Search

Abstract Background Motor vehicle crashes remain the leading cause of teen deaths in spite of preventive efforts. Prevention strategies could be advanced through new analytic approaches that allow us to better conceptualize the complex processes underlying teen crash risk. This may help policymakers design appropriate interventions and evaluate their impacts. Methods System Dynamics methodology was used as a new way of representing factors involved in the underlying process of teen crash risk. Systems dynamics modeling is relatively new to public health analytics and is a promising tool to examine relative influence of multiple interacting factors in predicting a health outcome. Dynamics models use explicit statements about the process being studied and depict how the elements within the system interact; this usually leads to discussion and improved insight. A Teen Driver System Model was developed by following an iterative process where causal hypotheses were translated into systems of differential equations. These equations were then simulated to test whether they can reproduce historical teen driving data. The Teen Driver System Model that we developed was calibrated on 47 newly-licensed teen drivers. These teens were recruited and followed over a period of 5-months. A video recording system was used to gather data on their driving events (elevated g-force, near-crash, and crash events) and miles traveled. Results The analysis suggests that natural risky driving improvement curve follows a course of a slow improvement, then a faster improvement, and finally a plateau: that is, an S-shaped decline in driving events. Individual risky driving behavior depends on initial risk and driving exposure. Our analysis also suggests that teen risky driving improvement curve is created endogenously by several feedback mechanisms. A feedback mechanism is a chain of variables interacting with each other in such a way they form a closed path of cause and effect relationships. Conclusions Teen risky driving improvement process is created endogenously by several feedback mechanisms. The model proposed in the present article to reflect this improvement process can spark discussion, which may pinpoint to additional processes that can benefit from further empirical research and result in improved insight.

With the introduction of autonomous and self-driving cars, innovative research is needed to ensure safety and reliability on the road. This work introduces a solution to understand vehicle behaviour based on sensors data. The behaviour is classified according to driving events. Understanding driving events can play a significant role in road safety and estimating the expense and risks of driving a vehicle. Rather than relying on the distance and time driven, driving events can provide a more accurate measure of vehicle driving consumption. This measure will become valuable as more ride-sharing applications are introduced to roads around the world. Estimating driving events can also help better design the road infrastructure to reduce congestion, energy consumption and pollution. By sharing data from official vehicles and volunteers, crowd sensing can be used to better understand congestion and road safety. This work studies driving events and proposes using machine learning to classify these events into different categories. The acquired data is collected using embedded mobile device motion sensors to train machine learning algorithms to classify the events. [ABSTRACT FROM AUTHOR]

With detailed sensor and visual data from automobiles, a data-driven model can learn to classify crash-related events during a drive. We propose a neural network model accepting time-series vehicle sensor data and forward-facing videos as input for learning classification of crash-related events and varying types of such events. To elaborate, a novel recurrent neural network structure is introduced, namely, denoising gated recurrent unit with decay, in order to deal with time-series automobile sensor data with missing value and noises. Our model detects crash and near-crash events based on a large set of time-series data collected from naturalistic driving behavior. Furthermore, the model classifies those events involving pedestrians, a vehicle in front, or a vehicle on either side. The effectiveness of our model is evaluated with more than two thousand 30-s clips from naturalistic driving behavior data. The results show that the model, including sensory encoder with denoising gated recurrent unit with decay, visual encoder, and attention mechanism, outperforms gated recurrent unit with decay, gated CNN, and other baselines not only in event classification and but also in event-type classification. [ABSTRACT FROM AUTHOR]

There is an increasing interest in technology-based solutions that can assist drivers in reducing their risk of involvement in road crashes. Previous studies showed that driving events produced by in-vehicle data recorders (IVDR) are applicable for identification of unsafe driving patterns, while combined examinations of driving events and road infrastructure characteristics are rare. This study explored the relationship between the IVDR-driving events, road characteristics and crashes, to examine a potential of the events for predicting crashes and identification of high-risk locations on the road network. The study database included 3500 segments of the interurban roads in Israel, for which the automatically produced IVDR events were matched with road infrastructure characteristics and crashes. Negative-binomial regression models were adjusted for the relationships between road characteristics and driving events, and subsequently, between events and crashes, given the exposure. Significant impacts were found, yet various event types showed different relations to the infrastructure characteristics and different effects on crashes, on various road types. Better road conditions were associated with a decrease in “braking” events and an increase in the “speed alert” events, where road layout constraints and junction proximity were associated with an opposite effect on events. “Braking” and total events showed better potential for predicting crashes on single-carriageway roads, with a positive link to crashes, where for other road types the “speed alert” events were stronger related to crashes, but with a negative link. The heterogeneity of findings indicates a need in further research of the above relationship, with a particular focus on definitions of driving events produced by the IVDR or other technologies. [ABSTRACT FROM AUTHOR]

Durability is the capacity of a machine to maintain the functionality during its intended service life, under the recommended conditions and the prescribed levels of maintenance. A durable machine has to be designed with a sufficient fatigue resistance that can be accomplished only if its service loads are well known. Tractors are characterized by few large load oscillations and these are of most interest in the evaluation of the fatigue damage accumulated by the structural elements. These oscillations are caused by driving events, defined as any major change in vehicle speed, attitude or operating condition. In this paper, a signal processing methodology to automatically identify the highly damage events for axle housings is described. An agricultural tractor was equipped with four Wheel Force Transducers (WFTs) and a CAN-Bus based data logger in order to acquire axle housing loads and the driver’s operation signals. Signals were acquired for three different applications as ploughing, subsoiling and implement transportation. The loading events were identified through the analysis of the pairwise Time-Varying Correlation Coefficient (TVCC) between the driver’s operation signals and the WFT signals. A driving event was classified as a pattern of TVCCs and for each, the Time-Varying Pseudo-Damage (TVPD) was calculated. Using this methodology, it has been found that the most damaging event for the axle housings occur during the headland turn due to full load accelerations and vertical load transfer between front and rear axles. For these reasons in-field operations were more damaging than off-road transportation. The outlined signal processing methodology should be considered as a first step into the development of a health monitoring methodology for tractors. [ABSTRACT FROM AUTHOR]

Road traffic injuries are a serious concern in emerging economies. Their death toll and economic impact are shocking, with 9 out of 10 deaths occurring in low or middle-income countries; and road traffic crashes representing 3% of their gross domestic product. One way to mitigate these issues is to develop technology to effectively assist the driver, perhaps making him more aware about how her (his) decisions influence safety. Following this idea, in this paper we evaluate computational models that can score the behavior of a driver based on a risky-safety scale. Potential applications of these models include car rental agencies, insurance companies or transportation service providers. In a previous work, we showed that Genetic Programming (GP) was a successful methodology to evolve mathematical functions with the ability to learn how people subjectively score a road trip. The input to this model was a vector of frequencies of risky maneuvers, which were supposed to be detected in a sensor layer. Moreover, GP was shown, even with statistical significance, to be better than six other Machine Learning strategies, including Neural Networks, Support Vector Regression and a Fuzzy Inference system, among others. A pending task, since then, was to evaluate if a more detailed comparison of different strategies based on GP could improve upon the best GP model. In this work, we evaluate, side by side, scoring functions evolved by three different variants of GP. In the end, the results suggest that two of these strategies are very competitive in terms of accuracy and simplicity, both generating models that could be implemented in current technology that seeks to assist the driver in real-world scenarios.

Automatic recognition of driving events, e.g., approaching roundabouts, is important both for the truck design process based on simulated road data and for advanced driver assistance systems. However, the problem faced is extremely challenging as only in-vehicle driving data must be used, whereas the number of driving events is usually quite large. In this paper, we propose a learning-based driving events classification method, which is trained and tested with a real driving events database. The proposed method includes definition of driving events relevant to our final application, selection of discriminating features, and classification, using two machine-learning techniques, namely, decision trees and linear logistic regression. We then introduce the digital road concept. This consists of simulated road data used in the truck design process to quantify the behavior of a truck, particularly in terms of fuel consumption. While a digital road typically contains far less driving information, we show that we can still apply the proposed driving events recognition models learnt on real driving data and pave the way for a more realistic assessment of truck characteristics via simulation tools. [ABSTRACT FROM PUBLISHER]

Motor vehicles greatly influence human life but are also a major cause of death and road congestion, which is an obstacle to future economic development. We believe that by learning driving patterns, useful navigation support can be provided for drivers. In this paper, we present a simple and reliable method for the recognition of driving events using hidden Markov models (HMMs), popular stochastic tools for studying time series data. A data acquisition system was used to collect longitudinal and lateral acceleration and speed data from a real vehicle in a normal driving environment. Data were filtered, normalized, segmented, and quantified to obtain the symbolic representation necessary for use with discrete HMMs. Observation sequences for training and evaluation were manually selected and classified as events of a particular type. An appropriate model size was selected, and the model was trained for each type of driving events. Observation sequences from the training set were evaluated by multiple models, and the highest probability decides what kind of driving event this sequence represents. The recognition results showed that HMMs could recognize driving events very accurately and reliably. [ABSTRACT FROM PUBLISHER]

Background: Motor vehicle crashes remain the leading cause of teen deaths in spite of preventive efforts. Prevention strategies could be advanced through new analytic approaches that allow us to better conceptualize the complex processes underlying teen crash risk. This may help policymakers design appropriate interventions and evaluate their impacts.Methods: System Dynamics methodology was used as a new way of representing factors involved in the underlying process of teen crash risk. Systems dynamics modeling is relatively new to public health analytics and is a promising tool to examine relative influence of multiple interacting factors in predicting a health outcome. Dynamics models use explicit statements about the process being studied and depict how the elements within the system interact; this usually leads to discussion and improved insight. A Teen Driver System Model was developed by following an iterative process where causal hypotheses were translated into systems of differential equations. These equations were then simulated to test whether they can reproduce historical teen driving data. The Teen Driver System Model that we developed was calibrated on 47 newly-licensed teen drivers. These teens were recruited and followed over a period of 5-months. A video recording system was used to gather data on their driving events (elevated g-force, near-crash, and crash events) and miles traveled.Results: The analysis suggests that natural risky driving improvement curve follows a course of a slow improvement, then a faster improvement, and finally a plateau: that is, an S-shaped decline in driving events. Individual risky driving behavior depends on initial risk and driving exposure. Our analysis also suggests that teen risky driving improvement curve is created endogenously by several feedback mechanisms. A feedback mechanism is a chain of variables interacting with each other in such a way they form a closed path of cause and effect relationships.Conclusions: Teen risky driving improvement process is created endogenously by several feedback mechanisms. The model proposed in the present article to reflect this improvement process can spark discussion, which may pinpoint to additional processes that can benefit from further empirical research and result in improved insight. [ABSTRACT FROM AUTHOR]