Method for evaluating driving fatigue of truck driver on long straight road in plateau high-altitude area

Abstract

The invention discloses a method for evaluating driving fatigue of a truck driver on a long straight road in a plateau high-altitude area. The method comprises the following steps: S1, constructing a driving simulation experiment model including a day and night environment and a low-visibility scene; s2, collecting subjective and objective data of a driver; s3, performing visualization and statistical analysis on the data, and identifying day and night differences; and S4, establishing a fatigue probability model based on logistic regression, and respectively constructing fatigue evaluation models in daytime and night scenes. By means of multi-dimensional data fusion, day and night parting modeling, plateau environment simulation and the like, accurate and efficient evaluation of the fatigue state of the truck driver on the plateau long straight road is achieved, and the method has high practical value and popularization prospects.

Description

Technical Field

[0001] This invention relates to the field of driver fatigue detection technology, specifically to a method for assessing driver fatigue in truck drivers on long, straight roads in high-altitude and plateau regions. Background Technology

[0002] In highway networks, long, straight sections are widely used due to their good visibility, high traffic efficiency, and low energy consumption. However, these sections often feature monotonous alignments and repetitive environments, easily leading to insufficient visual stimulation and driver distraction, thus inducing fatigue driving and becoming a potential cause of highway traffic accidents. Large trucks, in particular, are a high-risk group for fatigue driving due to their long travel times and high driving intensity. Fatigue driving refers to the phenomenon where, during prolonged continuous driving, drivers experience physiological and psychological dysfunction, resulting in decreased operational ability and misjudgment, significantly increasing the risk of accidents. Research shows that fatigue is not only related to driving time but is also influenced by external factors such as the monotony of the road environment and insufficient visual stimulation. To improve road traffic safety, it is necessary to systematically study the causes and characteristics of driver fatigue, identify the behavioral patterns of drivers in perception, judgment, and operation under different fatigue states, and verify the effectiveness of fatigue intervention strategies through simulation experiments.

[0003] For example, Chinese invention patent application CN120783323A (publication date October 14, 2025) discloses a method and system for detecting facial fatigue in tractor drivers based on deep learning. It mainly includes three stages: data-driven cloud space simulation, cloud band synthesis guided by scattering laws, and dataset construction based on imaging mechanisms. The method includes: acquiring facial images of the driver and constructing a facial feature dataset; constructing a facial feature detection model based on the facial feature dataset; configuring a facial key point dataset of the tractor driver and constructing a facial key point detection model; configuring the constructed facial feature detection model and facial key point detection model as the pre-processing and post-processing steps of the facial fatigue detection model, thereby constructing the facial fatigue detection model to detect the driver's facial fatigue features and determine the fatigue state. This invention, by constructing a detection model for the facial fatigue features of tractor drivers, detects the facial features of tractor drivers while they are driving the tractor, thereby determining the fatigue value, which improves accuracy compared to manual judgment.

[0004] Chinese invention patent application CN120612722A (published on September 9, 2025) provides a driver fatigue detection system and method for intelligent driving. By monitoring the driver's facial expressions and driving behavior in real time, the system can accurately assess the driver's fatigue state. When a high level of fatigue is detected, the system will automatically activate the corresponding intelligent driving mode. For example, in the first-level intelligent driving mode, the system will guide the vehicle to a safe location to stop, avoiding traffic accidents caused by driver fatigue and greatly ensuring driving safety.

[0005] Chinese invention patent application CN120705677A (published on September 26, 2025) provides a method for identifying driver fatigue levels that considers the loss of driving ability. It uses multi-feature fusion for fatigue driving identification and detection, avoiding over-reliance on a single or few features to determine fatigue status, thus improving model recognition performance and robustness. This invention takes into account the individual differences among drivers during fatigue driving and implements special processing for various features to reduce or even eliminate model classification errors caused by individual differences.

[0006] Current research on driver fatigue largely focuses on facial features, physiological and psychological factors, or broad perspectives, generally lacking specific studies targeting long, straight sections of highways. Furthermore, existing methods have limitations in environmental adaptability, data collection dimensions, and model discrimination, failing to fully consider the impact of diurnal environmental changes, individual differences, and multi-source data fusion on fatigue assessment. Therefore, there is an urgent need for a driver fatigue assessment method that can adapt to the unique road environments of high-altitude plateaus and incorporates multi-source physiological and behavioral data to improve the accuracy and practicality of fatigue identification. Summary of the Invention

[0007] To address the above technical problems, this invention provides a method for assessing driver fatigue in trucks traveling on long, straight roads in high-altitude plateau regions, comprising the following steps: Step S1, Driving simulation experiment: Set different vehicle types and define driving behaviors on the road model according to the simulation requirements; set the time transition mechanism from day to night, define the nighttime environmental parameters and set low visibility scenarios on specific road sections to complete the establishment of daytime and nighttime environmental scene models; Step S2, Data Collection: Data is collected through both subjective and objective data collection methods; Step S3: Data visualization and descriptive statistical analysis: The data collected in step S2 is visualized and analyzed in the form of a three-line table, and paired t-tests are performed on each data indicator for daytime and nighttime to analyze the statistical differences between daytime and nighttime data. Step S4: Establish the relationship between various statistical indicators and fatigue: Select vehicle speed, vehicle lateral deviation, blink frequency, fixation duration, pupil area, and steering wheel angle as independent variables, and driver fatigue level as dependent variable to establish a relationship model between road environment and driving fatigue for driving fatigue state detection; in the relationship model, 0 represents no fatigue and 1 represents fatigue.

[0008] Furthermore, the road model described in step S1 includes long straight road sections, low visibility sandy road sections, and wetland speed-limited road sections.

[0009] Furthermore, step S1 includes the following steps: Step S110: Establish a road simulation model: Import the road design drawings into the software to generate a road network, set the number of road lanes, the width of the central median, the direction of vehicle traffic and the design speed of the road network, and add traffic signs and directional signs to the road network. Step S120: Establish a landscape model: Set up landscapes on both sides of the road in the road model. Set up tree models on both sides of the road at about 5km and 10km to remind the driver of the position of the driver in the road model. Near the end of the road model, set up a checkpoint and a speed reduction sign to remind the driver that the driving simulation road is about to end. Step S130: Establish vehicle model: Set up trucks and cars that are driving normally on the road model, and set the vehicle behavior; in the daytime simulated road section, set up oncoming vehicles in the opposite lane, mainly trucks, with a speed of 100km / h; in the nighttime simulated road section, no other vehicles are set up. Step S140: Set the nighttime environment: When the driver drives the vehicle for 200m in the simulated road section, the time gradually changes from daytime to nighttime; the ambient time is 20:00 at night; in the last 2000m section, a foggy section is set to reduce the visibility in the simulated scene to simulate a sandy road section.

[0010] Furthermore, the specific method for subjective data collection in step S2 is as follows: Step S201, Subject Recruitment and Screening: Male individuals who have driven large vehicles are invited to participate in the experiment, covering different age groups, driving experience, and driver's license types; Step S202, Basic Information Collection: Collect statistics on the subject's age, driving experience, driver's license type, current vehicle type, and manual / automatic transmission driving status; Step S203, Driving Behavior Survey and Statistics: Collect data on the subjects' highway driving frequency, weekly driving frequency, and average driving time on weekdays; calculate the subjects' average annual and past three years driving mileage to assess driving proficiency; Step S204, Traffic Accident Investigation and Statistics: Record whether minor or major traffic accidents have occurred, and analyze driving safety.

[0011] Furthermore, the specific method for objective data collection in step S2 is as follows: data collected by instruments and equipment and data obtained by experimental personnel observing the driving behavior of the subjects. The objective data includes vehicle data, eye tracker data, hand data, and heart rate data.

[0012] Furthermore, the vehicle data includes vehicle speed and lateral offset distance. The specific method for collecting the vehicle data is as follows: real-time data of the simulated driving vehicle is detected using A-Lab software, and after the experiment, the data is calculated and summarized into a TXT format data document.

[0013] Furthermore, the eye tracker data includes the subject's blink frequency, blink duration, pupil size, and gaze direction. The specific method for collecting the eye tracker data is as follows: using the DikablisPro head-mounted eye tracker, the subject's eyes are recorded in video and in a first-person perspective, and the eye tracker data is recorded and exported as specific data in TXT format.

[0014] Furthermore, the hand data includes the tightness of gripping the steering wheel, the degree of tension and sweating, and the speed of steering wheel rotation; the specific method for collecting the hand data is as follows: the experimenter observes the subject's hand state during simulated driving and compiles the specific data in combination with the subject's questionnaire results.

[0015] Furthermore, the heart rate data is ECG heart rate monitor data, and the heart rate data acquisition method specifically involves: real-time monitoring of the subject's heart rate using an ECG heart rate monitor, and finally analyzing the heart rate data and outputting it as a TXT file.

[0016] Furthermore, in step S4, the dependent variable is standardized using the following formula: , Where μ is the mean and σ is the standard deviation. For the standardized dependent variable, The dependent variable before standardization.

[0017] Furthermore, the relationship model described in step S4 is an ordered multi-class Logistic regression model.

[0018] Furthermore, the formula for the Logistic regression model described in step S4 is: , Wherein: This represents the probability of a driver experiencing a certain level of fatigue. j = 1, 2, 3; x = (x1, x2, ..., xi), T is the characteristic independent variable; These are the regression coefficients of the model; The model intercept; The formula for calculating the probability of a driver being fatigued to a certain degree is: In the formula: y represents the driver's fatigue level; j represents the fatigue level grade, j = 1, 2, 3, ...; This is the i-th factor affecting driver fatigue.

[0019] Compared with existing technologies, the advantages and effects of this application are as follows: 1. This application proposes a simulation modeling method for high-altitude Gobi highway sections, integrating physiological indicators such as eye tracking and heart rate monitoring, combined with driving behavior data such as vehicle speed and lateral deviation, and subjective questionnaire feedback to construct a multi-dimensional fatigue driving analysis system. Breaking through the limitations of traditional single-data-dimensional research, this method achieves accurate identification of fatigue driving states through cross-validation of multi-source data, providing systematic data support for the quantitative study of the correlation between road landscape design and driving state, and significantly improving the reliability and comprehensiveness of the analysis results.

[0020] 2. This application uses a t-test to quantitatively analyze the significant differences in driving indicators (vehicle speed, pupil area, etc.) under daytime and nighttime scenarios, and for the first time establishes an independent daytime / nighttime logistic regression model to reveal the influence mechanism of fatigue driving under different lighting conditions. This fills the gap in the quantitative analysis of the impact of daytime and nighttime environmental differences on fatigue driving in existing research. The model results provide data support for the differentiated design of nighttime road landscapes (such as enhancing visual stimuli), making the research conclusions more aligned with the needs of actual traffic scenarios.

[0021] 3. This application establishes an ordered multi-category logistic regression model using six driving indicators, including vehicle speed, lateral deviation, and blink frequency, as independent variables to achieve a mathematical and statistical quantitative judgment of fatigue driving status. Compared with traditional subjective assessment methods, this model automatically calculates the fatigue probability through objective data, providing a standardized judgment basis for intelligent driving systems; the correlation analysis of variables in the model (such as the positive correlation between vehicle speed and fatigue) directly guides the optimization of road landscape parameters, providing a scientific design guide for improving driving safety.

[0022] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the preferred embodiments of this application are described in detail below with reference to the accompanying drawings.

[0023] The above and other objects, advantages and features of this application will become more apparent to those skilled in the art from the following detailed description of specific embodiments in conjunction with the accompanying drawings. Attached Figure Description

[0024] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In all drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to scale.

[0025] in: Figure 1 This is a flowchart illustrating a method for assessing driver fatigue in truck drivers on long, straight roads in high-altitude regions, as described in this application. Figure 2 This is a schematic diagram of objective data for a method to assess driver fatigue on long straight roads in high-altitude plateau regions, as described in this application. Figure 2 A represents real-time image data of the eye; Figure 2 B is a real-time heart rate monitoring graph; Figure 3 This is a visualization analysis diagram of vehicle speed for a method to assess driver fatigue on long straight roads in high-altitude plateau regions, as described in this application. Figure 3 A is a graph showing the change in driver speed during the daytime scene; Figure 3 B is a graph showing the change in driver speed during nighttime scenes; Figure 4 This is a visualization analysis diagram of vehicle lateral deviation in a method for assessing driver fatigue on long straight roads in high-altitude plateau regions, as described in this application. Figure 4 A is a diagram showing the lateral offset of vehicles in a daytime scene; Figure 4 B is the horizontal offset image of the night scene; Figure 5 This is a visualization analysis of blink frequency in a method for assessing driver fatigue on long straight roads in high-altitude plateau regions, as described in this application. Figure 5 A is a statistical chart of driver blinking frequency in daytime scenarios; Figure 5 B is a statistical chart showing the frequency of blinking by drivers in nighttime scenarios; Figure 6This is a visualization analysis diagram of the fixation time in a method for assessing driver fatigue on long straight roads in high-altitude plateau regions, as described in this application. Figure 6 A is a statistical chart showing the duration of driver gaze in daytime scenes; Figure 6 B is a statistical chart showing the duration of sustained gaze by drivers in nighttime scenes; Figure 7 This is a visual analysis diagram of hand features for a method to assess driver fatigue on long straight roads in high-altitude plateau regions, as described in this application. Figure 7 A shows the changes in steering wheel angle during the daytime scene; Figure 7 B is a diagram showing the change in steering wheel angle during nighttime scenes. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. In the following description, specific details such as specific configurations and components are provided merely to help fully understand the embodiments of this application. Therefore, those skilled in the art should understand that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this application. In addition, for clarity and brevity, descriptions of known functions and structures are omitted in the embodiments.

[0027] It should be understood that the phrase "an embodiment" or "this embodiment" throughout the specification means that a specific feature, structure, or characteristic related to the embodiment is included in at least one embodiment of this application. Therefore, "an embodiment" or "this embodiment" appearing throughout the specification does not necessarily refer to the same embodiment. Furthermore, these specific features, structures, or characteristics can be combined in any suitable manner in one or more embodiments.

[0028] Furthermore, reference numerals and / or letters may be repeated in different examples within this application. Such repetition is for the purpose of simplification and clarity and does not in itself indicate a relationship between the various embodiments and / or settings discussed.

[0029] In this article, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can mean: A exists alone, B exists alone, and A and B exist simultaneously. The term " / and" describes another type of relationship between related objects, indicating that two relationships can exist. For example, A / and B can mean: A exists alone, and A and B exist alone. In addition, the character " / " in this article generally indicates that the related objects before and after it have an "or" relationship.

[0030] In this article, the term "at least one" is merely a description of the relationship between related objects, indicating that there can be three relationships. For example, "at least one of A and B" can mean: A exists alone, A and B exist simultaneously, or B exists alone.

[0031] It should also be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion.

[0032] Example 1 This embodiment introduces a method for assessing driver fatigue in truck drivers on long, straight roads in high-altitude plateau regions.

[0033] Please refer to Figure 1 As shown, Figure 1 This is a flowchart illustrating a method for assessing driver fatigue in trucks traveling on long, straight roads in high-altitude regions, as described in this application.

[0034] Step S1, Driving simulation experiment: Set different vehicle types and define driving behaviors on the road model according to the simulation requirements; set the time transition mechanism from day to night, define the nighttime environmental parameters and set low visibility scenarios on specific road sections to complete the establishment of daytime and nighttime environmental scene models; Step S2, Data Collection: Data is collected through both subjective and objective data collection methods; Step S3: Data visualization and descriptive statistical analysis: The data collected in step S2 is visualized and analyzed in the form of a three-line table, and paired t-tests are performed on each data indicator for daytime and nighttime to analyze the statistical differences between daytime and nighttime data. Step S4: Establish the relationship between various statistical indicators and fatigue: Select vehicle speed, vehicle lateral deviation, blink frequency, fixation duration, pupil area, and steering wheel angle as independent variables, and driver fatigue level as dependent variable to establish a relationship model between road environment and driving fatigue for driving fatigue state detection; in the relationship model, 0 represents no fatigue and 1 represents fatigue.

[0035] The technical effect of this embodiment is that it provides a complete fatigue assessment process framework, clarifies the entire process from experimental modeling to fatigue probability output, and lays the foundation for the refinement and optimization of subsequent embodiments.

[0036] Example 2 Based on Example 1, this example discloses a further design of step S1 in a method for assessing driver fatigue on long straight roads in high-altitude plateau regions.

[0037] Furthermore, the road model described in step S1 includes long straight road sections, low visibility sandy road sections, and wetland speed-limited road sections.

[0038] Furthermore, step S1 includes the following steps: Step S110: Establish a road simulation model: First, import the road CAD design drawings into SILAB software. Draw the road centerline in segments according to the CAD drawings and design the road speed. Then, set the number of lanes and the width of the central median strip, connect each lane segment, and set the vehicle traffic direction. Next, draw the road markings. According to the road design drawings, add lane edge lines, carriageway boundaries, and central median strip indicators to the road network. Finally, according to the road design drawings, using the station numbers as references, add traffic signs, directional signs, etc., to the road network.

[0039] Step S120: Establish a landscape model: Based on the road design drawings, set up the landscape on both sides of the road, adding relevant trees, buildings, grass, etc. According to the actual road environment data, the simulated road section is mainly composed of Gobi desert with few man-made landscapes. The simulated road section is a highway, a two-way four-lane road with a lane width of 3.75 meters and a 3-meter-wide central median. The simulated road section is approximately 15km long. In setting up the road model: tree models are placed on both sides of the road at approximately 5km and 10km to remind drivers of their position in the road model; a checkpoint with a speed reduction sign is set near the end of the road model to remind drivers that the simulated road is about to end.

[0040] Step S130: Establish vehicle models: According to the simulation experiment requirements, set up vehicles such as trucks and cars that are driving normally on the road network, and set the vehicle behavior to simulate real driving conditions. The simulated vehicles are large trucks. The speed limit for large trucks on highways is 80~100km / h, and they drive in the rightmost lane. In the daytime simulated road section, oncoming vehicles are set up, mainly large trucks, with a speed of 100km / h. In the nighttime simulated road section, no other vehicles are set up.

[0041] Step S140: Set the nighttime environment: When the driver drives the vehicle for about 200m in the simulated road section, the time gradually changes from daytime to nighttime. The following road sections are all nighttime driving sections, with the environment time being 20:00 at night. In the last section of about 2000m, a foggy section is set to reduce the visibility in the simulated scene to simulate a sandstorm road section.

[0042] The technical effect of this embodiment is that by constructing a highly realistic road model using SILAB software, and combining landscape, vehicle, and day / night environment settings, the driving scenario on a long straight road in the plateau is realistically reproduced, thereby improving the reliability and representativeness of the experimental data.

[0043] Example 3 Based on Embodiment 1 or 2, this embodiment discloses a further design of step S2 of a method for assessing driver fatigue on long straight roads in high-altitude plateau regions.

[0044] Furthermore, the specific method for subjective data collection in step S2 is as follows: Step S201, Subject Recruitment and Screening: Male individuals who have driven large vehicles are invited to participate in the experiment, covering different age groups, driving experience, and driver's license types (such as A1, A2, B2, C1). Step S202, Basic Information Collection: Collect statistics on the subject's age, driving experience, driver's license type, current vehicle type (heavy-duty trailer / large truck / medium-duty truck, etc.) and manual / automatic transmission driving status; Step S203, Driving Behavior Survey and Statistics: Collect data on the subjects' highway driving frequency, weekly driving frequency, and average driving time on weekdays; calculate the subjects' average annual and past three years driving mileage to assess driving proficiency; Step S204, Traffic Accident Investigation and Statistics: Record whether minor or major traffic accidents have occurred, and analyze driving safety.

[0045] Please refer to Figure 2. Figure 2 This is a schematic diagram of objective data for a method to assess driver fatigue on long straight roads in high-altitude plateau regions, as described in this application.

[0046] Furthermore, the specific method for objective data collection in step S2 is as follows: data collected by instruments and equipment and data obtained by experimental personnel observing the driving behavior of the subjects. The objective data includes vehicle data, eye tracker data, hand data, and heart rate data.

[0047] Furthermore, the vehicle data includes vehicle speed and lateral offset distance. The specific method for collecting the vehicle data is as follows: real-time data of the simulated driving vehicle is detected using A-Lab software, and after the experiment, the data is calculated and summarized into a TXT format data document.

[0048] Furthermore, the eye tracker data includes the subject's blink frequency, blink duration, pupil size, and gaze direction. The specific method for collecting the eye tracker data is as follows: using the DikablisPro head-mounted eye tracker, the subject's eyes are recorded in video and in a first-person perspective, and the eye tracker data is recorded and exported as specific data in TXT format.

[0049] Furthermore, the hand data includes the tightness of gripping the steering wheel, the degree of tension and sweating, and the speed of steering wheel rotation; the specific method for collecting the hand data is as follows: the experimenter observes the subject's hand state during simulated driving and compiles the specific data in combination with the subject's questionnaire results.

[0050] Furthermore, the heart rate data is ECG heart rate monitor data, and the heart rate data acquisition method specifically involves: real-time monitoring of the subject's heart rate using an ECG heart rate monitor, and finally analyzing the heart rate data and outputting it as a TXT file.

[0051] The technical effect of this embodiment is that it comprehensively collects multi-source data such as driver basic information, driving behavior, eye movement, and heart rate to ensure the comprehensiveness of fatigue assessment and consideration of individual differences.

[0052] Example 4 Based on Example 3, this example discloses a further design of step S3 in a method for assessing driver fatigue on long straight roads in high-altitude plateau regions.

[0053] (1) Visual analysis of vehicle speed Vehicle speed data is divided into daytime simulated driving speed data and nighttime simulated driving speed data. The data selection interval for both sets is from the first valid data point at the start of the vehicle's journey to the last valid data point at the checkpoint at the end of the road simulation. The average speed per minute for the corresponding vehicle is then selected from this data interval, with each data point spaced one minute apart.

[0054] Please refer to Figure 3. Figure 3This is a visualization analysis diagram of vehicle speed for a method to assess driver fatigue on long straight roads in high-altitude plateau regions, as described in this application.

[0055] Overall, in the driving simulation experiments under both daytime and nighttime scenarios, the speed changes of all drivers generally followed a parabolic trend, with the apex of the parabola, i.e., the highest speed, occurring in the latter part of the simulated driving. This demonstrates that in simulated desert environments with relatively simple terrain and few signs, drivers gradually develop and experience increased fatigue as driving time increases, leading to a lack of attention to speed changes and resulting in continuously rising speeds. When there are obvious speed limit signs, toll booths, or other buildings in the environment, or in environments with reduced visibility, drivers will control their speed to ensure driving safety. Under normal conditions, drivers are energetic, and their acceleration and deceleration are more frequent, resulting in more complex fluctuations in speed. For fatigued drivers, vehicle operation decreases, and they may even maintain the same speed for extended periods.

[0056] A paired t-test of daytime and nighttime driving speeds yields the following results: Standard deviation (difference): Standard error (SE): t-statistic: Degrees of freedom (df): The p-value was calculated to be 0.003 (<0.05), indicating that there is a statistically significant difference between the daytime and nighttime data.

[0057] (2) Visual analysis of vehicle lateral offset Please refer to Figure 4. Figure 4 This is a visualization analysis diagram of the lateral displacement of a truck driver in a method for assessing driver fatigue on long straight roads in high-altitude areas, as described in this application.

[0058] The driving simulation test road is a two-way four-lane road, with each lane being 3.75m wide. Since vehicles need to change lanes within one minute before the simulation begins, vehicles will initially experience significant deviations. Regarding the lateral deviations of the vehicles, we divided the data collection and analysis into two parts: daytime and nighttime scenarios. We used the maximum absolute value of the lateral deviation distance per minute for data filtering.

[0059] Overall, the drivers had good control over the vehicle's direction of travel because they were all skilled drivers who were familiar with the vehicle and could ensure safe driving, with minimal lateral deviations.

[0060] A paired t-test of daytime and nighttime vehicle offset data yields the following results: Standard deviation (difference): sd = 0.327, Standard error (SE): t-statistic: Degrees of freedom (df): The calculated p-value was 0.006 (<0.05), indicating a statistically significant difference between daytime and nighttime vehicle offset data.

[0061] (3) Visual analysis of eye features Eye feature analysis mainly uses three aspects—blink frequency analysis, fixation time analysis, and pupil area analysis—to make a correlation judgment on whether the driver is experiencing fatigue driving.

[0062] 1) Blinking frequency analysis The raw data obtained from the driving simulation experiment were organized and analyzed. Data from experimenters with obvious characteristics were selected for presentation. The data were classified into daytime and nighttime simulation scenarios, and the duration of eye-opening and eye-closing was integrated and statistically analyzed according to the blinking frequency per minute.

[0063] Please refer to Figure 5. Figure 5 This is a visualization analysis of blink frequency in a method for assessing driver fatigue on long straight roads in high-altitude plateau regions, as described in this application.

[0064] Comparing driving data from daytime and nighttime simulated scenarios for each driver reveals that the duration of eye-opening was shorter in the nighttime scenario compared to the daytime scenario. This indicates that daytime conditions allow for greater driver focus, and the good lighting effectively reduces driver fatigue. Furthermore, the blinking frequency was significantly higher in the nighttime scenario compared to the daytime scenario, suggesting that darker driving environments at night are more likely to cause eye fatigue, leading to a significant increase in blinking frequency. Analysis of the daytime and nighttime data also shows that the subjects' eye-closing time was less than one second (recorded as one second during statistical analysis), with no prolonged periods of eye-closing. This demonstrates that while eye fatigue occurs during short periods of driving, it does not lead to severe driver fatigue.

[0065] According to literature, the blinking frequency of a driver in a normal, alert state is 17-19 times, while the blinking frequency in a fatigued state is 20-23 times. By comparing the blinking frequencies of drivers, the corresponding periods of alertness and fatigue can be obtained, and thus the probability of drivers experiencing fatigue driving in the experiment can be determined.

[0066] Table 1. Mean and standard deviation of driver blink frequency in daytime and nighttime scenarios. Combined variance: , t-value: t≈-0.459; With 20 degrees of freedom and a p-value of ≈0.653 > 0.05, there is no statistically significant difference in blink frequency between daytime and nighttime.

[0067] 2) Fixation time analysis Based on the raw data collected by the eye tracker, the data was processed and divided into daytime and nighttime scenes. For each subject, 50 fixation duration data points longer than 1 second were selected for each scene to form a data sample for analysis.

[0068] Please refer to Figure 6. Figure 6 This is a visualization analysis of the fixation time of a method for assessing driver fatigue on long straight roads in high-altitude plateau regions, as described in this application.

[0069] Based on the analysis of the sample data, the average continuous gaze time of each driver in the daytime scenario was 2.824 seconds, while the average continuous gaze time of each driver in the nighttime scenario was 2.727 seconds. This shows that the driver's gaze time is slightly longer in the daytime scenario than in the nighttime scenario, and their attention is more focused.

[0070] A t-test was performed on the sustained gaze time of drivers in daytime and nighttime scenarios. The mean was 1058.2 with a standard deviation of 987.5 during the day and 928.6 with a standard deviation of 1203.4 at night. Welch's t-test yielded t = 1.12, p = 0.263 (p > 0.05). Therefore, it was concluded that there was no significant difference in the mean gaze time of drivers between daytime and nighttime.

[0071] 3) Pupil area analysis Pupil area can effectively reflect a driver's fatigue level. Pupil data of all 20 drivers were collected and screened. The average pupil area per minute of each driver was used as the characteristic data. The fatigue response of drivers during driving was analyzed as driving time increased. Data of drivers with more obvious characteristics were selected as examples for display.

[0072] Please refer to Figure 6. Figure 6 This is a visualization analysis of the fixation time of a method for assessing driver fatigue on long straight roads in high-altitude plateau regions, as described in this application.

[0073] A comparison of pupil size during the day and at night shows that in nighttime scenarios, drivers have larger pupils due to the dimmer ambient light. Therefore, prolonged periods of keeping the eyes open can easily lead to eye fatigue and consequently, drowsy driving.

[0074] The difference in pupil size between drivers in daytime and nighttime scenarios was analyzed, and the specific data are shown in the table below.

[0075] Table 2. Data on the difference in pupillary facial area between drivers in daytime and nighttime scenarios. Comparing the differences in pupil area between daytime and nighttime scenarios reveals that the difference is greater in nighttime scenarios, indicating that the driver's attention fluctuates more significantly at night and that nighttime scenarios have a greater impact on the driver's focus while driving.

[0076] A paired t-test was performed on the daytime and nighttime vehicle pupil area data. The mean value for the daytime group was approximately 1500, and the mean value for the nighttime group was approximately 2900. The t-value was approximately 20.5, the degrees of freedom were 336, and the p-value was <0.0001. Therefore, there is a statistically significant difference in the daytime and nighttime vehicle pupil area data.

[0077] (4) Visual analysis of hand features Hand feature analysis mainly focuses on data collected by both machines and observers, analyzing the relationship between changes in the driver's grip on the steering wheel, the steering wheel rotation angle, and fatigue driving.

[0078] 1) Analysis of the tightness of gripping the steering wheel Based on real-time observations of drivers' hand characteristics in simulated driving environments, it was found that most drivers initially gripped the steering wheel firmly in the upper-middle area when first encountering the simulated vehicle. However, as the simulated driving time increased, the drivers' hands gradually relaxed, gripping the steering wheel more lightly in the lower-middle area. This indicates that as drivers become more familiar with the simulated vehicle's interior environment, they gradually relax, their driving tension decreases, and consequently, the degree of grip on the steering wheel decreases from tight to loose.

[0079] 2) Steering wheel rotation angle analysis Since the simulated road consists entirely of straight sections and large-radius circular curves, with each curve having a radius of approximately 5600m, the entire simulated road is treated as a long straight section. This allows for the collection and analysis of data on the driver's steering wheel angle. According to relevant literature, based on angular velocity calculations, a driver with a steering wheel angle ≤0.03° is likely in a state of alertness, while a driver with a steering wheel angle ≥0.05° is likely in a state of fatigue.

[0080] First, because steering wheel rotation angles can be positive or negative, the absolute value of the collected raw data is added to obtain the rotation angle data. Second, the data of each person's steering wheel rotation angle is filtered by taking the maximum absolute value per minute to obtain the steering wheel rotation angle data of each driver throughout the entire driving process. Taking the data of the subjects whose data changes are representative as an example, line graphs of relevant data in daytime and nighttime scenarios are drawn respectively.

[0081] Please refer to Figure 7. Figure 7 This is a visual analysis diagram of hand features used in a method for assessing driver fatigue on long straight roads in high-altitude plateau regions, as described in this application.

[0082] Comparing data from daytime and nighttime scenarios reveals that there is a higher likelihood of larger steering wheel angles at night, making drowsy driving more likely.

[0083] A paired t-test of daytime and nighttime vehicle offset data yields the following results: Standard deviation (difference): Standard error (SE): t-statistic: Degrees of freedom (df): The calculated p-value is 0.004 (<0.05), indicating a statistically significant difference between daytime and nighttime vehicle offset data.

[0084] (5) Visual analysis of heart rate characteristics Based on real-time observations of the driver's heart rate changes and analysis of the corresponding ECG heart rate monitor output data, it was found that the entire road segment had no sharp turns or vehicle intersections, the modeled route was linear and approximately a long straight line, and the lane width was 3.75 meters, which fully met the vehicle's driving needs. Therefore, the driver's mindset was relatively relaxed, without any tension, resulting in a regular and stable heart rate without significant changes.

[0085] The technical effect of this embodiment is that by using methods such as paired t-tests, the statistical differences of various indicators under day and night conditions are systematically analyzed, providing data support for model construction and enhancing the scientific nature of the conclusions.

[0086] Example 5 Based on Example 4, this example discloses a further design of step S4 in a method for assessing driver fatigue on long straight roads in high-altitude plateau regions.

[0087] Furthermore, based on the above analysis, an analysis of the differences between daytime and nighttime data for the dependent variable of each model revealed significant differences in driving speed, vehicle lateral deviation, pupil area change, and steering wheel rotation angle, while blinking frequency and fixation time did not show significant differences. Therefore, it is necessary to establish separate models for daytime and nighttime scenarios to minimize the error of the established models. The formula is: , Where μ is the mean and σ is the standard deviation. For the standardized dependent variable, The dependent variable before standardization.

[0088] Furthermore, the relationship model described in step S4 includes a daytime scene model and a nighttime scene model.

[0089] (1) Results and analysis of daytime scene model For daytime scene data, a logistic regression model was developed using Python to output the following key statistical indicators.

[0090] Table 3 Key Statistical Indicators of Daytime Scene Model Results The final determined daytime scene model is as follows: Where P1 is the probability of fatigue in a daytime scene (Y=1), x1 is the driving speed, x2 is the lateral deviation of the vehicle, x3 is the blink frequency, x4 is the duration of sustained fixation, x5 is the pupil area, and x6 is the steering wheel angle. The Logit function is: .

[0091] From the daytime scene model, we can see that: All independent variables were significant (P < 0.05), indicating a statistically significant effect on fatigue probability.

[0092] The regression coefficients for driving speed, blink frequency, and steering wheel angle are all positive, indicating that as the values ​​of the above dependent variables gradually increase, the probability of fatigue also gradually increases, showing a positive correlation.

[0093] The regression coefficients for vehicle lateral deviation, sustained fixation time, and pupil area were all negative, indicating that as these values ​​gradually increase, the probability of fatigue gradually decreases, showing a negative correlation.

[0094] The frequency of blinking has the greatest impact (coefficient = 1.10), while the steering wheel angle has the least impact (coefficient = 0.28).

[0095] (2) Results and analysis of nighttime scene model For nighttime scene data, a logistic regression model was developed using Python to output the following key statistical indicators.

[0096] Table 4 Key Statistical Indicators of Night Scene Model Results The final night scene model was determined to be as follows: , Where P2 is the probability of fatigue in a nighttime scene (Y=1), x1 is the driving speed, x2 is the vehicle's lateral deviation, x3 is the blink frequency, x4 is the duration of sustained fixation, x5 is the pupil area, and x6 is the steering wheel angle. The Logit function is: .

[0097] From the night scene model, we can see that: All independent variables were significant (P < 0.05), indicating a statistically significant effect on fatigue probability.

[0098] The regression coefficients for driving speed, blink frequency, sustained gaze duration, and steering wheel angle were all positive, indicating that as the values ​​of these dependent variables gradually increase, the probability of fatigue also gradually increases, showing a positive correlation.

[0099] The regression coefficients for vehicle lateral offset and pupil area are both negative, indicating that as these values ​​gradually increase, the probability of fatigue gradually decreases, showing a negative correlation.

[0100] The frequency of blinking had the greatest impact (coefficient = 1.20), while the duration of sustained fixation had the least impact (coefficient = 0.33).

[0101] (3) Logistic regression model This application uses an ordered multi-class logistic regression model to detect driver fatigue. The principle of ordered multi-class logistic regression is to sequentially divide the multiple categories of the dependent variable into multiple binary logistic regressions to establish the model. Vehicle speed, lateral deviation, blink frequency, fixation duration, pupil area, and steering wheel angle are selected as independent variables, and the driver's fatigue level is selected as the dependent variable. A model is established to represent the relationship between monotonic road environment and driver fatigue, where 0 represents no fatigue and 1 represents fatigue.

[0102] When the number of levels of the dependent variable is greater than 2, it is not possible to simply combine two levels into a binary logistic regression. A multi-class logistic regression model must be considered, which can be achieved by fitting a logistic regression model with one fewer level than the number of dependent variables. For example, consider a dependent variable with 4 levels, taking values ​​1, 2, 3, and 4, whose corresponding probabilities are... , , , For four independent variables, three models are fitted (accumulated model), and the probabilities of the corresponding value levels are: , , , Based on the above formula, we can obtain: , , , , In the formula: βi are the coefficients of the independent variables in the model, i = 1, 2, 3, ...; αi is the model intercept, i = 1, 2, 3, ...; xi is the i-th independent variable, i = 1, 2, 3, ... . Ordered multinomial logistic regression is a combination of multiple binary regressions. For the dependent variable with four levels, the logistic transformation is performed using the following methods: The cumulative probability of these three dependent variables taking ordered values. The cumulative probability can be calculated using the formula above. Since the explained variable (dependent variable) "driver fatigue level" is multiple and ordered, the model form used is as follows: , In the formula: y represents the driver's fatigue level; j represents the fatigue level grade, j = 1, 2, 3, ...; This is the i-th factor affecting driver fatigue.

[0103] Establish a cumulative logistic model: , In the formula: This represents the probability of a driver experiencing a certain level of fatigue. j = 1, 2, 3; x = (x1, x2, ..., xi)T are the characteristic independent variables; These are the regression coefficients of the model; This is the model intercept.

[0104] get and After estimating the parameters, the probability of a specific case (y = j) occurring can be obtained using the following formula: .

[0105] The technical advantages of this embodiment are as follows: This application uses six driving indicators, including vehicle speed, lateral deviation, and blink frequency, as independent variables to establish an ordered multi-category logistic regression model, achieving a mathematical and statistical quantitative judgment of fatigue driving status. Compared with traditional subjective assessment methods, this model automatically calculates the fatigue probability through objective data, providing a standardized judgment basis for intelligent driving systems; the correlation analysis of variables in the model (such as the positive correlation between vehicle speed and fatigue) directly guides the optimization of road landscape parameters, providing a scientific design guide for improving driving safety.

[0106] The above description is merely a preferred embodiment of the present invention and does not limit the scope of protection of the present invention. Various modifications and variations are possible with respect to the present invention. Any changes, modifications, substitutions, integrations, and parameter alterations to these embodiments within the spirit and principles of the present invention fall within the scope of protection of the claims of the present invention.

Claims

1. A method for assessing driver fatigue in truck drivers on long, straight roads in high-altitude plateau regions, characterized in that, Includes the following steps: Step S1, Driving simulation experiment: Set different vehicle types and define driving behaviors on the long straight road model according to the scenario requirements; set the time transition mechanism from day to night, define the nighttime environment parameters and set low visibility scenarios on specific road sections to complete the establishment of daytime and nighttime environment scenario models; Step S2, Data Collection: Data is collected through both subjective and objective data collection methods; Step S3: Data visualization and descriptive statistical analysis: The data collected in step S2 is visualized and analyzed in the form of a three-line table, and paired t-tests are performed on each data indicator for daytime and nighttime to analyze the statistical differences between daytime and nighttime data. Step S4: Establish the relationship between various statistical indicators and fatigue: Select vehicle speed, vehicle lateral deviation, blink frequency, fixation duration, pupil area, and steering wheel angle as independent variables, and driver fatigue level as dependent variable to establish a relationship model between road environment and driving fatigue for driving fatigue state detection.

2. The method for assessing driver fatigue of truck drivers on long straight roads in high-altitude plateau regions according to claim 1, characterized in that, The dependent variable is standardized using the following formula: ， Where μ is the mean and σ is the standard deviation. For the standardized dependent variable, The dependent variable before standardization.

3. The method for assessing driver fatigue on long straight roads in high-altitude plateau regions according to claim 2, characterized in that, The relationship model described in step S4 is an ordered multi-class Logistic regression model.

4. The method for assessing driver fatigue on long straight roads in high-altitude plateau regions according to claim 2, characterized in that, The formula for the Logistic regression model is: ， Wherein: This represents the probability of a driver experiencing a certain level of fatigue. , j = 1, 2, 3; x = (x1, x2, ..., xi), T is the characteristic independent variable; These are the regression coefficients of the model; The model intercept; The formula for calculating the probability of a driver being fatigued to a certain degree is: In the formula: y represents the driver's fatigue level; j represents the fatigue level grade, j = 1, 2, 3, ...; This is the i-th factor affecting driver fatigue.

5. The method for assessing driver fatigue on long straight roads in high-altitude plateau regions as described in claims 3 and 4, characterized in that, Step S1 includes the following steps: Step S110: Establish a road simulation model: Import the road design drawings into the software to generate a road model, set the number of road lanes, the width of the central median, the direction of vehicle traffic and the design road speed, and add traffic signs and directional signs to the road. Step S120: Establish a landscape model: Set up landscapes on both sides of the road in the road model. Set up tree models on both sides of the road at about 5km and 10km to remind the driver of the position of the driver in the road model. Near the end of the road model, set up a checkpoint and a speed reduction sign to remind the driver that the driving simulation road is about to end. Step S130: Establish vehicle model: Set up trucks and cars that are driving normally on the road model, and set the vehicle behavior; in the daytime simulated road section, set up oncoming vehicles in the opposite lane, and in the nighttime simulated road section, no other vehicles are set up. Step S140: Set the nighttime environment: When the driver drives the vehicle for 200m in the simulated road section, the time gradually changes from daytime to nighttime; the ambient time is 20:00 at night; in the last 2000m section, a foggy section is set to reduce the visibility in the simulated scene to simulate a sandy road section.

6. The method for assessing driver fatigue of truck drivers on long straight roads in high-altitude plateau regions as described in claims 3 and 4, is characterized in that... The specific method for subjective data collection in step S2 is as follows: Step S201, Subject Recruitment and Screening: Male individuals who have driven large vehicles are invited to participate in the experiment, covering different age groups, driving experience, and driver's license types; Step S202, Basic Information Collection: Collect statistics on the subject's age, driving experience, driver's license type, current vehicle type, and manual / automatic transmission driving status; Step S203, Driving Behavior Survey and Statistics: Collect data on the subjects' highway driving frequency, weekly driving frequency, and average driving time on weekdays; calculate the subjects' average annual and past three years driving mileage to assess driving proficiency; Step S204, Traffic Accident Investigation and Statistics: Record whether minor or major traffic accidents have occurred, and analyze driving safety.

7. The method for assessing driver fatigue of truck drivers on long straight roads in high-altitude plateau regions according to claim 6, characterized in that, The specific method for objective data collection in step S2 is as follows: data collected by instruments and equipment and data obtained by experimental personnel observing the driving behavior of the subjects. The objective data includes vehicle data, eye tracker data, hand data and heart rate data.

8. The method for assessing driver fatigue of truck drivers on long straight roads in high-altitude plateau regions according to claim 7, characterized in that, The vehicle data includes vehicle speed and lateral offset distance; the eye tracker data includes the subject's blink frequency, blink duration, pupil size, and gaze direction.

9. The method for assessing driver fatigue of truck drivers on long straight roads in high-altitude plateau regions according to claim 7, characterized in that, The hand data includes the tightness of the grip on the steering wheel, the degree of tension and sweating, and the speed of steering wheel rotation.

10. The method for assessing driver fatigue of truck drivers on long straight roads in high-altitude plateau regions according to claim 8 or 9, characterized in that, The heart rate data mentioned is from an ECG heart rate monitor.

Images