There is no agreement on the relation between driving parameters and drivers’ subjective states. A linear as well as a threshold relationship for different subjective variables and driving parameters has been put forward. In this study we investigate the relationship between time headway and the ratings of risk, task difficulty, effort, and comfort. Knowledge about this interrelation may advance the development of adaptive cruise control and autonomous driving and can add to the discussion about driver behavior models. An earlier study (Lewis-Evans, De Waard, & Brookhuis, 2010) found a threshold effect for drivers’ ratings of subjective variables for time headways between 0.5 and 4.0 s at a speed of 50 km/h. This study aims to replicate the threshold effect and to expand the findings to time headways at different speeds. A new measure for criticality was added as a categorical variable, indicating the controllability of a driving situation to give indications for the appliance of time headway in adaptive cruise control systems. Participants drove 24 short routes in a driving simulator with predefined speed and time headway to a leading vehicle. Time headway was varied eightfold (0.5–4 s in 0.5 s increments) and speed was varied threefold (50, 100, 150 km/h). A threshold effect for the ratings of risk, task difficulty, effort, and comfort was found for all three different speeds. Criticality proved to be a useful variable in assessing the preferred time headway of drivers.

The continuous motorization of traffic has led to a sustained increase in the global number of road related fatalities and injuries. To counter this, governments are focusing on enforcing safe and law-abiding behavior in traffic. However, especially in developing countries where the motorcycle is the main form of transportation, there is a lack of comprehensive data on the safety-critical behavioral metric of motorcycle helmet use. This lack of data prohibits targeted enforcement and education campaigns which are crucial for injury prevention. Hence, we have developed an algorithm for the automated registration of motorcycle helmet usage from video data, using a deep learning approach. Based on 91,000 annotated frames of video data, collected at multiple observation sites in 7 cities across the country of Myanmar, we trained our algorithm to detect active motorcycles, the number and position of riders on the motorcycle, as well as their helmet use. An analysis of the algorithm's accuracy on an annotated test data set, and a comparison to available human-registered helmet use data reveals a high accuracy of our approach. Our algorithm registers motorcycle helmet use rates with an accuracy of -4.4% and +2.1% in comparison to a human observer, with minimal training for individual observation sites. Without observation site specific training, the accuracy of helmet use detection decreases slightly, depending on a number of factors. Our approach can be implemented in existing roadside traffic surveillance infrastructure and can facilitate targeted data-driven injury prevention campaigns with real-time speed. Implications of the proposed method, as well as measures that can further improve detection accuracy are discussed.

Automated detection of motorcycle helmet use through video surveillance can facilitate efficient education and enforcement campaigns that increase road safety. However, existing detection approaches have a number of shortcomings, such as the inabilities to track individual motorcycles through multiple frames, or to distinguish drivers from passengers in helmet use. Furthermore, datasets used to develop approaches are limited in terms of traffic environments and traffic density variations. In this paper, we propose a CNN-based multi-task learning (MTL) method for identifying and tracking individual motorcycles, and register rider specific helmet use.We further release the HELMET dataset, which includes 91,000 annotated frames of 10,006 individual motorcycles from 12 observation sites in Myanmar. Along with the dataset, we introduce an evaluation metric for helmet use and rider detection accuracy, which can be used as a benchmark for evaluating future detection approaches.We show that the use of MTL for concurrent visual similarity learning and helmet use classification improves the efficiency of our approach compared to earlier studies, allowing a processing speed of more than 8 FPS on consumer hardware, and a weighted average F-measure of 67.3% for detecting the number of riders and helmet use of tracked motorcycles. Our work demonstrates the capability of deep learning as a highly accurate and resource efficient approach to collect critical road safety related data.

Maladaptive driving is an important source of self-inflicted accidents and this driving style could include high speeds, speeding violations, and poor lateral control of the vehicle. The literature suggests that certain groups of drivers, such as novice drivers, males, highly motivated drivers, and those who frequently experience anger in traffic, tend to exhibit more maladaptive driving patterns compared to other drivers. Remarkably, no coherent framework is currently available to describe the relationships and distinct influences of these factors. We conducted two studies with the aim of creating a multivariate model that combines the aforementioned factors, describes their relationships, and predicts driving performance more precisely. The studies employed different techniques to elicit emotion and different tracks designed to explore the driving behaviors of participants in potentially anger-provoking situations. Study 1 induced emotions with short film clips. Study 2 confronted the participants with potentially anger-inducing traffic situations during the simulated drive. In both studies, participants who experienced high levels of anger drove faster and exhibited greater longitudinal and lateral acceleration. Furthermore, multiple linear regressions and path-models revealed that highly motivated male drivers displayed the same behavior independent of their emotional state. The results indicate that anger and specific risk characteristics lead to maladaptive changes in important driving parameters and that drivers with these specific risk factors are prone to experience more anger while driving, which further worsens their driving performance. Driver trainings and anger management courses will profit from these findings because they help to improve the validity of assessments of anger related driving behavior.

In this study the location of vehicle to vehicle distance thresholds for self-reported subjective risk and comfort was researched. Participants were presented with ascending and descending time headway sequences in a driving simulator. This so called method of limits of ascending and descending stimuli (Gouy, Diels, Reed, Stevens, & Burnett, 2012) was refined to efficiently determine individual thresholds for stable time headways with a granularity of 0.1 s. Time headway thresholds were researched for 50, 100, and 150 km/h in a city, rural, and highway setting. Furthermore, thresholds for self-driving (level 0 automation: NHTSA, 2013) were compared with thresholds for the experience of subjective risk and comfort in assisted driving, similar to adaptive cruise control (level 1 automation). Results show that preferred individual time headways vary between subjects. Within subjects however, time headway thresholds do not significantly differ for different speeds. Furthermore we found that there was no significant difference between time headways of self-driving and distance-assisted driving. The relevance of these findings for the development of adaptive cruise control systems, autonomous driving and driver behavior modelling is discussed.