Evaluating the Effect of Fatigue on Driver’s Performance

IFAC PapersOnLine 58-9 (2024) 263–268 ScienceDirect ScienceDirect Available online at www.sciencedirect.com 2405-8963 Copyright © 2024 The Authors. This is an open access article under the CC BY-NC-ND license . Peer review under responsibility of International Federation of Automatic Control. 10.1016/j.ifacol.2024.07.407 10.1016/j.ifacol.2024.07.407 2405-8963 Evaluating the Effect of Fatigue on Driver's Performance M. Jirgl*, S. Sediva**, Z. Bradac*** Department of Control and Instrumentation, Faculty of Electrical Engineering and Communication, Brno University of Technology, Technicka 3082/12, 616 00 Brno, Czech Republic *(e-mail: [email protected]), **(e-mail: [email protected]), **(e-mail: [email protected]) Abstract: The paper presents an experimental cybernetic approach for the evaluation of drivers’ performance in connection with fatigue based on mathematical modelling and description of human behavior during a simple lane-changing task. Data are acquired via the in-house developed car driving simulator under defined conditions. Statistical evaluation of the used parameters for 50 active drivers is presented as an initial data set for the determination of the described drivers’ performance decision-making system built on fuzzy logic. The system was then used for calculating the DPI (Driver performance index) as a self-determined relative measure of drivers’ performance. Then, an example of applying the method with three volunteers influenced by fatigue is presented. All of these participants reached lower DPI in the case of fatigue induced by a 3-hour-long simulated loading drive on a highway. Although the results cannot be statistically significantly interpreted, they indicate a potential for further and deeper research. Keywords: Human driver model, drivers’ performance, fatigue, car driving simulator, fuzzy logic. 1 INTRODUCTION Driver fatigue is the same risk in driving a car as alcohol. Fatigue, sleepiness, or loss of attention are responsible for up to a third of traffic accidents (Philip and Åkerstedt, 2006). Crashes caused by fatigue are usually characterized by a loss of driver control over the vehicle, which leads to an inadvertent change in the vehicle's direction of travel and often to an absence of braking response. Human fatigue cannot be measured in practice and there is not even a single methodology for evaluating the degree of the fatigue. Due to fatigue, reaction times are longer, or interventions are inadequate. Specifically, for the driver, this is a decrease in attention. The vigilance decreases, on the other hand, a reaction delay and an information processing time increase, which influence an overall driver’s performance (Zhang, Z., 2022). Driver fatigue is often divided into physiological fatigue, sleepiness, and mental fatigue (Hu and Lodewijks, 2020). Physiological fatigue can be further divided into muscle fatigue, which manifests itself mainly in feelings of pain, and sensory fatigue (for example, visual fatigue), which in many cases a person does not perceive or feel. These are subjective feelings of an individual that are difficult to quantify. Physiological fatigue is mainly caused by a long-term stereotyped activity. The drowsiness slows reactions, reduces perception, and impairs decision-making skills. Mental fatigue is often manifested by a lack of interest in the activity being performed or an attempt to interrupt work. These are manifestations of a lack of concentration on the activity being performed. For example, monotonous driving on the highway often leads to accidents related to drowsiness, and the geometry of the road and the environment along the road can lead to mental fatigue of the driver (Hu and Lodewijks, 2020). A very dangerous manifestation of fatigue is microsleep as a manifestation of a critical decrease in attention, which occurs due to excessive mental load, mostly during monotonous activities. Microsleep has a highly individual character. It is influenced both genetically and by the general condition of the individual, his state of health, age, and training (Vysoký, 2001). The goal of this paper is to present an approach for driver performance assessment and try to apply this method also for fatigue assessment. Thus, the following section brings an overview of the current methods for fatigue detection. Then, the used technique for driver’s performance assessment is presented. The final section includes an example of applying the described approach for fatigue assessment and summarizes the obtained results. 2 FATIGUE DETECTION TECHNIQUES There are several methods designed for the detection of driver fatigue. The main classification of methods is based on obtaining information from the parameters of the driver (measurement of his physiological data) or from the parameters of the car (Satheesh and Basha, 2023). The distribution of techniques is shown in Fig. 1. A detailed overview of the methods used for the detection and prediction of driver fatigue or drowsiness can be found, for example, in the cited literature (Shaik, 2023; Khadkikar, Banginwar et al., 2023; Nudzikova and Slanina, 2016; Otahalova et al., 2012). 2.1 Measurement of the driver's physiological parameters Driver fatigue monitoring is possible based on data acquisition from sensors placed on the driver's body. By measuring heart rate, pulse rate, brain activity and other physiological variables, the condition of the driver can be evaluated. Three basic methods are used to identify driver fatigue – Evaluating the Effect of Fatigue on Driver's Performance M. Jirgl*, S. Sediva**, Z. Bradac*** Department of Control and Instrumentation, Faculty of Electrical Engineering and Communication, Brno University of Technology, Technicka 3082/12, 616 00 Brno, Czech Republic *(e-mail: [email protected]), **(e-mail: [email protected]), **(e-mail: [email protected]) Abstract: The paper presents an experimental cybernetic approach for the evaluation of drivers’ performance in connection with fatigue based on mathematical modelling and description of human behavior during a simple lane-changing task. Data are acquired via the in-house developed car driving simulator under defined conditions. Statistical evaluation of the used parameters for 50 active drivers is presented as an initial data set for the determination of the described drivers’ performance decision-making system built on fuzzy logic. The system was then used for calculating the DPI (Driver performance index) as a self-determined relative measure of drivers’ performance. Then, an example of applying the method with three volunteers influenced by fatigue is presented. All of these participants reached lower DPI in the case of fatigue induced by a 3-hour-long simulated loading drive on a highway. Although the results cannot be statistically significantly interpreted, they indicate a potential for further and deeper research. Keywords: Human driver model, drivers’ performance, fatigue, car driving simulator, fuzzy logic. 1 INTRODUCTION Driver fatigue is the same risk in driving a car as alcohol. Fatigue, sleepiness, or loss of attention are responsible for up to a third of traffic accidents (Philip and Åkerstedt, 2006). Crashes caused by fatigue are usually characterized by a loss of driver control over the vehicle, which leads to an inadvertent change in the vehicle's direction of travel and often to an absence of braking response. Human fatigue cannot be measured in practice and there is not even a single methodology for evaluating the degree of the fatigue. Due to fatigue, reaction times are longer, or interventions are inadequate. Specifically, for the driver, this is a decrease in attention. The vigilance decreases, on the other hand, a reaction delay and an information processing time increase, which influence an overall driver’s performance (Zhang, Z., 2022). Driver fatigue is often divided into physiological fatigue, sleepiness, and mental fatigue (Hu and Lodewijks, 2020). Physiological fatigue can be further divided into muscle fatigue, which manifests itself mainly in feelings of pain, and sensory fatigue (for example, visual fatigue), which in many cases a person does not perceive or feel. These are subjective feelings of an individual that are difficult to quantify. Physiological fatigue is mainly caused by a long-term stereotyped activity. The drowsiness slows reactions, reduces perception, and impairs decision-making skills. Mental fatigue is often manifested by a lack of interest in the activity being performed or an attempt to interrupt work. These are manifestations of a lack of concentration on the activity being performed. For example, monotonous driving on the highway often leads to accidents related to drowsiness, and the geometry of the road and the environment along the road can lead to mental fatigue of the driver (Hu and Lodewijks, 2020). A very dangerous manifestation of fatigue is microsleep as a manifestation of a critical decrease in attention, which occurs due to excessive mental load, mostly during monotonous activities. Microsleep has a highly individual character. It is influenced both genetically and by the general condition of the individual, his state of health, age, and training (Vysoký, 2001). The goal of this paper is to present an approach for driver performance assessment and try to apply this method also for fatigue assessment. Thus, the following section brings an overview of the current methods for fatigue detection. Then, the used technique for driver’s performance assessment is presented. The final section includes an example of applying the described approach for fatigue assessment and summarizes the obtained results. 2 FATIGUE DETECTION TECHNIQUES There are several methods designed for the detection of driver fatigue. The main classification of methods is based on obtaining information from the parameters of the driver (measurement of his physiological data) or from the parameters of the car (Satheesh and Basha, 2023). The distribution of techniques is shown in Fig. 1. A detailed overview of the methods used for the detection and prediction of driver fatigue or drowsiness can be found, for example, in the cited literature (Shaik, 2023; Khadkikar, Banginwar et al., 2023; Nudzikova and Slanina, 2016; Otahalova et al., 2012). 2.1 Measurement of the driver's physiological parameters Driver fatigue monitoring is possible based on data acquisition from sensors placed on the driver's body. By measuring heart rate, pulse rate, brain activity and other physiological variables, the condition of the driver can be evaluated. Three basic methods are used to identify driver fatigue – Evaluating the Effect of Fatigue on Driver's Performance M. Jirgl*, S. Sediva**, Z. Bradac*** Department of Control and Instrumentation, Faculty of Electrical Engineering and Communication, Brno University of Technology, Technicka 3082/12, 616 00 Brno, Czech Republic *(e-mail: [email protected]), **(e-mail: [email protected]), **(e-mail: [email protected]) Abstract: The paper presents an experimental cybernetic approach for the evaluation of drivers’ performance in connection with fatigue based on mathematical modelling and description of human behavior during a simple lane-changing task. Data are acquired via the in-house developed car driving simulator under defined conditions. Statistical evaluation of the used parameters for 50 active drivers is presented as an initial data set for the determination of the described drivers’ performance decision-making system built on fuzzy logic. The system was then used for calculating the DPI (Driver performance index) as a self-determined relative measure of drivers’ performance. Then, an example of applying the method with three volunteers influenced by fatigue is presented. All of these participants reached lower DPI in the case of fatigue induced by a 3-hour-long simulated loading drive on a highway. Although the results cannot be statistically significantly interpreted, they indicate a potential for further and deeper research. Keywords: Human driver model, drivers’ performance, fatigue, car driving simulator, fuzzy logic. 1 INTRODUCTION Driver fatigue is the same risk in driving a car as alcohol. Fatigue, sleepiness, or loss of attention are responsible for up to a third of traffic accidents (Philip and Åkerstedt, 2006). Crashes caused by fatigue are usually characterized by a loss of driver control over the vehicle, which leads to an inadvertent change in the vehicle's direction of travel and often to an absence of braking response. Human fatigue cannot be measured in practice and there is not even a single methodology for evaluating the degree of the fatigue. Due to fatigue, reaction times are longer, or interventions are inadequate. Specifically, for the driver, this is a decrease in attention. The vigilance decreases, on the other hand, a reaction delay and an information processing time increase, which influence an overall driver’s performance (Zhang, Z., 2022). Driver fatigue is often divided into physiological fatigue, sleepiness, and mental fatigue (Hu and Lodewijks, 2020). Physiological fatigue can be further divided into muscle fatigue, which manifests itself mainly in feelings of pain, and sensory fatigue (for example, visual fatigue), which in many cases a person does not perceive or feel. These are subjective feelings of an individual that are difficult to quantify. Physiological fatigue is mainly caused by a long-term stereotyped activity. The drowsiness slows reactions, reduces perception, and impairs decision-making skills. Mental fatigue is often manifested by a lack of interest in the activity being performed or an attempt to interrupt work. These are manifestations of a lack of concentration on the activity being performed. For example, monotonous driving on the highway often leads to accidents related to drowsiness, and the geometry of the road and the environment along the road can lead to mental fatigue of the driver (Hu and Lodewijks, 2020). A very dangerous manifestation of fatigue is microsleep as a manifestation of a critical decrease in attention, which occurs due to excessive mental load, mostly during monotonous activities. Microsleep has a highly individual character. It is influenced both genetically and by the general condition of the individual, his state of health, age, and training (Vysoký, 2001). The goal of this paper is to present an approach for driver performance assessment and try to apply this method also for fatigue assessment. Thus, the following section brings an overview of the current methods for fatigue detection. Then, the used technique for driver’s performance assessment is presented. The final section includes an example of applying the described approach for fatigue assessment and summarizes the obtained results. 2 FATIGUE DETECTION TECHNIQUES There are several methods designed for the detection of driver fatigue. The main classification of methods is based on obtaining information from the parameters of the driver (measurement of his physiological data) or from the parameters of the car (Satheesh and Basha, 2023). The distribution of techniques is shown in Fig. 1. A detailed overview of the methods used for the detection and prediction of driver fatigue or drowsiness can be found, for example, in the cited literature (Shaik, 2023; Khadkikar, Banginwar et al., 2023; Nudzikova and Slanina, 2016; Otahalova et al., 2012). 2.1 Measurement of the driver's physiological parameters Driver fatigue monitoring is possible based on data acquisition from sensors placed on the driver's body. By measuring heart rate, pulse rate, brain activity and other physiological variables, the condition of the driver can be evaluated. Three basic methods are used to identify driver fatigue – Evaluating the Effect of Fatigue on Driver's Performance M. Jirgl*, S. Sediva**, Z. Bradac*** Department of Control and Instrumentation, Faculty of Electrical Engineering and Communication, Brno University of Technology, Technicka 3082/12, 616 00 Brno, Czech Republic *(e-mail: [email protected]), **(e-mail: [email protected]), **(e-mail: brad[email protected]) Abstract: The paper presents an experimental cybernetic approach for the evaluation of drivers’ performance in connection with fatigue based on mathematical modelling and description of human behavior during a simple lane-changing task. Data are acquired via the in-house developed car driving simulator under defined conditions. Statistical evaluation of the used parameters for 50 active drivers is presented as an initial data set for the determination of the described drivers’ performance decision-making system built on fuzzy logic. The system was then used for calculating the DPI (Driver performance index) as a self-determined relative measure of drivers’ performance. Then, an example of applying the method with three volunteers influenced by fatigue is presented. All of these participants reached lower DPI in the case of fatigue induced by a 3-hour-long simulated loading drive on a highway. Although the results cannot be statistically significantly interpreted, they indicate a potential for further and deeper research. Keywords: Human driver model, drivers’ performance, fatigue, car driving simulator, fuzzy logic. 1 INTRODUCTION Driver fatigue is the same risk in driving a car as alcohol. Fatigue, sleepiness, or loss of attention are responsible for up to a third of traffic accidents (Philip and Åkerstedt, 2006). Crashes caused by fatigue are usually characterized by a loss of driver control over the vehicle, which leads to an inadvertent change in the vehicle's direction of travel and often to an absence of braking response. Human fatigue cannot be measured in practice and there is not even a single methodology for evaluating the degree of the fatigue. Due to fatigue, reaction times are longer, or interventions are inadequate. Specifically, for the driver, this is a decrease in attention. The vigilance decreases, on the other hand, a reaction delay and an information processing time increase, which influence an overall driver’s performance (Zhang, Z., 2022). Driver fatigue is often divided into physiological fatigue, sleepiness, and mental fatigue (Hu and Lodewijks, 2020). Physiological fatigue can be further divided into muscle fatigue, which manifests itself mainly in feelings of pain, and sensory fatigue (for example, visual fatigue), which in many cases a person does not perceive or feel. These are subjective feelings of an individual that are difficult to quantify. Physiological fatigue is mainly caused by a long-term stereotyped activity. The drowsiness slows reactions, reduces perception, and impairs decision-making skills. Mental fatigue is often manifested by a lack of interest in the activity being performed or an attempt to interrupt work. These are manifestations of a lack of concentration on the activity being performed. For example, monotonous driving on the highway often leads to accidents related to drowsiness, and the geometry of the road and the environment along the road can lead to mental fatigue of the driver (Hu and Lodewijks, 2020). A very dangerous manifestation of fatigue is microsleep as a manifestation of a critical decrease in attention, which occurs due to excessive mental load, mostly during monotonous activities. Microsleep has a highly individual character. It is influenced both genetically and by the general condition of the individual, his state of health, age, and training (Vysoký, 2001). The goal of this paper is to present an approach for driver performance assessment and try to apply this method also for fatigue assessment. Thus, the following section brings an overview of the current methods for fatigue detection. Then, the used technique for driver’s performance assessment is presented. The final section includes an example of applying the described approach for fatigue assessment and summarizes the obtained results. 2 FATIGUE DETECTION TECHNIQUES There are several methods designed for the detection of driver fatigue. The main classification of methods is based on obtaining information from the parameters of the driver (measurement of his physiological data) or from the parameters of the car (Satheesh and Basha, 2023). The distribution of techniques is shown in Fig. 1. A detailed overview of the methods used for the detection and prediction of driver fatigue or drowsiness can be found, for example, in the cited literature (Shaik, 2023; Khadkikar, Banginwar et al., 2023; Nudzikova and Slanina, 2016; Otahalova et al., 2012). 2.1 Measurement of the driver's physiological parameters Driver fatigue monitoring is possible based on data acquisition from sensors placed on the driver's body. By measuring heart rate, pulse rate, brain activity and other physiological variables, the condition of the driver can be evaluated. Three basic methods are used to identify driver fatigue – Evaluating the Effect of Fatigue on Driver's Performance M. Jirgl*, S. Sediva**, Z. Bradac*** Department of Control and Instrumentation, Faculty of Electrical Engineering and Communication, Brno University of Technology, Technicka 3082/12, 616 00 Brno, Czech Republic *(e-mail: [email protected]), **(e-mail: [email protected]), **(e-mail: [email protected]) Abstract: The paper presents an experimental cybernetic approach for the evaluation of drivers’ performance in connection with fatigue based on mathematical modelling and description of human behavior during a simple lane-changing task. Data are acquired via the in-house developed car driving simulator under defined conditions. Statistical evaluation of the used parameters for 50 active drivers is presented as an initial data set for the determination of the described drivers’ performance decision-making system built on fuzzy logic. The system was then used for calculating the DPI (Driver performance index) as a self-determined relative measure of drivers’ performance. Then, an example of applying the method with three volunteers influenced by fatigue is presented. All of these participants reached lower DPI in the case of fatigue induced by a 3-hour-long simulated loading drive on a highway. Although the results cannot be statistically significantly interpreted, they indicate a potential for further and deeper research. Keywords: Human driver model, drivers’ performance, fatigue, car driving simulator, fuzzy logic. 1 INTRODUCTION Driver fatigue is the same risk in driving a car as alcohol. Fatigue, sleepiness, or loss of attention are responsible for up to a third of traffic accidents (Philip and Åkerstedt, 2006). Crashes caused by fatigue are usually characterized by a loss of driver control over the vehicle, which leads to an inadvertent change in the vehicle's direction of travel and often to an absence of braking response. Human fatigue cannot be measured in practice and there is not even a single methodology for evaluating the degree of the fatigue. Due to fatigue, reaction times are longer, or interventions are inadequate. Specifically, for the driver, this is a decrease in attention. The vigilance decreases, on the other hand, a reaction delay and an information processing time increase, which influence an overall driver’s performance (Zhang, Z., 2022). Driver fatigue is often divided into physiological fatigue, sleepiness, and mental fatigue (Hu and Lodewijks, 2020). Physiological fatigue can be further divided into muscle fatigue, which manifests itself mainly in feelings of pain, and sensory fatigue (for example, visual fatigue), which in many cases a person does not perceive or feel. These are subjective feelings of an individual that are difficult to quantify. Physiological fatigue is mainly caused by a long-term stereotyped activity. The drowsiness slows reactions, reduces perception, and impairs decision-making skills. Mental fatigue is often manifested by a lack of interest in the activity being performed or an attempt to interrupt work. These are manifestations of a lack of concentration on the activity being performed. For example, monotonous driving on the highway often leads to accidents related to drowsiness, and the geometry of the road and the environment along the road can lead to mental fatigue of the driver (Hu and Lodewijks, 2020). A very dangerous manifestation of fatigue is microsleep as a manifestation of a critical decrease in attention, which occurs due to excessive mental load, mostly during monotonous activities. Microsleep has a highly individual character. It is influenced both genetically and by the general condition of the individual, his state of health, age, and training (Vysoký, 2001). The goal of this paper is to present an approach for driver performance assessment and try to apply this method also for fatigue assessment. Thus, the following section brings an overview of the current methods for fatigue detection. Then, the used technique for driver’s performance assessment is presented. The final section includes an example of applying the described approach for fatigue assessment and summarizes the obtained results. 2 FATIGUE DETECTION TECHNIQUES There are several methods designed for the detection of driver fatigue. The main classification of methods is based on obtaining information from the parameters of the driver (measurement of his physiological data) or from the parameters of the car (Satheesh and Basha, 2023). The distribution of techniques is shown in Fig. 1. A detailed overview of the methods used for the detection and prediction of driver fatigue or drowsiness can be found, for example, in the cited literature (Shaik, 2023; Khadkikar, Banginwar et al., 2023; Nudzikova and Slanina, 2016; Otahalova et al., 2012). 2.1 Measurement of the driver's physiological parameters Driver fatigue monitoring is possible based on data acquisition from sensors placed on the driver's body. By measuring heart rate, pulse rate, brain activity and other physiological variables, the condition of the driver can be evaluated. Three basic methods are used to identify driver fatigue – Evaluating the Effect of Fatigue on Driver's Performance M. Jirgl*, S. Sediva**, Z. Bradac*** Department of Control and Instrumentation, Faculty of Electrical Engineering and Communication, Brno University of Technology, Technicka 3082/12, 616 00 Brno, Czech Republic *(e-mail: [email protected]), **(e-mail: [email protected]), **(e-mail: [email protected]) Abstract: The paper presents an experimental cybernetic approach for the evaluation of drivers’ performance in connection with fatigue based on mathematical modelling and description of human behavior during a simple lane-changing task. Data are acquired via the in-house developed car driving simulator under defined conditions. Statistical evaluation of the used parameters for 50 active drivers is presented as an initial data set for the determination of the described drivers’ performance decision-making system built on fuzzy logic. The system was then used for calculating the DPI (Driver performance index) as a self-determined relative measure of drivers’ performance. Then, an example of applying the method with three volunteers influenced by fatigue is presented. All of these participants reached lower DPI in the case of fatigue induced by a 3-hour-long simulated loading drive on a highway. Although the results cannot be statistically significantly interpreted, they indicate a potential for further and deeper research. Keywords: Human driver model, drivers’ performance, fatigue, car driving simulator, fuzzy logic. 1 INTRODUCTION Driver fatigue is the same risk in driving a car as alcohol. Fatigue, sleepiness, or loss of attention are responsible for up to a third of traffic accidents (Philip and Åkerstedt, 2006). Crashes caused by fatigue are usually characterized by a loss of driver control over the vehicle, which leads to an inadvertent change in the vehicle's direction of travel and often to an absence of braking response. Human fatigue cannot be measured in practice and there is not even a single methodology for evaluating the degree of the fatigue. Due to fatigue, reaction times are longer, or interventions are inadequate. Specifically, for the driver, this is a decrease in attention. The vigilance decreases, on the other hand, a reaction delay and an information processing time increase, which influence an overall driver’s performance (Zhang, Z., 2022). Driver fatigue is often divided into physiological fatigue, sleepiness, and mental fatigue (Hu and Lodewijks, 2020). Physiological fatigue can be further divided into muscle fatigue, which manifests itself mainly in feelings of pain, and sensory fatigue (for example, visual fatigue), which in many cases a person does not perceive or feel. These are subjective feelings of an individual that are difficult to quantify. Physiological fatigue is mainly caused by a long-term stereotyped activity. The drowsiness slows reactions, reduces perception, and impairs decision-making skills. Mental fatigue is often manifested by a lack of interest in the activity being performed or an attempt to interrupt work. These are manifestations of a lack of concentration on the activity being performed. For example, monotonous driving on the highway often leads to accidents related to drowsiness, and the geometry of the road and the environment along the road can lead to mental fatigue of the driver (Hu and Lodewijks, 2020). A very dangerous manifestation of fatigue is microsleep as a manifestation of a critical decrease in attention, which occurs due to excessive mental load, mostly during monotonous activities. Microsleep has a highly individual character. It is influenced both genetically and by the general condition of the individual, his state of health, age, and training (Vysoký, 2001). The goal of this paper is to present an approach for driver performance assessment and try to apply this method also for fatigue assessment. Thus, the following section brings an overview of the current methods for fatigue detection. Then, the used technique for driver’s performance assessment is presented. The final section includes an example of applying the described approach for fatigue assessment and summarizes the obtained results. 2 FATIGUE DETECTION TECHNIQUES There are several methods designed for the detection of driver fatigue. The main classification of methods is based on obtaining information from the parameters of the driver (measurement of his physiological data) or from the parameters of the car (Satheesh and Basha, 2023). The distribution of techniques is shown in Fig. 1. A detailed overview of the methods used for the detection and prediction of driver fatigue or drowsiness can be found, for example, in the cited literature (Shaik, 2023; Khadkikar, Banginwar et al., 2023; Nudzikova and Slanina, 2016; Otahalova et al., 2012). 2.1 Measurement of the driver's physiological parameters Driver fatigue monitoring is possible based on data acquisition from sensors placed on the driver's body. By measuring heart rate, pulse rate, brain activity and other physiological variables, the condition of the driver can be evaluated. Three basic methods are used to identify driver fatigue – Copyright © 2024 The Authors. This is an open access article under the CC BY-NC-ND license ( https://creativecommons.org/licenses/by-nc-nd/4.0/ ) 264 M. Jirgl et al. / IFAC PapersOnLine 58-9 (2024) 263–268 electroencephalography (EEG), electrooculography (EOG) and electrocardiography (ECG) (Barua et al., 2019). The frequently used EEG method uses the measurement of brain activity when the nature of the electrical signals in the brain changes when the driver's attention drops. EOG is a technique for measuring the potential that is between the front and back of the human eye and is used to capture eye movement. Heart rate can be evaluated e.g. using the EKG. Figure 1. Classification of driver fatigue detection techniques. Fatigue or sleepiness can also be detected from certain facial signals (Flores et al., 2010). For example, rapid and continuous blinking, head movement or repeated yawning can indicate driver fatigue. Using a camera that scans the driver's face, the obtained data is evaluated and the degree of drowsiness of the driver is determined. 2.2 Measuring the vehicle parameters Methods based on obtaining specific vehicle parameters assume that the way a tired driver drives a vehicle is different from the way a non-tired driver drives a vehicle. Any change in the measured parameters that exceeds a certain defined threshold value may indicate an increased probability of a tired driver. Among the parameters that can be detected are the position and movement of the steering wheel, deviations from the position in the lane, movement of the pedals, etc. Sensing the angle of the steering wheel is a common and popular method of evaluating driver fatigue (Chai et al., 2019; Lu et al., 2021) similar to measuring the position of the car in the lane (Katyal et al.2014). Measuring a vehicle parameters, particularly the steering wheel angle and position in the driving lane, is a base also for our study, where data are acquired via a car driving simulator. Based on the repeated driver’s testing (e.g. within different stages of fatigue), the deviations of parameters defining a current driver’s performance can be evaluated. An advantage of the presented approach is its non-invasive nature because it evaluates only driving data logged from the testing scenario and does not require any additional sensors. Although it operates only as an offline analysis, it might be an interesting extension to the existing methods summarized in Fig. 1. 3 MEASURING AND EVALUATION OF DRIVER’S PERFORMANCE This study is based on an approach for driver’s performance assessment published in (Jirgl et al., 2022). The principle lies in employing the cybernetic approach, where interaction between a driver and a car is described as a standard feedback control loop. It enables to modelling of a driver as a controller, and a car as a controlled element. The dynamical properties of this control loop can be then evaluated on the data acquired during a specific testing scenario. For these purposes, we developed an in-house car driving simulator – CDS. 3.1 Car driving simulator – CDS The car driving simulator – CDS involves the CDS software developed in Unreal Engine 4, and the hardware platform, see Fig. 2. It is equipped with a 49” Samsung CHG90 QLED gaming display, and with a joystick mediating control input, namely, a Logitech G920 steering wheel with pedals. The maximal rotation of the steering wheel is 900°, and the resolution corresponds to about 0.1°. Figure 2. The workstation with developed CDS for driver testing. The reason for developing an in-house CDS was to requirements on the definition of testing scenario (including the measuring conditions) and data acquisition under a sufficient sampling rate. The CDS provides data sampled by 100 Hz. These data from each testing scenario are logged for the individual drivers under a unique ID into a .csv file and Driver fatigue detection techniques Based on driver Behaviroral parameter based Eye blinking Eye closure Eye movement Head movement Facial Expression Yawn Physiological parameter based ECG EEG Pulse EMG Based on vehicle Lane detection Steering wheel movement Acceleration M. Jirgl et al. / IFAC PapersOnLine 58-9 (2024) 263–268 265 electroencephalography (EEG), electrooculography (EOG) and electrocardiography (ECG) (Barua et al., 2019). The frequently used EEG method uses the measurement of brain activity when the nature of the electrical signals in the brain changes when the driver's attention drops. EOG is a technique for measuring the potential that is between the front and back of the human eye and is used to capture eye movement. Heart rate can be evaluated e.g. using the EKG. Figure 1. Classification of driver fatigue detection techniques. Fatigue or sleepiness can also be detected from certain facial signals (Flores et al., 2010). For example, rapid and continuous blinking, head movement or repeated yawning can indicate driver fatigue. Using a camera that scans the driver's face, the obtained data is evaluated and the degree of drowsiness of the driver is determined. 2.2 Measuring the vehicle parameters Methods based on obtaining specific vehicle parameters assume that the way a tired driver drives a vehicle is different from the way a non-tired driver drives a vehicle. Any change in the measured parameters that exceeds a certain defined threshold value may indicate an increased probability of a tired driver. Among the parameters that can be detected are the position and movement of the steering wheel, deviations from the position in the lane, movement of the pedals, etc. Sensing the angle of the steering wheel is a common and popular method of evaluating driver fatigue (Chai et al., 2019; Lu et al., 2021) similar to measuring the position of the car in the lane (Katyal et al.2014). Measuring a vehicle parameters, particularly the steering wheel angle and position in the driving lane, is a base also for our study, where data are acquired via a car driving simulator. Based on the repeated driver’s testing (e.g. within different stages of fatigue), the deviations of parameters defining a current driver’s performance can be evaluated. An advantage of the presented approach is its non-invasive nature because it evaluates only driving data logged from the testing scenario and does not require any additional sensors. Although it operates only as an offline analysis, it might be an interesting extension to the existing methods summarized in Fig. 1. 3 MEASURING AND EVALUATION OF DRIVER’S PERFORMANCE This study is based on an approach for driver’s performance assessment published in (Jirgl et al., 2022). The principle lies in employing the cybernetic approach, where interaction between a driver and a car is described as a standard feedback control loop. It enables to modelling of a driver as a controller, and a car as a controlled element. The dynamical properties of this control loop can be then evaluated on the data acquired during a specific testing scenario. For these purposes, we developed an in-house car driving simulator – CDS. 3.1 Car driving simulator – CDS The car driving simulator – CDS involves the CDS software developed in Unreal Engine 4, and the hardware platform, see Fig. 2. It is equipped with a 49” Samsung CHG90 QLED gaming display, and with a joystick mediating control input, namely, a Logitech G920 steering wheel with pedals. The maximal rotation of the steering wheel is 900°, and the resolution corresponds to about 0.1°. Figure 2. The workstation with developed CDS for driver testing. The reason for developing an in-house CDS was to requirements on the definition of testing scenario (including the measuring conditions) and data acquisition under a sufficient sampling rate. The CDS provides data sampled by 100 Hz. These data from each testing scenario are logged for the individual drivers under a unique ID into a .csv file and Driver fatigue detection techniques Based on driver Behaviroral parameter based Eye blinking Eye closure Eye movement Head movement Facial Expression Yawn Physiological parameter based ECG EEG Pulse EMG Based on vehicle Lane detection Steering wheel movement Acceleration include e.g. time vector – t, X and Y coordinates on the scenario map, distance from the center of the lane – e, and steering wheel angle (the driver’s control actions) – u. A more detailed description of the CDS can be found in (Michalik et al., 2021). 3.2 Dynamical properties of a human driver CDS offers several testing scenarios representing different driving situations under specific conditions. The Step response scenario, see (Michalik et al., 2021), was employed in the frame of our study. This scenario implements a simple lanechanging task under a defined car speed (here 90 km/h) kept by the speed limiter. The demand to change a lane is visualized by a green arrow in front of the driver’s view, as well as by an implemented lane assistant. A driver reacts to this visual stimulus by a corresponding control action in the form of a steering wheel rotation. More detailed information about the testing procedure can be found in (Jirgl et al., 2022). As stated at the beginning of this section, such kind of interaction between a driver and a simulated car can be represented, from a cybernetic point of view, as a standard feedback control loop. Based on the measured (logged) data representing the control action, u, as a response to the visual stimulus, and the control error, e, the dynamical properties of a driver can be then evaluated using methods known from the theory of automatic control. The basis for this description is a suitable model of a human (driver) behavior. In the context of this paper, the linear model in the form of the transfer function (1) was used: 𝐻𝐻(𝑠𝑠)=𝐾𝐾𝑠𝑠 (𝑇𝑇2𝑠𝑠2+2𝜉𝜉𝑇𝑇𝑠𝑠+1) 𝑒𝑒−𝑠𝑠𝑠𝑠 , (1) where K – driver’s gain, T – time constant [s], ξ – damping, τ – reaction delay [s], s – the Laplace operator. The model, therefore, contains 4 parameters representing the dynamical properties of the driver. The gain, K, together with a time constant, T, are related in a certain way to the speed, or dynamics, of a driver’s response. The damping, ξ, is connected with oscillations observable in the response. The reaction delay, τ, defines the time needed to process the visual information and to react using the appropriate control action. The reason for selecting the model structure of (1) is discussed e.g. in (Jirgl et al., 2022). All the mentioned parameters simultaneously participate in a resulting control quality; it can be objectively evaluated using integral quality criteria defined in linear absolute or quadratic form as: 𝐽𝐽𝑎𝑎𝑎𝑎𝑠𝑠=∫|𝑒𝑒(𝜏𝜏)|𝑑𝑑𝜏𝜏 𝑡𝑡 0 , (2) 𝐽𝐽𝑞𝑞=∫𝑒𝑒(𝜏𝜏)2𝑑𝑑𝜏𝜏 𝑡𝑡 0, (3) where e(t) is the control error (in this case the distance from the center of the desired lane). The higher value of the quality criterion means a worse controller setting i.e. it corresponds to a lower quality of a driver’s response. 𝐿𝐿1 and 𝐿𝐿2 norms of derivative of the driver’s control action – 𝑢𝑢′(𝑡𝑡), see (4) and (5), define the dynamics of the control actions and correspond with the resulting comfort of a ride (they are denoted as energy criteria in the frame of this study) 𝐿𝐿1=∫|𝑢𝑢′(𝜏𝜏)|𝑑𝑑𝜏𝜏 𝑡𝑡 0 , (4) 𝐿𝐿2=√∫|𝑢𝑢′(𝜏𝜏)|2𝑑𝑑𝜏𝜏 𝑡𝑡 0, (5) The idea behind the evaluation using these parameters is that a higher value of 𝐿𝐿1 or 𝐿𝐿2 norms are related to a less comfortable and smooth ride. 3.3 Driver’s performance assessment The parameters defined in the previous section were evaluated based on data from repeated testing by Step response scenario with 50 active drivers (men and women) in the age range from 18 to 60 years. The histograms of the evaluated parameters are in Fig. 3. The principle of the driver performance assessment lies in the comparison of a currently identified set of the parameters (K, T, ξ, τ, 𝐽𝐽𝑎𝑎𝑎𝑎𝑠𝑠, 𝐽𝐽𝑞𝑞, 𝐿𝐿1, 𝐿𝐿1) for a tested driver with the histograms and consequent interpretation of the values. The results of the study (Jirgl et al., 2022) based on the dataset of 30 drivers indicated that the described approach could be advantageously used for an objective assessment of drivers’ performance. However, based on this analysis, it was found that a driver's performance from the perspective of his/her abilities or skills cannot be decided solely based on a selected parameter, but it is always a set of parameters that must be considered, which makes the final assessment rather complicated. It led to the design and development of a decision-making system for objective assessment of driver performance via a sole indicator – the driver performance index. 3.4 Driver performance index Driver performance index (DPI) is a self-designed criterion whose value from 0 to 100% is obtained on the basis of an algorithm deciding on the input parameters (K, T, ξ, τ, 𝐽𝐽𝑎𝑎𝑎𝑎𝑠𝑠, 𝐽𝐽𝑞𝑞, 𝐿𝐿1, 𝐿𝐿1). Since the algorithm must be able to consider different, often conflicting, criteria, it does not mean that achieving a higher value must necessarily mean higher quality (e.g. at the cost of a greater amount of energy used, thus an uncomfortable ride). However, in general, a higher value of the DPI corresponds to better performance of the driver from the point of view of his abilities or skills. This can subsequently be used for a relative comparison of the drivers' performance or for assessing the performance of the selected driver during repeated measurements. The DPI evaluator was implemented as a fuzzy decision system. The reason for choosing the fuzzy system is its ability to smoothly transition between evaluated states and also the 266 M. Jirgl et al. / IFAC PapersOnLine 58-9 (2024) 263–268 a) b) possibility of considering the degree of membership to the interval of values of individual evaluated parameters. Since the complexity of the fuzzy system increases with the number of input variables, there was an effort to minimize the number of these inputs and select the most important ones. By analyzing the influence of individual parameters on the driver's performance or abilities, only a sub-group of the parameters was selected, see Tab. 1. The universe for the individual inputs, as well as the intervals corresponding to the so-called “ideal driver” were defined based on the histograms in Fig. 3. Table 1. Selected inputs of the DPI evaluator. Parameter Universe Ideal values Reaction delay τ (s) (0.1−2) (0,1−0,3) The time constant T (s) (0−1,5) (0−0,2) The a bsolute criterion of quality 𝐽𝐽𝑎𝑎𝑎𝑎𝑎𝑎 (0−10) (0−2,5) The quadratic criterion of quality 𝐽𝐽𝑎𝑎𝑎𝑎𝑎𝑎 (0−7) (0−1,5) “Energy” criterion 𝐿𝐿1 (0−1) (0−0,005) “Energy” criterion 𝐿𝐿2 (0−1) (0−0,003) Using the mentioned input parameters, a cascade fuzzy system (fuzzy tree) of the Mamdani type (including FIS 1, FIS 2, and FIS 3) was designed and implemented, see Fig. 4. Figure 4. A structure of the Fuzzy DPI evaluator. The subsystems FIS 1, FIS 2, and FIS 3 have several input variables mapped with 2 or 3 trapezoidal membership functions on the corresponding universe. The membership functions of the output variables are defined as singletons. The control surfaces of the individual subsystems are in Fig. 5. All the measured data representing the tested group of drivers were processed using the DPI evaluator and the results are shown in Fig. 6. The histogram in Fig. 6 presents a distribution of the DPI parameter for the tested group of 50 drivers. For each of the drivers, the results corresponding with his/her current state can be compared with the DPI histogram, and/or with the histograms in Fig. 3. It brings e.g. a possibility to evaluate the driving performance of a given driver relatively within the tested group. 4 ASSESSMENT OF THE EFFECT OF FATIGUE ON DRIVER’S PERFORMANCE From the repeated testing of the selected driver, it is possible to evaluate any deviations in his/her behavior based on changes in the evaluated parameters covered by the DPI. These changes can be caused by many factors, where, for example, fatigue also plays a significant role. This study focuses on an experiment employing the results of a driver’s performance assessment for an effective and objective evaluation of his/her fatigue. The approach is, similarly to the methods shown in Fig. 1, based on measuring the steering wheel angle, and position of the car in the lane in the frame of a specific testing scenario. However, a novelty of this approach is that we try to decide about driver fatigue via evaluation of the dynamics of his/her responses to a visual stimulus. Application of the method to 3 volunteers is demonstrated as an example. Basic information about the tested group is in Tab. 2. The testing procedure was similar to the one described in the previous section. However, in this case, the first round of testing was followed by a 3-hour long drive on the simulator in the Long-distance drive scenario intended purely to induce fatigue. Right after that, the second round of testing was applied. Figure 3. Example of the identified driver behavioral model parameters (a) and criteria values (b) for the tested group of 50 drivers. Model parameters (FIS 1) Quality/ Energy (FIS 2) DPI (FIS 3) (0-100) % DPI ቂ 𝑇𝑇 𝜏𝜏 ቃ ൦ 𝐽𝐽 𝑙𝑙𝑙𝑙𝑙𝑙 𝐽𝐽𝑘𝑘𝑘𝑘 𝐿𝐿1 𝐿𝐿2൪ output 2 (0−1) output 1 (0 – 1) M. Jirgl et al. / IFAC PapersOnLine 58-9 (2024) 263–268 267 a) b) possibility of considering the degree of membership to the interval of values of individual evaluated parameters. Since the complexity of the fuzzy system increases with the number of input variables, there was an effort to minimize the number of these inputs and select the most important ones. By analyzing the influence of individual parameters on the driver's performance or abilities, only a sub-group of the parameters was selected, see Tab. 1. The universe for the individual inputs, as well as the intervals corresponding to the so-called “ideal driver” were defined based on the histograms in Fig. 3. Table 1. Selected inputs of the DPI evaluator. Parameter Universe Ideal values Reaction delay τ (s) (0.1−2) (0,1−0,3) The time constant T (s) (0−1,5) (0−0,2) The absolute criterion of quality 𝐽𝐽𝑎𝑎𝑎𝑎𝑎𝑎 (0−10) (0−2,5) The quadratic criterion of quality 𝐽𝐽𝑎𝑎𝑎𝑎𝑎𝑎 (0−7) (0−1,5) “Energy” criterion 𝐿𝐿1 (0−1) (0−0,005) “Energy” criterion 𝐿𝐿2 (0−1) (0−0,003) Using the mentioned input parameters, a cascade fuzzy system (fuzzy tree) of the Mamdani type (including FIS 1, FIS 2, and FIS 3) was designed and implemented, see Fig. 4. Figure 4. A structure of the Fuzzy DPI evaluator. The subsystems FIS 1, FIS 2, and FIS 3 have several input variables mapped with 2 or 3 trapezoidal membership functions on the corresponding universe. The membership functions of the output variables are defined as singletons. The control surfaces of the individual subsystems are in Fig. 5. All the measured data representing the tested group of drivers were processed using the DPI evaluator and the results are shown in Fig. 6. The histogram in Fig. 6 presents a distribution of the DPI parameter for the tested group of 50 drivers. For each of the drivers, the results corresponding with his/her current state can be compared with the DPI histogram, and/or with the histograms in Fig. 3. It brings e.g. a possibility to evaluate the driving performance of a given driver relatively within the tested group. 4 ASSESSMENT OF THE EFFECT OF FATIGUE ON DRIVER’S PERFORMANCE From the repeated testing of the selected driver, it is possible to evaluate any deviations in his/her behavior based on changes in the evaluated parameters covered by the DPI. These changes can be caused by many factors, where, for example, fatigue also plays a significant role. This study focuses on an experiment employing the results of a driver’s performance assessment for an effective and objective evaluation of his/her fatigue. The approach is, similarly to the methods shown in Fig. 1, based on measuring the steering wheel angle, and position of the car in the lane in the frame of a specific testing scenario. However, a novelty of this approach is that we try to decide about driver fatigue via evaluation of the dynamics of his/her responses to a visual stimulus. Application of the method to 3 volunteers is demonstrated as an example. Basic information about the tested group is in Tab. 2. The testing procedure was similar to the one described in the previous section. However, in this case, the first round of testing was followed by a 3-hour long drive on the simulator in the Long-distance drive scenario intended purely to induce fatigue. Right after that, the second round of testing was applied. Figure 3. Example of the identified driver behavioral model parameters (a) and criteria values (b) for the tested group of 50 drivers. Model parameters (FIS 1) Quality/ Energy (FIS 2) DPI (FIS 3) (0-100) % DPI ቂ 𝑇𝑇 𝜏𝜏 ቃ ൦ 𝐽𝐽 𝑙𝑙𝑙𝑙𝑙𝑙 𝐽𝐽𝑘𝑘𝑘𝑘 𝐿𝐿1 𝐿𝐿2 ൪ output 2 (0−1) output 1 (0 – 1) Figure 5. Control the surface of the individual fuzzy subsystems (FIS 1 – upper, FIS 2 – 2 middle, FIS 3 – bottom). For an objective and comprehensive evaluation of changes in the driver's performance, the DPI evaluation system described in the previous subsection was used. The results are presented in Fig. 7. Figure 6. Histogram of the DPI values for the tested group of 50 drivers. Table 2. Basic information about the tested subjects. Driver ID Sex Age Driving license (years) Mileage per year (km/year) No. 90 M 35 17 10 000 No. 91 F 28 11 8 000 No. 92 M 41 22 15 000 Figure 7. The DPI values for the group of drivers before and after the Long-distance drive scenario. The results indicate that the DPI value decreased in all cases after the long drive, even quite significantly in the case of driver No. 91. In such a case, it is appropriate to investigate the most significant differences, see Tab. 3. Table 3. Summary of the most influenced parameters evaluated before and after the Long-distance drive scenario. Driver No. 90 τ (s) T (s) 𝑱𝑱𝒂𝒂𝒂𝒂𝒂𝒂 𝑱𝑱𝒒𝒒 𝑳𝑳𝟏𝟏 𝑳𝑳𝟐𝟐 before 0.38 0.72 3.67 2.57 0.06 0.09 after 0.60 0.54 3.56 2.47 0.11 0.12 Driver No. 91 τ (s) T (s) 𝑱𝑱𝒂𝒂𝒂𝒂𝒂𝒂 𝑱𝑱𝒒𝒒 𝑳𝑳𝟏𝟏 𝑳𝑳𝟐𝟐 before 0.53 0.39 3.14 1.61 0.23 0.17 after 0.54 0.29 3.34 1.62 0.61 0.39 Driver No. 92 τ (s) T (s) 𝑱𝑱𝒂𝒂𝒂𝒂𝒂𝒂 𝑱𝑱𝒒𝒒 𝑳𝑳𝟏𝟏 𝑳𝑳𝟐𝟐 before 0.49 0.49 3.84 2.33 0.13 0.14 after 0.60 0.38 3.78 2.28 0.19 0.17 268 M. Jirgl et al. / IFAC PapersOnLine 58-9 (2024) 263–268 The results shown in Tab. 3 indicate, that certain changes occurred in individual parametersbetween the first and second testing drive. In all cases, the reaction delay τ increased, although only slightly for driver No. 92, which is directly related to fatigue. On the other hand, in all cases the value of the time constant T related to lower inertia decreased, but the values of the so-called energy criteria (𝐿𝐿1, and 𝐿𝐿2) increased more significantly; it means that the driver uses more energy for the control action (steering), and these control actions are not as calm and gradual. 5 CONCLUSIONS The goal of this paper was to present an experimental cybernetic approach for the evaluation of drivers’ performance and investigate its possibilities for fatigue detection. The method is based on data representing human behavior during a simple lane-changing task. These data are acquired via an inhouse developed car driving simulator. The idea of the cybernetic approach is to describe the human controller by an appropriate model and to evaluate standard metrics of the control loop summarized in section 3.2. All the parameters were evaluated on a testing group counting 50 active drivers, see Fig. 3. These data were employed as an initial data set for determining and tuning the described drivers’ performance decision-making system built on fuzzy logic. The testing procedure was then repeatedly applied to three volunteers before and after a 3-hour-long simulated loading drive on the highway inducing fatigue. In all three cases, the drivers’ performance decreased after a loading drive as Fig. 7 and Tab. 3 present. Although the results cannot be statistically significantly interpreted, they indicate a potential for further and deeper research. Future work will focus on measuring a broader dataset under fatigue influence in combination with frequently used methods (or criteria) for fatigue detection and verifying the presented results. ACKNOWLEDGMENT This work was supported by the Internal Science fund of Brno University of Technology under Grant No. FEKT-S-23-8451. REFERENCES Barua, S., Ahmed, M. U. et. al. (2019). Automatic driver sleepiness detection using EEG, EOG and contextual information. Expert Systems with Applications, volume 115, pages 121-135. Chai, M., et. al. (2019). Drowsiness monitoring based on steering wheel status. Transportation Research Part D 66, pages 95-103. Flores, M. J., Armingol, J. M. and de la Excalera, A. (2010). Real-time warning system for driver drowsiness detection using visual information. Journal of Intelligent & Robotic Systems, volume 59, pages 103-125. Hu, X. and Lodewijks, G. (2020). Detecting fatigue in car drivers and aircraft pilots by using non-invasive measures: The value of differentiation of sleepiness and mental fatigue. Journal of Safety Research, volume 72, pages 173-187. Jirgl, M., Fiedler, P. and Bradac, Z. (2022). Human Driver Performance Assessment based on HiLCPS Concept. In IFAC-PapersOnLine. pp. 345-350. Katyal, Y., Alur, S. and Dwivedi, S. (2014). Safe driving by detection lane discipline and driver drowsiness. IEEE International Conference on Advanced Communication Control and Computing Technologies (ICACCCT), pages 1008-1012. Khadkikar, P., Banginwar, R. et. al. (2023). A comparative study of driver drowsiness detection techniques. Proceedings of the International Conference on Sustainable Computing and Smart Systems (ICSCSS 2023), pages 911-915. Lu, J., Zheng, X. et. al. (2021). Can steering wheel detect your driving fatigue? IEEE Transaction on Vehicular Technology, volume 70, issue 6. Michalik, D., Jirgl, M., Arm, J. and Fiedler, P. (2021). An Unreal Engine 4-Based Vehicle Driving Simulator and its Application in Driver Behavior Analysis. Safety, 7 (2), 1-15. Nudzikova, P., and Slanina, Z. (2016). User Identification by Biometric Methods. In Advances in Intelligent Systems and Computing (pp. 181–190). doi:10.1007/978-3-31927644-1_17. Otahalova, T., Slanina, Z., and Vala, D. (2012). Embedded sensors system for real time biomedical data acquisition and analysis. IFAC Proceedings Volumes, 45(7), 261– 264. doi:10.3182/20120523-3-cz-3015.00050. Philip, P. and Åkerstedt, T. (2006). Transport and industrial safety, how are they affected by sleepiness and sleep restriction? Sleep Medicine Reviews, volume 10, issue 5, pages 347-356. Satheesh, N. and Basha, J. (2023). Review on various real time driver drowsiness detection techniques. AIP Conference Proceedings. Shaik, E. (2023). A systematic review on detection and prediction of driver drowsiness. Transportation Research Interdisciplinary Perspectives, volume 21. Vysoký, P. (2001). Central fatigue identification of human operator. Neutral network world, volume 11, issue 5, pages 525-535. Zhang, Z. (2022). Study on the influence of driving fatigue on drivers’ effective reaction time. 10th International Conference on Traffic and Logistic Engineering, ICTLE 2022, pages 89-94.