Maen Qaseem Ghadi

A machine learning framework for predicting the traffic accident severity under class imbalance conditions is predented in Amman, Jordan. The methodology began with a preprocessing pipeline consisting of IQR-based (Interquartile range) outlier removal, label encoding, feature scaling, and Synthetic Minority Over-sampling Technique (SMOTE). Sixteen model configurations were considered, comprising four ensemble learning algorithms (XGBoost, LightGBM, Random Forest, CatBoost) under oversampling, with Optuna-based Bayesian hyperparameter optimisation. The result indicated that the class balancing using SMOTE had a greater influence on the predictive performance than hyperparameter optimisation, specifically for CatBoost model. Feature importance showed that vehicles involved, collision type, and speed are the strongest predictors of accident severity.

Urban Jordan has witnessed substantial traffic growth in recent years. This study examines the long-term and dynamic nexus between crash rates in urban Jordan cities and macroeconomic explanatory variables. Three types of crashes have been studied including, vehicle rollover, vehicle-pedestrian crashes, and vehicle-vehicle crashes against the following explanatory variables; the proportion of the population living in urban areas, GDP growth rate, the total length of road networks, vehicle number growth rate, and spaces of the newly added building. Accordingly, a case study method has been used, covering the timeframe from 2004 to 2022, by employing the autoregressive distributed lag (ARDL) technique. The ARDL estimates showed the presence of long-run cointegration between variables for all types of crashes. Furthermore, the ARDL estimates on the long- and short-run indicate a varying effect of the considered explanatory variables by crash type. The results indicate that the expansion of the urban population and GDP growth rate play a significant role in determining crash rate in urban cities. Moreover, this study urges decision-makers to monitor the impact of expected migration on traffic safety, whether from the countryside or neighbouring countries experiencing crises.

This study adopts an innovative nonlinear autoregressive distributed lag (N-ARDL) approach and compares it with the standard ARDL and the traditional ordinary least-squared (OLS) regressions. These approaches are used to examine the long-run and dynamic relationships between the accident severity in urban Jordan cities and the following explanatory variables: proportion of population living in urban areas, GDP growth rate, total length of road networks, vehicle ownership growth rate and spaces of newly added buildings. The case study covers the period from 2004 to 2022. The standard and nonlinear ARDL estimates showed the presence of long-run co-integration between variables. Moreover, the NARDL estimates on the long- and short-run indicate a varying effect of the considered explanatory variables on accident severity at the positive and negative partial sum. In general, GDP growth positively affects accident severity in the short run. Whereas, the expansion of road length presents a positive and negative impact on the upward and downward partial sum. In contrast, the explanatory variables differed in their impact on accident severity on each side of asymmetry. The variability of the results obtained by the nonlinear ARDL model helps suggest different policies for different high and low levels of traffic safety.

This research studies the impact of a number of climate change features on the severity of traffic accidents in Jordan. This case study is based on monthly data collected from six meteorological stations. The relationship between climate and accident severity is delicate and hard to observe in the short term; therefore, three validation techniques were used: feature selection, correlation testing, and time series modeling. In the first two techniques, Recursive Feature Elimination with Random Forest and a correlation matrix were employed to select the features with the greatest influence on the target variable. Applying the feature selection algorithm highlighted rainfall as the feature with the greatest influence on the target variable, followed by CO₂ emissions and average temperatures. For a more precise explanation, the selected features were used as inputs for the Autoregressive Distributed Lag (ARDL) approach. Unlike conventional linear models, ARDL can handle a non-linear time series dataset, which is fit for our case study. The developed model has confirmed the positive impact of average rainfalls and mean temperatures on accident risk in the long run and CO₂ emissions in the short run. This finding highlights the importance of implementing adaptive strategies and policies to strengthen resilience against the growing effects of climate change on road safety in Jordan.

With a full fleet of autonomous vehicles (AV) in the future, the concept of signal-free intersections has attracted transportation researchers. Multiple studies have investigated different methods and protocols to facilitate vehicle decision-making at intersections. Most of these methods have relied on strong centralized or inter-vehicle communications. This research aims to develop a new protocol for managing vehicle movement at four-leg intersections assuming a fully automated fleet. The main concept of the proposed protocol is to maintain a continuous movement of vehicles entering the intersection, without stopping in queues, by controlling their sequence of movements. In the methodology a dynamic occupancy grid (DOG) approach is applied by initially dividing the intersection into dynamic grids (i.e. cells). The cells are in a virtual movement, emanating away from each lane-group without overlapping, like a gear machine. Each vehicle sits on a specific cell to traverse the intersection safely at a predetermined time. In other words, vehicles approaching an intersection must register their speed and position by certain sensors, and in return receive the appropriate acceleration and speed to finally be allocated to the suitable moving grids. The efficiency of the applied protocol was demonstrated by a practical example that presented a higher intersection capacity value exceeding the ideal saturation flow rate in some cases. This reflects the efficiency of the implemented protocol and its applicability to different low to high traffic flows. Moreover, the protocol shows more flexibility in dealing with different weather, geometric and traffic conditions.

This research aims to identify and interpret the main traffic characteristics affecting the pavement condition index (PCI) of asphalt within parking areas. The study also examines the effect of severity and type of asphalt pavement distress on parking decisions. Previous literature examines the impact of the environment on asphalt pavement deterioration. This paper will focus on the shape of traffic operations within a parking area. The methodology includes applying a multilevel mixed regression model to reveal the hidden hierarchical relationship between traffic activities and the different types of pavement distresses and then discovering their effect on asphalt PCI. The resulting models showed that parking lots are aging faster than the common traveling roads due to the increased impact of axial loads on their surfaces at low and static speeds and some distinctive traffic characteristics. Moreover, some types of pavement distress, such as alligator cracks, potholes and depressions, appear more widely than others on the parking lot surface and significantly influence the parking decision. The conclusion of the research may significantly impact pavement design and maintenance management for parking lots.

Background: Encrypting functions of vehicle internal networks makes the lives of third parties more difficult since, in most cases, the meaning of messages carried by the CAN bus is not disclosed Objectives: This paper proposes a reverse engineering method to discover, to a large extent, the semantics of CAN messages in a vehicle internal network. Methods: A filtering mechanism has been applied that includes several statistical processes to interpret the codes of CAN messages. The speed change function of a vehicle has been chosen as an example to be followed in the development steps of this approach to predict the motion mechanism of the vehicle. The selected codes were verified by developing a multilevel model that relates the hierarchical relationship between the bytes and IDs and their impact on the speed factor. Results: The most influential IDs and bytes on vehicle speed functions were: ID 512, ID 520, ID 664, B2, B4, and B6, respectively. Conclusion: The selected codes used to model the observed speed do not mean they all share the speed function, but there is a good possibility that at least some fulfill this function. However, with some optimization, the same methodology can be applied to detect other semantic messages in the CAN network based on the expected data type.