Attention is a cognitive process essential for safe driving that requires the processing of relevant stimuli while simultaneously suppressing a large number of distractors. Visual distractions, in particular, pose a high risk due to the inherently visual nature of driving and the similarity to relevant stimuli. In this paper, we propose a protocol to evaluate the efects of diferent visual distractors on driving attention using a virtual reality (VR) driving simulator. The protocol is composed of multiple driving tasks in a controlled VR environment that presents relevant stimuli that resemble stop signs and irrelevant distractors of varying similarity. By measuring drivers' reaction times and lane keeping under diferent experimental conditions, we aim to evaluate how the presence and similarity of distractions afect attention. The use of a virtual reality simulator allows safe and reproducible testing, overcoming the ethical and practical challenges of driving studies in real-world.
CEUR ceur-ws.org
Driving is a complex task susceptible to various influencing factors, with distractions emerging as a significant safety concern. These distractions, which can be caused from both visual and acoustic stimuli, exert a more relevant impact when visual, due to the inherent visual demands Italy ∗Corresponding author. distractors into real-world driving scenarios, or practical ones associated with accurately measuring driving performances. For these reasons, recent years have seen a surge of interest in using virtual reality (VR) environments to overcome these challenges, ofering immersive experiences while avoiding the ethical pitfalls of real-world evaluations.
Virtual reality driving simulators encompass a spectrum of configurations, ranging from basic
setups with desktop computers and steering wheels [
With this work, we propose a protocol to evaluate attention and the efect of diferent kinds of distractors and stimuli while driving in a simulated environment.
The evaluation of driving attention has been a critical focus within transportation research, with investigations spanning both real-life driving scenarios and simulator-based studies.
Real-life driving studies ofer a direct and ecologically valid assessment of driving attention. Researchers can use various data collection methods, including eye-tracking devices, physiological measurements, and in-vehicle cameras, to estimate the driver’s attention.
Eye-tracking glasses have been long used to measure the movements of a driver’s gaze
and to examine the impact of distractions on drivers’ visual attention in real-world trafic
conditions [
However, conducting real-life driving studies presents challenges such as ethical concerns, safety issues, and the dificulty of controlling experimental variables.
Simulator-based studies provide a controlled and reproducible environment for the systematic
manipulation of driving conditions while monitoring attention. In addition, virtual reality
driving simulators provide a platform to study attentional processes without exposing participants
to the risks associated with real-world driving. Attention has been studied with diferent types
of driving simulators. The complexity can range from very simple simulators consisting of
normal gaming devices, such as steering wheels and pedal boards [
Despite these considerations, driving simulators have been used extensively in transportation
research. For example, they have been used to evaluate drivers’ attention [
Another example of the possible use of a simulator is the manipulation of the cognitive workload through task complexity and environmental conditions. For example, Son et. al. studied the efects of cognitive workload on the behavior of older drivers during simulated driving [24]. Cantin et. al. also investigated the efects of age and driving complexity on the mental workload in a driving simulation [25].
While driving, we are constantly exposed to various stimuli, such as trafic lights, road signs, other vehicles, and pedestrians. Simultaneously, numerous distractors, such as advertisements or phone notifications, must be ignored to ensure safe driving. However, this task is challenging, as distractors can often resemble important stimuli. Additionally, driving itself is a complex activity that requires sustained attention.
In this work, we propose a protocol to assess the impact of diferent types of distractors on driving attention. Specifically, we investigate distractors that vary in similarity to a crucial stimulus, resembling a stop sign. The evaluation is conducted using a driving simulator, which provides the necessary control over stimuli and distractors, something that would be nearly impossible to achieve in real-world driving conditions. While a real-world setup might ofer higher ecological validity, the use of a simulator allows precise control and analysis of various factors, which is critical for scientific research.
The virtual environment features an extra-urban two-lane road within a rural landscape. The driving simulation uses a first-person perspective and takes place during day-time. Taskirrelevant lateral road signs and other moving vehicles are rendered in the environment. The road circuit forms a loop, predominantly rectilinear with sharper bends at its extremities, allowing drivers to re-enter the route.
Our experimental activity consists of three diferent kinds of blocks, called pure, mixedeasy, and mixed-hard. Each kind of block corresponds to a driving session taking place in the environment described above. During each driving session a participant is subject to diferent (a) (b) kinds of event, also called trials. Diferent blocks are composed of diferent kinds and number of trials.
The pure block is composed of 60 trials. In this case, trials can be of two kinds: target-present and target-absent. During a target-present trial, one of the stimuli shown in Figure 1, designed to resemble a classic stop sign, is presented. Half of the participants is shown the stimulus from Figure 1a, while the other half is shown the stimulus from Figure 1b. This approach ensures that the experiment is not biased toward a specific configuration, such as the red portion appearing in either the lower or upper part of the stimulus.
We ask participants to respond to the stimulus by pressing the brake pedal. When responding correctly, the sign disappears and thus the trial ends; however, this also happens after 2 seconds in absence of any response. During a target-absent trial no stimulus is shown to the participant. This serves to break any expectation of the presentation of a trial. A new trial takes place after a random time interval sampled from an uniform distribution between 4 and 6 seconds. A pure block is composed of 30 target-present and 30 target absent trials.
Mixed blocks work like the pure block, but are composed of 120 trials in total, subdivided in 30 target-present trials, 30 target-absent trials, 30 distractor-present trials, and 30 distractor-present and target-present trials
During distractor-present trials, a distractor is displayed. A mixed block can be classified as either “hard” or “easy”. In the “hard” condition, the distractor is one of the sign from Figure 1 (that not used as stimulus), closely resembling the target. In the “easy” condition, the distractor is less similar to the target, featuring lighter colors and lacking any white sections, as shown in Figure 2.
During distractor-present and target present trials both a distractor and the target are shown.
Regardless of the type of block, the target and the distractors appear in the upper part of the screen, either on the left or on the right.
During each experimental session, participants complete a pure block, a mixed hard block, and a mixed easy block, lasting 7, 21, and 21 minutes, respectively. The order of the blocks is randomized and varies for each participant. The session also includes a 2-minute training phase to allow participants to familiarize themselves with the simulator, along with two short breaks between the blocks.
To measure the performance of the participants, we assess the following metrics: 1. Reaction time – the time it takes for participants to respond to stimuli, that is, the time it takes to press the brake when a stimulus appears. 2. Lane keeping – the ability of participants to maintain their position within the lane, reflecting their overall control and focus during the driving task. Lane keeping is assessed using the “wheel error” metric, which measures the diference between the angle of the front wheels and the angle required to follow the optimal trajectory provided by the simulator. The standard deviation of these measurements, calculated within a rolling window, serves as the indicator of lane-keeping performance. 3. Accuracy of responses – the correctness of the participant’s reactions to stimuli, particularly in distinguishing between relevant stimuli and distractors. 4. Error rates – the frequency of incorrect responses or missed stimuli, which provides insight into how well participants can ignore distractors while focusing on important driving cues.
The simulator setup we propose is rather minimalistic. We use a gaming pc powered by an Intel Core i7-10700K 3.8 GHz, Nvidia Geforce RTX 3080 (10GByte), and 32 GByte of RAM, with three 24 inches widescreen monitors (1920 x 1080), arranged in a semi-circular configuration. The lateral monitors are positioned at 120° relative to the central monitor to realistically simulate both central and peripheral fields of view, allowing for a comprehensive horizontal field of view of 210°. Car controls are simulated using standard racing game equipment, including a Logitech G29 Driving Force steering wheel, pedal board and gear shift lever (which is not used during this experiment). The driver sit in an adjustable ofice chair.
We use the CarnetSoft driving simulation software (version 7.1), which has previously been
employed in research on visual attention and inhibitory mechanisms during driving [
The virtual environment depicts an extra-urban, two-lane road in a daytime rural setting, with drivers experiencing right-hand trafic from a first-person perspective. The road features task-irrelevant lateral road signs and other vehicles traveling in the same direction. Audiovisual stimulation matched the environment to enhance realism and engagement. The entire setup is implemented in a darkened and controlled laboratory, ensuring isolation from external stimuli and consistent experimental environment.
We utilize questionnaires to evaluate various aspects of the simulation and to assess participant characteristics. We evaluate driving habits of the participants using both open and multiplechoice questions. More specifically, participants are asked to indicate how many years it passed since they obtained their driving license, if they are currently driving, and, in case, how many times they drive per week and how many kilometers they travel on average. Finally, we ask about the usage of vision correction methods.
Moreover, sleep quality is evaluated using open questions regarding the average sleeping hours per night and, more specifically, how many hours were slept and – if any – how many awakenings occurred the night before the experiment. Furthermore, we ask participants to indicate how they rate the quality of their sleep that night on a 10-point Likert scale (from 1 = very poor to 10 = very good). Ultimately, we ask them to indicate if they assumed any stimulating substance (e.g., cofee) within the two hours preceding the experiment.
To evaluate previous video-games experience we use a translation in Italian of the questionnaire presented by Stichcombe et al. [26]. This brief questionnaire includes open and multiple-choice questions, as well as Likert-scale questions on video-games expertise - in general - and on racing video-games expertise - in particular.
The experience of presence – here intended as the construction of a spatial-functional mental model of the virtual environment within an embodied cognition framework [27] – is measured by adapting four selected items taken from the Igroup Presence Questionnaire (IPQ). More specifically, the selected items are intended to measure the following constructs: one item for Involvement [28], two items for Experienced Realism [29, 30], and one item for General Presence [28]. All elements are evaluated on a 7-point Likert scale; the anchors of the extremities are defined by labels adapted to the content of the element itself, in a range from -3 (defining low experience of presence) to + 3 (defining high experience of presence).
To assess whether the typical symptoms of simulator sickness influence driving performance, we ask participants to complete an adapted version of the Simulator Sickness Questionnaire (SSQ) [31]. It comprises 16 items - concerning the main experienced oculomotor, disorientation, and nausea-related symptoms when using a virtual simulator - to be evaluated on a 5-point Likert scale.
This paper presents a protocol for assessing the efects of visual distractors on driving attention using a driving simulator. The proposed methodology evaluates how distractors that resemble trafic signs influence attention and performance in a safe, reproducible setting, overcoming the limitations of real-world studies. By incorporating diferent distractor dificulty levels, the protocol reliably tests attention under various cognitive loads. It is cost-efective and accessible, and is adaptable for further investigation with additional sensors like eye-tracking or physiological sensors. This framework supports research into visual distractions’ impact on driving, contributing to road safety advancements. [15] A. Facchin, S. Rocca, V. Strina, L. Vacchi, C. Lega, S. Fontana, et al., Assessing the impact of selective attention in a minimalist virtual reality driving simulator: An analysis of perceived experience and motion sickness, in: CEUR WORKSHOP PROCEEDINGS, volume 3576, CEUR-WS, 2023, pp. 42–52. [16] D. Sportillo, A. Paljic, L. Ojeda, Get ready for automated driving using virtual reality,
Accident Analysis & Prevention 118 (2018) 102–113. [17] C. Merenda, C. Suga, J. Gabbard, T. Misu, Efects of vehicle simulation visual fidelity on assessing driver performance and behavior, in: 2019 IEEE Intelligent Vehicles Symposium (IV), 2019, pp. 1679–1686. doi:10.1109/IVS.2019.8813863. [18] R. A. Wynne, V. Beanland, P. M. Salmon, Systematic review of driving simulator validation studies, Safety science 117 (2019) 138–151. [19] R. Ezzati Amini, C. Al Haddad, D. Batabyal, I. Gkena, B. De Vos, A. Cuenen, T. Brijs, C. Antoniou, Driver distraction and in-vehicle interventions: A driving simulator study on visual attention and driving performance, Accident Analysis & Prevention 191 (2023) 107195. URL: https://www.sciencedirect.com/science/article/pii/S0001457523002427. doi:https://doi.org/10.1016/j.aap.2023.107195. [20] H. C. Lee, A. H. Lee, D. Cameron, Validation of a driving simulator by measuring the visual attention skill of older adult drivers, The American Journal of Occupational Therapy 57 (2003) 324–328. [21] H. Antonson, S. Mårdh, M. Wiklund, G. Blomqvist, Efect of surrounding landscape on driving behaviour: A driving simulator study, Journal of Environmental Psychology 29 (2009) 493–502. [22] M. Cassarino, M. Maisto, Y. Esposito, D. Guerrero, J. S. Chan, A. Setti, Testing attention restoration in a virtual reality driving simulator, Frontiers in psychology 10 (2019) 250. [23] S. Soares, S. Ferreira, A. Couto, Driving simulator experiments to study drowsiness: A systematic review, Trafic injury prevention 21 (2020) 29–37. [24] J. Son, Y. Lee, M.-H. Kim, Impact of trafic environment and cognitive workload on older drivers’ behavior in simulated driving, International Journal of Precision Engineering and Manufacturing 12 (2011) 135–141. [25] V. Cantin, M. Lavallière, M. Simoneau, N. Teasdale, Mental workload when driving in a simulator: Efects of age and driving complexity, Accident Analysis & Prevention 41 (2009) 763–771. [26] A. Stinchcombe, Y. Kadulina, C. Lemieux, R. Aljied, S. Gagnon, Driving is not a game: Video game experience is associated with risk-taking behaviours in the driving simulator, Computers in Human Behavior 69 (2017) 415–420. [27] T. Schubert, F. Friedmann, H. Regenbrecht, The experience of presence: Factor analytic insights, Presence: Teleoperators & Virtual Environments 10 (2001) 266–281. [28] M. Slater, M. Usoh, Representations systems, perceptual position, and presence in immersive virtual environments, Presence: Teleoperators & Virtual Environments 2 (1993) 221–233. [29] C. M. Hendrix, Exploratory studies on the sense of presence in virtual environments as a function of visual and auditory display parameters, Master’s thesis, Citeseer, 1994. [30] B. G. Witmer, M. J. Singer, Measuring presence in virtual environments: A presence questionnaire, Presence 7 (1998) 225–240. [31] 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.