Eva Michelaraki
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Eva Michelaraki has a total of 50 co-authors. Here are the top co-authors by number of shared publications:
- George Yannis
National Technical Univ. of Athens
14 shared publications - Christos Katrakazas
Department of Transportation Planning and Engineering, National Technical University of Athens, 5 Iroon Polytechniou Str., 157 73 Athens, Greece
5 shared publications George Yannis
National Technical University of Athens, Department of Transportation Planning and Engineering, 5 Heroon Polytechniou str., Athens 15773, Greece
5 shared publications- Thodoris Garefalakis
Department of Transportation Planning and Engineering, National Technical University of Athens, 5 Iroon Polytechniou Str., 157 73 Athens, Greece
5 shared publications Christos Katrakazas
4 shared publications- Tom Brijs
Transportation Research Institute (IMOB), UHasselt, Martelarenlaan 42, 3500 Hasselt, Belgium
4 shared publications
Latest Publications
- Open Access
- 31 October 2026
- Published by Springer Nature
- p. 549-557
Structural Equation Model Analysis for the Identification of the Safety Tolerance Zone
Task complexity refers to the dynamic road and environmental conditions influencing driving, while coping capacity reflects a driver’s ability to manage these challenges. The concept of the Safety Tolerance Zone (STZ) describes the point at which self-regulated control is considered safe, where the demands of the driving task are balanced with the driver’s ability to cope. This study endeavours to quantify the relationship between task complexity and coping capacity in identifying the STZ. Towards that end, a driving simulator experiment involving 55 participants was conducted and the most reliable indicators were collected and analysed. Generalized Linear Models were performed to examine the impact of road, vehicle and driver characteristics on crash risk, while Structural Equation Models were applied to assess the relationship between latent and observable variables of task complexity, coping capacity and crash risk. Results indicated that increased task complexity significantly elevated crash risk, while greater coping capacity mitigated it. It was also revealed that task complexity and coping capacity were inter-related with a positive correlation, implying that drivers coping capacity increases as the complexity of driving task increases. Findings highlighted the importance of adaptive driver support systems in maintaining drivers within the STZ under varying road conditions.
- 31 October 2026
- Published by Springer Nature
- p. 549-557
Open AccessStructural Equation Model Analysis for the Identification of the Safety Tolerance Zone
Task complexity refers to the dynamic road and environmental conditions influencing driving, while coping capacity reflects a driver’s ability to manage these challenges. The concept of the Safety Tolerance Zone (STZ) describes the point at which self-regulated control is considered safe, where the demands of the driving task are balanced with the driver’s ability to cope. This study endeavours to quantify the relationship between task complexity and coping capacity in identifying the STZ. Towards that end, a driving simulator experiment involving 55 participants was conducted and the most reliable indicators were collected and analysed. Generalized Linear Models were performed to examine the impact of road, vehicle and driver characteristics on crash risk, while Structural Equation Models were applied to assess the relationship between latent and observable variables of task complexity, coping capacity and crash risk. Results indicated that increased task complexity significantly elevated crash risk, while greater coping capacity mitigated it. It was also revealed that task complexity and coping capacity were inter-related with a positive correlation, implying that drivers coping capacity increases as the complexity of driving task increases. Findings highlighted the importance of adaptive driver support systems in maintaining drivers within the STZ under varying road conditions.
- Open Access
- 31 October 2026
- Published by Springer Nature
- p. 585-593
Assessing Traffic Safety of Vehicle-Pedestrian Interaction at Intersections, Mid-Block Crossings and Jaywalking
Pedestrians consist a large part of vulnerable road users in transportation networks with unpredictable behaviour and different motion profiles and patterns. Their interaction with vehicular traffic is a major pillar in the field of transportation and it is vital to maintain high safety levels in any type of interaction due to pedestrians’ high vulnerability. Numerous studies have dealt with these issues and been focused on evaluating traffic safety levels at the encounter points using different indicators and surrogate safety measures (SSMs). Scope of this work is to present the findings of an extensive literature review summarizing the different indicators used for assessing the risk levels of these interactions. The review includes interactions at intersections, mid-block crosswalks and jaywalking using real time collected data, virtual experiments and driving simulators. Apart from these issues, the review revealed critical values of these indicators for categorizing the safety levels of the interactions as well for identifying the thresholds the pedestrians perceive for deciding to cross the road segment. The results showed that time-, distance- and speed based indicators are the main categories of parameters used for describing interactions and assessing their safety levels.
- 31 October 2026
- Published by Springer Nature
- p. 585-593
Open AccessAssessing Traffic Safety of Vehicle-Pedestrian Interaction at Intersections, Mid-Block Crossings and Jaywalking
Pedestrians consist a large part of vulnerable road users in transportation networks with unpredictable behaviour and different motion profiles and patterns. Their interaction with vehicular traffic is a major pillar in the field of transportation and it is vital to maintain high safety levels in any type of interaction due to pedestrians’ high vulnerability. Numerous studies have dealt with these issues and been focused on evaluating traffic safety levels at the encounter points using different indicators and surrogate safety measures (SSMs). Scope of this work is to present the findings of an extensive literature review summarizing the different indicators used for assessing the risk levels of these interactions. The review includes interactions at intersections, mid-block crosswalks and jaywalking using real time collected data, virtual experiments and driving simulators. Apart from these issues, the review revealed critical values of these indicators for categorizing the safety levels of the interactions as well for identifying the thresholds the pedestrians perceive for deciding to cross the road segment. The results showed that time-, distance- and speed based indicators are the main categories of parameters used for describing interactions and assessing their safety levels.
- Open Access
- 31 October 2026
- Published by Springer Nature
- p. 558-566
Examination of Driving Behaviour Patterns Among Last-Mile Delivery Riders in Greece
Road crashes remain a major challenge, with vulnerable road users comprising more than half of all traffic-related fatalities and injuries. Among these groups, last-mile delivery professionals are at particularly high risk due to their constant presence on urban streets and frequent exposure to hazardous driving conditions. This study examined the driving behaviour patterns among last-mile delivery riders in Greece. Towards that end, a stated preference survey was conducted with a sample of 100 riders aged 18 to 65. A binary logit discrete choice model was developed to evaluate the relative influence of various factors in different hypothetical scenarios. The results revealed that driving behaviour, trip time and salary type were significant determinants of participants' choices. Respondents expressed a strong preference for flat salaries, emphasizing the value of stable income over performance-based compensation. Policymakers should consider hybrid salary structures combining flat pay with performance incentives to balance stability and productivity.
- 31 October 2026
- Published by Springer Nature
- p. 558-566
Open AccessExamination of Driving Behaviour Patterns Among Last-Mile Delivery Riders in Greece
Road crashes remain a major challenge, with vulnerable road users comprising more than half of all traffic-related fatalities and injuries. Among these groups, last-mile delivery professionals are at particularly high risk due to their constant presence on urban streets and frequent exposure to hazardous driving conditions. This study examined the driving behaviour patterns among last-mile delivery riders in Greece. Towards that end, a stated preference survey was conducted with a sample of 100 riders aged 18 to 65. A binary logit discrete choice model was developed to evaluate the relative influence of various factors in different hypothetical scenarios. The results revealed that driving behaviour, trip time and salary type were significant determinants of participants' choices. Respondents expressed a strong preference for flat salaries, emphasizing the value of stable income over performance-based compensation. Policymakers should consider hybrid salary structures combining flat pay with performance incentives to balance stability and productivity.
- Open Access
- 29 October 2026
- Published by Springer Nature
- p. 983-991
Investigation of Users’ Preferences on Smart Public Services and Green Mobility Hubs
Mobility hubs integrate shared and active mobility into existing public transport systems. De-spite an extensive public transportation network, private vehicle dependency remains high. The aim of this study was to investigate users’ preferences on smart public services and green mobility hubs. To achieve this objective, a stated choice survey was developed and distributed in a sample of 152 participants in two stations, Voula and Irakleio that correspond to different public transport modes, tram and metro. Results indicated a generally positive attitude towards the development of smart and green mobility hubs. The survey revealed variations in commuter preferences between the two locations, with green spaces and smart charging benches being highly valued. Well-designed mobility hubs can enhance sustainability, efficiency and livability in urban areas. The proposed interventions aimed to promote sustainable urban mobility, reduce environmental impact and improve quality of life in the studied areas and beyond.
- 29 October 2026
- Published by Springer Nature
- p. 983-991
Open AccessInvestigation of Users’ Preferences on Smart Public Services and Green Mobility Hubs
Mobility hubs integrate shared and active mobility into existing public transport systems. De-spite an extensive public transportation network, private vehicle dependency remains high. The aim of this study was to investigate users’ preferences on smart public services and green mobility hubs. To achieve this objective, a stated choice survey was developed and distributed in a sample of 152 participants in two stations, Voula and Irakleio that correspond to different public transport modes, tram and metro. Results indicated a generally positive attitude towards the development of smart and green mobility hubs. The survey revealed variations in commuter preferences between the two locations, with green spaces and smart charging benches being highly valued. Well-designed mobility hubs can enhance sustainability, efficiency and livability in urban areas. The proposed interventions aimed to promote sustainable urban mobility, reduce environmental impact and improve quality of life in the studied areas and beyond.
- Open Access
- 1 August 2026
- Published by MDPI AG in Infrastructures
- Vol. 11 (8) , 266
Evaluating Risky Driving Behavior Using a Naturalistic Driving Dataset: A Hybrid Modelling Approach
Driver behavior is a critical factor in road safety, contributing to the majority of traffic crashes. The i-DREAMS project introduced the concept of a Safety Tolerance Zone (STZ) to enhance driving safety through real-time and post-trip interventions. This study develops and evaluates three hybrid machine learning models—(i) Deep Neural Network–Random Forest (DNN-RF), (ii) Convolutional Neural Network–Long Short-Term Memory (CNN-LSTM), and (iii) Recurrent Neural Network–AdaBoost (RNN-AdaBoost)—to classify risky driving behavior into three safety levels using naturalistic driving data from Belgium and the UK. The dataset includes 69 drivers, 15,389 trips, and 265,512 min of driving data. Among the models tested, the DNN-RF model demonstrated the highest accuracy, reaching 98% in Belgium and 97% in the United Kingdom, outperforming other approaches. Feature importance analysis identified harsh acceleration and braking as the most critical factors in Belgium, while total trip distance and harsh acceleration were predominant in the UK. To enhance model transparency, we applied the Local Interpretable Model-agnostic Explanations (LIME) algorithm, providing valuable insights into model predictions. The findings support the potential of hybrid deep learning models in improving road safety by accurately detecting risky driving behaviors. These insights can inform targeted interventions and driver assistance technologies to mitigate crash risks and promote safer driving practices.
- 1 August 2026
- Published by MDPI AG in Infrastructures
- Vol. 11 (8) , 266
Open AccessEvaluating Risky Driving Behavior Using a Naturalistic Driving Dataset: A Hybrid Modelling Approach
Driver behavior is a critical factor in road safety, contributing to the majority of traffic crashes. The i-DREAMS project introduced the concept of a Safety Tolerance Zone (STZ) to enhance driving safety through real-time and post-trip interventions. This study develops and evaluates three hybrid machine learning models—(i) Deep Neural Network–Random Forest (DNN-RF), (ii) Convolutional Neural Network–Long Short-Term Memory (CNN-LSTM), and (iii) Recurrent Neural Network–AdaBoost (RNN-AdaBoost)—to classify risky driving behavior into three safety levels using naturalistic driving data from Belgium and the UK. The dataset includes 69 drivers, 15,389 trips, and 265,512 min of driving data. Among the models tested, the DNN-RF model demonstrated the highest accuracy, reaching 98% in Belgium and 97% in the United Kingdom, outperforming other approaches. Feature importance analysis identified harsh acceleration and braking as the most critical factors in Belgium, while total trip distance and harsh acceleration were predominant in the UK. To enhance model transparency, we applied the Local Interpretable Model-agnostic Explanations (LIME) algorithm, providing valuable insights into model predictions. The findings support the potential of hybrid deep learning models in improving road safety by accurately detecting risky driving behaviors. These insights can inform targeted interventions and driver assistance technologies to mitigate crash risks and promote safer driving practices.
- Open Access
- 26 July 2026
- Published by Elsevier in Accident Analysis & Prevention
- Vol. 236, 108684
Implementation of an integrated model for understanding the impact of task complexity and coping capacity on crash risk
- 26 July 2026
- Published by Elsevier in Accident Analysis & Prevention
- Vol. 236, 108684
Open AccessImplementation of an integrated model for understanding the impact of task complexity and coping capacity on crash risk
- Open Access
Emerging Solutions and Technologies for Smart Mobility and Vehicle Safety in Transportation
The rapid evolution of transportation technologies and the growing integration of artificial intelligence (AI) are transforming the landscape of road safety and smart mobility
Open AccessEmerging Solutions and Technologies for Smart Mobility and Vehicle Safety in Transportation
The rapid evolution of transportation technologies and the growing integration of artificial intelligence (AI) are transforming the landscape of road safety and smart mobility
- Open Access
- 10 December 2025
- Published by MDPI AG in Infrastructures
- Vol. 10 (12) , 342
Comparative Analysis of Stated Preference Data for Identifying Driving Behaviour Patterns of Last-Mile Delivery Professionals
The role of last-mile delivery professionals is becoming increasingly vital in modern urban logistics, driven by the rapid expansion of e-commerce and rising consumer expectations for fast and reliable services. This study aimed to analyse the decision-making patterns of last-mile delivery professionals through stated preference data. To achieve this, a stated-preference questionnaire was conducted with 333 riders aged 18–65 from Croatia, Cyprus, Greece, Italy and Slovenia. A random parameter logit (RPL) model was applied to evaluate the influence of factors such as driving behaviour, delivery time and salary type on decision-making in hypothetical scenarios. Results showed that driving behaviour, trip duration and salary type significantly affected respondents’ preferences. Participants displayed a strong preference for flat salaries, indicating the importance of income stability over performance-based pay. Driving behaviour was also crucial, as respondents favoured legal and safe practices. Interestingly, while shorter delivery times were generally preferred, several scenarios revealed a tolerance for longer durations, possibly reflecting perceived benefits such as safer routes or reduced stress. Comparative analyses also revealed regional differences in vehicle use, work patterns and safety perceptions. The study highlights the need for tailored training programs on safety compliance, route optimization and time management, alongside hybrid salary structures balancing stability and productivity.
- 10 December 2025
- Published by MDPI AG in Infrastructures
- Vol. 10 (12) , 342
Open AccessComparative Analysis of Stated Preference Data for Identifying Driving Behaviour Patterns of Last-Mile Delivery Professionals
The role of last-mile delivery professionals is becoming increasingly vital in modern urban logistics, driven by the rapid expansion of e-commerce and rising consumer expectations for fast and reliable services. This study aimed to analyse the decision-making patterns of last-mile delivery professionals through stated preference data. To achieve this, a stated-preference questionnaire was conducted with 333 riders aged 18–65 from Croatia, Cyprus, Greece, Italy and Slovenia. A random parameter logit (RPL) model was applied to evaluate the influence of factors such as driving behaviour, delivery time and salary type on decision-making in hypothetical scenarios. Results showed that driving behaviour, trip duration and salary type significantly affected respondents’ preferences. Participants displayed a strong preference for flat salaries, indicating the importance of income stability over performance-based pay. Driving behaviour was also crucial, as respondents favoured legal and safe practices. Interestingly, while shorter delivery times were generally preferred, several scenarios revealed a tolerance for longer durations, possibly reflecting perceived benefits such as safer routes or reduced stress. Comparative analyses also revealed regional differences in vehicle use, work patterns and safety perceptions. The study highlights the need for tailored training programs on safety compliance, route optimization and time management, alongside hybrid salary structures balancing stability and productivity.
- Open Access
- 3 September 2025
- Published by Springer Nature
- p. 147-161
Investigating the Effect of Driver-Vehicle-Environment Interaction with Risk Through Naturalistic Driving Data
While mobility and safety of drivers are challenged by behavioral changes, the increasingly complex road environment has placed a higher demand on their adaptability. The ultimate goal of this paper was to identify the impact that the balance between task complexity and coping capacity had on crash risk. Towards that aim, an integrated model for understanding the effect of the inter-relationship of task complexity and coping capacity with risk was developed. A vast library of data from a naturalistic driving experiment was created in three countries (i.e., Belgium, UK and Germany) to investigate the most prominent driving behavior indicators available, including speeding, headway, overtaking, duration, distance and harsh events. In order to fulfil the aforementioned objectives, exploratory analysis, such as Generalized Linear Models (GLMs) were developed, and the most appropriate variables associated to the latent variable “task complexity” and “coping capacity” were estimated from the various indicators. Additionally, Structural Equation Models (SEMs) were used to explore how the model variables were inter-related, allowing for both direct and indirect relationships to be modelled. The analyses revealed that higher task complexity levels lead to higher coping capacity by drivers. Additionally, the effect of task complexity on risk was greater than the impact of coping capacity in Belgium and Germany, while mixed results were observed in the UK.
- 3 September 2025
- Published by Springer Nature
- p. 147-161
Open AccessInvestigating the Effect of Driver-Vehicle-Environment Interaction with Risk Through Naturalistic Driving Data
While mobility and safety of drivers are challenged by behavioral changes, the increasingly complex road environment has placed a higher demand on their adaptability. The ultimate goal of this paper was to identify the impact that the balance between task complexity and coping capacity had on crash risk. Towards that aim, an integrated model for understanding the effect of the inter-relationship of task complexity and coping capacity with risk was developed. A vast library of data from a naturalistic driving experiment was created in three countries (i.e., Belgium, UK and Germany) to investigate the most prominent driving behavior indicators available, including speeding, headway, overtaking, duration, distance and harsh events. In order to fulfil the aforementioned objectives, exploratory analysis, such as Generalized Linear Models (GLMs) were developed, and the most appropriate variables associated to the latent variable “task complexity” and “coping capacity” were estimated from the various indicators. Additionally, Structural Equation Models (SEMs) were used to explore how the model variables were inter-related, allowing for both direct and indirect relationships to be modelled. The analyses revealed that higher task complexity levels lead to higher coping capacity by drivers. Additionally, the effect of task complexity on risk was greater than the impact of coping capacity in Belgium and Germany, while mixed results were observed in the UK.
- Open Access
- 3 September 2025
- Published by Springer Nature
- p. 215-221
Exploitation of Naturalistic Driving Data to Estimate Crash Risk Through Machine Learning Techniques
Despite continuous investment in road and vehicle safety, as well as improvements in technology standards, the total amount of road crashes has been increasing over the last decades. The aim of the current study was to identify the most critical precursors of risk from both the task complexity and the coping capacity side. To that end, data collected from 30 drivers who participated in a naturalistic driving experiment for four months in UK were collected and analyzed. Several Structural Equation Models (SEMs) were applied in order to explore the effect between the latent variables of task complexity, coping capacity and risk. Results indicated that demographic characteristics, such as gender and age had a negative correlation with coping capacity indicating that male drivers and especially elderly people had a lower level of coping capacity, Moreover, better general driving skills were associated with higher coping capacity. It was also revealed that vehicle strain (increased vehicle age) along with type of fuel and trip difficulty were associated with higher task complexity levels. Overall, it is important to consider the specific factors and context involved when assessing the relationship between task complexity, coping capacity and risk.
- 3 September 2025
- Published by Springer Nature
- p. 215-221
Open AccessExploitation of Naturalistic Driving Data to Estimate Crash Risk Through Machine Learning Techniques
Despite continuous investment in road and vehicle safety, as well as improvements in technology standards, the total amount of road crashes has been increasing over the last decades. The aim of the current study was to identify the most critical precursors of risk from both the task complexity and the coping capacity side. To that end, data collected from 30 drivers who participated in a naturalistic driving experiment for four months in UK were collected and analyzed. Several Structural Equation Models (SEMs) were applied in order to explore the effect between the latent variables of task complexity, coping capacity and risk. Results indicated that demographic characteristics, such as gender and age had a negative correlation with coping capacity indicating that male drivers and especially elderly people had a lower level of coping capacity, Moreover, better general driving skills were associated with higher coping capacity. It was also revealed that vehicle strain (increased vehicle age) along with type of fuel and trip difficulty were associated with higher task complexity levels. Overall, it is important to consider the specific factors and context involved when assessing the relationship between task complexity, coping capacity and risk.