Next Article in Journal
Relationship between Self-Reported Neighborhood Safety and Happiness and Life Satisfaction among Women in Low-Middle Income Countries
Previous Article in Journal
Assessing School Travel Safety in Scotland: An Empirical Analysis of Injury Severities for Accidents in the School Commute
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Safety
Volume 8
Issue 2
10.3390/safety8020030
safety-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles thumb_up
...
Endorse textsms
...
Comment
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Effects of Automation and Fatigue on Drivers from Various Age Groups

by
Sadegh Arefnezhad
*,
Arno Eichberger
and
Ioana Victoria Koglbauer
Institute of Automotive Engineering, Graz University of Technology, 8010 Graz, Austria
*
Author to whom correspondence should be addressed.
Safety 2022, 8(2), 30; https://doi.org/10.3390/safety8020030
Submission received: 22 January 2022 / Revised: 29 March 2022 / Accepted: 6 April 2022 / Published: 11 April 2022
Browse Figures
Review Reports Versions Notes

Abstract

:
This study explores how drivers are affected by automation when driving in rested and fatigued conditions. Eighty-nine drivers (45 females, 44 males) aged between 20 and 85 years attended driving experiments on separate days, once in a rested and once in a fatigued condition, in a counterbalanced order. The results show an overall effect of automation to significantly reduce drivers’ workload and effort. The automation had different effects, depending on the drivers’ conditions. Differences between the manual and automated mode were larger for the perceived time pressure and effort in the fatigued condition as compared to the rested condition. Frustration was higher during manual driving when fatigued, but also higher during automated driving when rested. Subjective fatigue and the percentage of eye closure (PERCLOS) were higher in the automated mode compared to manual driving mode. PERCLOS differences between the automated and manual mode were higher in the fatigued condition than in the rested condition. There was a significant interaction effect of age and automation on drivers’ PERCLOS. These results are important for the development of driver-centered automation because they show different benefits for drivers of different ages, depending on their condition (fatigued or rested).

1. Introduction

1.1. Motivation

According to recent reports, driver drowsiness is one of the major causes of traffic accidents [1], thus research on driver drowsiness in manual and automated driving is important for road safety. The National Highway Traffic Safety Administration (NHTSA) has announced that 72,000 drowsiness-related car accidents were reported by police from 2009 to 2013 [2]. The American Automobile Association (AAA) has also reported that 40% of the drivers that participated in a study admitted to falling asleep while driving at least once in their life, whereas 27% of them fell asleep in the previous month [3]. Dingus et al. [4] have also reported that about 22–24% of car crashes or near-crashes are related to the drivers’ drowsiness.
In addition to the impairing influence of drowsiness in manual driving, monitoring the drivers’ state and their ability to control the car is one of the main requirements of the third level of automated vehicles [5]. At this automated level, the driver will still be responsible for the car’s performance and they should act safely and promptly in case of automated system faults or complex traffic scenarios [6,7]. Notwithstanding research findings [8] showing that learning can interact with fatigue, leading to shorter reaction times, most previous studies demonstrated that drowsiness has a significant influence to increase drivers’ reaction time for braking or steering maneuvers in risky situations to prevent accidents [9,10,11]. It may not be safe to hand over the car’s control to the drowsy drivers in automated cars which can increase the risk of accidents.

1.2. Effects of Human Factors in Driver Drowsiness

1.2.1. Effects of Automation, Age, and Gender

Multiple benefits are considered to emerge from driving automation such as reducing drivers’ workload and improving the safety and efficiency of road transportation [12,13]. A reduction in the cognitive and physical workload associated with a driving task is considered to reduce drivers’ effort and make driving less prone to errors [14,15]. Research has been dedicated to adapting the level of automation to the level of the operator’s workload and avoiding both, a too high and a too low effort [16,17,18,19]. When the workload becomes too low, the driving performance is also expected to decrease because drivers may become disengaged from the task. Thus, driving automation can also potentially reduce drivers’ situational awareness [12]. There is a relationship between workload and fatigue, meaning that high workload is a predictor of increased fatigue [20].
Research shows significant interaction effects of drivers’ age and gender in relation to automation (see for example [21,22,23]). An assessment of the autonomous emergency braking (AEB) system on a test-driving area using a pedestrian-dummy as a trigger showed that female drivers perceive the AEB braking as being less safe than men did [24,25]. This effect was confirmed by simulated driving tests [21].
An assessment with the Adaptive Cruise Control (ACC) on the road showed significant gender differences in the estimation of the gap between vehicles, perceived usability, and comfort when driving with the ACC [25]. Drivers with an age of 60–75 years selected a slower ACC speed than the drivers in their 20s and 30s [22]. Additionally, drivers with an age of 40–49 years selected a slower ACC speed than those aged 30–39 years [22]. The same study also shows that automation influences drivers’ perceived workload and this effect is mediated by drivers’ age. Muslim et al. [23] investigated age and gender differences in interacting with an automated driving system (ADS) with four distinct levels of automation: The system could only keep the lane and continue automated driving at a slow speed (20 km/h) (ADS-level 1), or the system requested the driver to take over the vehicle control and change lanes manually (ADS-level 2), or the system requested the driver’s permission to execute the lane-change maneuver automatically (ADS-level 3) or the system informed the driver that an automatic lane-change maneuver will start in 6 s and the driver could veto the maneuver execution (ADS-level 4). Muslim et al. [23] found significant effects of the level of automation, gender, and age on drivers’ reaction time. Younger female drivers had longer RT than older female drivers when using ADS-3, and longer RT than younger males when driving with ADS-4.
Given the potential benefits of automation for drivers, and for the safety and efficiency of road transportation, it is necessary to consider the age and gender aspects when designing and assessing automated systems.

1.2.2. Effects of Fatigue, Age, and Gender

In a survey study with 1000 Australian drivers, Obst et al. [26] found that 80% of drivers across all age and gender groups drove when sleepy. In addition, many drivers reported that they frequently drove when sleepy, despite being aware of the risks. Most frequently, young to middle-aged drivers and male drivers reported driving while sleepy [26]. Akerstedt and Kecklund [27] studied the influence of age and gender on the risk of being involved in driving accidents. According to their outcomes, the risk of accidents for the young drivers (18–24 years old) is 5 to 10 times higher in the late-night than during the morning. Moreover, the risk of having an accident during the night is two times higher for men than for women. However, Soares et al. [28] reported that male and older drivers are less prone to feel drowsy than other drivers based on the subjective rating of drowsiness using the Karolinska Sleepiness Scale (KSS). The vulnerability of young drivers to sleepiness while driving at nighttime was also studied in [29] using electroencephalogram (EEG) signals to measure the influence of sleepiness on the power of related EEG subbands. This study shows that young drivers can experience extreme levels of sleepiness more rapidly than older drivers during a long drive.
Ǻkerstedt et al. [30] investigated the effects of driving home from a night shift as compared to a normal shift using a driving simulator. Driving home from the night shift was associated with larger lateral deviation, longer eye closure duration, increased subjective sleepiness, and more frequent incidents as compared to driving after the day shift. In another simulator study, Hargutt et al. [31] showed that drivers’ fatigue leads to a loss of attention as indicated by subjective ratings and increased reaction times. Drivers tried to increase speed and, thus, the difficulty of the driving task to regulate their attention [31].
Ahlström et al. [32] studied the influence of partially automated driving (level 2) on fatigue in four different driving tests: daytime-manual, daytime-automated, nighttime-manual, and nighttime-automated. In their study, different sleepiness indicators included subjective ratings on the KSS, blink durations, the PERcentage of eyelid CLOSure (PERCLOS), pupil diameter, heart rate, and recordings of 64 EEG channels. Their results show that the drivers felt more fatigued in the automated tests. The drowsiness effects related to automated driving were stronger during the night as compared to daytime driving tests. In addition, the effect of performing a secondary task (performing a quiz test) on drowsiness during automated driving was studied in [33], where drowsiness in drivers was measured using eyelid movement data. Drivers participated in three different tests: (1) manual driving (no automation), (2) automated driving (hands-off), (3) automated driving, and additionally performing a secondary task. According to the results, the drowsiness index increased in the first and second tests while it remained approximately constant during the third type of tests which highlights the importance of performing a secondary task for keeping the drivers alert during automated driving.
In a review of 10 driver sleepiness studies, Scarpelli et al. [34] concluded that older drivers are less vulnerable to sleep loss and sleepiness-related driving impairments than young adults. For example, Campagne et al. [35] compared the frequency of driving errors and the vigilance of older drivers (60–70 years) and younger drivers (20–30 years and 40–50 years) during long, monotonous night driving sessions in the simulator. They found that driving errors were more frequent in young drivers. Younger drivers also showed a positive correlation between the alpha power of the EEG considered to indicate low vigilance and the number of driving errors. In older drivers, no correlation was found between driving errors and their level of vigilance assessed through the EEG [35]. Lowden et al. [29] also compared young (18–24 years) and older drivers (55–64 years) during simulated morning and evening driving sessions. They found that, although subjective sleepiness scores from the KSS were higher during night driving in both age groups, the younger drivers had higher scores than the older drivers.
Although research shows interdependencies between automation and fatigue [32], automation and workload [12,13], automation and age [23], workload and fatigue [20], workload and age [22], fatigue, age, and gender [27,28,29], there is a gap in understanding how these variables vary simultaneously. This study aims to address these interactions.

2. Research Questions

This study investigates the effects of automation, drivers’ age, gender, and fatigue conditions (e.g., fatigued due to sleep deprivation, extended wakefulness or rested) on drivers’ workload, subjective and objective fatigue, and reaction time. In addition, in an exploratory approach, correlations between drivers’ age, subjective and objective fatigue, and effort are analyzed for both driving modes (automated and manual) and experimental conditions (fatigued and rested).

3. Materials and Methods

3.1. Participants

Ninety-two drivers volunteered for participation in the study. Data from 3 participants were incomplete. Thus, the data of eighty-nine active drivers (45 females, 44 males) were used in this study. Descriptive data for the age and driving activity of the participants are presented in Table 1. There were no significant differences between the gender and age groups related to the driving activity in the past 12 months. Informed consent was obtained from participants before the experiments, and they received a compensation of EUR 50 after finishing the experiment. The study was conducted according to the ethical guidelines of the Declaration of Helsinki and the General Data Protection Regulation of the European Union. The study protocol was approved by the Ethics Committee of the Medical University of Graz in vote 30–409 ex 17/18 dated 3 August 2018.

3.2. Equipment

This study uses the dataset collected in the WACHSens project which was a collaborative project of these partners: (1) Human Research Institute Weiz, (2) Graz University of Technology, (3) apptec Factum Vienna, and (4) AVL UK. The Automated Driving Simulator of Graz (ADSG) at the Institute of Automotive Engineering (Graz University of Technology) was used to perform the driving tests. The ADSG is shown in Figure 1.
In the ADSG, eight LCD panels simulate the driving track at an angle of 180 degrees. A rear screen is added to the simulator that can be observed in the inner mirror of the car cabin. To improve the immersion feature of the ADSG, the engine, rolling and wind noises were also generated. The vibration of the car chassis and seats were produced by four bass shakers. The haptic feedback of the steering wheel was produced by the Sensodrive™ steering wheel system (https://www.sensodrive.de/, accessed on 9 April 2022). The vehicle dynamic of the car was simulated using the full vehicle software AVL-VSM™ (https://www.avl.com/-/avl-vsm-vehicle-simulation, accessed on 9 April 2022) that also calculates the engine speed and torque for the noise generators. To simulate the automated driving mode in the ADSG, two different driver assistance systems were implemented: (1) Adaptive Cruise Control (ACC) and (2) Lane-Keeping Assist (LKA). The ACC and LKA control the longitudinal and lateral vehicle dynamic during the automated driving tests, respectively. Before the automated tests, the driver only needs to turn on both ACC and LKA using a provided human–machine interface. Thus, the driver gives no input to control the car during the automated test. A “coffee cup” icon appeared three times randomly during the test on the head-up display (HUD). The icon disappeared when the drivers pressed the steering wheel. A pressure-sensitive foil was placed inside the steering wheel that sensed pressure inserted from the driver’s hands. Thus, the driver did not need to reach out his hands unless they were grabbing the steering wheel. Our previous studies explain more details of the driving simulator [36,37].

3.3. Dependent and Independent Variables

The workload was self-assessed by the drivers using the NASA Task Load Index [39] with six subscales: mental demand, physical demand, temporal demand, performance, effort, and frustration. Each NASA TLX scale ranged from 0 (very low) to 10 (very high). Subjective fatigue was assessed using the Karolinska Sleepiness Scale (KSS) ranging from 1 (very alert) to 9 (very sleepy, effort to stay awake, fighting sleep) [40].
The PERCLOS was calculated during driving as the proportion of the cumulative time when the eyes are more than 80% closed during a 1 min interval [41]. For example, if eyes are cumulatively more than 80% closed for 15 s, then the PERCLOS will be equal to 15 60 or 0.25. If the eyes are never closed more than 80% the PERCLOS will be zero during the corresponding time window and if they are always more than 80% closed the PERCLOS will be 1. Previous studies show that PERCLOS is an accurate indicator of driver sleepiness [42,43]. In addition, the reaction time (RT) of the drivers to a “coffee cup” icon that appeared at three random times on the head-up display (HUD) was recorded and analyzed. The icon disappeared when the drivers pressed the steering wheel.
Independent variables were the condition of the driver (fatigued versus rested), the driving mode (manual versus automated), and the participant’s age and gender. Each participant attended driving experiments on two days, once in a rested and once in a fatigued condition, in a counterbalanced order to avoid the order of presentation artifacts. Counterbalancing means that the participant samples (age and gender groups) were divided in half, with one half completing the two conditions (fatigued versus rested; manual versus automated driving) in one order and the other half completing the conditions in the reverse order. In the fatigued condition, the drivers were either in a state of extended wakefulness (16 h awake) or sleep deprivation (4 h sleep in the previous night). Generally, adults need between 7 and 9 h of sleep per night [44].
During each test day, the participants drove for 30 min on a motorway in each, an automated and a manual mode, in a counterbalanced order. In the automated driving mode, the drivers used the Adaptive Cruise Control and Lane-Keeping Assist.

3.4. Experimental Procedure

The drivers were instructed to drive according to rules and regulations. Each participant received a familiarization session with the driving simulator and practiced the reaction prompted by the coffee cup icon. The participants were instructed to react fast to the coffee cup icon that was randomly displayed during driving. Before the test, the participants filled out the KSS questionnaire that was used as a baseline. After the driving test, the participants filled out the NASA-TLX and KSS questionnaires.

3.5. Data Analysis

Repeated-measures analysis of variance was calculated for analyzing the effects of automation, condition, gender, and age. There were two within-subjects factors: condition (with two levels fatigued and rested) and driving mode (manual and automated). There were two between-subjects factors: gender (female and male) and age group (20–49 years and 50–85 years). Data transformation by l n ( x + 1 ) was used before calculating the ANOVA. In addition, non-parametric coefficients of correlations have been calculated using the raw data. For the statistical tests, alpha was set at 0.05.

4. Results

4.1. Automation Effects on Drivers’ Workload

The results show significant effects of the driving mode on the mental demand [F(1,85) = 53.54, p < 0.0001, η2 = 0.39], physical demand [F(1,85) = 56.14, p < 0.0001, η2 = 0.40], temporal demand [F(1,85) = 42.28, p < 0.0001, η2 = 0.33], performance [F(1,85) = 46.21, p < 0.0001, η2 = 0.35], and effort [F(1,85) = 53.07, p < 0.0001, η2 = 0.38], as rated by the drivers. Frustration was not significantly influenced by the driving mode. As Table 2 shows, the drivers perceived a lower workload and performance during automated driving as compared to manual driving. In the supplementary information, Table S1 presents the mean and standard deviation of drivers’ workload for different age and gender groups.
As illustrated in Figure 2, the interaction term between condition and automation was significant for drivers’ perceived temporal demand (time pressure) [F(1,85) = 12.01, p < 0.001, η2 = 0.12], effort F[(1,85) = 3.96, p < 0.05, η2 = 0.05], and frustration [F(1,85) = 9.14, p < 0.003, η2 = 0.10]. The effects on mental demand, physical demand, and perceived performance did not reach statistical significance and are not illustrated. Temporal demand was significantly affected by drivers’ age [F(1,85) = 5.04, p < 0.03, η2 = 0.06]. Subjective time pressure was lower in the age group 20–49 (M = 0.754, SD = 0.078) as compared to the age group 50–85 (M = 0.994, SD = 0.073). Other workload aspects were not significantly affected by age. The interaction term age*gender did not reach statistical significance.

4.2. Automation Effects on Drivers’ PERCLOS

As Table 2 shows, all parameters of drivers’ PERCLOS were lower in manual as compared to automated driving, which also means less fatigue, since PERCLOS is an accepted indicator. These differences reached statistical significance [minimum PERCLOS F(1,71) = 5.08, p < 0.03, η2 = 0.07; maximum PERCLOS F(1,71) = 25.56, p < 0.0001, η2 = 0.30; median PERCLOS F(1,71) = 21.76, p < 0.0001, η2 = 0.23; mean PERCLOS F(1,71) = 26.46, p < 0.0001, η2 = 0.27 and PERCLOS variance F(1,71) = 6.76, p < 0.01, η2 = 0.08].
As illustrated in Figure 3, the interaction term condition between and automation was significant for drivers’ mean, median, and maximum PERCLOS [mean PERCLOS F(1,71) = 8.43, p < 0.005, η2 = 0.11; median PERCLOS F(1,71) = 7.21, p < 0.009, η2 = 0.09 and maximum PERCLOS F(1,71) = 7.21, p < 0.009, η2 = 0.09]. The effects on minimum PERCLOS and variance PERCLOS did not reach statistical significance. In the supplementary information, Table S2 presents the mean and standard deviation of drivers’ PERCLOS for different age and gender groups.

4.3. Automation Effects on Drivers’ Subjective Fatigue

The analysis of KSS ratings shows that drivers felt significantly more fatigued after driving in the automated mode [Mean = 1.895, SD = 0.032] and manual mode [Mean = 1.85, SD = 0.032], as compared to the baseline values before the tests [Mean = 1.561, SD = 0.03], [F(2,170) = 81.74, p < 0.0001, η2 = 0.49]. The interaction term condition and automation was also significant [F(2,170) = 15.39, p < 0.0001, η2 = 0.15], (see also Figure 4). As illustrated in Figure 4, the baseline KSS was higher in the fatigued as compared to the rested condition, because the drivers were already fatigued when they started the experiment. This was expected if the participants followed the instruction for the fatigued part of the experiment. In the supplementary information, Table S3 presents the mean and standard deviation of drivers’ KSS scores for different age and gender groups.

4.4. Automation Effects on Drivers’ Reaction Time

Driving automation slowed down the reaction time (RT) of drivers to icons that appeared at random intervals on the head-up display during driving (see Table 2). This effect was significant for various parameters [minimum RT F(1,44) = 13.73, p < 0.001, η2 = 0.24; maximum RT F(1,44) = 7.79, p < 0.008, η2 = 0.15; median RT F(1,44) = 18.28, p < 0.0001, η2 = 0.29; mean RT F(1,44) = 11.49, p < 0.001, η2 = 0.21], except the RT variance. The interaction term between automation and condition did not reach significance for the RT. In the supplementary information, Table S4 presents the mean and standard deviation of drivers’ reaction time for different age and gender groups.

4.5. Effects of Drivers’ Age and Gender

Interaction terms automation–gender and the automation–age group did not have a statistically significant effect on drivers’ workload, their reaction time to the icons on the head-up-display, or the subjective fatigue scores on the KSS. The interaction term between automation and age group had a significant effect on maximum PERCLOS [F(1,71) = 5.24, p < 0.025, η2 = 0.069], with differences in PERCLOS between automated and manual mode being larger in the age group 20–49 years as compared to the group 50–85 years (see Figure 5). Effects of age, gender, or fatigue condition did not significantly influence drivers’ reaction time. In the supplementary information, Tables S1–S4 present the mean and standard deviation of drivers’ workload, PERCLOS, KSS, and reaction time for different age and gender groups, respectively.

4.6. Correlations between Age, Effort, and Fatigue

In addition, correlations between subjective effort, subjective fatigue, age, and different PERCLOS parameters have been calculated. The details of these correlations are presented in Table S5 of the supplementary information. The results show a significant positive correlation between the scores of subjective effort and fatigue related to automated driving in the rested (Rho = 0.428, p < 0.0001, N = 89) and fatigued condition (Rho = 0.288, p < 0.003, N = 89). Similarly, there were significant positive correlations between the scores of subjective effort and fatigue related to manual driving in the rested (Rho = 0.439, p < 0.0001, N = 89) and fatigued condition (Rho = 0.432, p < 0.0001, N = 89). When driving with automation in the rested condition, only correlations between subjective fatigue and minimum PERCLOS reached statistical significance. In the fatigued condition, there were significant correlations between subjective fatigue and maximum, median, mean PERCLOS, and PERCLOS variance both in the automated and manual driving modes. In the fatigued condition, age correlates negatively with the maximum, median, mean PERCLOS, and PERCLOS variance when driving with automation. Similarly, when driving manually in the fatigued condition, age correlates negatively with the maximum PERCLOS and PERCLOS variance.

5. Discussion

Research shows that driver drowsiness is one of the major causes of traffic accidents [1]. This explorative study aimed to investigate how drivers are affected by automation when driving in rested and fatigued conditions. Overall, the results show a positive effect of automation to significantly reduce drivers’ perceived workload and effort, confirming the findings of previous works in [12,13,14]. Nevertheless, the results of this study show that automation has different effects, depending on if the drivers are rested or fatigued. The temporal demand experienced by drivers and their effort was higher in the manual mode than in the automated mode, and these differences were larger when fatigued as compared to the rested condition. An inverse effect was observed for frustration. This was similar in both conditions when driving automation was used. However, in manual driving mode frustration was higher when drivers were in a fatigued state as compared to the rested condition. The higher subjective effort was associated with higher subjective fatigue in both driving modes, automated and manual.
Despite positive effects to reduce drivers’ workload and effort, driving automation that releases the human from steering and speed control as used in this study, also increases drivers’ drowsiness, meaning that drivers can be less vigilant and prepared to react. Although high effort is a known contributor to fatigue, decreasing drivers’ effort under a certain level can have a negative effect [15,44]. The results also show that during automated driving, drivers’ subjective fatigue, the percentage of eye closure (PERCLOS), and reaction times were also higher as compared to manual driving. The PERCLOS was higher in the automated mode than in manual mode, and these differences were larger when fatigued as compared to the rested condition. Although the reaction to the icon is different from taking over the control of a vehicle, a similar delay could be inferred for take-over. The slower reaction time in the automated mode as compared to manual mode needs to be considered in the design of the human–machine interaction for the third level of automated vehicles [5]. At level three, the driver is expected to act safely and promptly in case of automated system faults or complex traffic scenarios [6,7]. For this, a longer handover time is recommended.
As expected, in the fatigued condition, higher fatigue scores were associated with higher values of the maximum, median, mean PERCLOS, and PERCLOS variance in both driving modes. However, drivers’ age added interesting effects. In the age group 50–85 years, the maximal PERCLOS was higher than in the age group 20–49 years when driving manually. Nevertheless, in the age group 50–85 years, the maximal PERCLOS was lower than in the age group 20–49 years when using automation. The automation eliminates the physical load associated with the driving task and resources can focus on monitoring. Thus, this effect is particularly beneficial for drivers from the older age group, as demonstrated by the positive effect of automation to reduce drivers’ PERCLOS. As the results show, older age was associated with lower values of the maximum, median, mean PERCLOS, and PERCLOS variance during automated driving in the fatigued condition. In addition, older age was associated with lower values of the maximum PERCLOS and PERCLOS variance also during manual driving in the fatigued condition. This is well in line with the findings that older drivers are less vulnerable to sleep loss than young adults [27,32,33], and older drivers are less prone to become drowsy [28]. In the rested condition, these correlations between age and PERCLOS did not reach statistical significance, probably because this relationship manifests only in a fatigued state. However, more research is needed to clarify this finding. The younger drivers seem to be more alert when the task load is higher (e.g., manual mode). This interpretation could be supported by previous findings showing that drivers tried to increase speed and, thus, the difficulty of the driving task to improve their alertness in a fatigued state [28]. Nevertheless, when interpreting these results in combination with the negative correlation between age and subjective fatigue when driving fatigued in both manual and automated modes, the question arises if age is the sole cause of this effect, or if lifestyles also play a role. This aspect needs further investigation and could be addressed in future studies. Although gender effects did not reach statistical significance, it is a recommended good practice to apply a gender-relevant methodology and address possible gender-specific needs and preferences in research [21,22,45].
These results are important because they show emergent effects and interdependencies of automation, fatigue, task demand, and effort. In addition, these results show how automation affects drivers of different ages when driving in both fatigued and rested conditions. Understanding drivers’ subjective experience and reactions are essential for the development of driver-centered automation. Monitoring drivers’ state and their ability to control the car is one of the main requirements of the third level of automated vehicles [5]. Our results show that a slower reaction time needs to be accounted for in developing systems for the third level of automation. At this level, the driver is expected to take over the control of the vehicle in complex traffic scenarios or cases of automation failures [6,7]. In addition, the results of this study show that drivers’ age and condition (e.g., fatigue) are variables that will also need consideration in the development of adaptive automation. Thus, the scope of adaptive automation research focusing on workload and effort [16,17,18,19] could be enlarged to include other variables such as age and drowsiness state.
It was not possible to collect a dataset from drivers of all of all ages in this study, thus the analysis is limited to two broader age groups (20–49 and 50–85 years). There is no consensus yet in the literature on the age groups to be considered. Other studies on driver drowsiness describe age groups such as young (18 ± 24 years), intermediate (24 ± 45), middle-aged (45 ± 64), old (65+) [27], or 25–34 years, 35–44 years, 55–64 years, 65 years, and over [26]. Some studies of driver drowsiness only present the average values (e.g., mean age of 36.2 ± 16.4 years) [28], or the range (e.g., between 26 and 65 years) and mean 37.95 ± 12.81 years [8]. Studies of age and automation describe age groups in decades [21,22] or considered drivers younger and older than 45 years [23]. Sleep studies describe age groups in decades [44] or five-year age groups. For example, Ohayon et al. [46] performed a meta-analysis of sleep across the human lifespan using data from 65 studies performed with 3577 participants divided in five-year age groups.

6. Conclusions

For several years, the market introduction of conditional vehicle automation according to SAE J3016 level 3 was obstructed due to undefined procedures for vehicle certification. It has been especially difficult to monitor the driver’s ability to handle take-over requests when the automation system recognizes inappropriate system functions outside of the operational design domain. Therefore, the UN regulation No. 157 for Automated Lane Keeping Systems has been developed and approved; however, this procedure focuses on monitoring the driver’s gaze, without taking into account the driver’s actual mental performance in operating an automated vehicle and the related human factors.
The results of this study show that although automation generally has a positive effect to reduce drivers’ workload, other factors such as drivers’ age and condition also play a significant role. The optimum levels of effort and automation support depend also on drivers’ age and condition (fatigued or rested), which modulate the effect of automation. The study also showed that drivers in the automated vehicle mode showed slower reaction times, which would affect take-over procedures. We recommend that the design of future systems needs to compensate for the driver’s age, condition, and reaction time for proper human–machine interaction.

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/safety8020030/s1, Table S1. Workload self-ratings of female and male drivers from both age groups related to manual and automated driving in the rested and fatigued conditions; Table S2. PERCLOS of female and male drivers from both age groups related to manual and automated driving in the rested and fatigued conditions; Table S3. KSS self-ratings of female and male drivers from both age groups related to the baseline before-test, manual and automated driving in the rested and fatigued conditions; Table S4. Reaction time (RT) of female and male drivers from both age groups related to manual and automated driving in the rested and fatigued conditions; Table S5. Non-parametric Spearman’s coefficients of correlation between subjective fatigue, age, and PERCLOS (* Indicates significant correlations at p < 0.05).

Author Contributions

Conceptualization, A.E. and I.V.K.; methodology, A.E. and I.V.K.; investigation, A.E. and S.A.; data curation, S.A.; writing—original draft preparation, S.A. and I.V.K.; writing—review and editing, A.E. and I.V.K.; project administration, A.E.; funding acquisition, A.E.; All authors have read and agreed to the published version of the manuscript.

Funding

This project titled WACHSens was carried out by Human Research Institut für Gesundheit-stechnologie und Präventionsforschung GmbH, Graz University of Technology, AVL Powertrain UK Limited, and Factum apptec ventures GmbH. It was funded by the Austrian Research Promotion Agency (FFG) via the Future Mobility Program (grant no. 860875).

Informed Consent Statement

Informed consent was obtained from all subjects involved in the study.

Acknowledgments

The authors acknowledge Lena Wagner, who assisted in data preparation as well as Matthias Frühwirth and Clemens Kaufmann for their assistance in the experimental design and data preparation. Open Access Funding by the Graz University of Technology.

Conflicts of Interest

The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

References

  1. Tefft, B.C. Prevalence of Motor Vehicle Crashes Involving Drowsy Drivers, United States, 2009–2013; AAA Foundation for Traffic Safety: Washington, DC, USA, 2014; Available online: http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.684.6347&rep=rep1&type=pdf (accessed on 31 October 2021).
  2. National Highway Traffic Safety Administration. Drowsy Driving 2015 DOT HS 812 446; U.S. Department of Transportation: Washington, DC, USA, 2017. Available online: https://crashstats.nhtsa.dot.gov/Api/Public/ViewPublication/812446 (accessed on 31 October 2021).
  3. Tefft, B.C. The Prevalence and Impact of Drowsy Driving; AAA Foundation for Traffic Safety: Washington, DC, USA, 2010; Available online: https://aaafoundation.org/wp-content/uploads/2018/02/2010DrowsyDrivingReport.pdf (accessed on 31 October 2021).
  4. Dingus, T.A.; Neale, V.L.; Garness, S.A.; Hanowski, R.J.; Keisler, A.S.; Lee, S.E.; Perez, M.A.; Robinson, G.S.; Belz, S.M.; Casali, J.G.; et al. Impact of Sleeper Berth Usage on Driver Fatigue; Final Project Report; Virginia Tech Transportation Institute: Blacksburg, VA, USA, 2001; Available online: https://vtechworks.lib.vt.edu/bitstream/handle/10919/55096/PB2002107930.pdf?sequence=1&isAllowed=y (accessed on 31 October 2021).
  5. Hecht, T.; Feldhütter, A.; Radlmayr, J.; Nakano, Y.; Miki, Y.; Henle, C.; Bengler, K. A Review of Driver State Monitoring Systems in the Context of Automated Driving. In Proceedings of the 20th Congress of the International Ergonomics Association (IEA 2018), Volume VI, Transport ergonomics and Human Factors (TEHF), Aerospace Human Factors and Ergonomics, Florence, Italy, 26–30 August 2018; Bagnara, S., Tartaglia, R., Albolino, S., Alexander, T., Fujita, Y., Eds.; Springer: Cham, Switzerland, 2018; pp. 398–408, ISBN 978-3-319-96073-9. [Google Scholar]
  6. Serban, A.; Poll, E.; Visser, J. A Standard Driven Software Architecture for Fully Autonomous Vehicles. In Proceedings of the 2018 IEEE International Conference on Software Architecture Companion (ICSA-C), Seattle, WA, USA, 30 April–4 May 2018. [Google Scholar] [CrossRef] [Green Version]
  7. Hirz, M.; Walzel, B. Sensor and object recognition technologies for self-driving cars. Comput.-Aided Des. Appl. 2018, 15, 501–508. [Google Scholar] [CrossRef] [Green Version]
  8. Kreuzmair, C.; Gold, C.; Meyer, M.-L. The influence of driver fatigue on take-over performance in highly automated vehicles. In Proceedings of the 25th International Technical Conference on the Enhanced Safety of Vehicles (ESV), Detroit, MI, USA, 6 June 2017. [Google Scholar]
  9. Guo, M.; Li, S.; Wang, L.; Chai, M.; Chen, F.; Wei, Y. Research on the Relationship between Reaction Ability and Mental State for Online Assessment of Driving Fatigue. Int. J. Environ. Res. Public Health 2016, 13, 1174. [Google Scholar] [CrossRef] [PubMed]
  10. Corfitsen, M. Tiredness and visual reaction time among young male nighttime drivers: A roadside survey. Accid. Anal. Prev. 1994, 26, 617–624. [Google Scholar] [CrossRef]
  11. Philip, P.; Sagaspe, P.; Moore, N.; Taillard, J.; Charles, A.; Guilleminault, C.; Bioulac, B. Fatigue, sleep restriction and driving performance. Accid. Anal. Prev. 2005, 37, 473–478. [Google Scholar] [CrossRef]
  12. De Winter, J.C.; Happee, R.; Martens, M.H.; Stanton, N.A. Effects of adaptive cruise control and highly automated driving on workload and situation awareness: A review of the empirical evidence. Transp. Res. Part F Traffic Psychol. Behav. 2014, 27, 196–217. [Google Scholar] [CrossRef] [Green Version]
  13. Vollrath, M.; Krems, J.F. Verkehrspsychologie: Ein Lehrbuch für Psychologen, Ingenieure und Informatiker, 1. Auflage Transportation Psycholgy: A Manual for Psychologists, Engineers and Computer Scientists, 1st ed.; Kohlhammer Verlag: Stuttgart, Germany, 2011; ISBN 9783170295612. [Google Scholar]
  14. Hoedemaeker, M.; Brookhuis, K.A. Behavioural adaptation to driving with an adaptive cruise control (ACC). Transp. Res. Part F Traffic Psychol. Behav. 1998, 1, 95–106. [Google Scholar] [CrossRef]
  15. Stanton, N.A.; Marsden, P. From fly-by-wire to drive-by-wire: Safety implications of automation in vehicles. Saf. Sci. 1996, 24, 35–49. [Google Scholar] [CrossRef] [Green Version]
  16. Dijksterhuis, C.; Stuiver, A.; Mulder, B.; Brookhuis, K.A.; de Waard, D. An adaptive driver support system: User experiences and driving performance in a simulator. Hum. Factors 2012, 54, 772–785. [Google Scholar] [CrossRef]
  17. Koglbauer, I.; Braunstingl, R. Applications of Cardiac and Electrodermal Activity Assessment in Aviation. In Aviation Psychology: Applied Methods and Techniques; Koglbauer, I., Biede-Straussberger, S., Eds.; Hogrefe: Boston, MA, USA, 2021; pp. 141–162. ISBN 9780889375888. [Google Scholar]
  18. Hockey, G.R.J.E.; Gaillard, A.W.K.E.; Burov, O. Operator Functional State: The Assessment and Prediction of Human Performance Degradation in Complex Tasks; IOS Press: Amsterdam, The Netherlands, 2003; ISBN 9781586033620. [Google Scholar]
  19. Prinzel III, L.J.; Parasuraman, R.; Freeman, F.G.; Scerbo, M.W.; Mikulka, P.J.; Pope, A.T. Effects of a Psychophysiological System for Adaptive Automation on Performance, Workload, and the Event-Related Potential P300 Component. Human Factors 2003, 45, 601–614. [Google Scholar] [CrossRef]
  20. Fan, J.; Smith, A.P. The Impact of Workload and Fatigue on Performance. In International Symposium on Human Mental Workload: Models and Applications; Longo, L., Leva, M., Eds.; Springer: Cham, Switzerland, 2017. [Google Scholar] [CrossRef]
  21. Koglbauer, I.; Holzinger, J.; Eichberger, A.; Lex, C. Autonomous emergency braking systems adapted to snowy road conditions improve drivers’ perceived safety and trust. Traffic Inj. Prev. 2018, 19, 332–337. [Google Scholar] [CrossRef] [Green Version]
  22. Koglbauer, I.; Holzinger, J.; Eichberger, A.; Lex, C. Drivers’ Interaction with Adaptive Cruise Control on Dry and Snowy Roads with Various Tire-Road Grip Potentials. J. Adv. Transp. 2017, 2017, 5496837. [Google Scholar] [CrossRef] [Green Version]
  23. Muslim, H.; Itoh, M.; Liang, C.K.; Antona-Makoshi, J.; Uchida, N. Effects of gender, age, experience, and practice on driver reaction and acceptance of traffic jam chauffeur systems. Sci. Rep. 2021, 11, 17874. [Google Scholar] [CrossRef] [PubMed]
  24. Koglbauer, I.V.; Eichberger, A.; Lex, C.; Holzinger, J.; Schinko, C.; Ullrich, T. A model for subjective evaluation of automated vehicle control. In Proceedings of the 18th International Symposium on Aviation Psychology (ISAP 2015), Dayton, OH, USA, 4–7 May 2015. Wright State University Conferences and Events. [Google Scholar]
  25. Koglbauer, I.; Eichberger, A.; Lex, C.; Bliem, N.; Sternat, A.; Holzinger, J.; Schinko, C.; Battel, M. Bewertung von Fahrerassistenzsystemen von Nicht Professionellen Fahrerinnen und Fahrern im Realversuch [Assessment of Driver Assistance Systems with Amateur Drivers in Real Driving Tests]. In Humanwissenschaftliche Beiträge zur Sicherheit und Ökologie des Verkehrs. [Human Scientific Contributions to the Safety and Ecology of Transportation]; Chaloupka-Risser, C., Ed.; INFAR: Salzburg, Austria, 2015; pp. 59–70. [Google Scholar]
  26. Obst, P.; Armstrong, K.; Smith, S.; Banks, T. Age and gender comparisons of driving while sleepy: Behaviours and risk perceptions. Transp. Res. Part F Traffic Psychol. Behav. 2011, 14, 539–542. [Google Scholar] [CrossRef] [Green Version]
  27. Akerstedt, T.; Kecklund, G. Age, gender and early morning highway accidents. J. Sleep Res. 2001, 10, 105–110. [Google Scholar] [CrossRef]
  28. Soares, S.; Monteiro, T.; Lobo, A.; Couto, A.; Cunha, L.; Ferreira, S. Analyzing Driver Drowsiness: From Causes to Effects. Sustainability 2020, 12, 1971. [Google Scholar] [CrossRef] [Green Version]
  29. Lowden, A.; ANUND, A.; Kecklund, G.; Peters, B.; Akerstedt, T. Wakefulness in young and elderly subjects driving at night in a car simulator. Accid. Anal. Prev. 2009, 41, 1001–1007. [Google Scholar] [CrossRef]
  30. Akerstedt, T.; Peters, B.; ANUND, A.; Kecklund, G. Impaired alertness and performance driving home from the night shift: A driving simulator study. J. Sleep Res. 2005, 14, 17–20. [Google Scholar] [CrossRef]
  31. Hargutt, V.; Hoffmann, S.; Vollrath, M.; Krüger, H. Compensation for drowsiness & fatigue-a driving simulation study. In Proceedings of the International Conference on Traffic and Transport Psychology ICTTP, Bern, Switzerland, 4–7 September 2000. [Google Scholar]
  32. Ahlström, C.; Zemblys, R.; Jansson, H.; Forsberg, C.; Karlsson, J.; ANUND, A. Effects of partially automated driving on the development of driver sleepiness. Accid. Anal. Prev. 2021, 153, 106058. [Google Scholar] [CrossRef]
  33. Schömig, N.; Hargutt, V.; Neukum, A.; Petermann-Stock, I.; Othersen, I. The Interaction Between Highly Automated Driving and the Development of Drowsiness. Procedia Manuf. 2015, 3, 6652–6659. [Google Scholar] [CrossRef] [Green Version]
  34. Scarpelli, S.; Alfonsi, V.; Gorgoni, M.; Camaioni, M.; Giannini, A.M.; de Gennaro, L. Age-Related Effect of Sleepiness on Driving Performance: A Systematic-Review. Brain Sci. 2021, 11, 1090. [Google Scholar] [CrossRef]
  35. Campagne, A.; Pebayle, T.; Muzet, A. Correlation between driving errors and vigilance level: Influence of the driver’s age. Physiol. Behav. 2004, 80, 515–524. [Google Scholar] [CrossRef] [PubMed]
  36. Schinko, C.; Peer, M.; Hammer, D.; Pirstinger, M.; Lex, C.; Koglbauer, I.; Eichberger, A.; Holzinger, J.; Eggeling, E.; Fellner, D.W.; et al. Building a Driving Simulator with Parallax Barrier Displays. In Proceedings of the 11th Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications. International Conference on Computer Graphics Theory and Applications, Rome, Italy, 27–29 February 2016; pp. 281–289, ISBN 978-989-758-175-5. [Google Scholar]
  37. Lex, C.; Hammer, D.; Pirstinger, M.; Peer, M.; Samiee, S.; Schinko, C.; Ullrich, T.; Battel, M.; Holzinger, J.; Koglbauer, I.; et al. Multidisciplinary Development of a Driving Simulator with Autostereoscopic Visualization for the Integrated Development of Driver Assistance Systems. In Proceedings of the Conference for Combustion, Emissions, Particulates and Testing, Dresden, Germany, 26–27 November 2015. [Google Scholar]
  38. Arefnezhad, S.; Eichberger, A.; Frühwirth, M.; Kaufmann, C.; Moser, M.; Koglbauer, I.V. Driver Monitoring of Automated Vehicles by Classification of Driver Drowsiness Using a Deep Convolutional Neural Network Trained by Scalograms of ECG Signals. Energies 2022, 15, 480. [Google Scholar] [CrossRef]
  39. Hart, S.G.; Staveland, L.E. Development of NASA-TLX (Task Load Index): Results of Empirical and Theoretical Research. In Human Mental Workload; Hancock, P.A., Meshkati, N., Eds.; North-Holland: Amsterdam, The Netherlands, 1988; pp. 139–183. ISBN 9780444703880. [Google Scholar]
  40. Akerstedt, T.; Gillberg, M. Subjective and objective sleepiness in the active individual. Int. J. Neurosci. 1990, 52, 29–37. [Google Scholar] [CrossRef] [PubMed]
  41. Dinges, D.; Grace, R. PERCLOS: A Valid Psychophysiological Measure of Alertness as Assessed by Psychomotor Vigilance; Federal Highway Administration, Office of Motor Carriers: Washington, DC, USA, 1998; Report No. FHWA-MCRT-98-006. [Google Scholar] [CrossRef]
  42. Sommer, D.; Golz, M. Evaluation of PERCLOS based current fatigue monitoring technologies. In Proceedings of the 2010 Annual International Conference of the IEEE Engineering in Medicine and Biology, Buenos Aire, Argentina, 31 August–4 September 2010; pp. 4456–4459. [Google Scholar] [CrossRef]
  43. Junaedi, S.; Akbar, H. Driver Drowsiness Detection Based on Face Feature and PERCLOS. J. Phys. Conf. Ser. 2018, 1090, 12037. [Google Scholar] [CrossRef]
  44. Chaput, J.P.; Dutil, C.; Sampasa-Kanyinga, H. Sleeping hours: What is the ideal number and how does age impact this? Nat. Sci. Sleep 2018, 10, 421–430. [Google Scholar] [CrossRef] [Green Version]
  45. Schiebinger, L. Gendered innovations: Harnessing the creative power of sex and gender analysis to discover new ideas and develop new technologies. Triple Helix 2014, 1, 1–17. [Google Scholar] [CrossRef] [Green Version]
  46. Ohayon, M.M.; Carskadon, M.A.; Guilleminault, C.; Vitiello, M.V. Meta analysis of quantitative sleep parameters from childhood to old age in healthy individuals: Developing normative sleep values across the human lifespan. Sleep 2004, 27, 1255–1273. [Google Scholar] [CrossRef]
Safety 08 00030 g001 550
Figure 1. Automated Driving Simulator of Graz (ADSG). During the WACHSens project and to cancel the disturbing noise of the surrounding area and adjust the indoor temperature, ADSG is placed inside an insulating wooden cube. This figure is reprinted from our previous open-access work [38].
Figure 1. Automated Driving Simulator of Graz (ADSG). During the WACHSens project and to cancel the disturbing noise of the surrounding area and adjust the indoor temperature, ADSG is placed inside an insulating wooden cube. This figure is reprinted from our previous open-access work [38].
Safety 08 00030 g001
Safety 08 00030 g002 550
Figure 2. Effects of automation on drivers’ workload in rested and fatigued conditions. Error bars indicate SD.
Figure 2. Effects of automation on drivers’ workload in rested and fatigued conditions. Error bars indicate SD.
Safety 08 00030 g002
Safety 08 00030 g003 550
Figure 3. Effects of automation on drivers’ PERCLOS in rested and fatigued conditions. Error bars indicate SD.
Figure 3. Effects of automation on drivers’ PERCLOS in rested and fatigued conditions. Error bars indicate SD.
Safety 08 00030 g003
Safety 08 00030 g004 550
Figure 4. Subjective fatigue scores in different conditions and driving modes. Error bars indicate SD.
Figure 4. Subjective fatigue scores in different conditions and driving modes. Error bars indicate SD.
Safety 08 00030 g004
Safety 08 00030 g005 550
Figure 5. Driving automation had different effects in various age groups. Error bars indicate SD.
Figure 5. Driving automation had different effects in various age groups. Error bars indicate SD.
Safety 08 00030 g005
Table
Table 1. Age and driving activity of the participants.
Table 1. Age and driving activity of the participants.
GroupGroup SizeAge
(Years)		Driving Activity in the Past 12 Months
(Thousand Kilometers)
DataNMeanSDMeanSD
Women 20–49 years2331.2610.4614.04313.645
Men 20–49 years1829.449.9415.24414.843
Both genders 20–49 years4130.4610.1514.57114.015
Women 50–85 years2160.297.1313.01917.449
Men 50–85 years2762.529.0117.81411.021
Both genders 50–85 years4861.548.2315.71714.232
Women total4449.2918.8316.78712.591
Men total4545.1117.1713.55415.400
Table
Table 2. Workload, PERCLOS, and reaction time (RT) in automated and manual modes.
Table 2. Workload, PERCLOS, and reaction time (RT) in automated and manual modes.
Driving ModeAutomated DrivingManual Driving
Dependent MeasureMeanSDMeanSD
Mental Demand0.9500.0621.3730.060
Physical Demand0.6700.0601.1300.055
Temporal Demand0.7090.0581.0380.060
Performance1.3430.0761.7470.046
Effort1.0470.0711.4940.058
Frustration1.2210.0751.2200.068
Minimal PERCLOS0.0080.0020.0050.001
Maximal PERCLOS0.1980.0140.1330.010
Median PERCLOS0.0500.0070.0250.004
Mean PERCLOS0.0620.0070.0340.004
Variance PERCLOS0.0050.0010.0030.001
Minimal RT0.9400.0430.8140.038
Maximal RT1.3220.0841.1010.058
Median RT1.0830.0550.9320.040
Mean RT1.1540.0660.9700.046
Variance RT0.6260.1670.3520.105
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Share and Cite

MDPI and ACS Style

Arefnezhad, S.; Eichberger, A.; Koglbauer, I.V. Effects of Automation and Fatigue on Drivers from Various Age Groups. Safety 2022, 8, 30. https://doi.org/10.3390/safety8020030

AMA Style

Arefnezhad S, Eichberger A, Koglbauer IV. Effects of Automation and Fatigue on Drivers from Various Age Groups. Safety. 2022; 8(2):30. https://doi.org/10.3390/safety8020030

Chicago/Turabian Style

Arefnezhad, Sadegh, Arno Eichberger, and Ioana Victoria Koglbauer. 2022. "Effects of Automation and Fatigue on Drivers from Various Age Groups" Safety 8, no. 2: 30. https://doi.org/10.3390/safety8020030

APA Style

Arefnezhad, S., Eichberger, A., & Koglbauer, I. V. (2022). Effects of Automation and Fatigue on Drivers from Various Age Groups. Safety, 8(2), 30. https://doi.org/10.3390/safety8020030

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop