risk transmission paths, extract all transmission links in the driving risk transmission path, and each transmission link corresponds to a characteristic change event. The characteristic change event includes the event type, the event occurrence time, and the human factors or road conditions that affect the event.

[0133] From the set of driving risk transmission paths, each driving risk transmission path is selected sequentially according to the order of its path identifier. For each selected path, all transmission links are extracted one by one. Each transmission link corresponds to a characteristic change event, and the event type is divided into human factor characteristic change events and road condition characteristic change events. For example, "delayed driver operation reaction" is a human factor characteristic change event, and "reduction in road surface friction coefficient" is a road condition characteristic change event. The event occurrence time refers to the relative time of the characteristic change event in the entire risk transmission process. The time of occurrence of the initial triggering factor is used as the base time, and the occurrence time of subsequent events is the time offset relative to the base time. The human factor or road condition factor affected by the event refers to the human factor factor (such as the driver's reaction ability, the escort's prompting behavior) or road condition factor (such as road surface condition, road traffic condition) directly affected by the characteristic change event.

[0134] Step S142: Analyze the degree of influence of each characteristic change event on vehicle driving parameters, including driving speed, driving direction, braking distance, and steering angle. By comparing the changes in vehicle driving parameters before and after the event, determine the independent influence value of each characteristic change event on each vehicle driving parameter. Normalize each independent influence value of each characteristic change event, and calculate the combined normalized independent influence values ​​to obtain the single, dimensionless parameter influence value of the characteristic change event.

[0135] For each characteristic change event, the changes in vehicle driving parameters before and after the event are simulated using a vehicle dynamics model. For example, for the road condition characteristic change event of "reduced road surface friction coefficient," the braking distance changes when the vehicle brakes with the same brake pedal force before and after the friction coefficient decreases; the increase in braking distance is the independent impact value of this event on braking distance. For the human factor characteristic change event of "delayed driver operation response," the steering angle changes before and after the response delay under the same road conditions; the deviation in steering angle is the independent impact value of this event on steering angle. The independent impact values ​​of each vehicle driving parameter are normalized separately, using the min-max normalization method to map them to the range of 0-1. During the comprehensive calculation, weights are assigned to each vehicle driving parameter according to its importance in different risk scenarios. For example, in high-speed driving scenarios, braking distance and driving speed have larger weights, while driving direction and steering angle have relatively smaller weights. The normalized independent impact values ​​are multiplied by their corresponding weights and then summed to obtain the parameter impact value of the characteristic change event. The larger this value, the greater the comprehensive impact of the event on the vehicle driving parameters.

[0136] Step S143: Accumulate the parameter influence values ​​of all transmission links in the transmission order to obtain the total parameter influence value of the driving risk transmission path. During the accumulation process, each parameter influence value is weighted according to the position weight of the transmission link in the path. The weight coefficient value of the link at the front end of the path is less than the weight coefficient value of the link at the back end of the path.

[0137] The further downstream a transmission link is in the path, the more direct its impact on the final driving risk outcome, and therefore, the larger its weighting coefficient. The positional weights are determined using an exponentially increasing method; for example, the weighting coefficient for the first transmission link is 0.1, the second 0.2, the third 0.4, the fourth 0.8, and so on. The specific weighting coefficient values ​​are adjusted according to the total number of transmission links in the path, ensuring that the sum of all weighting coefficients is 1. The parameter impact value of each transmission link is multiplied by its corresponding positional weighting coefficient, and then all weighted parameter impact values ​​are summed to obtain the total parameter impact value of the driving risk transmission path. This value reflects the overall degree of influence of the entire risk transmission path on vehicle driving parameters.

[0138] Step S144: Extract the transmission speed characteristics of the driving risk transmission path. The transmission speed characteristics are obtained by calculating the average time interval between adjacent transmission links. The smaller the average time interval value, the faster the transmission speed value corresponding to the transmission speed characteristics.

[0139] The time interval between adjacent transmission links refers to the time of event occurrence in the later transmission link minus the time of event occurrence in the earlier transmission link. The average time interval of all adjacent transmission links in the calculation path is the quantitative indicator of transmission speed. The smaller the average time interval, the faster the transition speed between links in the risk transmission process, and the more rapidly the risk develops; conversely, the larger the average time interval, the slower the risk develops.

[0140] Step S145: Based on the total parameter impact value and transmission velocity characteristics, query the preset impact range mapping table. The impact range mapping table stores the risk impact range descriptions corresponding to different total parameter impact value ranges and transmission velocity ranges. The risk impact range descriptions include the affected vehicle system and the affected driving area.

[0141] The impact range mapping table is pre-established based on extensive accident data analysis and vehicle system simulation results. The table is divided into multiple intervals based on the total parameter impact value (e.g., 0-0.3, 0.3-0.6, 0.6-1.0) and multiple intervals based on the average time interval of the transmission speed characteristics (e.g., greater than 5 seconds, 3-5 seconds, less than 3 seconds). Each combination of the total parameter impact value interval and the transmission speed interval corresponds to a risk impact range description. For example, a combination of a total parameter impact value interval of 0.6-1.0 and a transmission speed interval of less than 3 seconds might correspond to a risk impact range description of "affecting the vehicle's braking system and steering system, affecting the current lane and the adjacent left lane." Based on the total parameter impact value and transmission speed characteristics of the current driving risk transmission path, the interval combination is determined, and then the corresponding risk impact range description is retrieved from the impact range mapping table.

[0142] Step S146: Analyze the risk outcome types in the driving risk transmission path, determine the secondary risk types that the risk outcome may cause, and supplement the description of the risk impact range found in the query with the secondary risk types to form risk impact range information. The secondary risk types include vehicle loss of control risk, rear-end collision risk, and lane departure risk.

[0143] Secondary risks that may arise from risk outcome types such as "vehicle collision" include vehicle loss of control risk (the vehicle loses control after a collision) and fuel leakage risk (the fuel tank is damaged due to a collision). Secondary risk types are determined by analyzing the characteristics and potential subsequent developments of each risk outcome type. Then, the risk impact range description retrieved from the impact range mapping table is supplemented based on the secondary risk type. For example, if the original description is "affects the vehicle's braking system and steering system, affecting the current lane and the adjacent left lane," and the risk outcome type is "vehicle collision" and the secondary risk type is "vehicle loss of control risk," then the supplemented risk impact range information would be "affects the vehicle's braking system, steering system, and vehicle body structure system, affecting the current lane, the adjacent left lane, and the central median strip area, potentially causing vehicle loss of control risk."

[0144] Step S147: Repeat steps S141 to S146 above to extract path features and determine the risk impact range for each driving risk transmission path in the set of driving risk transmission paths, so that each driving risk path corresponds to unique risk impact range information.

[0145] Following steps S141 to S146, each path in the set of driving risk transmission paths is processed sequentially. Each path undergoes steps such as transmission link extraction, influence degree analysis of feature change events, calculation of total parameter influence value, transmission speed feature extraction, influence range mapping query, and supplementation of secondary risk types, ultimately obtaining unique risk influence range information for each path.

[0146] Step S148: Associate and store the path identifier of each driving risk transmission path with the corresponding risk impact range information to form a path impact range association table. The path impact range association table also includes the total parameter impact value and transmission speed characteristics of each path.

[0147] The path impact range association table is stored in the form of a database table, containing fields for path identifier, total parameter impact value, transmission speed characteristic, and risk impact range information. For each traffic risk transmission path, the path identifier, total parameter impact value, transmission speed characteristic, and corresponding risk impact range information are filled into the corresponding fields in the table, achieving associated storage. Through this association table, information such as the risk impact range, total parameter impact value, and transmission speed characteristic corresponding to any path identifier can be quickly retrieved.

[0148] Step S150: Based on the risk impact range corresponding to each driving risk transmission path, generate a driving risk dynamic early warning instruction containing a description of the risk transmission direction and risk impact range, and send the driving risk dynamic early warning instruction to the driving control terminal to trigger the driving control terminal to perform an early warning prompt operation.

[0149] Step S151: Extract the path identifier, risk impact range information, total parameter impact value, and transmission speed characteristics of all driving risk transmission paths from the path impact range association table.

[0150] Through database query operations, the path identifier, risk impact range information, total parameter impact value, and transmission speed characteristic field values ​​of all records are read from the path impact range association table. The above data is loaded into memory to form a data list for subsequent priority sorting and early warning instruction generation.

[0151] Step S152: Prioritize all driving risk transmission paths according to the magnitude of the total parameter impact value. The larger the total parameter impact value, the higher the ranking position of the path priority. For paths with the same total parameter impact value, perform a secondary ranking based on the transmission speed characteristic. The faster the transmission speed value, the higher the ranking position of the path priority.

[0152] A multi-keyword sorting algorithm is used. First, the total parameter influence value is used as the first sorting key, and the parameters are sorted in descending order. When the total parameter influence values ​​are the same, the transmission speed feature is used as the second sorting key. The smaller the average time interval of the transmission speed feature, the faster the transmission speed. Therefore, the parameters are sorted in ascending order of the average time interval (i.e., the faster the transmission speed, the higher the ranking). After sorting, a sequence of traffic risk transmission paths is obtained, arranged from high to low priority.

[0153] Step S153: Select a preset number of driving risk transmission paths as core warning paths in order of priority from high to low. The preset number is determined according to the complexity of the current driving scenario.

[0154] The predetermined number of paths is related to the complexity of the current driving scenario, which is assessed by comprehensively considering factors such as road type (e.g., highway, urban road, rural road), traffic flow, and weather conditions. For example, in a complex scenario involving highways, heavy traffic, and heavy rain, the predetermined number is set to 10; in a general scenario involving urban roads, moderate traffic flow, and sunny weather, the predetermined number is set to 5; and in a simple scenario involving rural roads, low traffic flow, and cloudy weather, the predetermined number is set to 3. The predetermined number of paths are selected as core warning paths from the ranked sequence of driving risk transmission paths, in descending order of priority.

[0155] Step S154: Extract the risk transmission path of each core early warning path, determine the characteristic change direction of each transmission link in the risk transmission path, and connect the characteristic change directions in the transmission order to obtain the risk transmission direction of the core early warning path. The risk transmission direction is represented by a sequence of direction vectors, and each direction vector corresponds to the characteristic change direction of a transmission link.

[0156] For each core warning path, extract all its transmission links, i.e., the sequence of characteristic change events. Analyze the direction of characteristic change for each characteristic change event; for example, the characteristic change direction of "decrease in road surface friction coefficient" is "negative change" (friction coefficient value decreases), and the characteristic change direction of "increased driver reaction delay" is "positive change" (delay time increases). Use direction vectors to represent the direction of characteristic change; for example, "+1" represents positive change, "-1" represents negative change, and "0" represents no significant change (in practical applications, there may be more direction definitions depending on the characteristic type). Arrange the direction vectors of each transmission link in the transmission order to form a direction vector sequence, which is the risk transmission direction of the core warning path. For example, if the transmission links of a core warning path are "decrease in road surface friction coefficient → increased driver reaction delay → increased braking distance," its corresponding direction vector sequence is [-1, +1, +1].

[0157] Step S155: Extract the affected vehicle systems and affected driving areas from the risk impact range information corresponding to each core early warning path, integrate the affected vehicle systems and affected driving areas into a risk impact range description, and use structured language to describe the risk impact range to determine the specific impact mode of each affected object.

[0158] The risk impact range information includes the affected vehicle systems (such as braking systems, steering systems, body structure systems, etc.) and the affected driving areas (such as the current lane, the adjacent left lane, the central median strip area, etc.). After extracting this information, it is described using structured language, which includes elements such as object, impact method, and impact degree. For example, the affected vehicle system is "braking system," the impact method is "decreased braking performance," and the impact degree is "severe"; the affected driving area is "current lane," the impact method is "vehicle trajectory deviation," and the impact degree is "moderate." Combining these elements forms a risk impact range description, such as "Braking system: severely reduced braking performance; Steering system: moderately reduced steering accuracy; Current lane: moderate vehicle trajectory deviation; Adjacent left lane: risk of vehicle collision exists."

[0159] Step S156: Integrate the path identifier, risk transmission direction, and risk impact range description of each core early warning path to form a single path early warning information. Each single path early warning information includes a path identifier field, a transmission direction field, and an impact range field.

[0160] The single-path warning information uses a JSON-formatted data structure. The path identifier field stores the path identifier string, the propagation direction field stores the direction vector sequence (e.g., [-1,+1,+1]), and the impact range field stores a structured text describing the risk impact range. For example, a single-path warning information might look like this: {"Path Identifier":"R-4-CZ","Propagation Direction":[-1,+1,+1],"Impact Range":"Braking System: Braking efficiency severely reduced; Steering System: Steering accuracy moderately reduced; Current Lane: Vehicle trajectory moderately deviated; Adjacent Left Lane: Risk of vehicle collision exists"}.

[0161] Step S157: Summarize the single-path warning information corresponding to all core warning paths, add the current warning generation time and warning validity period information, and generate a dynamic warning instruction for driving risks. The dynamic warning instruction for driving risks is written in a format that can be parsed by the driving control terminal, and includes an instruction header, instruction body, and instruction tail. The instruction header contains an instruction type identifier, the instruction body contains all single-path warning information, and the instruction tail contains checksum information.

[0162] The current warning generation time is the exact time the system generates the warning instruction, accurate to the second. The warning validity period is determined based on the transmission speed characteristics of the risk transmission path and the current driving speed. The faster the transmission speed and the higher the driving speed, the shorter the warning validity period, for example, set to 5-10 minutes. The driving risk dynamic warning instruction adopts a custom binary protocol format. The instruction header contains the instruction type identifier (e.g., 0x01 indicates a driving risk dynamic warning instruction), instruction length, and other information; the instruction body contains the current warning generation time, the warning validity period, and single-path warning information for all core warning paths (JSON format data is serialized); the instruction tail contains a CRC checksum, used by the driving control terminal to verify the integrity of the received instruction.

[0163] Step S158: The generated dynamic driving risk warning command is sent to the driving control terminal. After receiving the dynamic driving risk warning command, the driving control terminal parses the single-path warning information in the command body, and determines the corresponding warning prompt method based on the parsed risk transmission direction and risk impact range description. The determined warning prompt method is executed, and the warning execution time, warning prompt method, and driver's response to the warning are recorded to form a warning execution log. The warning execution log is used for subsequent evaluation of the warning effect. The warning prompt methods include voice prompts, light prompts, and seat vibration prompts. Different risk impact range descriptions correspond to different warning prompt combinations.

[0164] The vehicle's internal CAN bus or Ethernet transmits dynamic warning commands for driving risks to the vehicle control terminal. Upon receiving the command, the vehicle control terminal first parses the command header and performs a CRC check. If the check passes, it parses the command body to extract the current warning generation time, warning validity period, and single-path warning information. Based on the risk transmission direction (direction vector sequence) and risk impact range description (structured language expression) in the single-path warning information, the warning prompt method is determined. For example, when the risk impact range description includes "severe impact on braking system" and "severe deviation of vehicle trajectory in current lane," the warning prompt method combination is: voice prompt (repeatedly playing "Attention! Braking system efficiency severely reduced, vehicle trajectory deviated, slow down immediately!"), light prompt (flashing red brake malfunction light and turn indicator lights on the instrument panel), and seat vibration prompt (alternating high-intensity vibration on the left and right sides of the seat). After the warning prompt is executed, the vehicle control terminal records the warning execution time (accurate to the second), the warning prompt method used (the specific combination and parameters of voice, light, and seat vibration), and the driver's response to the warning (such as whether to press the brake pedal, whether to turn the steering wheel, whether to reduce the driving speed, etc.), forming a warning execution log, which is stored in the terminal's local storage for subsequent data analysis to evaluate the warning effect.

[0165] In one exemplary embodiment, a dynamic warning system for driving risks based on the fusion of human factors and road conditions is provided. This system can be a terminal, server, etc., and its internal structure diagram can be as follows: Figure 2As shown, this dynamic warning system for driving risks based on the fusion of human factors and road conditions includes a processor, memory, input / output interface, communication interface, display unit, and input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interface. The processor provides computational and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system and computer programs. The internal memory provides the environment for the operation of the operating system and computer programs in the non-volatile storage medium. The input / output interface is used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, near-field communication, or other technologies. When the computer program is executed by the processor, it implements a dynamic warning method for driving risks based on the fusion of human factors and road conditions. The display unit is used to form a visually visible image and can be a display screen, projection device, or virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device can be a touch layer covering the display screen, or a button, trackball, or touchpad set on the shell of the dynamic warning system for driving risks based on the fusion of human factors and road conditions. It can also be an external keyboard, touchpad, or mouse, etc.

[0166] It should be noted that, in order to simplify the description of the present invention and thus help to understand one or more embodiments of the invention, multiple features may sometimes be grouped into one embodiment, drawing or description thereof in the foregoing description of the embodiments of the present invention.

Claims

1. A dynamic early warning method for driving risks based on the fusion of human factors and road conditions, characterized in that, The method includes: The system collects dynamic status data of drivers, escorts, and real-time road status data during the driving process. Based on the collected dynamic status data of drivers and escorts, it generates a sequence of dynamic evolution of human factors and a sequence of real-time changes in road conditions. The interaction between each state node in the dynamic evolution sequence of human factors and each road condition node in the real-time change sequence of road conditions is analyzed, and a two-way correlation model between human factors and road conditions is established. Based on the aforementioned bidirectional correlation model between human factors and road conditions, risk triggering conditions under different combinations of state nodes and road condition nodes are extracted, and a set of driving risk transmission paths is identified. For each driving risk transmission path in the set of driving risk transmission paths, path features are extracted, and the risk impact range corresponding to each driving risk transmission path is determined based on the extracted path features; Based on the risk impact range corresponding to each driving risk transmission path, a dynamic driving risk warning instruction is generated, which includes a description of the risk transmission direction and the risk impact range. The dynamic driving risk warning instruction is then sent to the driving control terminal, triggering the driving control terminal to perform a warning prompt operation.

2. The method for dynamic early warning of driving risks based on the fusion of human factors and road conditions according to claim 1, characterized in that, The process involves collecting dynamic state data of the driver, escort, and real-time road conditions during the driving process. Based on the collected dynamic state data of the driver and escort, a dynamic evolution sequence of human factors is generated; based on the collected real-time road condition data, a real-time change sequence of road conditions is generated. This includes: The driver's facial dynamic image data is collected by the image acquisition device installed in the driver's seat at preset time intervals. The driver's eye state data and facial expression data are extracted from the collected facial dynamic image data. The eye state data records the driver's eyelid closure frequency and eyeball rotation trajectory, and the facial expression data records the driver's mouth corner change range and facial muscle contraction state. By wearing physiological sensing devices on the driver, the driver's physiological dynamic data is collected synchronously. The physiological dynamic data records the driver's heart rate change data, respiratory rate data, and skin conductance response data. The collected driver's eye state data, facial expression data and physiological dynamic data are integrated to obtain the driver's dynamic state data. The audio and image acquisition devices installed in the carriage collect the escort's voice interaction data and movement posture data. The voice interaction data records the content of the escort's voice communication with the driver and the changes in voice tone. The movement posture data records the escort's body movement range and direction. The collected escort's voice interaction data and movement posture data are integrated to obtain the escort's dynamic state data. The collected dynamic status data of drivers are segmented according to time sequence, with each segment corresponding to a time window. The core features of the dynamic status data of drivers within each time window are extracted to form driver status nodes. Similarly, the collected dynamic status data of escorts are segmented according to the same time window, and the core features of the dynamic status data of escorts within each time window are extracted to form escort status nodes. The driver status node and the escort status node corresponding to the same time window are associated and bound to form a human factor status node. All human factor status nodes are connected in sequence according to the time window to generate a dynamic evolution sequence of human factor status. By using road image acquisition equipment installed on the outside of the vehicle, dynamic image data of the road surface is collected. From the dynamic image data of the road surface, data on the status of road surface coverings and data on the damage status of the road surface are extracted. The data on the status of road surface coverings records the type and coverage area of ​​the road surface coverings, while the data on the damage status of the road surface records the distribution of cracks and the location of potholes. By installing environmental sensing devices on vehicles, dynamic environmental data around the road is collected. The dynamic environmental data around the road records the movement trajectory of obstacles around the road, changes in light around the road, and precipitation around the road. By using vehicle positioning equipment and traffic information receiving equipment, dynamic traffic data of the road is collected. The dynamic traffic data of the road records the lane occupancy, the speed of adjacent vehicles, and changes in road traffic signals. The collected data on road surface cover status, road surface damage status, road surrounding environment dynamics, and road traffic dynamics are segmented according to time sequence. The segmented time window is consistent with the time window for generating human condition status nodes. The core features of the above data within each time window are extracted to form road condition nodes. All road condition nodes are connected in sequence according to the time window to generate a real-time road condition change sequence.

3. The method for dynamic early warning of driving risks based on the fusion of human factors and road conditions according to claim 1, characterized in that, The analysis of the interaction between each state node in the dynamic evolution sequence of human factors and each road condition node in the real-time change sequence of road conditions establishes a two-way correlation model between human factors and road conditions, including: All human-factor state nodes are extracted from the dynamic evolution sequence of human-factor state, and each human-factor state node includes a driver state sub-node and an escort state sub-node. All road condition nodes are extracted from the real-time road condition change sequence, and each road condition node includes a road surface state sub-node, a road surrounding environment state sub-node, and a road traffic state sub-node. Each human factor status node is paired with the road condition node corresponding to the same time window to form multiple pairs of human factor road condition nodes. Each pair of human factor road condition nodes includes a human factor status node and a corresponding road condition node. For each pair of human-factor and road condition nodes, feature decomposition is performed on the human-factor state nodes and road condition nodes. Driver operation tendency features, driver fatigue features, and escort prompt features are extracted from the human-factor state nodes. Road resistance features, road vision interference features, and road traffic conflict features are extracted from the road condition nodes. Calculate the degree of interaction between driver's operational tendencies and road resistance characteristics, driver fatigue characteristics and road visual interference characteristics, and escort personnel's prompting characteristics and road traffic conflict characteristics for the same person at road condition nodes. The calculated levels of the three interaction effects are standardized, and the standardized levels of the three interaction effects are weighted and integrated to obtain the overall correlation strength value of each pair of human-road condition node pairs. In the weighted integration process, corresponding weight coefficients are assigned according to the importance of each interaction effect level under different driving scenarios. By comparing the overall correlation strength values ​​of human factors and road conditions node pairs in adjacent time windows according to the order of time windows, analyzing the changing trend of the overall correlation strength values, identifying the time windows in which the overall correlation strength values ​​change abruptly, and marking the human factors and road conditions node pairs corresponding to the abrupt change time windows as key node pairs; Extract the changes in human state characteristics and road condition characteristics from all key node pairs, analyze the triggering effect of changes in human state characteristics on changes in road condition characteristics, and analyze the feedback effect of changes in road condition characteristics on changes in human state characteristics to form a description of two-way influence relationship. Based on the description of the bidirectional influence relationship of all key node pairs and the overall association strength value of non-key node pairs, a bidirectional association model of human factors and road conditions is constructed, which includes node association rules, feature interaction rules, and time evolution rules. The node association rules define the association conditions between different types of human factor state nodes and road condition nodes, the feature interaction rules define the interaction mode between human factor features and road condition features, and the time evolution rules define the change pattern of the association relationship over time.

4. The method for dynamic early warning of driving risks based on the fusion of human factors and road conditions according to claim 1, characterized in that, Based on the aforementioned human-road-condition bidirectional correlation model, risk triggering conditions under different combinations of state nodes and road condition nodes are extracted, and a set of driving risk transmission paths is identified, including: All node association rules are extracted from the bidirectional human-factor and road condition association model. Each node association rule corresponds to a combination of human-factor state node type and road condition node type. For each combination of human factor state node type and road condition node type, analyze the possible abnormal human factor characteristics and abnormal road condition characteristics under this combination. Abnormal human factor characteristics include delayed driver operation response, driver inattention, and untimely prompts from escort personnel. Abnormal road condition characteristics include reduced road surface friction coefficient, sudden appearance of obstacles around the road, and reduction of road traffic lanes. Each human factor characteristic anomaly is combined with the corresponding road condition characteristic anomaly to form multiple sets of anomaly combination scenarios. Each set of anomaly combination scenarios includes one human factor characteristic anomaly and one corresponding road condition characteristic anomaly. For each abnormal combination scenario, trigger condition analysis is performed to determine the initial triggering factors that lead to the risk under the abnormal combination scenario, and at the same time, the order in which the initial triggering factors cause subsequent characteristic changes is determined. Based on the initial triggering factors and the sequence of feature changes, a risk transmission chain is constructed for each abnormal combination scenario. The risk transmission chain starts from the initial triggering factor, connects sequentially to the triggered human factor feature changes or road condition feature changes, and ends at the possible driving risk outcome. All the risk transmission chains constructed are matched with the time evolution rules in the bidirectional correlation model of human factors and road conditions to verify whether the time interval of each link in the risk transmission chain meets the time threshold requirements in the time evolution rules, and risk transmission chains that do not meet the time threshold requirements are eliminated. For the verified risk transmission chain, supplement the missing feature change links in the chain so that the risk transmission chain fully reflects the transmission process from the initial triggering factor to the driving risk result; All the completed risk transmission chains were classified and organized into three groups according to the type of risk transmission initiation factors: human-induced triggering, road condition-induced triggering, and human-induced and road condition-induced triggering. The risk transmission chains in each group constitute the driving risk transmission subset under that type. All driving risk transmission subsets are integrated to form a driving risk transmission path set containing different types of risk transmission paths. Each driving risk transmission path corresponds to a unique path identifier in the set. The path identifier includes information on the type of initiating factor, the number of transmission links, and the type of risk outcome.

5. The method for dynamic early warning of driving risks based on the fusion of human factors and road conditions according to claim 1, characterized in that, The step of extracting path features for each driving risk transmission path in the set of driving risk transmission paths, and determining the risk impact range corresponding to each driving risk transmission path based on the extracted path features, includes: Select a driving risk transmission path from the set of driving risk transmission paths, extract all transmission links in the driving risk transmission path, and each transmission link corresponds to a feature change event. The feature change event includes the event type, the time of the event, and the human factors or road conditions that affect the event. The impact of each characteristic change event on vehicle driving parameters, including driving speed, driving direction, braking distance, and steering angle, is analyzed. By comparing the changes in vehicle driving parameters before and after the event, the independent impact value of each characteristic change event on each vehicle driving parameter is determined. The independent impact values ​​of each characteristic change event are normalized separately, and the normalized independent impact values ​​are combined to obtain the single, dimensionless parameter impact value of the characteristic change event. The parameter impact values ​​of all transmission links are accumulated in the order of transmission to obtain the total parameter impact value of the driving risk transmission path. During the accumulation process, each parameter impact value is weighted according to the position weight of the transmission link in the path. The weight coefficient value of the link at the front end of the path is less than the weight coefficient value of the link at the back end of the path. The transmission speed characteristics of the driving risk transmission path are extracted. The transmission speed characteristics are obtained by calculating the average time interval between adjacent transmission links. The smaller the average time interval value, the faster the transmission speed value corresponding to the transmission speed characteristics. Based on the total parameter impact value and transmission speed characteristics, a preset impact range mapping table is queried. The impact range mapping table stores the risk impact range descriptions corresponding to different total parameter impact value ranges and transmission speed ranges. The risk impact range descriptions include the affected vehicle system and the affected driving area. The risk outcome types in the driving risk transmission path are analyzed to determine the secondary risk types that the risk outcomes may cause. The risk impact range descriptions obtained from the query are supplemented with the secondary risk types to form risk impact range information. The secondary risk types include vehicle loss of control risk, rear-end collision risk, and lane departure risk. Repeat the above steps to extract path features and determine the risk impact range for each driving risk transmission path in the set of driving risk transmission paths, so that each driving risk path corresponds to unique risk impact range information; The path identifier of each driving risk transmission path is associated with and stored with the corresponding risk impact range information to form a path impact range association table. The path impact range association table also includes the total parameter impact value and transmission speed characteristics of each path.

6. The method for dynamic early warning of driving risks based on the fusion of human factors and road conditions according to claim 1, characterized in that, The process involves combining the risk impact range corresponding to each driving risk transmission path to generate a dynamic driving risk warning instruction containing a description of the risk transmission direction and risk impact range. This instruction is then sent to the driving control terminal, triggering the terminal to perform a warning notification operation, including: Extract the path identifier, risk impact range information, total parameter impact value, and transmission speed characteristics of all driving risk transmission paths from the path impact range association table; All driving risk transmission paths are prioritized based on the magnitude of the total parameter impact value. The larger the total parameter impact value, the higher the priority of the path. For paths with the same total parameter impact value, a secondary sort is performed based on the transmission speed characteristic. The faster the transmission speed, the higher the priority of the path. A preset number of driving risk transmission paths are selected as core early warning paths in descending order of priority. The preset number is determined based on the complexity of the current driving scenario. Extract the risk transmission path of each core early warning path, determine the characteristic change direction of each transmission link in the risk transmission path, and connect the characteristic change directions in the transmission order to obtain the risk transmission direction of the core early warning path. The risk transmission direction is represented by a sequence of direction vectors, and each direction vector corresponds to the characteristic change direction of a transmission link. Extract the affected vehicle systems and affected driving areas from the risk impact range information corresponding to each core early warning path, integrate the affected vehicle systems and affected driving areas into a risk impact range description, and use structured language to describe the risk impact range to determine the specific impact mode of each affected object; The path identifier, risk transmission direction, and risk impact range description of each core early warning path are integrated to form a single path early warning information. Each single path early warning information includes a path identifier field, a transmission direction field, and an impact range field. All core warning paths are summarized, and the current warning generation time and warning validity period information are added to generate a dynamic warning instruction for driving risks. The dynamic warning instruction for driving risks is written in a format that can be parsed by the driving control terminal, and includes an instruction header, instruction body, and instruction tail. The instruction header contains an instruction type identifier, the instruction body contains all single-path warning information, and the instruction tail contains relevant information. The generated dynamic driving risk warning command is sent to the driving control terminal. After receiving the command, the terminal parses the single-path warning information in the command body and determines the corresponding warning prompt method based on the risk transmission direction and risk impact range description. The determined warning prompt method is then executed, and the warning execution time, warning prompt method, and driver's response are recorded to form a warning execution log. The warning execution log is used for subsequent evaluation of the warning effect. Warning prompt methods include voice prompts, light prompts, and seat vibration prompts. Different risk impact range descriptions correspond to different combinations of warning prompts.

7. The method for dynamic early warning of driving risks based on the fusion of human factors and road conditions according to claim 2, characterized in that, The dynamic status data of drivers collected in chronological order is segmented, with each segment corresponding to a time window. The core features of the driver's dynamic status data within each time window are extracted to form driver status nodes. Similarly, the dynamic status data of security guards collected in the same time window is segmented, and the core features of the security guard's dynamic status data within each time window are extracted to form security guard status nodes, including: The duration of the time window is determined based on the vehicle's speed. The faster the speed, the shorter the time window; the slower the speed, the longer the time window. Starting from the start time of data collection, the collected dynamic status data of drivers is segmented according to the determined time window duration to obtain multiple segments of driver dynamic status data. Each segment of driver dynamic status data corresponds to a time window identifier. Statistical analysis was performed on the eye state data in each segment of driver dynamic state data. The average value of eyelid closure frequency and the coverage of eyeball rotation trajectory within the time window were calculated. The average value and coverage were used as the core features of eye state. The facial expression data in the segmented data of the driver's dynamic state were analyzed, and the maximum and minimum values ​​of the change in the corner of the mouth and the duration of the facial muscle contraction were extracted. The maximum, minimum and duration were taken as the core features of facial expression. The physiological dynamic data in the segmented data of the driver's dynamic state were statistically analyzed, and the fluctuation amplitude of heart rate change data, the average value of respiratory rate data, and the peak value of skin conductance response data were calculated. The fluctuation amplitude, average value, and peak value were used as the core physiological dynamic features. The core features of eye state, facial expression, and physiological dynamics are standardized respectively. The standardized core features of eye state, facial expression, and physiological dynamics are then integrated into a feature vector to form the driver state node corresponding to the time window. The driver state node includes the time window identifier and the corresponding feature vector. Using the same time window duration, the collected dynamic status data of the escorts are segmented to obtain multiple segments of escort dynamic status data. Each segment of escort dynamic status data corresponds to the same time window identifier as the driver's status node. The voice interaction data in each segment of the escort's dynamic status data is analyzed to extract the number of keywords and the number of tone changes in the voice communication content. The number of keywords and the number of tone changes are taken as the core features of voice interaction. The motion posture data in the segmented dynamic state data of the escort personnel were analyzed, and the average value of the limb movement amplitude and the angle between the direction of the movement and the direction of vehicle travel were calculated. The average value and the angle were used as the core features of the motion posture. The core features of voice interaction and the core features of action posture are standardized separately. The standardized core features of voice interaction and the core features of action posture are integrated into a feature vector to form the escort status node corresponding to the time window. The escort status node contains the same time window identifier and corresponding feature vector as the driver status node.

8. The method for dynamic early warning of driving risks based on the fusion of human factors and road conditions according to claim 3, characterized in that, The calculation of the interaction between driver's operational tendencies and road resistance characteristics, the interaction between driver fatigue characteristics and road visibility interference characteristics, and the interaction between escort personnel's prompting characteristics and road traffic conflict characteristics for the same person at road condition nodes include: Driver operation tendency features are extracted from the human state nodes of the same human-factor road condition node pair. These driver operation tendency features include steering operation frequency, braking operation frequency, and acceleration operation frequency. Road resistance features are extracted from the road condition nodes of the same node pair. These road resistance features include road gradient, road surface friction coefficient, and road smoothness. An interaction matrix is ​​established between driver operation tendency characteristics and road resistance characteristics. The rows of the interaction matrix represent the parameters of driver operation tendency characteristics, and the columns represent the parameters of road resistance characteristics. Each element in the matrix is ​​used to store the interaction influence value of the corresponding parameter pair. Based on historical driving data, the correlation coefficient between the standardized steering operation frequency and the standardized road slope is calculated as the interaction value; the correlation coefficient between the standardized braking operation frequency and the standardized road surface friction coefficient is calculated as the interaction value; the correlation coefficient between the standardized acceleration operation frequency and the standardized road smoothness is calculated as the interaction value. All interaction influence values ​​in the interaction matrix are normalized so that all values ​​are in the same numerical range. Then the trace of the interaction matrix is ​​calculated, and the trace value is used as the degree of interaction between driver operation tendency characteristics and road resistance characteristics. Driver fatigue features are extracted from the human state nodes of the same human-cause road condition node pair. Driver fatigue features include the proportion of eyelid closure time, heart rate variability coefficient, and reaction delay time. Road vision interference features are extracted from the road condition nodes of the same node pair. Road vision interference features include the amplitude of light intensity change, visibility caused by precipitation, and the range of obstruction by surrounding objects. Construct a correlation function between driver fatigue characteristics and road vision interference characteristics. The input of the correlation function is the parameters of driver fatigue characteristics and the parameters of road vision interference characteristics, and the output is the interaction influence value. The standardized eyelid closure duration percentage and the standardized light intensity variation amplitude are input into the first submodule of the correlation function to calculate their combined effect value. The combined effect value is the result of a function trained on historical data that outputs a dimensionless evaluation value. The standardized heart rate variability coefficient and the standardized visibility caused by precipitation are input into the second submodule of the correlation function to calculate their coupled effect value. The coupled effect value is the result of a function trained on historical data that outputs a dimensionless evaluation value. The standardized reaction delay time and the standardized occlusion range of surrounding objects are input into the third submodule of the correlation function to calculate their superimposed effect value. The superimposed effect value is the result of a function trained on historical data that outputs a dimensionless evaluation value. The output values ​​of the first submodule, the second submodule, and the third submodule are weighted and summed to obtain the degree of interaction between driver fatigue characteristics and road vision interference characteristics. Extract escort personnel prompt features from the human state nodes of the same human-cause road condition node pair. The escort personnel prompt features include prompt frequency, prompt accuracy, and prompt response time. Extract road traffic conflict features from the road condition nodes of the same node pair. The road traffic conflict features include the approach speed of adjacent vehicles, the number of lane occupancy conflicts, and the frequency of traffic signal changes. An impact assessment model is established for the characteristics of escort personnel prompts and road traffic conflicts. The impact assessment model includes an input layer, a processing layer, and an output layer. The input layer receives various parameters of the two types of features, the processing layer performs interactive calculations on the parameters, and the output layer outputs the degree of interactive impact. The input layer inputs the standardized alert frequency and the standardized approach speed of adjacent vehicles into the first computing unit of the processing layer to calculate the mitigation coefficient of the alert frequency on the risk of adjacent vehicle approach. The mitigation coefficient is the result of a function calculation based on historical data and outputting dimensionless coefficients. The input layer inputs the standardized prompt accuracy and the standardized number of lane occupancy conflicts into the second processing unit of the processing layer to calculate the lane occupancy conflict avoidance efficiency of the prompt accuracy. The avoidance efficiency is the result of a function calculation based on historical data training and outputting dimensionless coefficients. The input layer inputs the standardized prompt response time and the standardized traffic signal change frequency into the third processing unit of the processing layer to calculate the response time's effectiveness in responding to traffic signal change risks. The effectiveness is a function calculation result based on historical data training, outputting dimensionless coefficients. The processing layer comprehensively evaluates the output results of the first, second, and third operation units to determine the proportion of positive and negative impacts. The output layer calculates the degree of interaction between the escort officer prompting feature and the road traffic conflict feature based on the proportion of positive and negative impacts.

9. A dynamic early warning system for driving risks based on the fusion of human factors and road conditions, characterized in that, include: processor; A machine-readable storage medium for storing machine-executable instructions of the processor; The processor is configured to execute the dynamic warning method for driving risks based on the fusion of human factors and road conditions as described in any one of claims 1 to 8 by executing the machine-executable instructions.

10. A computer program product, characterized in that, The computer program product includes machine-executable instructions stored in a computer-readable storage medium. The processor of the dynamic warning system for driving risks based on the fusion of human factors and road conditions reads the machine-executable instructions from the computer-readable storage medium and executes the machine-executable instructions, causing the dynamic warning system for driving risks based on the fusion of human factors and road conditions to perform the dynamic warning method for driving risks based on the fusion of human factors and road conditions as described in any one of claims 1 to 8.

Images