Abstracts Track 2023

Area 1 - Connected Vehicles

Nr: 4

Road Asset Management Considering Connected and Automated Vehicles: An Overview, Opportunities, and Challenges


Adelino L. Ferreira

Abstract: Connected and automated vehicles (CAVs) have the potential to significantly impact transportation systems in terms of mobility, environment, safety, and the economy. These vehicles rely on a range of sensors and cameras to detect road signs and lane markings, as well as to scan their surroundings, and they are connected to other vehicles and infrastructure. Previous research has highlighted the need for asset management processes to manage the intrinsic aspects of CAVs more effectively, intending to improve performance, resource utilization, and budget planning. The main objective of this research study is to understand the impact of CAVs on road asset management. To do so, firstly it is provided an overview of asset and road asset management, followed by an introduction to CAVs. Then it was explored how road design may evolve with the widespread adoption of CAVs, considering the potential improvements and resource savings in road construction. Finally, future research opportunities, challenges, and relevant topics are outlined.

Area 2 - Intelligent Vehicle Technologies

Nr: 5

Predicting Road Traffic Accident Severity in the Eastern Province of Saudi Arabia Using Machine Learning Algorithms


Hussam I. Hijazi and Hassan M. Al-Ahmadi

Abstract: Road traffic safety is a crucial issue that affects millions of people worldwide. Traffic accidents can cause severe injuries, disabilities, and even fatalities, making it one of the leading causes of death globally. In Saudi Arabia, a significant portion of deaths is caused by road traffic accidents (RTA), accounting for 9.19% or 12,317 fatalities in 2020. Eastern Province is one of the most densely populated regions in the country, covering an area of around 672,522 square kilometers and having a population of over five million residents. Due to increased motorization since the 1970s oil boom, there has been a surge in road accidents in the area. This study examined different machine learning algorithms to predict crash severity in the Eastern Province of Saudi Arabia, using data accidents from 2009 to 2021. The data was prepared for modeling by cleaning it, imputing missing values, and grouping it. The geographic and temporal distribution of the data was examined to ensure its reliability. Correlation analysis was used to assess the strength of the relationship between two sets of items, and new features were constructed from existing ones using a feature construction process. To address the imbalance of the dataset, Synthetic Minority Oversampling Technique (SMOTE) was used as an oversampling technique. Data transformation was also employed to create new features using transformation functions on existing ones. The most informative machine learning features were selected using a recursive feature elimination with the cross-validation (RFECV) wrapper technique. The final dataset used in this study contained a total of 31,728 traffic accidents that were deemed valid, with 7,293 fatalities and 24,435 injuries. The study used four common classification algorithms, Logistic Regression (LR), Decision Tree Classifier (DTC), Random Forest Classifier (RFC), and eXtreme Gradient Boosting (XGBoost), which were trained and evaluated. Optimizing the models' performance was enhanced through several preprocessing techniques, hyperparameter tuning with RandomizedSearchCV, and a custom refit strategy. Preprocessing techniques were employed to convert categorical data into numerical data using one of three methods (ordinal encoding, one-hot encoding, and dummy encoding), and scaling numerical data using one of four scaling strategies (without scaling, standardization, normalization, and standardization then normalization). The approach also involved feature selection or using all features. Cross-validation was used to ensure generalization. A custom refit strategy was implemented to avoid overfitting. The performance of the model using the best parameters was evaluated using seven different metrics. The DTC and XGBoost models exhibited the best performance, with the choice of which model to use depending on the desired balance between performance and execution time. For projects requiring high performance and greater computational resources, the XGBoost model may be the optimal choice. The LR or DTC model may be more appropriate if computational resources are limited, and a quick solution is needed. Exploring different preprocessing strategies is worth considering since they can significantly affect the model's performance. Applying these algorithms to predict crash severity in Saudi Arabia's Eastern Province can significantly improve road safety, inform targeted safety strategies, and reduce societal and economic costs of traffic accidents.

Nr: 7

Model-Based Traffic State Estimation for Link Traffic Using Onboard Vehicle Camera


Tanay Rastogi, Michele Simoni and Anders Karlström

Abstract: Traffic State Estimation (TSE) refers to the process of estimating traffic states, namely flow, density and speed in real-time for a specific road network. TSE is a critical process for various applications, including traffic management systems and intelligent transportation systems. Conventional TSE methods have relied on data from either stationary sensors like loop detectors or mobile sensors like GPS-equipped floating car data. However, both methods have limitations. Stationary sensors are sparsely located, making it difficult to obtain a complete picture of the traffic flow, while mobile sensors require prior knowledge about the distribution of vehicles on the road network. To address these limitations, this research article proposes a new approach that uses lightweight cameras, such as dash-cam or in-vehicle embedded cameras, to collect vehicle trajectories from street-view video sequences captured from an onboard camera on a moving vehicle travelling over a link. The proposed method utilizes deep learning-based models like YOLO v5 object detector and StrongSORT multi-object tracker to identify and track vehicles. The distance of the detected vehicle is calculated using GPS metadata from the camera and photogrammetry algorithms, for each time-stamped frame in the video sequence. Based on this approach, it is possible to derive partial vehicle trajectories on the link, which can be aggregated in space-time density cells. A novel TSE method based on Genetic Algorithm (GA) using Cell Transmission Model (CTM) is proposed to derive densities in unobserved cells. We propose an algorithm where first, the parameters of the link triangular Fundamental Diagram (FD) are estimated, and then, employed for predicting the density values for unobserved cells based on simulation. The proposed TSE method is tested on simulated traffic data generated from the SUMO traffic simulator. The test involved 140 different traffic scenarios for a bi-directional link of 100 meters with 1 lane in each direction, with varying traffic speed and density on the link. Preliminary testing of the TSE method’s performance, measured through Root Mean Square Error (RMSE), shows promising results for simulated scenarios. There are ongoing efforts to test the proposed TSE method on street-view video sequences collected from an urban street in Stockholm. The proposed approach represents an opportunity for an alternative TSE method based on trajectories collected from vehicles with onboard camera.

Area 3 - Data Analytics

Nr: 3

2PCube: Artificial Intelligence and Digital Twin Powered Predictive Maintenance Solution for Electric Mobility


Krishna Kumar and Iyimide Shola-Shittu

Abstract: In recent years, there has been significant thrust by industries, R&D organizations and universities worldwide in developing electric powered vehicles (including cars, drones and aircraft) in order to reduce carbon emission. However, their developments and usages suffer from severe issues ranging from battery failures to their overall reliability. The paper examines all these issues and challenges facing electric vehicles, and proposes artificial intelligence and digital twin-powered failure analysis and predictive maintenance solution, 2PCube for electric mobility. This solution involves an innovative autonomous multi-agent framework with a hybrid method wherein the data-driven method is combined with the model-based method. Reinforcement learning is considered along with convolutional neural networks (CNN) layers, long short-term memory (LSTM) and conventional neuron layers for autonomous fault detection, isolation and prognosis of electric vehicle subsystems. The selection of hyperparameters of the hybrid model is automated and optimized via genetic algorithm, particle swarm optimization and differential evolution. Furthermore, predictive maintenance of all subsystems (fault detection, isolation, prognosis and recovery) of electric vehicles with a focus on the propulsion subsystem will be examined to achieve high reliability, sustainability, and longer periods of their operations. Finally, the proposed methodology will be discussed for future electric mobility failure and predictive maintenance.

Area 4 - Smart Mobility and Sustainable Transport Services

Nr: 6

Behavioral and Equity Effects of the Discontinuation of an Exemption from Congestion Charges: Commuter Crossing Time Shifts


Fatemeh Naqavi, Emma Engström and Joel Franklin

Abstract: This paper studies the impact of congestion charges on mobility equity using a natural experiment. Congestion pricing has been introduced in several large cities to mitigate negative externalities from road transport such as congestion, noise, and air pollution, as well as global climate emissions. However, there are concerns that this policy could influence equity of access. In this study, we use a natural experiment in Stockholm, where older low-emission vehicles, previously exempt, began to be subject to congestion pricing in August 2012. We examine data on cordon crossings during the same two-week period in 2012 and 2013, allowing us to study the change in behavior of Green Vehicle (GV) owners as an effect of an ending of an exemption to charges in 2012, and thus to pinpoint the effects of the policy. The data account for the time and date of crossings, and the postal code of the car owners, which we use to compare impacts across income groups. We analyze crossing time data before and after the end of the exemption by graphical studies of histograms and by non-parametric statistical tests to determine differences across the two years. The results indicate that the number of crossings during peak hours decreased in 2013, particularly for low-income earners compared to high-income earners. The findings suggest that congestion charges affect low-income earners more, highlighting the disproportional impacts of the policy on different groups and raising questions about equity.