Studying crash avoidance maneuvers prior to an impact considering different types of driver’s distractions

Causal mediation analysis of the impacts of distracted driving on crash injury risks

Distracted driving can pose great risks to road traffic safety. Although there is a rich body of literature devoted to identifying the statistical association between distracted driving and crash risks, few are available to examine the causal effect mechanism of distracted driving. Thus, the study attempts to conduct the causal mediation analysis to reveal the impact mechanism of distracted driving on crash injury risks, in which various hazardous driving actions are used as the mediators between driver distraction and crash injuries. Sensitivity analysis is also carried out to validate the underlying assumption of causal mediation analysis. The analytic results indicate that 1) distracted driving can lead to a higher likelihood of hazardous driving actions such as failing to yield, disobeying traffic control devices, driving left of lane center, and failing to stop in assured clear distance, 2) both the driver distraction and hazardous actions are the contributory factors to the severe crash injuries, and 3) distracted driving is identified to have significant mediation effects on crash injury risks. The study confirms the causal mediation effects of distracted driving on crash injury risks, which can serve to propose specific safety countermeasures to mitigate the crash injury risks.

An Extended Car-Following Model Considering the Drivers’ Characteristics under a V2V Communication Environment

In intelligent transportation systems, vehicles can obtain more information, and the interactivity between vehicles can be improved. Therefore, it is necessary to study car-following behavior during the introduction of intelligent traffic information technology. To study the impacts of drivers’ characteristics on the dynamic characteristics of car-following behavior in a vehicle-to-vehicle (V2V) communication environment, we first analyzed the relationship between drivers’ characteristics and the following car’s optimal velocity using vehicle trajectory data via the grey relational analysis method and then presented a new optimal velocity function (OVF). The boundary conditions of the new OVF were analyzed theoretically, and the results showed that the new OVF can better describe drivers’ characteristics than the traditional OVF. Subsequently, we proposed an extended car-following model by combining V2V communication based on the new OVF and previous car-following models. Finally, numerical simulations were carried out to explore the effect of drivers’ characteristics on car-following behavior and fuel economy of vehicles, and the results indicated that the proposed model can improve vehicles’ mobility, safety, fuel consumption, and emissions in different traffic scenarios. In conclusion, the performance of traffic flow was improved by taking drivers’ characteristics into account under the V2V communication situation for car-following theory.

Real-time accident detection: Coping with imbalanced data

Detecting accidents is of great importance since they often impose significant delay and inconvenience to road users. This study compares the performance of two popular machine learning models, Support Vector Machine (SVM) and Probabilistic Neural Network (PNN), to detect the occurrence of accidents on the Eisenhower expressway in Chicago. Accordingly, since the detection of accidents should be as rapid as possible, seven models are trained and tested for each machine learning technique, using traffic condition data from 1 to 7 min after the actual occurrence. The main sources of data used in this study consist of weather condition, accident, and loop detector data. Furthermore, to overcome the problem of imbalanced data (i.e., underrepresentation of accidents in the dataset), the Synthetic Minority Oversampling TEchnique (SMOTE) is used. The results show that although SVM achieves overall higher accuracy, PNN outperforms SVM regarding the Detection Rate (DR) (i.e., percentage of correct accident detections). In addition, while both models perform best at 5 min after the occurrence of accidents, models trained at 3 or 4 min after the occurrence of an accident detect accidents more rapidly while performing reasonably well. Lastly, a sensitivity analysis of PNN for Time-To-Detection (TTD) reveals that the speed difference between upstream and downstream of accidents location is particularly significant to detect the occurrence of accidents.