Driving is a complex cognitive task that is prone to distractions, especially visual ones. Eficient visual selective attention is critical for safety; therefore, it is necessary to study the efects of distraction on driving. Conducting experiments on real-life driving behaviour raises serious practical and ethical concerns; therefore, we investigate the use of virtual reality in this area. In particular, we are exploring the use of low-cost virtual reality simulators using commonly available hardware and simple AI techniques to ensure easy reproducibility of our experiments.
CEUR ceur-ws.org
Driving is a complex cognitive task influenced by various factors, including distractions. Distractions can stem from both visual and acoustic sources, with visual distractions generally having a more significant impact due to the visual nature of driving. The allocation of attention resources to multiple driving tasks is a delicate balance, which can be disrupted by distractors, such as dashboard lights, road signs or advertisements.
To enhance driving safety and prevent distraction-related accidents, eficient visual selective
attention is essential [
© 2021 Copyright for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
divided attention and sustained attention, play crucial roles in safe driving [
Driving simulators designed for virtual reality encompass a broad spectrum of configurations, ranging from simple setups using desktop computers equipped with a steering wheel, to highly sophisticated systems replicating car interiors with multi-axial vibration and shaking mechanisms. Additionally, the adoption of VR headsets further enhance the immersive potential of these simulators.
While there may be a natural inclination to consistently opt for the most advanced and immersive equipment, this choice carries several potential drawbacks. Foremost among these is the prohibitive cost associated with top-tier equipment, which can significantly impede widespread adoption. Reproducibility stands as an essential principle in scientific investigation, particularly within clinical research contexts. Consequently, the adoption of excessively expensive equipment should be judiciously considered, as it has the potential to compromise the reproducibility of experiments by other researchers.
The hardware used is not the only factor that influences the immersion and realism of a simulation. The software components also play a decisive role. In the context of driving simulators, besides the graphics, the behaviour of other artificial agents can also strongly influence the experience. For example, road users or pedestrians with realistic behaviour, that react realistically to the user’s actions, can greatly increase the realism of a simulation. Hence, artificial intelligence (AI) techniques of various kinds, from “traditional” techniques to recent deep learning-based techniques, play a crucial role as they are used both to interpret the user’s behaviour and to select the best actions for the autonomous agents.
Considering the importance of reducing the efects of distractors, we designed a study to
evaluate the efect of brain stimulation techniques on driving activity. Functional imaging
studies have highlighted the role of the dorsal frontoparietal attention network in managing
distractions while driving [
Following each session of the experimental procedure, participants were asked to complete questionnaires regarding their subjective experiences of immersion within the simulation and any sensations of sickness they may have encountered.
With this work we want to clarify whether a simplified simulation, akin to the one used in our study, still provides a sense of immersion and whether it efectively reduces the potential disadvantages of more immersive simulators, such as motion sickness. We believe that our ifndings can ofer valuable insights for fellow researchers considering the optimal selection of simulation equipment for future investigations within this domain. While the main goal of the experiment was to evaluate the efects of diferent kinds of neuromodulation on attention while driving, in this article we focus only on the efectiveness of our simulation setup. Therefore, the research question we want to answer with this work are: RQ1: Is our simple simulator efective in terms of realism, presence and involvement? RQ2: Does our simple simulator produce motion sickness?
In Section 2 we describe the experimental setup, the simulation we developed, the hardware, and the simulation software we used. Section 3 introduces the questionnaires we used to evaluate presence, realism and involvement of our simulation setup and the motion sickness it produces. Finally, Section 4 discuss the results of the questionnaires.
We recruited two groups of participants of diferent ages. The first one is composed of 27 healthy participants, with an average age of 24.7 years (Standard Deviation, SD = 2.6), spanning an age range from 21 to 30. This group consisted of 25 right-handed individuals and 2 left-handed individuals. Recruitment occurred through the Sona System of University of Milano - Bicocca. The inclusion criteria encompassed individuals aged between 20 and 35 years, possessing a valid driving license for a minimum of two years, normal or corrected-to-normal vision, and normal hearing, as confirmed by self-reporting. Exclusion criteria entailed a history of neurological or psychiatric disorders, epileptic seizures, intracranial metallic implants, cardiac diseases, or substance abuse or dependence. These criteria align with established safety guidelines for noninvasive brain imaging techniques [14, 15, 16]
The second group is composed of 22 older individuals with a mean age of 68.7 years (Standard Deviation, SD = 2.5), spanning an age from 64 to 73 years old, with no left-handed participant. Inclusion criteria are the same of the previous group, with the exception of age, which had to be above 64 years old. Table 1 summarizes the characteristics of the two groups of participants.We wanted to make sure that the only significant diference between the two groups was age. Otherwise, other characteristics could influence the results. Therefore, we used a t-test to evaluate any statistically significant diference between the two groups. The p-value of such a test is given in Table 1. The only significant diferences concern age and number of years with a driving licence. These diferences are to be expected, of course, since the two groups difer in terms of age. Apart from age, we can confirm that the two groups are similar in terms of other factors that might afect attention while driving, such as sleeping time and the number of hours or kilometres driven per week.
We secured written informed consent from all participants, adhering to the principles outlined in the Declaration of Helsinki. Ethical approval was granted by the University of Milano-Bicocca Ethical Committee (605/2021; 27/04/2021).
Participants had to perform a driving simulation task, which employed an adapted version of paradigms previously utilized by Karthaus et al. [17, 18].
The simulation environment hardware comprises a set of components, including a computer, three screens, a chair, driving pedals, and a steering wheel. The arrangement of these components was thoughtfully designed to emulate the field of view experienced during real-world driving, with a visual representation depicted in Figure 1. It has to be noted that our setup also captures the lateral field of view, which we consider crucial for instilling a sense of realism into the simulation. Although the main part of the simulation typically occupy the central screen area, we believe that this panoramic representation contributes significantly to the overall immersive experience.
To replicate a car’s control interface, we employed standard gaming peripherals for the pedals, steering wheel, and gearshift. In keeping with simplicity and accessibility, we utilized a standard chair for seating. To enhance participants’ immersion within the environment, a curtain panel was used to separate them from the supervising researcher, and ambient lighting was dimmed during the driving sessions.
Our selection of this setup followed a meticulous evaluation of various alternatives. While we acknowledge that our configuration may not match the high degree of realism ofered by the most recent alternatives, it boasts crucial advantages: cost-efectiveness and ease of replication. We firmly believe that these attributes are indispensable for ensuring the protocol’s reproducibility and its applicability across diverse contexts.
In terms of realism, the least realistic component within our experimental setup is the seating, consisting of a standard fixed chair. Although alternative solutions exist that mimic authentic car seats (often resembling sports car seats), our choice remains significantly more cost-efective and widely available.
Furthermore, we deliberately avoided the use of head-mounted displays, such as virtual reality visors, as an alternative to conventional screens. While these devices provide an exceptionally immersive experience, they do come with potential issues, notably motion sickness, which can aflict individuals unfamiliar with such technology. Considering that our target demographic encompasses elderly individuals, many of whom may not have prior experience with such equipment, we opted for the use of standard monitors arranged in a semi-circular configuration.
For the simulation software, we selected the well-established CarnetSoft driving simulator, a platform which have already been used in similar experiments [17, 18].
Since the final goal of our experiment was to compare the efects of the diferent types of tDCS, each participant took part in three driving sessions. Before each session, the participant underwent a neuromodulation session of one of the following types: conventional tDCS, Focal High Definition tDCS, or a placebo tDCS (sham treatment). In this paper, we focus on the eficacy of our simulation setup, rather than on the neuromodulatory efects. Therefore, we report results of the sham treatment only.
The simulation replicates a highway in the countryside. Within this context, users navigate a predetermined route that traverses the highway, takes an exit, and subsequently re-enters the highway. This path forms a continuous loop, repeated multiple times within a single session. Consequently, it features a diverse range of elements, including wide-radius turns, sharp bends, and nearly straight stretches. While the path is never perfectly linear, the curvature is imperceptible for extended sections.
Furthermore, the simulation incorporates vehicular trafic, consisting of cars, lorries, and motorbikes. Given the highway setting, the are no pedestrians, bicycles, or intersecting roads. The composition and behavior of the trafic, including actions such as following or overtaking the user, are randomized. This inclusion of trafic, altough with simple dynamics, and the use of a non-straight path sets our work apart from prior experiments [17, 18], which used simpler kinds of environments.
The participant’s control over the vehicle is constrained, intentionally limiting simulation variability to ensure reproducibility. Specifically, the participant’s car adheres to a predetermined path, automatically maintaining a constant distance from a leading vehicle. Participants controlled the steering wheel to keep the car within its lane, while also responding to specific stimuli and suppressing others. The stimuli we used are: Braking: the stopping lights of the leading car turn on. The participant has to respont to this stimulus by pressing the braking pedal; Sign: a sign, reproducing those found in highways, appear at the top center of the screen. The sign may represent either a city or a country. A participant must respond only to one variant of this stimulus (which one is chosen at random), by operating a lever on the steering wheel.
A single simulation session lasts approximately 25 minutes, plus 5 minutes for practicing before the first session. During the simulation, the participant is presented with: • 72 braking stimuli; • 72 sign stimuli, of which 50% are “go” stimuli (i.e., the participant must respond to them), and 50% are “no-go”; • 72 combined stimuli consisting of a braking and a sign; the signs are divided into 50% “go”and 50% “no-go”.
An example of combined stimulus is shown in Figure 2. The time interval between two stimuli is drawn from a random uniform distribution between 6 and 8 seconds.
After each session, participants completed two questionnaires. The first consists of a selection of questions from the IGroup Presence Questionnaire (IPQ)”, which aims to assess the participants’ perceived characteristics of the simulated environments [19]. The IPQ consists of a series of questions designed to measure diferent aspects of a simulation: general presence, spatial presence, involvement and experienced realism. The questions we selected are listed in Table 2 along with statistics on the responses of the two groups of participants.
The second questionnaire is the ”Simulator Sickness Questionnaire”, which measured participants’ sickness experienced during the simulated driving activity [20]. It consists of 16 items that assess diferent aspects of sickness on a scale from “None” (equivalent to a score of 0) to “Severe” (equivalent to a score of 3) and is divided into three non-mutually exclusive categories: Nausea, Oculomotor and Disorientation. The score for each category is the sum of the scores of the items belonging to that category. These are then summarised by the “Total” score, which is
while navigating in the virtual world? (i.e. sounds, room temperature, other people, etc.)
seem consistent with your real world experience ?
-0.1 0.7 1.0 0.3 0.3 -0.4 0.4 n.s. n.s. <0.05 n.s. a weighted sum of the components. The scores of the SSQ questionnaire are summarised in Table 3, while the symptoms that the questionnaire assesses are listed in Table 4 along with the categories to which they belong.
The hardware used for our simulation is remarkably simple, widely available, and inexpensive, especially when compared with more complex alternatives. Furthermore, from a software perspective, we have not implemented any complex behaviours for the trafic agents. Since our simulation takes place on a highway, there are of course no pedestrians. The actors in the scenario are cars, trucks, and motorbikes, each of which is assigned a randomly generated speed and occasionally overtakes the participant’s vehicle. Consequently, the AI responsible for trafic management is similarly straightforward.
We believe that simplicity has distinct advantages. Nevertheless, it is essential to evaluate whether this simplicity has negative efects on the experimental activity, and, hence to answer to RQ1. For this reason, we statistically analysed the responses to the IPQ, to evaluate the simulation in terms of realism, presence and involvement, and to identify any diference between the two age groups in their perceptions of the simulation. The column p-value” in Table 2 shows the p-value of a Wilcoxon test used to assess a statistically relevant diference between the responses of the two groups. We could only detect a diference in the third question, which assesses the realism of the simulation: older participants perceived the simulation as less consistent with the real experience than younger participants. On the other hand, we could not find any diference between the answers to the second question, which assesses the same characteristic, that is, realism. Therefore, we believe that this point needs further investigation in future studies. Considering that the possible responses to the IPQ range from −3 to 3, we can say that the simulation is not considered particularly immersive, but it is not considered to be unengaging either. Further studies should move towards a comparison with more complex simulations, to assess whether the added complexity is a benefit to the perceived experience.
As for RQ2, symptoms in the categories “nausea” and “disorientation” of the SSQ questionnaire are generally minor and do not difer significantly between the two groups (Table 3). However, it should be noted the high standard deviation in the category “disorientation” for the young group, which corresponds to a high variability of symptoms. More research is needed to understand the reason why some participants experience a moderate level of disorientation, especially considering that our simulation is not very diferent from a normal video game that young users are likely to have experience with.
Although oculomotor disorders are not severe, they are less negligible in the young group, with a significant diference from the elderly group. This may be surprising, as we assume that younger people are much more familiar with virtual reality and video games. Therefore, one would expect fewer symptoms in this group. However, this is not the first experiment to find that older people sufer less from motion sickness than younger people [ 21]. Similar to the perceived realism in the IPQ, this point also deserves further investigation.
We designed an experiment, based on a driving simulator, to assess driving attention in two groups of subjects of diferent ages and to evaluate whether two diferent types of direct current brain stimulation can be used to increase attention. Our simulation, which is characterised by its simplicity in terms of hardware and software, provides a low-cost and accessible platform for studying human responses in a controlled driving environment. While we believe that simplicity ofers distinct advantages in terms of accessibility and cost-efectiveness, it is critical to further evaluate whether this simplicity has any adverse efects on experimental outcomes. We evaluated participants’ experiences using two well-known questionnaires to assess the immersiveness of the simulation and motion sickness. Our statistical analysis of participant responses measured by the IPQ questionnaire revealed that older participants perceived the simulation to be less consistent with real-world experiences compared to their younger counterparts. This finding highlights the need for further investigation into the realism of our simulation and suggests that future studies should explore how additional complexity might enhance the perceived experience.
In relation to the SSQ questionnaire, we found that symptoms related to “nausea” and “disorientation” were generally very low and did not difer significantly between age groups. Further research is needed to understand why some young participants experienced moderate disorientation, given their likely familiarity with video games and virtual reality.
Interestingly, our study found that oculomotor disorders were more pronounced in the young group, despite the assumption that they were more familiar with virtual environments. This observation is consistent with previous research suggesting that younger people are more prone to motion sickness than older individuals. This phenomenon should be investigated further to uncover underlying factors that contribute to these findings. [13] J. D. Cosman, P. V. Atreya, G. F. Woodman, Transient reduction of visual distraction following electrical stimulation of the prefrontal cortex, Cognition 145 (2015) 73–76. [14] M. A. Nitsche, D. Liebetanz, A. Antal, N. Lang, F. Tergau, W. Paulus, Modulation of cortical excitability by weak direct current stimulation–technical, safety and functional aspects, Supplements to Clinical neurophysiology 56 (2003) 255–276. [15] S. Rossi, M. Hallett, P. M. Rossini, A. Pascual-Leone, S. of TMS Consensus Group, et al., Safety, ethical considerations, and application guidelines for the use of transcranial magnetic stimulation in clinical practice and research, Clinical neurophysiology 120 (2009) 2008–2039. [16] P. M. Rossini, D. Burke, R. Chen, L. G. Cohen, Z. Daskalakis, R. Di Iorio, V. Di Lazzaro, F. Ferreri, P. Fitzgerald, M. S. George, et al., Non-invasive electrical and magnetic stimulation of the brain, spinal cord, roots and peripheral nerves: Basic principles and procedures for routine clinical and research application. an updated report from an ifcn committee, Clinical neurophysiology 126 (2015) 1071–1107. [17] M. Karthaus, E. Wascher, S. Getzmann, Efects of visual and acoustic distraction on driving behavior and eeg in young and older car drivers: a driving simulation study, Frontiers in aging neuroscience 10 (2018) 420. [18] M. Karthaus, E. Wascher, M. Falkenstein, S. Getzmann, The ability of young, middle-aged and older drivers to inhibit visual and auditory distraction in a driving simulator task, Transportation research part F: trafic psychology and behaviour 68 (2020) 272–284. [19] T. Schubert, F. Friedmann, H. Regenbrecht, The experience of presence: Factor analytic insights, Presence: Teleoperators & Virtual Environments 10 (2001) 266–281. [20] R. S. Kennedy, N. E. Lane, K. S. Berbaum, M. G. Lilienthal, Simulator sickness questionnaire: An enhanced method for quantifying simulator sickness, The international journal of aviation psychology 3 (1993) 203–220. [21] A. T. Dilanchian, R. Andringa, W. R. Boot, A pilot study exploring age diferences in presence, workload, and cybersickness in the experience of immersive virtual reality environments, Frontiers in Virtual Reality 2 (2021) 736793.