Effect of time pressure on steering control of the drivers in a car-following situation

The current study focused on analyzing steering control of the drivers during a car-following situation under increasing time pressure conditions. A driving simulator experiment was conducted on ninety-two participants to assess steering performance measures. Five different steering control measures: Variability in Steering Angle (VSA), Steering Reversal Rate (SRR), Steering Speed (SS), Stability of Steering Control (SSC), and Maximum Steering Swerve (MSS) were examined under No Time Pressure (NTP), Low Time Pressure (LTP), and High Time Pressure (HTP) driving conditions. Repeated measures ANOVA (for continuous data) and Friedman’s test (count data) with post-hoc analysis and Generalized Estimating Equation (GEE) modeling technique were used to investigate the influence of time pressure and different predictor variables. The statistical analysis showed that time pressure driving conditions significantly affected steering control of the drivers. The pairwise comparison of time pressure conditions revealed that HTP significantly affected most of the steering control measures as compared to LTP. Further, a GEE model also exhibited similar results where steering control measures were substantially influenced by HTP as compared to LTP. Moreover, in addition to time pressure conditions, demographic characteristics showed significant influence on steering control measures. The GEE model results showed that female drivers performed 13% more steering corrections (5° SRR) which led to better SSC by 124.44% than male drivers. Additionally, it was discovered that young-aged and experienced drivers took extra steering efforts to control lateral position of the vehicle by increasing 53.50% SS and 1% SRR compared to middle-aged and inexperienced drivers. The findings from the current study revealed that drivers undergo fast and abrupt steering maneuvers under time pressure conditions. The research approach demonstrated in the current study can be beneficial to discriminate minimum requirement of steering efforts and set-up threshold values for various steering evasion techniques to control and maintain safe lateral position during car-following maneuvers.     

Exploring behavioral validity of driving simulator under time pressure driving conditions of professional drivers

Driving simulators have become an important tool in human factors research, given that they are appropriately validated. Therefore, this study aims to explore the behavioral (absolute and relative) validity of a fixed-base driving simulator by analyzing different driving behavior measures such as speed, longitudinal acceleration, lateral acceleration, and brake pedal force. Thirty professional drivers participated in the experiment and the data was collected in real and simulated worlds under No Time Pressure (NTP) and Time Pressure (TP) driving conditions. Initially, comparative analyses were conducted on different driving behavior measures using Wilcoxon-signed rank test to examine absolute validity of the driving simulator. Finally, Generalized Linear Mixed (GLM) models were developed for computing the effective distance between real and simulated worlds by quantifying the parameters and for establishing relative validity. In general, the continuous profiles of driving behavior measures followed similar trends in real and simulated worlds and comparative analyses indicated relative validity of the driving simulator. The GLM models showed significant interaction effect of driving environments (real-world and simulated world) and driving conditions (NTP and TP) where high driving speed, high brake pedal force, and low lateral acceleration were observed in simulated world under TP driving condition than real-world under TP driving condition. Overall, the statistical analyses showed qualitative correspondence (relative validity) of the driving behavior measures in between real and simulated worlds. The findings from the current study showed expediency of the driving simulator and its effectiveness in conducting research on human factors and driver safety.     

Investigating the influence of time pressure on overtaking maneuvers and crash risk

The research conducted on overtaking maneuver for evaluating drivers’ safety showed adverse effects of urgency on driving performance and decision making. Therefore, a driving simulator study was designed to examine driving performance of the drivers and its implication on overtaking and crash probabilities under increasing time pressure conditions. Eighty-eight participants data were analyzed in the current study. Three different time pressure conditions: No Time Pressure (NTP), Low Time Pressure (LTP), and High Time Pressure (HTP) were considered for analyzing driving performance of the drivers while executing overtaking maneuvers. The driving performance was assessed using minimum time-to-line crossing and coefficient of variation in speed to dissect the safety margin adopted by the drivers while overtaking the lead vehicle. Further, minimum time-to-line crossing and coefficient of variation in speed were considered as explanatory variables to investigate their influence on overtaking and crash probabilities. Parametric survival analysis and Generalized Linear Mixed Models (GLMM) were used to assess the driving performance, overtaking and crash probabilities. The parametric survival analysis showed that minimum time-to-line crossing reduced by 36.7% and 63.8% in LTP and HTP driving conditions, respectively. The GLMM results revealed that coefficient of variation in speed increased by 3.437% in HTP (no significant effect in LTP) as compared to NTP driving conditions. Further, the GLMM results showed that overtaking and crash probabilities decreased with increment in minimum time-to-line crossing and coefficient of variation in speed values. Additionally, it was observed that male drivers took risky decisions than female drivers. Nevertheless, the comparative analysis revealed that male drivers were less prone to crashes than female drivers. Overall, it can be inferred that the drivers take risky decisions with increment in time pressure to complete the driving task, even at the expense of their own safety which exposed them to high likelihood of crashes.     

The influence of music genres on the driving behaviour of young drivers and their visual scanning of the environment

Recently, distractions have been recognized as a significant risk factor in road traffic. This simulator study aims to investigate how different music genres affect the driving behaviour of young drivers and their visual scanning of the environment in urban settings. The genres considered were Croatian pop, foreign pop, classical music, metal, and Balkan folk music, while on one road section there was no music. The research sample consisted of 61 participants (44 males and 17 females) with a mean age of 24.58 years and a mean driving experience of 5.25 years. The influence of music on the drivers’ behaviour was analysed on the basis of the data collected from the driving simulator, eye tracking glasses, and structured observation during driving sessions. It was found that the highest average speed (around 60 km/h) was recorded while the participants were listening to Balkan folk and metal music, while other music genres, as well as the “no music” condition, influenced driving speed in a similar way to one another and the participants drove at 50 km/h on average. Furthermore, the results suggest that the music genre also affects how drivers visually scan the environment (the number of gazes classified as fixations and number of road signs looked at). The findings obtained may be used in road safety work and practical recommendations and further research are discussed.     

Capturing the effects of texting on young drivers behaviour based on copula and Gaussian Mixture Models

This research effort aims to investigate the impact of texting on young drivers' behavior and safety based on data from driving simulator experiments, for different driving contexts, like motorways, urban and rural roads, during daytime and night, and for alternative weather conditions (‘clear sky’ and rain). The study offers a complete and comprehensive investigation of the effects of texting on driving behavior, able to provide evidence on policy-making. For the purposes of this study, a driving simulator experiment was carried out where 34 young participants drove predefined driving scenarios. Initially, multivariate copula analysis was used in order to explore statistical inferences among variables, especially since it retains a parametric specification for bivariate dependencies and allows testing of several parametric structures to characterize them. Secondly, alternative copula configurations were tested, which showed that texting and other road and environmental characteristics affect young drivers behavior and in particular more than one outcome can occur at the same time. Finally, Gaussian Mixture Modeling (GMM) was employed, demonstrating that the variables' pairs that presented the strongest correlations were lane departure and speed, as well as speed and reaction time. GMMs application showed that drivers using mobile phones who were involved in a collision presented a different driving behavior compared to the drivers who were occupied but were not involved in a collision.     

Evaluating the impact of real-time coaching programs on drivers overtaking cyclists

Innovative motor insurance schemes involve the use of on-board devices to collect kinematic driving data as part of the so-called ‘Pay-How-You-Drive’ schemes, which charge premiums based on drivers’ behavior. Some of these schemes also involve on-board coaching programs, which give real-time feedback to users.Here, we aimed to investigate the influence of motor insurance on-board real-time coaching programs on drivers’ behavior while overtaking cyclists, as motor vehicle/bicycle interactions are a relevant issue in road safety. The tested programs give real-time feedback to users on their acceleration, promoting smoother and safer driving styles.Data were collected with a driving simulator experiment involving 67 young drivers. The experiment was divided into two trials: in the first, participants drove as normally as possible without receiving any type of feedback; in the second, which took place one month later, they received feedback based on their driving behavior. Using data from the first trial, participants were clustered (k-mean approximation) into two groups, according to their driving style (aggressive vs. defensive). For each group, half of the drivers received contingent positive feedback (when a smooth driving event occurred) and the other half received contingent negative feedback (when a harsh driving event occurred). Feedback was presented in the form of auditory cues (for half of one group) or as visual cues (for the others). Thus, there were eight groups based on driving style, feedback type, and feedback modality.Multiple kinematic variables were studied with mixed ANOVA, and included not only clearance distances, speeds, and acceleration, but also the chosen overtaking strategy (accelerative vs. flying). Driving style, gender, car usage, feedback type and modality were considered as factors in the analysis.Results showed that the coaching programs had a significant positive effect, in terms of safety, reducing acceleration and speeds during the overtaking and inducing drivers to adopt the safer accelerative strategy. It was also particularly effective in improving the performance of aggressive drivers. These results are of high interest for real-world applications because they were obtained with a general-purpose coaching program; conversely, it might be impractical to develop dedicate programs for specific situations such as drivers overtaking cyclists.     

Keep calm,pay attention,and carry on: Anxiety and consciousness mediate the effect of,mindfulness on driving performance in young drivers

Road traffic crashes are currently one of the main causes of deaths in the world and many efforts have been made to develop effective interventions to reduce them. Mindfulness has risen as a method for improving mental and physical well-being and has been hypothesized as potentially beneficial for driving performance. This has led to some commercial ventures based on such hypothesis, despite that the empirical evidence backing up them is still limited. Besides, at the moment there is not yet a clear account of the specific mechanism underlying this proposals. So, it seems plausible that the relationship between mindfulness and driving performance is indirect, and that personality traits such as conscientiousness and neuroticism may play a mediation role between mindfulness and driving performance. A sample of 98 drivers between the ages of 19 and 29 completed questionnaires assessing mindfulness, anxiety trait and anxiety state, and the big five personality traits. Driving performance was assessed in a driving simulator. A mediation model was fitted with conscientiousness and neuroticism set as mediators of the effects of the relationship between the subject’s mindfulness levels and the driving performance.     

Interhemispheric transmission time in an auditory two-choice reaction task

An unimanual auditory choice reaction task was performed by 16 right-handed male subjects. Upon release of a central ready button subjects pressed a target button on their right or left side depending on the ear in which they heard a tone. A significant 'Ear' x 'Hand' interaction effect appeared on both reaction time (RT) and the first component of movement (MT₁), showing that uncrossed conditions give rise to faster responses than crossed conditions. Interhemispheric transmission time (IHTT) was estimated from the difference of response latency in the two types of conditions. No asymmetry in the speed of information transfer between the hemispheres was found. IHTT was 16 ms in the case of RT, which supported previous research with visual reaction tasks, and IHTT for MT₁ was 13 ms, which has not been reported before.     

Crossed-uncrossed difference (CUD) on reaction time and movement time in three two-choice experiments

The present study addresses the robustness and the reliability of the crossed-uncrossed difference (CUD) on a reaction time (RT) and a movement time (MT) component of a prescribed unimanual response to lateralized stimuli. Experiment 1 demonstrated positive CUDs both when a visual warning signal (WS) and an auditory reaction signal (RS) appeared on corresponding and non-corresponding sides of the body. Experiment 2 showed effect of handedness on CUD calculated among right-handers and left-handers. Experiment 3 investigated CUDs through five successive days, indicating that CUDs became steady and reliable although practice affected both RT and MT. All experiments indicated CUD on RT and MT. In addition, Experiments 1 and 2 demonstrated hemispheric asymmetries in favour of an anatomical theory while Experiment 3 did not show any asymmetries and supports an S-R compatibility theory.     

A comprehensive systematic review of the laboratory-based research investigating the influence of alcohol on driving behaviour

In the past thirty years, alcohol influence on drivers has been widely researched using driving simulators across the world. However, a critical evaluation of the existing evidence is required to project the current status and provide directions for future investigations. The present study conducts a comprehensive systematic review of the research examining the effects of alcohol on simulated driving performance. The literature search was conducted in academic databases such as Google Scholar, SCOPUS, Transportation Research Information Database (TRID), PubMed, and Web of science. In total, 1015 articles were identified from the initial database search along with 14 additional articles from other sources (Articles Plus, Cochrane Library, Medline, National Technical Information Services, National Highway Traffic Safety Administration, and Medline Grey Literature Report), out of which 110 articles were included in the systematic review. The majority of the studies were conducted in the United States of America (45%), followed by Australia (15%). These studies were reviewed based on the following aspects: type of driving simulator, study design, Blood Alcohol Concentration (BAC) or alcohol dose used in the experiments, driving environment characteristics, driving performance measures, data analysis techniques, and additional factors affecting driver behaviour under the influence of alcohol. Overall, the evidence regarding driving performance under the influence of alcohol shows that the majority of the previous studies have found a significant impact of alcohol on the various aspects of driving behaviour. The systematic review highlighted the methodological limitations observed in the previous driving simulator studies which should be acknowledged in future research.     

Driver's avoidance characteristics to hazardous situations: A driving simulator study

When analyzing the causes of an accident, it is critical to determine whether the driver could have prevented the accident. In previous studies on the reaction times of drivers, the definition and values of reaction times vary, so applying reaction time is difficult. In such analysis, the driver’s reaction time from perception is required to determine whether the driver could have prevented the accident, but past studies are difficult to utilize in accident analysis as reaction time measurements were taken after the occurrence of hazardous situations. In this study, 93 subjects from age groups ranging from 20 s to 40 s participated in an experiment inside a full-scale driving simulator, to determine reaction time values that can be practically applied to accident analysis. A total of 4 hazardous accident situations were reproduced, including driving over the centerline, pedestrian jaywalking, a vehicle cutting in, and intersection traffic signal violation. The Time-To-Collision (TTC) was 2.5 s and the driving speed was set to the common city road speed limits of 60 and 80 km/h. An eye tracker was used to determine the driver’s Saccade Latency (SL) during hazardous situations. Brake Reaction Time from Perception (BRT_P), Steer Reaction Time from Perception (SRT_P), and Driver Reaction Time from Perception (DRT_P) were derived, and the measurements were statistically analyzed to investigate differences by age group, gender, speed, and type of hazardous situation. Most participants were found to avoid collisions by braking first rather than steering for the presented hazardous situations, except for the cutting in situation. Also, to determine a reaction time that would cover most drivers, the 85th percentile of DRT_P was calculated. The 85th percentile of DRT_P was in the range of 0.550 – 0.800 s. Specifically for each hazardous situation, it was 0.650 s for driving over the centerline, 0.800 s for the pedestrian jaywalking, 0.660 s for cutting in, and 0.550 s for the intersection traffic signal violation. For all 4 hazardous situations combined, the 85th percentile of DRT_P was 0.646 s. The findings can be utilized to determine the driver’s likelihood of avoiding accidents when faced with similar hazardous situations.     

Driving mental workload and performance of ageing drivers

The World Health Organization stated that the global ageing population is increasing rapidly as well as the case of road accidents involving ageing drivers. This study presents the driving mental workload and performance model of ageing drivers in the context of real-time road driving. Twenty paid participants (ten males) with a mean age of 57.8 years old (SD = 2.7) and mean driving experience of 29.6 years (SD = 8.5) took part in driving experiments with three complexity levels of situation: simple situation (SS), moderately complex situation (MCS), and very complex situation (VCS). The subjective ratings using NASA Task Load Index (NASA-TLX), physiological measure using electroencephalogram, number of traffic violations (NTVs), speed variability, and reaction time of peripheral detection task were measured. The driving experiments reveal the following: (1) The subjective workload ratings on mean physical demand score were the highest. (2) The electroencephalogram results show that situation complexity had significant effects on theta relative power and alpha relative power of two channel locations (3) The highest mean NTVs was in VCS. (4) The mean speed variability in the MCS was significantly lower than that of in the SS and VCS. (5) The maximum reaction time was recorded in VCS while the minimum reaction time was recorded in the MCS. The overall driving performance score regression models were developed based on the strong correlation and linear relationship between mental workload and driving performance elements. The models may benefit as a reference for designers, manufacturers, developers, and policymakers in designing better driving environment for ageing drivers by integrating safety and transportation, thus, optimizing and sustaining the driving performance of ageing drivers.     

The ability of young,middle-aged and older drivers to inhibit visual and auditory distraction in a driving simulator task

Driver distraction is one major cause of road traffic accidents. In order to avoid distraction-related accidents it is important to inhibit irrelevant stimuli and unnecessary responses to distractors and to focus on the driving task, especially when unpredictable critical events occur. Since inhibition is a cognitive function that develops until young adulthood and decreases with increasing age, young and older drivers should be more susceptible to distraction than middle-aged drivers. Using a driving simulation, the present study investigated effects of acoustic and visual distracting stimuli on responses to critical events (flashing up brake lights of a car ahead) in young, middle-aged, and older drivers. The task difficulty was varied in three conditions, in which distractors could either be ignored (perception-only), or required a simple response (detection) or a complex Go-/NoGo-response (discrimination). Response times and error rates to the critical event increased when a simultaneous reaction to the distractor was required. This distraction effect was most pronounced in the discrimination condition, in which the participants had to respond to some of the distracting stimuli and to inhibit responses to some other stimuli. Visual distractors had a stronger impact than acoustic ones. While middle-aged drivers managed distractor inhibition even in difficult tasks quite well (i.e., when responses to distracting stimuli had to be suppressed), response times of young and old drivers increased significantly, especially when distractor stimuli had to be ignored. The results demonstrate the high impact of distraction on driving performance in critical traffic situations and indicate a driving-related inhibition deficit in young and old drivers.     

Response-level probability effects on reaction time: Now you see them,now you don't

Many reaction time (RT) experiments have tested for response-level probability effects. Their results have been mixed, which is surprising because psychophysiological studies provide clear evidence of motor-level changes associated with an anticipated response. A survey of the designs used in the RT studies reveals many potential problems that could conceal the effects of response probability. We report five new RT experiments testing for response-level probability effects with the most promising of the previous designs—that of Blackman (1972 Blackman, A. R. 1972. Influence of stimulus and response probability on decision and movement latency in a discrete choice reaction task. Journal of Experimental Psychology, 92: 128–133. doi:10.1037/h0032169[Crossref], [PubMed] , [Google Scholar])—and with new designs. Some of these experiments yield evidence of response-level probability effects, but others do not. It appears that response-level probability effects are present primarily in simple tasks with a strong emphasis on response preparation, possibly because participants only expend effort on response preparation in these tasks.     

The effects of adaptive cruise control (ACC) headway time on young-experienced drivers' overtaking tendency in a driving simulator

The preference to maintain a certain desired speed is perhaps the most prevalent explanation for why a driver of a manually driven car decides to overtake a lead vehicle. Still, the motivation for overtaking is also affected by other factors such as aggressiveness, competitiveness, or sensation-seeking caused by following another vehicle. Whether such motivational factors for overtaking play a role in partially automated driving is yet to be determined. This study had three goals: (i) to investigate whether and how a driver's tendency to overtake a lead vehicle changes when driving a vehicle equipped with an adaptive cruise control (ACC) system. (ii) To study how such tendencies change when the headway time configuration of the ACC system varies. (iii) To examine how the manipulation of the speed and speed variance of the lead vehicle affect drivers' tendencies to overtake a lead vehicle. We conducted two different experiments, where the second experiment followed the first experiment's results. In each experiment, participants drove three 10–12 min simulated drives under light traffic conditions in a driving simulator under manual and level one (L1) automation driving conditions. The automation condition included an ACC with two headway time configurations. In the first experiment, it was 1 sec and 3 secs, and in the second, it was 1 sec and 2 secs. Each drive included six passing opportunities representing three different speeds of the lead vehicle (−3 km/h, +3 km/h, +6 km/h relative to the participant), with or without speed variance. Results show that drivers tended to overtake a lead vehicle more often in manual mode than in automated driving modes. In the first experiment, ACC with a headway time of 1 sec led to more overtaking events than ACC with 3 secs headway time. In addition, the relative speed of the lead vehicle and its speed variability affected overtaking tendencies. In the second experiment, the relative speed of the lead vehicle and its speed variability affected overtaking tendencies only when interacting with each other and with driving configuration. When the speed of the lead vehicle was +3 km/h and included variability, more overtaking events occurred in manual mode than both automation modes. This work has shown that driving with ACC might help reduce overtaking frequencies and more considerable when the headway time is set to 3 secs.     

Effectiveness of the compensatory strategy adopted by older drivers: Difference between professional and non-professional drivers

It has been a controversial issue for the effect of ageing population on driving safety. Apparently, drivers’ physiological and cognitive performances deteriorate with age. However, older drivers may compensate for the elevated risk by adjusting their behaviors, known as compensatory strategy. Despite the extensive research on this topic, the compensatory strategy of older professional drivers is not well understood since many studies focused on the differences in compensatory behavior between older and young drivers. Professional drivers tend to be more skillful and able to cope with the unfavorable driving environments, thus presenting a higher capability to mitigate the risk. This study attempts to examine the compensatory behavior and its safety effect amongst older professional drivers, as compared to those of older non-professional drivers, using the driving simulator approach. In the driving simulator experiment, participants were asked to follow a leading vehicle for one hour, and two sudden brake events were presented. 41 (mid-aged and older) drivers completed the driving tests. Each participant was required to complete a car-following test, either under high or low traffic flow conditions. Performance indicators include driving capability (i.e. lateral control, longitudinal control, and brake reaction time) and compensatory behavior (i.e. average speed, and time headway). Additionally, two modified traffic conflict measures: time exposed time-to-collision (TET) and time integrated time-to-collision (TIT) are applied to indicate the traffic conflict risk. The random parameter Tobit models were estimated to measure the association between conflict risk and driver attributes, and random intercept models were used to assess other driving performance indicators. Results show that despite the impaired lateral control performance and longer brake reaction time of older drivers, the likelihood of severe traffic conflict of older drivers is lower than that of mid-aged drivers. Furthermore, though both older professional and older non-professional drivers adopted longer time headway, the reduction in the risk of severe traffic conflict is more profound among the older professional drivers. Such findings suggest that older professional drivers are more capable of mitigating the possible collision risk by adopting the compensatory strategy, as compared to older non-professional drivers. This justifies the existence of compound effect by the compensatory strategy of older driver and better driving skills of professional driver. This research provides useful insights into driver training and management strategies for employers, as older drivers would become a major cohort in the transportation industry.     

Dealing with time pressure

A sample of 250 people completed a questionnaire where they rated 30 statements describing their behavior and experiences during a recent time pressure situation. A factor analysis resulted in three factors: Personal Burden, Work Problems and Difficulties, and Challenge Orientation. People high on this last factor liked the excitement of being under pressure, were very absorbed in the task, and showed signs of being effective in dealing with the situation. Such individuals were having what has been identified as a flow experience often observed in sports and the arts. The results of this study suggest that effective coping with time pressure is related to taking a task-oriented attitude, avoiding self-preoccupation, and using time management skills to focus on successful achievement.     

Driving distraction at night: The impact of cell phone use on driving behaviors among young drivers

Currently, young drivers are more likely than other drivers to use cell phones while driving at night, which has become a major cause of road crashes. However, limited attention has been given to distracted nighttime driving. Therefore, the aim of this study was to explore the interaction effect of cell phone use and time of day (daytime and nighttime) on young drivers’ car-following performance. Forty-three young drivers engaged in a driving simulator experiment with a within-subject design that included three distractions (no distraction, talking and texting on a cell phone) and two times of day. This paper applied non-parametric tests to analyze the data and obtained the following results: (1) the standard deviation of lane position (SDLP) did not significantly differ at either time of day under no distraction, but it was significantly higher at night on straight roads and large-radius curves after introducing distractions. In addition, participants drove faster and gave less headway on small-radius curves under both distractions at night; (2) texting significantly increased the SDLP, while there was less lateral variation during the talking tasks than under no distraction on simple road sections; and (3) compared with the experienced drivers, the novice drivers drove faster during the talking tasks on small-radius curves, but there was no significant difference between groups during the texting tasks. These findings provide both theoretical and practical implications for related policy makers to enhance traffic safety.     

The differences in hazard response time and driving styles of violation-involved and violation-free taxi drivers

The present study aimed to investigate the relationships between taxi drivers’ traffic violations in past driving and two domains: driving skill (hazard perception skill) and driving style. Five hundred and fifty taxi drivers aged 25 – 59 were recruited to finish a video-based hazard perception test and the Chinese version of the Multidimensional Driving Style Inventory (MDSI). The relationships between hazard response time, driving style and traffic violations were examined, and the differences in hazard response times and driving styles of violation-involved drivers (n = 220) and violation-free drivers (n = 330) were compared. The results showed that taxi drivers’ traffic violations are closely related to their driving styles and hazard response time. Violation-involved drivers scored significantly higher in hazard response time and maladaptive driving styles (i.e., anxious, risky and angry styles) and lower in careful driving style than violation-free drivers. More importantly, drivers’ hazard response time and driving styles can effectively predict their violation involvement in the last 12 months with an overall classification accuracy of 66.4%. The findings provide evidence for the usefulness of video-based hazard perception tests and the MDSI in taxi driver testing and training.     

Identification of driving simulator sessions of depressed drivers: A comparison between aggregated and time-series classification

Depression has been found to significantly increase the probability of risky driving and involvement in traffic collisions. The majority of studies correlating depressive symptoms with driving, pursue to predict the differences in driving behavior if the driver has already been diagnosed. Little evidence can be found, however, on how mental and psychological disorders can be identified from driving data, and usually analyses utilize simple models and aggregated data. This study aims at utilizing microscopic data from a driving simulator to detect sessions belonging to “depressed” drivers by utilizing powerful machine learning classifiers. Driving simulator sessions from 11 older drivers with symptoms of depression and 65 healthy drivers were utilized towards that aim. Random Forests, an ensemble classifier, with proven efficiency among transportation applications, are then trained on highly disaggregated data describing the mean and standard deviation of speed and lateral or longitudinal acceleration of drivers in the simulator. The kinematic data were aggregated in 30-seconds, 1-minute and 5-minute intervals, but the corresponding time-series of the measurements were also taken into account. Furthermore, classifiers were treated with imbalanced learning techniques to address the scarcity of depressed drivers among the healthy. Time-series of mean speed and the standard deviation of longitudinal acceleration even with a duration of 30-seconds have proven to be the best predictors of driving sessions belonging to depressed drivers with a very low rate of false alarms. The results outperform previous approaches, and indicate that naturalistic driving data or deep learning could prove even more efficient in detecting depression.