Sara Ferreira

Driving simulators are widely used in driver sleepiness research. However, the novelty of first exposure to the simulator and broader experimental setting may alter the driver’s underlying sleepiness state, introducing systematic differences between sleepiness-related measurements obtained during the first and subsequent visits in repeated-visit designs. These exposure-related differences are referred to here as the first-visit effect. This study assessed first-visit effects, stabilization of sleepiness indicators across repeated exposures, and differences in patterns of change across repeated daytime and night-time visits. Twenty-six young male drivers completed six visits in a high-fidelity moving-base simulator: three daytime/full-sleep and three night-time/sleep-deprived visits, with three 30-min driving sessions per visit. All nine sleepiness indicators, spanning subjective, electroencephalographic, ocular, cardiac autonomic, vigilant-attention, and driving-performance domains, had lower estimated values at Visit 1 than at later visits. Statistically significant first-visit effects were observed across ocular, cardiac autonomic, vigilant-attention, and driving-performance domains. Seven indicators met the predefined stabilization criteria between Visits 1 and 5; the electroencephalographic indicators did not meet the criteria. The Holm-adjusted interaction tests provided no statistically significant evidence that patterns of change differed between daytime and night-time conditions. These findings indicate that Visit 1 and subsequent pre-stabilization visits should not automatically be treated as comparable with later visits. Experimental exposure history should therefore be considered when designing, analyzing, and interpreting driver-sleepiness studies. More broadly, the results suggest that findings obtained during a single experimental visit may not necessarily generalize to subsequent exposures, a concern because many driver sleepiness studies rely on a single laboratory visit.

Human factors substantially contribute to road crashes. However, assessing their effects is complex due to the influence of individual characteristics such as personality. Therefore, studies examining the relationship between personality traits and driving behaviour are essential. This study aimed to construct driver behaviour profiles based on these relationships within a large sample of Portuguese drivers. A community sample of 747 licensed drivers, aged under 75 and with at least three years of driving experience, completed an online survey. Instruments included the NEO-Five Factor Inventory-20, the Impulsivity and Sensation Seeking Scale, and the 24-item Driver Behaviour Questionnaire (DBQ). Firstly, multiple linear regressions were conducted considering the three driving behaviour dimensions of the DBQ to support the construction of the profiles. Results indicated that neuroticism, agreeableness, extraversion, impulsivity, and sensation seeking predicted infractions and aggressive driving. Neuroticism, conscientiousness, and impulsivity predicted non-intentional errors, while neuroticism, openness, conscientiousness, and impulsivity were associated with lapses. Even after controlling for age and gender, personality traits remained significant predictors. Secondly, four driver behaviour profiles were constructed using two alternative methodologies: an empirical approach and cluster analysis with k-means. Profiles built using the empirical approach resulted in four groups of drivers characterised by more easily identifiable driving behaviours: prudent, regular, distracted/forgetful, and aggressive drivers. The distracted/forgetful group showed a positive relationship to crash involvement. Overall, the study shows that the complex driver behaviour needs to be carefully grouped.

Driving simulators offer a safe and controlled way to study fatigue in truck drivers, but variation in scenario design and incomplete reporting limit reproducibility and cross-study comparison. This systematic review synthesized scenario parameters used in truck-driving simulators to induce fatigue-related reductions in alertness and identified recurring protocol patterns associated with interpretable fatigue-related change. Following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines and a prospectively registered protocol (PROSPERO CRD420261302272), systematic searches were conducted in February 2026 in Scopus, Web of Science, IEEE Xplore, PubMed, and ScienceDirect. Peer-reviewed original studies published in English were eligible if they involved truck drivers, used a driving simulator, reported fatigue-relevant scenario parameters, and measured at least one fatigue-related outcome; no restriction was applied to publication year. Twenty-three studies comprising 419 participants met the eligibility criteria and were synthesized narratively. Risk of bias was appraised using an adapted 11-item checklist for driving simulator experiments, developed with the National Heart, Lung, and Blood Institute (NHLBI) quality assessment tools as a reference framework. Across the qualitative evidence base, fatigue-related change was reported more consistently in protocols combining sustained time on task with low-variability driving demands, typically implemented through monotonous road environments and reduced traffic complexity. Effects were more readily interpretable when sessions were scheduled at night or after work shifts and when outcomes were assessed repeatedly during the drive. However, incomplete control or reporting of baseline sleep pressure, stimulant intake, counterbalancing, familiarization, simulator sickness, and outlier handling limited causal interpretation and confidence in cross-study comparison. Overall, the evidence supports recurring design patterns rather than a single optimal protocol and highlights the need for standardized scenario descriptions and minimum reporting requirements.

The continuous development of advanced driver assistance systems is paving the way for the deployment of autonomous vehicles. Until then, manual vehicles and partially automated vehicles (AVs) will co-exist. Past experience has shown that the deployment of new technology must consider users’ acceptability and adoption. This is the case of AVs, that must provide safety and comfortable travel experiences. To evaluate acceptability profiles towards AVs, this study analyses the determinant factors of AVs acceptability to identify different Portuguese population clusters. A questionnaire was developed to explore prospective users’ representations regarding benefits and expectations, risks and concerns, previous experience with automated driving technology, and preferred use cases for AVs. A cluster analysis was performed using the k-means algorithm and, after, chi-square tests characterized cluster membership. In the end, acceptability profiles were compared for different use cases using ANOVA post-hoc tests. Three clusters of prospective users were identified: objectors, ambivalents, and enthusiasts. Driving pleasure, safety, reliability of the technology, and data privacy are prevailing negative factors while improved road safety, reduced emissions, and non-driving tasks possibilities favor acceptability. Sociodemographic characteristics, like income, education, place of residence, and self-perception about the adoption of new technologies reflected the main differences between clusters.

Truck drivers play a pivotal role in global freight transport systems, yet their occupational and behavioral risk exposures make them a priority population in road safety research. This systematic review examines the factors influencing risky driving behaviors among truck drivers and their impacts on road safety outcomes. Following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines, the review aimed to identify hazardous driving behaviors, the internal and external factors contributing to these behaviors, and their consequences for traffic safety. Inclusion criteria targeted original research published in English between 2009 and 2024 specifically focused on truck driver behavior and road safety outcomes. Systematic searches across PubMed, Scopus, Web of Science, and IEEE Xplore yielded 104 studies meeting these criteria. The synthesis revealed prevalent risky behaviors—such as speeding, fatigue-related impairments, distracted driving, and substance use—driven by internal factors (e.g., health conditions, psychological stress) and external pressures (e.g., occupational demands, regulatory constraints). These behaviors were consistently associated with increased crash risk. Nonetheless, limitations including the exclusion of non-English studies, reliance on self-reported data, and lack of standardized metrics constrained cross-study comparability and generalizability. Effective interventions identified include fatigue management programs, driver monitoring technologies, and positive safety climates. Findings underscore the urgent need for evidence-based, multifaceted strategies to enhance truck driver safety and inform policy, industry practices, and future research.

Deep learning (DL) models show considerable promise in detecting driver drowsiness, a major contributor to road traffic crashes. This systematic review evaluates the performance, contexts of application, and implementation challenges of DL-based drowsiness detection systems. Conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines, the review includes peer-reviewed empirical studies published between 2015 and 2025 that develop and validate DL models using data collected in real or simulated driving environments. Studies were identified through systematic searches in PubMed, Scopus, Web of Science, ScienceDirect, and IEEE Xplore, last updated in March 2025. Due to methodological heterogeneity, findings are synthesized narratively. Eighty-one studies meet the inclusion criteria. Most employ Convolutional Neural Networks, Recurrent Neural Networks, or hybrid architectures and use behavioral, physiological, or multimodal inputs. Reported median values for accuracy and F1-score exceed 0.95 under both simulated and real-world conditions. However, studies frequently lack demographic diversity, standardized performance reporting, and robust validation protocols. Key limitations include limited dataset transparency, inconsistent evaluation metrics, and insufficient attention to ethical and privacy considerations. While DL models exhibit strong predictive performance, their real-world deployment remains limited by practical and methodological constraints. Future research should place emphasis on the development of inclusive datasets, the conduct of multi-context evaluations, the advancement of real-world deployment strategies, and the rigorous adherence to ethical standards.