Safety performance evaluation of freeway merging areas under autonomous vehicles environment using a co-simulation platform

Merging areas serve as the potential bottlenecks for continuous traffic flow on freeways. Traffic incidents in freeway merging areas are closely related to decision-making errors of human drivers, for which the autonomous vehicles (AVs) technologies are expected to help enhance the safety performance. However, evaluating the safety impact of AVs is challenging in practice due to the lack of real-world driving and incident data. Despite the increasing number of simulation-based AV studies, most relied on single traffic/vehicle driving simulators, which exhibit limitations such as inaccurate description of AV behavior using pre-defined driving models, limited testing modules, and a lack of high-fidelity traffic scenarios. To this end, this study addresses these challenges by customizing different types of car-following models for AVs on freeway and developing a software-in-the-loop co-simulation platform for safety performance evaluation. Specifically, the environmental perception module is integrated in PreScan, the decision-making and control model for AVs is designed by Matlab, and the traffic flow environment is established by Vissim. Such a co-simulation platform is supposed to be able to reproduce the mixed traffic with AVs to a large extent. By taking a real freeway merging scenario as an example, comprehensive experiments were conducted by introducing a single AV and multiple AVs on the mainline of freeway, respectively. The single AV experiment investigated the performance of different car-following models microscopically in the case of merging conflict. The safety and comfort of AVs were examined in terms of TTC and jerk, respectively. The multiple AVs experiment examined the safety impact of AVs on mixed traffic of freeway merging areas macroscopically using the developed risk assessment model. The results show that AVs could bring significant benefits to freeway safety, as traffic conflicts and risks are substantially reduced with incremental market penetration rates.

Applying the Accumulator model to predict driver's reaction time based on looming in approaching and braking conditions

INTRODUCTION

The prediction of when the driver will react to a change in the lead vehicle motion is critical for assessing rear-end crash risk using car-following models. Past studies have assumed constant reaction time and driver's continuous reaction. However, these assumptions are not valid as the driver's reaction time can vary in different car-following situations and the driver does not continuously react to the lead vehicle motion. Thus, this study predicted the driver's reaction time using the Wiedemann car-following model and the Accumulator model. The Accumulator model assumes the driver's start of reaction based on the accumulation of looming and thereby reflects the driver's intermittent reaction.

METHOD

Fifty drivers' behavior was observed using a driving simulator in two scenarios: (1) approach and follow a moving lead vehicle and (2) approach a stopped lead vehicle. The Accumulator model predicted the reaction times based on different looming variables (angular velocity and tau-inverse), lead vehicle type (car and truck), and lead vehicle brake lights (on or off).

RESULTS

The Accumulator model showed lower prediction errors of the reaction time than the Wiedemann model, which assumes reaction based on the fixed looming threshold. The Accumulator model predicted the reaction times more accurately when it was calibrated with the angular velocity due to width and height of lead vehicles. Moreover, the Accumulator model with tau-inverse produced the smallest prediction error of reaction times among different Accumulator models and the Wiedemann model when lead vehicle brake lights were on.

CONCLUSIONS

This study demonstrates that the Accumulator model is a promising method of predicting the driver's reaction time in car-following situations, which affects rear-end crash risk.

PRACTICAL APPLICATIONS

The Accumulator model can be incorporated into a car-following model for the prediction of reaction times and can estimate the rear-end collision risk of vehicles more accurately.

Modelling braking behaviour of distracted young drivers in car-following interactions: A grouped random parameters duration model with heterogeneity-in-means

Braking is an important characteristic of driving behaviour that has a direct relationship with rear-end collisions in a car-following task. Braking becomes more crucial when drivers' cognitive workload increases because of using mobile phones whilst driving. This study, therefore, investigates and compares the effects of using mobile phones whilst driving on braking behaviour. Thirty-two young licenced drivers, evenly split by gender, faced a safety-critical event, that is, leader's hard braking, in a car-following situation. Each participant drove the CARRS-Q Advanced Driving Simulator and was required to respond to a braking event in the simulated environment in three phone conditions: baseline (no phone conversation), handheld, and hands-free. A random parameters duration modelling approach is employed to (i) model drivers' braking (or deceleration) times using a parametric survival model, (ii) capture unobserved heterogeneity associated with braking times, and (iii) account for repeated experiment design. The model identifies the handheld phone condition as a random parameter whilst vehicle dynamics variables, hands-free phone condition, and driver-specific variables are found as fixed parameters. The model suggests that most distracted drivers (in the handheld condition) reduce their initial speeds more slowly than undistracted drivers, reflecting their delayed initial braking that may lead to abrupt braking to avoid a rear-end collision. Further, another group of distracted drivers exhibits faster braking (in the handheld condition), recognising the risk associated with mobile phone usage and delayed initial braking. Provisional licence holders are found to be slower in reducing their initial speeds than open licence holders, indicating their risk-taking behaviour because of their less experience and more sensitivity to mobile phone distraction. Overall, mobile phone distraction appears to impair the braking behaviour of young drivers, which poses significant safety concerns for traffic streams.

Using multicolor perceptual markings as a rear-end crash risk mitigator: A field investigation

Perceptual markings on roadways are prevailing countermeasures with substantial effectiveness for accident prevention, and a variety of alternatives and derivatives of them are developed to expect to receive an augmented performance of behavioral intervention and crash risk mitigation. However, the proper use of colors as a way of developing effective and innovative perceptual markings is seldomly recognized in-depth from the perspective of visual perceptual mechanism in behind. Given this, in this study, we introduced a kind of multicolor perceptual markings (MCPMs) pattern, i.e., one red marking follows one yellow marking ("1Y + 1R"), two red markings follow two yellow markings ("2Y + 2R"), and three red markings follow three yellow markings ("3Y + 3R"), and evaluated their effects on longitudinal and lateral driving behaviors and real-time safety benefits in car-following via a series of field investigation on a real-world expressway in China. The statistical analyses of the relative differences of speed (θ_v), distance headway (θ_d), time headway (θ_h), lateral movement (θ_p), and crash risk (η_mTTC and η_DRAC, developed from time-to-collision (TTC) and deceleration rate to avoid crash (DRAC)) suggest that, 1) the MCPMs could lead to substantial increases in car-following time and distance headways, and reduction in speed. The maximum time headway increase (0.61 s), speed reduction (1.42 m/s), and distance increase (3.6 m) were found in the condition of "1Y + 1R" compared with the baseline; 2) the MCPMs stabilized the lateral movement of vehicles on the lane at each observation section, and "1Y + 1R" yielded the best performance of lane-keeping; 3) the MCPMs yielded applaudable real-time safety benefits, which were believed to afford the drivers a better chance to accommodate their behaviors to a safer car-following status. The findings of this study suggest the MCPMs could be an especially applaudable form of perceptual markings, and could also be a critical reference of how to use colors in a better way for developing augmented perceptual markings.

Characterizing car-following behaviors of human drivers when following automated vehicles using the real-world dataset

As the market penetration rate of automated vehicles (AVs) increases, there will be a transition period when the traffic stream is composed of both AVs and human-driven vehicles (MVs) in the near future. However, the interactions between MVs and AVs, especially whether MVs will behave differently when following AVs compared to following MVs, have not been fully understood. Previous studies in this field mainly conducted traffic/numerical simulations or field experiments to investigate human drivers' behavior changes, but these approaches all have critical drawbacks such as simplified driving environments and limited sample sizes. To fill in the knowledge gap, this study uses the high-resolution (10 Hz) Waymo Open Dataset to reveal differences in car-following behaviors between MV-following-AV and MV-following-MV cases. Driving volatility measures, time headways and time-to-collision (TTC) are adopted to quantify and compare MV-following-AV and MV-following-MV interactions. The principal component analysis (PCA) is applied on the high-dimensional feature space, followed by the hierarchical clustering on the dimension-reduced feature set to categorize MV driving styles when following AVs. The comparison results indicate that MV-following-AV events have lower driving volatility in terms of velocity and acceleration/deceleration, smaller time headways and higher TTC values. Furthermore, the clustering results reveal that human drivers when following AVs exhibit four different car-following styles: high-velocity-non-aggressive, high-velocity-aggressive, low-velocity-non-aggressive, and low-velocity-aggressive. These findings highlight the vital importance of taking into account the heterogeneity of MV-following-AV behaviors when designing mixed traffic control algorithms and can be beneficial for AV fleet operators to improve their algorithms.

Multianticipation for string stable Adaptive Cruise Control and increased motorway capacity without vehicle-to-vehicle communication

Adaptive Cruise Control (ACC) systems have been expected to solve many problems of motorway traffic. Now that they are widespread, it is observed that the majority of existing systems are string unstable. Therefore, small perturbations in the speed profile of a vehicle are amplified for the vehicles following upstream, with negative impacts on traffic flow, fuel consumption, and safety. Increased headway settings provide more stable flow but at the same time it deteriorates the capacity. Substantial research has been carried out in the past decade on utilizing connectivity to overcome this trade-off. However, such connectivity solutions have to overcome several obstacles before deployment and there is the concrete risk that motorway traffic flow will considerably deteriorate in the meanwhile. As an alternative solution, the paper explores multianticipation without inter-vehicle communication, taking advantage of the recent advancements in the field of RADAR sensing. An analytical study is carried out, based on the most widely used model and parameter settings used to simulate currently available commercial ACC systems, comparing the transfer functions and step responses for the nominal and the multianticipative formulations. Then, a microsimulation framework is employed to validate our claim on different speed profiles. Analytical results demonstrate that multianticipation enhances stability without impacting traffic flow. On the contrary, the simulation study indicates that the multianticipative-ACC can produce higher road capacity even in the presence of external disturbances and for a wide range of calibrated parameters. Finally, optimality conditions for the tuning of the headway policy are derived from a Pareto optimization.

Quantifying effects of reverse linear perspective as a visual cue on vehicle and platoon crash risk variations in car-following using path analysis

Road markings are prevalent in practice as perceptual countermeasures to crashes, and a great deal of them have been used for speed reduction. However, there is rare seen any equivalent measures especially for distance control. More importantly, the visual perceptual mechanism of road markings on driving behaviors and crash risk is still blur. Given this, in the present study, we comprehensively quantified the effects of reverse linear perspective (RLP) from its origin as a visual cue, produced by a kind of transverse line markings on road, and explored the effects on car-following behaviors and crash risk variations by path analyses imbedded in a structural equations model, which was approximated with naturalistic driving and traffic flow data. In the model, multiple sources of observed factors in visual perception, driver behaviors, and traffic flow characteristics, and exogenous unobserved factors of distance risk perception, speed risk perception, and platoon risk status were comprehensively structured to explain the vehicle crash risk variation and the platoon crash risk variation. The results indicate that (1) distance risk perception, speed risk perception, and platoon risk status were well explanatory and predictive to vehicle crash risk variation and platoon crash risk variation; (2) the effects of reverse linear perspective as a visual cue on driving behaviors and crash risk variations in car-following were adequately quantified by its geometrical characteristics concerning distance perception; (3) the visual cue of reverse linear perspective in addition with initial distance, stopping sight distance, and the type of leading vehicles explained 33 % of the variance in distance risk perception; the temporal frequency, initial speed, and the type of following vehicles explained 23 % of the variance in speed risk perception; distance risk perception, speed risk perception, and platoon risk status combinedly explained 25 % and 22 % of the total variance in vehicle crash risk variation and platoon crash risk variation, respectively; (4) vehicle crash risk variation and platoon crash risk variation were equivalently specified by those observed explanatory factors. The findings of this study suggest the usefulness and importance of understanding the contribution of psychological factors on crash risk, and emphasize that the road markings can be an effective and readily practical countermeasure in easing traffic safety issues.

Wielding and evaluating the removal composition of common artefacts in EEG signals for driving behaviour analysis

Noninvasive EEG signals provide neural activity information at high resolution, of which human mental status can be properly detected. However, artefacts always exist in brain oscillatory EEG signals and thus impede the accuracy and reliability of relevant analysis, especially in real-world tasks. Moreover, the use of a mature artefact identification method cannot assure impeccable artefact separation; this leads to a trade-off between removing contaminated information and losing valuable information. This study addresses this problem by investigating a simulator-based driving behaviour analysis using a car-following scenario to correlate the EEG-based mental features with behavioural responses. The study develops an architecture for an artefact composition pool and proposes three integrated prediction models to evaluate the removal compositions of the EEG artefacts. Three errors (mean absolute, root mean square, mean absolute percentage) and R-squared index are considered for measuring the performance of the models. The results show that the best-performing composition outperformed the no-removal and all-removal cases by 11.75% and 4.28% improvements, respectively. Specifically, we investigate different common artefacts including eye blinks, horizontal eye movements, vertical eye movements, generic discontinuities and muscle artefacts. The gained knowledge on artefact removal, EEG spectral features and stimuli-response patterns can be further applied to properly manipulate real-world EEG signals and develop an effective brain-computer interface.

Long-term effectiveness of reverse linear perspective markings on crash mitigation in car-following: Evidence from naturalistic observations

Perceptual markings on roads are verified with short-term effectiveness for accident prevention. However, the long-term performance of them is seldomly investigated, which unintentionally impedes its more widely recognition and application as a low-cost and readily achievable countermeasure. Also, the previous perceptual markings were only tested for speed reduction effect, little is known concerning their influence on headway adjustment. Given this, in this study, we investigated the short-, medium-, and long-term performance of the reverse linear perspective markings (RLPMs) on driving behaviors and safety benefits in car-following. The RLPMs were a form of markings pattern that can produce reverse linear perspective visual information on the lane and lead to distance underestimation. The RLPMs were permanently installed on a straight and a curve segment of a freeway in China, and the naturalistic vehicle flow data one day, four months, one year, two years, and three years after the installation of the RLPMs were collected. The statistical analyses of general and sectional relative differences of speed, distance headway and time headway suggest that 1) the speed reduced and distance and time headways increased in short-, medium-, and long-term as compared with the baseline on both the straight and curve segments; 2) the long-term performance of RLPMs significantly weakened as compared with the short-term performance, yet sustained to 0.50 m/s in speed reduction, 3.77 m in distance headway increase, and 0.097 s in time headway increase on average within the observations in one year and above on the straight segment; similar sustained performance of 0.47 m/s in speed reduction, 2.60 m in distance headway increase, and 0.072 s in time headway increase were observed on the curve segment; 3) the RLPMs were tested to have positive and relatively endured effectiveness on mitigating crash risk in car-following measured by two surrogate safety indicators based on time-to-crash (TTC) and deceleration rate to avoid a crash (DRAC). The findings of this study suggest the RLPMs could be an especially applaudable form of perceptual markings as they are relatively effective in the long-term and are multifunctional in intervening speed, distance, headway, and crash risk. This study also emphasizes the challenge of more field tests and observations on the long-term performance of the perceptual markings, and the thorough considerations of the visual perception mechanism behind the markings to achieve an alternative solution to the long-term issue.

A data-driven feature learning approach based on Copula-Bayesian Network and its application in comparative investigation on risky lane-changing and car-following maneuvers

The era of 'Big Data' provides opportunities for researchers to have deep insights into traffic safety. By taking advantages of 'Big Data', this study proposes a data-driven method to develop a Copula-Bayesian Network (Copula-BN) using a large-scale naturalistic driving dataset with multiple features. The Copula-BN is able to explain the causality of a risky driving maneuver. As compared with conventional BNs, the Copula-BN developed in this study has the following advantages: the Copula-BN 1. Has a more rational and explainable structure; 2. Is less likely to be over-fitting and can attain more satisfactory prediction performance; and 3. Can handle not only discrete but also continuous features. In terms of technical innovations, Shapley Additive Explanation (SHAP) is used for feature selection, while Gaussian Copula function is employed to build the dependency structure of the Copula-BN. As for applications, the Copula-BNs are used to investigate the causality of risky lane-changing (LC) and car-following (CF) maneuvers, upon which the comparisons are made between the two essential but risky driving maneuvers. In this study, the Copula-BNs are developed based on the Second Highway Research Program (SHRP2) Naturalistic Driving Study (NDS) database. Upon network evaluation, the Copula-BNs for both risky LC and CF maneuvers demonstrate satisfactory structure performance and promising prediction performance. Feature inferences are conducted based on the Copula-BNs to respectively illustrate the causation of the two risky maneuvers. Several interesting findings related to features' contribution are discussed in this paper. To a certain extent, the Copula-BN developed using the data-driven method makes a trade-off between prediction and causality within the 'Big Data'. The comparison between risky LC and CF maneuvers also provides a valuable reference for crash risk evaluation, road safety policy-making, etc. In the future, the achievements of this study could be applied in Advanced Driver-Assistance System (ADAS) and accident diagnosis system to enhance road traffic safety.

Speed cascade adaptive control for hybrid electric vehicle using electronic throttle control during car-following process

Achieving robust longitudinal speed control for hybrid electric vehicles (HEVs) through precise position tracking of electric throttle control system (ETCS) can improve engine fuel economy and vehicle longitudinal speed performance. Whereas, nonlinearities resulting from friction, gearbox, and return springs of ETCS, uncertain system parameters related to production deviations and device aging, disturbance from the air flow fluctuation on the throttle plate, and unknown road grade and uncertain preceding vehicle acceleration make control design challenging. Aiming at this issue, a speed cascade control scheme considering car-following scenario is investigated for a parallel ETCS controlled HEV in this paper, of which contains a primary speed adaptive controller and a secondary electronic throttle adaptive nonlinear active disturbance rejection controller with the adaptive gains extended state observer. The distinction from the existing relevant literatures is that the inherent characteristics of nonlinearity and uncertainty in the ETCS and longitudinal velocity kinematics, and the car following scenarios are explicitly taken into account in the design of the cascade control for ETCS controlled HEVs. Both simulation and rapid-control-prototype (RCP) experimental results demonstrate the effectiveness and practicality of the proposed scheme and the advantages over other existing research strategies.

Predicting driver reaction time and deceleration: Comparison of perception-reaction thresholds and evidence accumulation framework

Prediction of driver reaction to the lead vehicle motion based on the perception-reaction time (PRT) is critical for prediction of rear-end crash risk. This study determines PRT at various spacings in approaching and braking conditions, and examines the association of PRT and deceleration rate with crash risk. For these tasks, a total of 50 drivers' behavior was observed in a driving simulator experiment with 4 different scenarios - reaction to a decelerating lead vehicle, reaction to a stopped lead vehicle, perception of a lead vehicle's speed change, and perception of a slow-moving lead vehicle. The study tested three hypotheses of PRT including perception and reaction thresholds and the evidence accumulation framework using a visual variable (tau-inverse). It was found that the drivers neither reacted after a specific PRT from the start of perception nor reacted at a specific value of tau-inverse. Rather, the drivers generally reacted when the accumulation of evidence (tau-inverse) over time reached a threshold. It was also found that the magnitude of deceleration rate depends on the tau-inverse at the start of braking and hence, higher crash risk was associated with higher level of urgency and insufficient brake force rather than longer PRT. This study demonstrates that the evidence accumulation framework is a promising method of predicting driver reaction in approaching and braking conditions for different types of lead vehicle, and the level of urgency is important for predicting the probability of crash.

Effects of peripheral transverse line markings on drivers' speed and headway choice and crash risk in car-following: A naturalistic observation study

Rear-end crashes are closely related to car-following situation of vehicles. Speeding and insufficient headway are the major reasons as the drivers have not enough time to react to a sudden brake from the leading vehicle. Perceptual countermeasures, like speed reduction markings, are widely used in practice for accident prevention, and are verified with substantial effectiveness. However, compared with its practical application, the perceptual countermeasures are rarely analyzed in depth from the perspective of drivers' visual perception where the meaning of "perceptual" actually dwells. In addition, its effect on drivers' headway (distance) choice is almost ignored in previous research. Given this, the present study explored the effects of a certain type of perceptual treatment, i.e., the peripheral transverse line markings (PTLMs), on drivers' choice of speed and headway (distance) in car-following by a series of on-road experiments. In the on-road experiments, temporary line markings were installed on a real-world freeway in China to shape the PTLMs. The intersection angle (α) and the longitudinal spacing (λ) of the PTLMs were manipulated to attempt to associate the line markings with drivers' visual perception. Results of general and sectional relative differences of time headway (η_h, θ_h), speed (η_v, θ_v), and distance (η_d, θ_d) suggests that 1) the speed was reduced, the distance and time headway were increased significantly after the installation of PTLMs when compared with the original condition; 2) a larger intersection angle (α) and a smaller longitudinal spacing (λ) of PTLMs could lead to a greater variations in speed and headway (distance); in particular, the PTLMs in a form of α=150°, λ=2m resulted in 0.44 s increase in time headway, 1.33 m/s reduction in speed, and 4.07 m increase in distance in maximum; 3) the real-time crash risk variations under the influence of PTLMs were evaluated by two modified and extended surrogate safety indicators. The effects of PTLMs were discussed and explained considering the influences of optical illusion on lane width narrowing, edge rate on speed and "discontinuity effect" on distance, respectively. The findings of this study provide theoretical support for the perceptual countermeasures and suggest comparative advantages of PTLMs in dealing with rear-end crashes by intervening drivers' speed and headway choice.

Distressed in the queue? Psychophysiological and behavioral evidence for two alternative car-following techniques

BACKGROUND

Nature offers numerous examples of animal species exhibiting harmonious collective movement. Unfortunately, the motorized Homo sapiens sapiens is not included and pays a price for it. Too often, drivers who simply follow other drivers are caught in the worst road threat after a crash: congestions. In the past, the solution to this problem has gone hand in hand with infrastructure investment. However, approaches such as the Nagoya Paradigm propose now to see congestion as the consequence of multiple interacting particles whose disturbances are transmitted in a waveform. This view clashes with a longlasting assumption ordering traffic flows, the rational driver postulate (i.e., drivers' alleged propensity to maintain a safe distance). Rather than a mere coincidence, the worldwide adoption of the safety-distance tenet and the worldwide presence of congestion emerge now as cause and effect. Nevertheless, nothing in the drivers' endowment impedes the adoption of other car-following (CF) strategies. The present study questions the a priori of safety-distance, comparing two elementary CF strategies, Driving to keep Distance (DD), that still prevails worldwide, and Driving to keep Inertia (DI), a complementary CF technique that offsets traffic waves disturbances, ensuring uninterrupted traffic flows. By asking drivers to drive DD and DI, we aim to characterize both CF strategies, comparing their effects on the individual driver (how he drives, how he feels, what he pays attention to) and also on the road space occupied by a platoon of DD robot-followers.

METHODS

Thirty drivers (50% women) were invited to adopt DD/DI in a driving simulator following a swinging leader. The design was a repeated measures model controlling for order. The CF technique, DD or DI, was the within-subject factor. Order (DD-DI / DI-DD) was the between-subjects factor. There were four blocks of dependent measures: individual driving performance (accelerations, decelerations, crashes, distance to lead vehicle, speed and fuel consumption), emotional dimensions (measures of skin conductance and self-reports of affective states concerning valence, arousal, and dominance), and visual behavior (fixations count and average duration, dwell times, and revisits) concerning three regions of the driving scene (the Top Rear Car -TRC- or the Bottom Rear Car -BRC- of the leading vehicle and the surrounding White Space Area -WSA). The final block concerned the road space occupied by a platoon of 8 virtual DD followers.

RESULTS

Drivers easily understood and applied DD/DI as required, switching back and forth between the two. Average speeds for DD/DI were similar, but DD drivers exhibited a greater number of accelerations, decelerations, speed variability, and crashes. Conversely, DI required greater CF distance, that was dynamically adjusted, and spent less fuel. Valence was similar, but DI drivers felt less aroused and more dominant. When driving DD visual scan was centered on the leader's BRC, whereas DI elicited more attention to WSA (i.e., adopting wider vision angles). In spite of DI requiring more CF distance, the resulting road space occupied between the leader and the 8th DD robot was greater when driving DD.

The impact of the connected environment on driving behavior and safety: A driving simulator study

The connected environment provides surrounding traffic information to drivers via different driving aids that are expected to improve driving behavior and assist in avoiding safety-critical events. These driving aids include speed advisory, car-following assistance, lane-changing support, and advanced information about possible unseen hazards, among many others. While various studies have attempted to examine the effectiveness of different driving aids discretely, it is still vague how drivers perform when they are exposed to a connected environment with vehicle-to-vehicle and vehicle-to-infrastructure communication capabilities. As such, the objective of this study is to examine the effects of the connected environment on driving behavior and safety. To achieve this aim, an innovative driving simulator experiment was designed to mimic a connected environment using the CARRS-Q Advanced Driving Simulator. Two types of driving aids were disseminated in the connected environment: continuous and event-based information. Seventy-eight participants with diverse backgrounds drove the simulator in four driving conditions: baseline (without driving aids), perfect communication (uninterrupted supply of driving aids), communication delay (driving aids are delayed), and communication loss (intermittent loss of driving aids). Various key driving behavior indicators were analyzed and compared across various routine driving tasks such as car-following, lane-changing, interactions with traffic lights, and giving way to pedestrians at pedestrian crossings. Results suggest that drivers in the perfect communication scenario maintain a longer time-to-collision during car-following, a longer time-to-collision to pedestrian, a lower deceleration to avoid a crash during lane-changing, and a lower propensity of yellow light running. Overall, drivers in the connected environment are found to make informed (thus better) decisions towards safe driving.

Identification of aggressive driving from naturalistic data in car-following situations

INTRODUCTION

Aggressive driving has been associated as one of the causes for crashes, sometimes with very serious consequences. The objective of this study is to investigate the possibility of identifying aggressive driving in car-following situations on motorways by simple jerk metrics derived from naturalistic data.

METHOD

We investigate two jerk metrics, one for large positive jerk and the other for large negative jerk, when drivers are operating the gas and brake pedal, respectively.

RESULTS

The results obtained from naturalistic data from five countries in Europe show that the drivers from different countries have a significantly different number of large positive and large negative jerks. Male drivers operate the vehicle with significantly larger number of negative jerks compared to female drivers. The validation of the jerk metrics in identifying aggressive driving is performed by tailgating (following a leading vehicle in a close proximity) and by a violator/non-violator categorization derived from self-reported questionnaires. Our study shows that the identification of aggressive driving could be reinforced by the number of large negative jerks, given that the drivers are tailgating, or by the number of large positive jerks, given that the drivers are categorized as violators. Practical applications: The possibility of understanding, classifying, and quantifying aggressive driving behavior and driving styles with higher risk for accidents can be used for the development of driver support and coaching programs that promote driver safety and are enabled by the vast collection of driving data from modern in-vehicle monitoring and smartphone technology.

Optimizing the safety-efficiency balancing of automated vehicle car-following

This paper proposes an approach to rationally set automated vehicles' car following behavior that explicitly balances between the competing considerations of safety (i.e. small probabilities of a high-consequence crash) and efficiency (guaranteed but small impacts on journey arrival time due to the choice of car following distance). The specification of safety and efficiency are both based on empirically supported concepts and data. In numerical analyses with empirical vehicle trajectories at two sites, we demonstrate intuitive response to systematic variation in numerical values selected as inputs, as well as whether the scope of the efficiency consideration is selfish or systemwide. The proposed balancing is aligned with the standard "Hand Rule" criterion to demonstrate that a duty of care has been met, in which a burden must be borne if it is less than the product of the probability of loss to a third party and the magnitude of loss. Thus the proposed approach is intended to be useful for designers of control algorithms for AVs to establish that they have met their duty of care, taking both safety and efficiency into account.

Structural equations modeling of real-time crash risk variation in car-following incorporating visual perceptual, vehicular, and roadway factors

In this study, we attempted to explain drivers' crash risk variation in car-following for crash avoidance considering the effects of drivers' visual perception, vehicle type, and horizontal curves, with a structural equations model. The model was built by incorporating drivers' speed risk perception and distance risk perception as latent variables. A series of on-road experiments was conducted on the curved segments of a freeway in China to collect naturalistic driving data to approximate the model. The results indicate that (1) the amount of variance in speed risk perception accounted for by the temporal and spatial frequency and the following vehicle type was 21%; (2) the amount of variance in distance risk perception accounted for by the temporal and spatial frequency, leading vehicle type, stopping sight distance (SSD), horizontal sightline offset (HSO), and radius was 29%; and (3) speed risk perception and distance risk perception explained 27% of the total variance in crash risk variation, which is significantly higher than previous similar results that were commonly limited to 10%. The results were explained from the perspective of the effect of line markings, vehicle type (size), and curves on driving behaviors, respectively. In addition, the difference between the effect of speed risk perception and distance perception on crash risk variation was discussed considering the direct and indirect origins of risk in driving. The findings suggests that the incorporation of visual perceptual, vehicular, and roadway factors and its relevant speed risk perception and distance risk perception can better explain the crash risk in car-following. This study also emphasized the possibility and the need of applying the line markings as a visual intervention to prevent the drivers from rear-end crashes on curves, which may provide new insights and be a new solution for roadway safety.

Quantitatively mining and distinguishing situational discomfort grading patterns of drivers from car-following data

Situational discomfort awareness plays an important role in decision making among drivers and has rarely been discussed in detail in previous research. An instrumented vehicle was used to collect car-following data from multiple drivers, thereby quantitatively examining situational discomfort grading patterns using a new discomfort grading method and the latent Dirichlet allocation model. In this process, the gas pedal data and speed difference data are particularly involved in the computation for providing broader meaning to discomfort and building more comprehensive situations. The results show that individual discomfort awareness varies between drivers. More importantly, the potential patterns of situational discomfort grading are extracted, which provides knowledge for characterizing drivers in the context of discomfort awareness. The knowledge achieved can be further applied to distinguish drivers and identify the typical comfort and discomfort zones. This study has great value for promoting investigations on traffic psychology and developing more effective and customized driver assistant systems.

Using naturalistic driving study data to investigate the impact of driver distraction on driver's brake reaction time in freeway rear-end events in car-following situation

INTRODUCTION

The rear-end crash is one of the most common freeway crash types, and driver distraction is often cited as a leading cause of rear-end crashes. Previous research indicates that driver distraction could have negative effects on driving performance, but the specific association between driver distraction and crash risk is still not fully revealed. This study sought to understand the mechanism by which driver distraction, defined as secondary task distraction, could influence crash risk, as indicated by a driver's reaction time, in freeway car-following situations.

METHOD

A statistical analysis, exploring the causal model structure regarding drivers' distraction impacts on reaction times, was conducted. Distraction duration, distraction scenario, and secondary task type were chosen as distraction-related factors. Besides, exogenous factors including weather, visual obstruction, lighting condition, traffic density, and intersection presence and endogenous factors including driver age and gender were considered.

RESULTS

There was an association between driver distraction and reaction time in the sample freeway rear-end events from SHRP 2 NDS database. Distraction duration, the distracted status when a leader braked, and secondary task type were related to reaction time, while all other factors showed no significant effect on reaction time.

CONCLUSIONS

The analysis showed that driver distraction duration is the primary direct cause of the increase in reaction time, with other factors having indirect effects mediated by distraction duration. Longer distraction duration, the distracted status when a leader braked, and engaging in auditory-visual-manual secondary task tended to result in longer reaction times.

PRACTICAL APPLICATIONS

Given drivers will be distracted occasionally, countermeasures which shorten distraction duration or avoid distraction presence while a leader vehicle brakes are worth considering. This study helps better understand the mechanism of freeway rear-end events in car-following situations, and provides a methodology that can be adopted to study the association between driver behavior and driving features.

Chosen risk level during car-following in adverse weather conditions

This study examines how precipitation, light conditions and surface conditions affect the drivers' risk perception. An indicator CRI (Chosen Risk Index) is defined, which describes the chosen risk level for drivers in a car-following situation. The dataset contains about 70 000 observations of driver behaviour and weather status on a rural road. Based on the theory of risk homeostasis and an assumption that driving behaviour in situations with daylight, dry road and no precipitation reflects drivers' target level of risk, generalised linear models (GLM) were estimated for cars and trucks separately to reveal the effect of adverse weather conditions on risk perception. The analyses show that both car and truck drivers perceive the highest risk when driving on snow covered roads. For car drivers, a snow covered road in combination with moderate rain or light snow are the factors which lowers the CRI the most. For trucks, snow cover and partially covered roads significantly lowers the CRI, while precipitation did not seem to impose any higher risk. Interaction effects were found for car drivers only.

Longitudinal control behaviour: Analysis and modelling based on experimental surveys in Italy and the UK

This paper analyses driving behaviour in car-following conditions, based on extensive individual vehicle data collected during experimental field surveys carried out in Italy and the UK. The aim is to contribute to identify simple evidence to be exploited in the ongoing process of driving assistance and automation which, in turn, would reduce rear-end crashes. In particular, identification of differences and similarities in observed car-following behaviours for different samples of drivers could justify common tuning, at a European or worldwide level, of a technological solution aimed at active safety, or, in the event of differences, could suggest the most critical aspects to be taken into account for localisation or customisation of driving assistance solutions. Without intending to be exhaustive, this paper moves one step in this direction. Indeed, driving behaviour and human errors are considered to be among the main crash contributory factors, and a promising approach for safety improvement is the progressive introduction of increasing levels of driving automation in next-generation vehicles, according to the active/preventive safety approach. However, the more advanced the system, the more complex will be the integration in the vehicle, and the interaction with the driver may sometimes become unproductive, or risky, should the driver be removed from the driving control loop. Thus, implementation of these systems will require the interaction of human driving logics with automation logics and then an enhanced ability in modelling drivers' behaviour. This will allow both higher active-safety levels and higher user acceptance to be achieved, thus ensuring that the driver is always in the control loop, even if his/her role is limited to supervising the automatic logic. Currently, the driving mode most targeted by driving assistance systems is longitudinal driving. This is required in various driving conditions, among which car-following assumes key importance because of the huge number of rear-end crashes. The increased availability of lower-cost information and communication technologies (ICTs) has enhanced the possibility of collecting copious and reliable car-following individual vehicle data. In this work, data collected from three different experiments, two carried out in Italy and one in the UK, are analysed and compared. The experiments involved 146 drivers (105 Italian drivers and 41 UK drivers). Data were collected by two instrumented vehicles. Our analysis focused on inter-vehicular spacing in equilibrium car-following conditions. We observed that (i) the adopted equilibrium spacing can be fitted using lognormal distributions, (ii) the adopted equilibrium spacing increases with speed, and (iii) the dispersion between drivers increases with speed. In addition, according to different headway thresholds (up to 1 second) a significant number of potentially dangerous behaviours is observed. Three different car-following paradigms are also applied to each of the experiments, and modelling parameters are calibrated and compared to obtain indirect confirmation about the observed similarities and differences in driving behaviour.

Impact of mobile phone use on car-following behaviour of young drivers

Multitasking, such as the concurrent use of a mobile phone and operating a motor vehicle, is a significant distraction that impairs driving performance and is becoming a leading cause of motor vehicle crashes. This study investigates the impact of mobile phone conversations on car-following behaviour. The CARRS-Q Advanced Driving Simulator was used to test a group of young Australian drivers aged 18-26 years on a car-following task in three randomised phone conditions: baseline (no phone conversation), hands-free and handheld. Repeated measure ANOVA was applied to examine the effect of mobile phone distraction on selected car-following variables such as driving speed, spacing, and time headway. Overall, drivers tended to select slower driving speeds, larger vehicle spacings, and longer time headways when they were engaged in either hands-free or handheld phone conversations, suggesting possible risk compensatory behaviour. In addition, phone conversations while driving influenced car-following behaviour such that variability was increased in driving speeds, vehicle spacings, and acceleration and decelerations. To further investigate car-following behaviour of distracted drivers, driver time headways were modelled using Generalized Estimation Equation (GEE). After controlling for various exogenous factors, the model predicts an increase of 0.33s in time headway when a driver is engaged in hands-free phone conversation and a 0.75s increase for handheld phone conversation. The findings will improve the collective understanding of distraction on driving performance, in particular car following behaviour which is most critical in the determination of rear-end crashes.