Emerging Cutting-Edge Technologies and Applications for Safer, Sustainable, and Intelligent Road Systems in Smart Cities: A Review

Abstract

This review paper explores the role of artificial intelligence (AI)-driven technologies in transforming road transportation systems within smart cities. Adopting a granular approach to the selected research, it examines the extent to which these technologies contribute to creating intelligent road networks, beginning with their integration into the conceptualization and design of road space. Through a comprehensive review of recently published indexed articles, the study addresses key questions regarding AI’s contribution to smart road systems and their ability to adapt during the transition toward sustainable, technology-enabled urban environments. Additionally, it investigates the boundaries between relevant disciplines, areas of overlap and integration, and the benefits of interdisciplinary dialogue in developing effective AI-driven approaches for the design, implementation, and management of smart urban road systems. The findings aim to guide future research, policymaking, and practical applications, ultimately enhancing urban mobility, quality of life, and user experience within smart city contexts. The scope of this research encompasses a wide range of stakeholders involved in transportation and related fields, fostering a multidisciplinary perspective on sustainable urban mobility.

1. Introduction

Roads are vital to urban design, shaping spaces through movement, connectivity, and accessibility. Their patterns influence economic, environmental, and street–quality outcomes. Organizing roads hierarchically helps efficiently manage various types of traffic, from local streets to major arterial roads [1]. Thoughtful road design, especially at intersections, should facilitate smooth and safe transitions between roads of similar or adjacent hierarchy levels, thereby streamlining traffic flow [2,3]. As traffic volume increases and vehicle speeds decline, urban planning and road design should adopt new assessment methods and tools to effectively address safety and infrastructure capacity challenges, adapting to the evolving needs of cities [4]. In the smart era, advanced management systems and technologies are essential for promoting sustainable transportation and developing safer, more efficient road networks [5,6]. On freeways, technology enables autonomous, high-speed travel, while in urban areas, it helps manage dense traffic, complex situations, and diverse mobility needs to support inclusive urban development [6,7]. Consequently, cities should implement essential tools and devices to address evolving traffic demands, building safer, resilient, interconnected, and digitally engaged communities that boost urban mobility, reduce congestion, and improve the overall user experience [5,8]. In this context, artificial intelligence (AI)-driven technological innovation can support the shift toward an intelligent era by enhancing urban planning and road design [8,9]. Using machine learning (ML) approaches—including deep learning (DL) and reinforcement learning (RL)—AI analyzes sensor and vehicle data to recognize patterns and predict outcomes, enabling real-time decisions related to congestion, safety, inefficiency, and sustainability [8,9,10]. While AI-driven analytics can forecast mobility challenges and suggest solutions, they also face hurdles such as computational complexity, data privacy, data quality, system integration, and ensuring model interpretability and reliability [11,12]. Integrating AI and Internet-of-Things (IoT) into road infrastructure is crucial for adapting to changing conditions and promoting sustainable transportation in smart cities [13]. These technologies enable intelligent, connected ecosystems that improve efficiency, safety, and quality of life [14,15]. However, challenges remain regarding successful integration, evaluation methods, and the readiness of key deployment components within existing infrastructure.

As society moves into a fully smart era, AI-driven technologies are increasingly enhancing road infrastructure and transportation management by improving data collection, communication, resilience to disruptions, and supporting public decision-making through contextual comparisons [5,14]. However, while pioneers and AI awareness campaigns are broadening understanding of AI’s applications across various fields [16,17], many outside the field perceive AI as complex and cryptic, leading to limited awareness of its impacts despite growing media coverage [18]. This highlights the need for improved public education and awareness to enable informed discussions on new ways of traveling, address concerns related to modern mobility, and support better decision-making [5,19]. Effective governance requires clear planning, integrated control centers with AI services, and specialized personnel [17]. Additionally, developing scenario analysis and innovative frameworks is vital for creating sustainable digital mobility solutions that incorporate emerging technologies and AI into modern road network planning and design [5].

Based on this premise, it is reasonable to question whether and to what extent the application of advanced AI-driven technologies can enhance the quality of life for road users and urban communities, as well as improve urban living conditions and mobility experiences. Therefore, the main objective of this review paper is to deepen the understanding of the implementation of emerging, cutting-edge AI-driven technologies in road transportation systems, particularly within the context of smart cities. Additionally, it aims to highlight key issues that require further investigation or attention through applied research and case studies. Key questions (KQs) have been developed as research objectives to explore emerging technologies, review recent advancements, and assess their potential for practical implementation in the road sector to improve transportation systems and foster innovation-driven development, as follows:

• KQ1: What is the role of AI-driven technologies, and to what extent do they contribute to making the road transportation system intelligent?
• KQ2: What are the boundaries between the disciplines of potential interest, the margins of overlap and integration, and the benefits of dialogue among them in effectively contributing to the development of an AI-driven approach suitable for the design, contextualization, construction, and management of a smart road system in cities?
• KQ3: Is it possible to assess the adaptive capacity of the road system and its components to handle changes during the transition to a technologically sustainable urban environment?

From this perspective, the scope of the research aims to include a broad range of stakeholders involved in transportation and related fields. Although inspired by established review procedures [20], the authors conducted a literature review through a structured paper selection process focused on exploring the implications and challenges of deploying cutting-edge AI-driven innovations, such as intelligent infrastructure, real-time traffic management and autonomous driving, to support development in road transportation systems and smart cities.

This review helped establish a multidisciplinary background, answer key questions, and analyze themes within the collected texts. To filter texts eligible for analysis, the Scopus database [21] was used to identify scientific articles aligned with the scope of this paper. Based on Google trends [22], as presented in Section 2, key terms such as “Road Transportation System,” “Smart Cities,” and “Artificial Intelligence” were combined to retrieve relevant documents from January 2013 to June 2025, covering a decade of research. These terms helped frame the KQs, classify papers, and support conceptual discussions. This approach effectively defined the joint research topic and ensured a broad time span to investigate innovative aspects. The search was limited to articles published in Scopus-indexed journals within the specified period, encompassing all subject areas available in the database.

Following this initial phase, a semantic analysis was conducted to interpret the selected articles [23]. This involved understanding research objectives, identifying key concepts, evaluating models and solutions through case studies, and assessing their performance. Based on content and relationships among papers, inclusion criteria were established to group studies for the literature review. Extracting information helped identify key themes (KTs), sub-themes (KSTs), and categorize the studies. While not exhaustive, this foundation facilitated the identification of potential gaps in the current literature. The contributions of this paper include developing a comprehensive literature review that integrates insights from various fields, deepening understanding at the intersection of road transportation, smart cities, and AI. It aims to identify interdisciplinary gaps to enhance integration and support the digital transition in road and network design, ultimately improving citizens’ quality of life. The study also highlights key achievements and lessons learned in implementing AI in transportation, providing valuable insights for practitioners and researchers to inform future projects and advance effective technology deployment and infrastructure development in smart cities.

The paper is structured with Section 2 including the materials, methods, and semantic analysis used to identify key themes and categorize studies, while Section 3 discusses the role of a multidisciplinary literature review along with major findings. The discussion is presented in Section 4, and the article concludes with Section 5.

2. Materials and Methods for Literature Review

This section outlines the literature review methodology, selecting Scopus publications from 2013 to mid-2025 using key terms such as “Road Transportation System,” “Smart Cities,” and “Artificial Intelligence”. A hypothesis-driven semantic analysis was conducted to identify themes, categorize research, and detect patterns, emphasizing key findings, gaps, and future research directions.

Building a Framework for a Multidisciplinary Knowledge Base

In line with the paper’s scope, Google search trends shown in Figure 1, covering January 2004 to 17 June 2025, indicate significant growth in topics such as “Road Transportation System,” “Smart Cities,” and “Artificial Intelligence” [22].

Overall, interest in these areas has steadily grown, with recent years showing a marked increase. While isolated peaks occurred around specific innovations, the trends reflect a consistent rise in focus on these fields over the past decade. The authors chose these three terms to help define the joint research topic while covering a broad time span to assess innovative elements within the study’s context. To ensure an accurate and manageable selection of sources, maintain consistency in the research method, and facilitate comparison between studies, the Scopus database was used to filter and search for articles aligned with the review’s specific focus [21]. The ‘article title, abstract, keywords’ function in Scopus was initially used to conduct targeted searches for documents from 2013 to 17 June 2025, focusing on combinations of “Road Transportation System” and “Artificial Intelligence,” as well as “Smart Cities” and “Artificial Intelligence.” This approach identified literature exploring AI’s relationship with these areas, highlighting key trends, innovations, and challenges. Figure 2 shows studies based on these key term pairs, with peer-reviewed documents, including journal articles, reported separately. Specifically, Figure 2a shows 1566 peer-reviewed documents and 590 journal articles related to “Artificial Intelligence” and “Road Transportation Systems”, while Figure 2b shows 5797 documents and 1828 articles for “Artificial Intelligence” and “Smart Cities.” These pairings enabled targeted analysis of shared themes and issues across the research domains.

A subsequent search using ‘article title, abstract, and keywords’ in Scopus for “Artificial Intelligence,” “Road Transportation System”, and “Smart Cities” identified 168 peer-reviewed documents and 68 journal articles as of 17 June 2025, as shown in Figure 3. This combined approach aimed to explore common themes across AI, road transportation system, and smart cities. From 2018 to 17 June 2025, studies were screened based on titles and abstracts to exclude duplicates, inconsistencies, or irrelevant records. Only articles identified through the joint search were included in the review, reflecting the growing research focus on these interconnected areas (see Figure 3). To ensure high publication standards, the authors excluded other types of studies indexed in Scopus (e.g., conference papers, book chapters, and so on), while including review papers for their insights. Additional studies from other sources or databases were omitted to ensure the reliability and relevance of the literature review. This approach minimized potential biases and focused the analysis on a curated set of journal articles, thereby enhancing the study’s validity and coherence [23]. Retrieving full texts enabled the selection of records aligned with the research objectives, resulting in a focused collection to address the KQs in Section 1.

Eligible articles were evaluated through semantic analysis of their texts. This analysis facilitated effective data organization, enabling a detailed synthesis of research by understanding objectives, identifying key concepts, and assessing the significance and effectiveness of the proposed models, methods, and technological solutions beyond surface features. It also revealed connections across articles and provided valuable insights into the research landscape of road transportation, smart cities, and AI. Inclusion criteria were established based on the relevance of the texts to the joint topic and involved forming thematic research groups through the identification of KTs and KSTs. Identifying two supporting KSTs within each KT improved understanding, contributed to building a foundational body of information, and enhanced the depth and granularity of the analysis. Extracting information from the articles facilitated their categorization into KTs and KSTs. The allocation process involved an iterative review of each article to extract relevant information, deepen understanding of their meaning, and clarify the context behind the KQs. Additionally, this method helped uncover subtle differences, enabling clearer communication of the findings and challenges. It also improved information retrieval by recognizing synonyms and contextual nuances, thereby increasing accuracy and reducing confirmation bias in theme identification [23]. Although not limited to this approach, the attribution of articles to each KT was organized by gradually adding one key theme at a time to the joint topic in the Scopus database. Eight KTs were identified to categorize the papers and structure the review. Each KT and its corresponding KSTs are detailed below, along with references to the relevant papers:

• Smart Infrastructure (KT1): Integrating road design with advanced technologies developed for sensors and big data facilitates communication with vehicles, promoting smarter and more responsive infrastructure. This theme encompasses articles grouped under Sustainable Smart Mobility and Infrastructure Development (KST1₁) [24,25,26,27,28] and Data-driven Smart Infrastructure Optimization (KST1₂) [10,29,30,31,32,33].
• Traffic Management (KT2): Developing new approaches and tools to coordinate and control traffic involves monitoring, signal adjustment, and smart technologies to optimize road networks. Articles were grouped under Real-time Traffic Monitoring (KST2₁) [34,35,36,37,38] and Predictive Analytics (KST2₂) [6,19,39,40,41,42,43].
• Autonomous Vehicles (AVs) (KT3): This topic emphasizes next-generation obstacle recognition and decision-making systems for navigation and control, aiming to improve vehicle interaction with transportation systems and ensure safer roads. Articles were grouped under sub-themes as Navigation and Control (KST3₁) [44,45,46] and Advanced Communication Technologies for AVs (KST3₂) [47,48,49,50,51].
• Safety Enhancements (KT4): The theme concerns user communities, focusing on how context influences driving experiences, including perspectives related to Autonomous Driving for Crash Prevention (KST4₁) [16,52,53,54,55,56,57,58] and Emergency Response (KST4₂) [59].
• Environmental Impact (KT5): The theme concerns deploying innovative methods and tools to develop solutions for reducing pollution and improving energy efficiency, thereby supporting sustainable transportation practices. It encompasses articles grouped under Emissions Reduction (KST5₁) [9,60] and Energy Efficiency (KST5₂) [61,62,63,64].
• User Experience (KT6): The theme focuses on developing and implementing advanced tools to offer personalized navigation and route suggestions based on traffic conditions, thereby enhancing driver satisfaction [65,66,67,68,69,70,71,72,73]. The articles were grouped into Enhanced Mobility (KST6₁) [65,66,67,68,69,70,71,72] and Personalized Services (KST6₂), which focus on individual customization [73].
• Road Maintenance (KT7): Advanced technologies for predictive maintenance and road issue detection aim to enhance safety and infrastructure durability. The sub-themes included both Predictive Maintenance (KST7₁) [12,74,75,76] and Smart Materials Integration and Performance Analysis (KST7₂) [77,78,79].
• Road Intersection (KT8): The theme focuses on designing human-centric road intersections integrated with an AI-powered, citizen-friendly traffic management system for crowded cities. Articles were grouped under Innovative Intersection Design through Simulation (KST8₁) [80,81,82] and Smart Intersection Management (KST8₂) [83,84].

The articles grouped by KTs and KSTs were compiled into a final set. One article was excluded because it focused on geopolitical issues related to AI deployment, which are unrelated to the primary research focus of the paper. As a result, 67 papers were included in the review. Some overlap among KTs remained because many studies had broad relevance to both the research topic and methodology across various themes, highlighting their cross-cutting significance. Although a hypothesis-driven semantic analysis could not be exhaustive, the outlined methodological trajectory helped identify threads across papers, lessons learned, and future research challenges. Figure 4 shows the flowchart of the methodology used for the review.

3. Literature-Informed Review

This section presents the multidisciplinary literature review, focusing on the KTs and KSTs identified through the joint search and semantic analysis of the texts outlined earlier. It categorizes the papers and highlights major findings. Figure 5 summarizes the structure of the review based on the KTs and KSTs.

3.1. Smart Infrastructure

The literature emphasizes smart infrastructure (KT1), which is closely related to road mobility in urban environments. It includes methods and applications focused on improving city functionality and quality of life through advanced technologies [6,13,19]. Thematic analysis highlighted its essential role in the smart era, stressing the importance of integrating innovative road infrastructure design with sensor and big data technologies to improve urban transportation performance [5,24,25,31]. Analysis of these texts identified 11 articles related to KT1, categorized by their focus, methodologies, and findings [10,24,25,26,27,28,29,30,31,32,33] (see

Table 1. Recent studies on Smart Infrastructure (KT1) across the identified KST1_1,2.

Authors (Country 1, Year)	Study Focus	Methodology	Findings and Challenges
Sustainable Smart Mobility and Infrastructure Development (KST11)
Alanazi (Saudi Arabia, 2023) [24]	Lessons from advanced countries for infrastructure development.	Benchmark analysis to identify literature-informed indicators.	Strategic roadmap but outdated data quickly becomes irrelevant.
Guerrieri and Parla (Italy, 2022) [25]	Recognizing pedestrians, cyclists and vehicles near tramways.	DL technique tested using real-world data.	Accurate user location (96%); processing speed need improvement.
Zheng et al. (China, 2022) [26]	Smart road infrastructure for traffic classification and monitoring.	Applying AI models to analyze real-time traffic patterns.	89% accuracy; adaptability across varied environments is challenging.
Kuru and Khan (UK, 2021) [27]	Integrating AVs into urban road infrastructure.	Dynamic modeling and scenario analysis.	Lessons learned for self-driving integration and city implementation.
Huang et al. (China, 2020) [28]	Applications in cooperative vehicle infrastructure systems in cities.	Simulating driving strategies through scenario analysis	Driving behavior should be tested under more realistic road designs.
Data-Driven Smart Infrastructure Optimization (KsT12)
Tay et al. (Malaysia, 2025) [10]	Traffic prediction and congestion assessment.	Implementing traffic rerouting and dynamic vehicle selection.	33% travel time reduction; issues of system scalability persist.
Ghani Khan et al. (Pakistan, 2023) [29]	Developing path selection algorithm with V2X exchange.	Literature review, crowdsensing, data analysis and simulation.	23% travel time reduction; concerns about network decentralization and scalability persist
Naveed et al. (Saudi Arabia, 2022) [30]	Developing a vision-based traffic system in virtual environment.	Wireless sensor data simulation and visual analytics applications.	Improved efficiency; issues of field-testing and real-world use persist.
Wang et al. (Saudi Arabia, 2021) [31]	Designing an AI-driven road target recognition system for smart city.	Target recognition algorithm design with advanced learning techniques application.	Further testing to manage data diversity on IoT sensors is required.
Hernández-Jiménez et al. (Mexico, 2019) [32]	Integrating driving technologies with intelligent road infrastructure	DL models for route decisions through simulation	Insights to develop smart urban roadway systems
Habibzadeh et al. (US & Canada, 2018) [33]	Advanced traffic prediction for smart mobility solutions.	ML and data analytics applications.	Despite improvements, managing big, fast-changing data is challenging.

¹ With multiple authors’ affiliations, either the country of the first author or the funding organization acknowledged as supporting the authors is indicated.

).

The first thematic group of papers focuses on Sustainable Smart Mobility and Infrastructure Development (KST1₁), highlighting recent progress, techniques, and approaches that promote eco-friendly, efficient, and safe urban transportation systems to support sustainable mobility [24,25,26,27,28]. In this area, Alanazi [24] conducted a comprehensive literature review to extract lessons from advanced countries such as South Korea, Singapore, and Japan to guide infrastructure development in Saudi Arabia. The study identified key strategies and 60 indicators in the transportation systems of top countries, emphasizing the importance of emerging technologies like autonomous driving, IoT, AI, and data analytics. The author developed a comprehensive framework to evaluate smart mobility infrastructure, creating a strategic roadmap for Saudi Arabia’s future improvements. However, rapid technological progress can quickly render data outdated, even by 2022. Additional challenges include expanding surveys and addressing broader sustainable development issues to incorporate innovations into policymaking for more resilient urban mobility solutions.

Integrating object detection systems into modern road design is crucial for advancing sustainable smart mobility and infrastructure, improving public transportation efficiency as cities modernize.

Guerrieri and Parla [25] showed that DL and computer vision can protect pedestrians, cyclists, and vehicles near tramways, with a detection rate of 96%. However, their method would benefit from further testing to explore its full potential, as experiments were only conducted at a single tramway in Palermo, Italy, intersecting a 24-m roundabout. This underscores the importance of additional validation in diverse settings to confirm its broader applicability and reliability for urban driver-assistance systems.

Implementing self-powered smart transportation skins for traffic classification is a key aspect of road infrastructure development, aiding cities in analyzing traffic patterns, improving urban planning, and facilitating smart city integration [26,27,28]. In this regard, Zheng et al. [26] developed a flexible, AI-enabled transportation skin with self-powered sensors for vehicle classification, achieving 89% accuracy. Despite its potential for data collection and traffic analysis, this approach relies on existing infrastructure support and raises concerns about sustainable energy supply, deployment effectiveness, and adaptability across diverse environments. These challenges underscore the need for further validation, particularly for large-scale, real-world implementation in varied urban settings. Similarly, Kuru and Khan [27] proposed a comprehensive framework for the seamless integration of AVs into existing road infrastructure, enhancing deployment efficiency through behavioral coordination within embedded centralized systems [5]. Simulations of 11 scenarios with increasing penetration levels of fully autonomous ground vehicles within mixed traffic showed that urban traffic mobility improves significantly beyond 70% AV penetration, suggesting that smart city traffic management can be achieved without extensive road infrastructure changes, supporting broader automation objectives. Meanwhile, the study by Huang et al. [28] advances sustainable smart mobility by proposing a cooperative control model for heterogeneous traffic involving manual and autonomous vehicles. The simulation tested an evaluation method for urban road systems that accounts for delays or pauses varying over time, enabling better prediction and control of the system’s behavior under different conditions. Despite insights promoting vehicle-infrastructure cooperation, the authors emphasized the importance of extensive real-world testing to validate practicality and effectiveness outside controlled environments, as this is an essential step toward developing intelligent, eco-friendly infrastructure in smart cities.

The literature on Data-driven Smart Infrastructure Optimization (KST1₂) focuses on methodologies, techniques, or frameworks that use real-time data, AI, sensor networks, and soft sensing techniques to enhance road performance and adaptability [10,29,30,31,32,33]. These approaches are vital for developing intelligent roads, enabling innovations such as adaptive routing [10,29,32], real-time traffic monitoring [30], autonomous vehicle (AV) navigation [31], and advanced traffic prediction [33], thereby supporting dynamic and intelligent urban mobility systems.

In this context, Tay et al. [10] developed a straightforward AI-driven dynamic re-routing system that selects vehicles based on congestion and travel costs, providing multiple near-optimal alternative routes under current road conditions. Simulations demonstrated a 33% reduction in total travel time. Despite its promising advantages, ongoing research and investment are necessary to develop more advanced algorithms capable of optimizing various types of road infrastructure on a larger, scalable scale. In turn, Ghani Khan et al. [29] emphasized the importance of stable internet connectivity for path selection and information transfer in transportation systems. They proposed a smart framework that uses smartphones as embedded sensors within temporary networks to enable optimal path selection by communicating with neighboring devices, resulting in a 23% reduction in travel time compared to conventional methods. Nonetheless, improving information transfer among fast-moving users requires flexible, adaptive techniques that need validation in real-world conditions. Aligned with the principles of adaptive routing, Hernández-Jiménez [32] developed a DL-based router that considers both local and global network conditions to make more intelligent routing decisions. Simulation results demonstrated that this router outperforms several well-known routing algorithms by reducing network load and the number of steps messages take within a reasonable time frame.

Further research emphasizes that real-time traffic monitoring, AV navigation, and advanced traffic prediction are essential for evolving roadway systems and smart urban development [30,31,33]. These technologies improve decision-making, traffic flow, and safety by offering real-time information and adaptable responses. Naveed et al. [30] integrated a visual analytics framework with wireless sensors to monitor and analyze road metrics, improving traffic prediction and demonstrating superior performance in virtual environments. While this highlights the potential of adaptive sensor-based road infrastructure, real-world testing is essential to confirm its effectiveness beyond simulated scenarios. Similarly, Wang et al. [31] developed a road target recognition algorithm that enhances navigation in smart city contexts by connecting vehicles with wearable devices and road cameras, leveraging RL for improved accuracy and real-time performance. Habibzadeh et al. [33] further explored the role of ML in soft sensing to manage large, redundant datasets, demonstrating that systematic application of machine intelligence can enhance data processing and decision-making in smart cities. Collectively, these studies underscore the promising potential of integrated technological solutions to optimize urban mobility, but further validation in real-world settings remains crucial for supporting ongoing urban growth.

Building on technological progress, the first research group designs advanced road infrastructures that require real-world testing, while the second emphasizes data-driven solutions using AI and real-time analytics, which also need further validation. Together, these approaches aim to seamlessly integrate spatial planning and road systems, enhancing sustainability, safety, maintenance, and user experience in smart traffic management. Continued research and field validation are essential to ensure the scalability and reliability of these innovative solutions [5,6].

3.2. Traffic Management

The second set of articles focuses on Traffic Management KT2, which involves developing new approaches and tools for coordinating and controlling traffic through monitoring, signal adjustments, and smart technologies to optimize road networks. Twelve articles were identified as relevant to KT2 based on their focus and methods [6,19,34,35,36,37,38,39,40,41,42,43], and they were organized into two sub-themes: Real-time Traffic Monitoring (KST2₁) [34,35,36,37,38], emphasizing immediate traffic detection, and Predictive Analytics (KST2₂) [6,19,39,40,41,42,43], which focuses on forecasting future traffic. This categorization clarifies different strategies within smart traffic management.

Table 2. Recent studies on Traffic Management (KT2) across the identified KST2_1,2.

Authors (Country 1, Year)	Study Focus	Methodology	Findings and Challenges
Real-time Traffic Monitoring (KST21)
Puzio et al. (Poland, 2025) [34]	AI and IoT solutions in Polish cities for traffic management.	Empirical approach integrating GIS, big data and ML	Smart systems improve mobility but require better behavioral data use.
Ventura et al. (Brazil, 2025) [35]	Comparing anomaly detection models for security reasons.	ML models’ performance in simulated network scenarios.	Validation in real vehicular networks remains limited.
Dadheech et al. (India, 2024) [36]	Developing a video analytics system using data from IoT sensors.	DL-based model for quick detection, and sensor fusion.	Enhancing data processing near data source and sensor integration.
Moumen et al. (Morocco, 2023) [37]	AI framework using IoT data to optimize road traffic forecasts.	Real-time data collection via IoT sensors analyzed with ML/DL	Improved predictions but traffic variability require further research.
Liu et al. (China, 2021) [38]	Developing a low-power road monitoring system for smart cities.	Functional requirements analysis and system architecture design	Efficient data management; further validation for system sustainability.
Predictive Analytics (KST22)
Mrad et al. (Tunisia, 2025) [39]	Exploring AI models to predict short-term motorway traffic data	Data decomposition, feature selection and signal analysis.	Improved predictions; data variability and stability require research.
Sheeba and Selvaganesan (India, 2024) [40]	Developing a smart traffic management system.	DL and optimization techniques for traffic management and decision-making	Testing data analysis methods to reduce reliance on centralization.
Robinsha and Amutha (India, 2024) [41]	Integrating IoT and ITS.	Designing a Velocious IoT architecture using federated learning to lead smooth travel.	Despite efficiency, issues concern interoperability and scalability.
Saleem et al. (Pakistan, 2024) [19]	Making road networks safer, more reliable for less congested in cities.	Using advanced data processing, and AI for traffic management.	96% accuracy; processing data and cybersecurity require research.
Musa et al. (Saudi Arabia, 2023) [42]	Developing a sustainable framework based on IoT and ITS	Combining real-time traffic data, AI sensors, and ML algorithms.	Improved traffic prediction, but issues of data heterogeneity and system integration persist.
Huang et al. (China, 2022) [43]	Smart solutions for parking, road systems and traffic management.	Case study and scenario analysis, future trend prediction.	Issues related to energy consumption, data exchange, system updates.
Ait Ouallane et al. (Marocco, 2022) [6]	Review recent research on smart urban traffic management system.	Literature review on congestion reduction methods.	Identified gaps to cover future research on sustainability issues.

¹ With multiple authors’ affiliations, either the country of the first author or the funding organization acknowledged as supporting the authors is indicated.

summarizes KT2 studies, highlighting their focus, methodology, findings, and challenges across KST2₁ and KST2₂.

Building on this theme of technological advancements in traffic management, Puzio et al. [34] assessed the effectiveness of AI- and IoT-based intelligent transportation systems (ITS) in major Polish cities and their impact on reducing congestion, emissions, and crashes. Their study, covering cities such as Warsaw, Krakow, Wroclaw, Gdansk, Poznan, Katowice, and Lodz, utilized crowdsourced mobility data, GIS mapping, big data analytics, and ML to analyze traffic patterns from 2020 to 2024. They also evaluated dynamic traffic lights and crash prediction models in national and EU-funded smart city initiatives, providing valuable insights for both theory and practice. Despite the benefits of these technologies, such as reduced emissions, enhanced safety, and increased energy efficiency, challenges like system digitization, interoperability, privacy, and IoT security remain. Future efforts should focus on integrating behavioral data, developing adaptive AI, and fostering stakeholder collaboration through standardized protocols to create secure, connected, and scalable urban traffic systems.

Similarly, Ventura et al. [35] evaluated anomaly detection models for vehicular networks, focusing on real-time identification of security threats and attack mitigation. Their study compared models within the broader ML [8]: a hybrid deep learning architecture combining Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) networks, often used for spatio-temporal data analysis, and Transformer-based Anomaly Detection (TranAD), designed to identify unusual patterns in time-series data. The study in [35] examined how the proportion of anomalies in training data influences detection accuracy, highlighting the importance of balancing sensitivity and generalization. Key challenges include noisy training data and the limitations of static models, emphasizing the need to develop adaptive, scalable intrusion detection systems for vehicular networks.

Complementary to the research above, Dadheech et al. [36] focused on improving real-time traffic monitoring through an efficient video analytics system that utilizes data from multiple IoT sensors. Their approach involved two main steps: first, processing data close to the source to reduce transmission volume; second, employing a powerful deep learning system called You Only Look Once (YOLO) for object detection in videos. Sensor fusion increased object detection accuracy by 2%, particularly under extreme weather conditions, effectively balancing speed and precision for smart city traffic analysis. Despite these promising results, challenges remain in managing large data volumes, ensuring low latency, and maintaining accuracy in adverse environmental conditions without delays or security issues. Future research should focus on enhancing data processing efficiency near collection points, improving sensor integration, and addressing privacy concerns to develop scalable, robust, and privacy-preserving traffic monitoring solutions. In addition, Moumen et al. [37] advanced the field by integrating IoT sensor data with ML and DL algorithms to enable both short-term and long-term traffic forecasts. Their framework emphasized real-time traffic data collection and processing, complemented by forecasting models to enable proactive traffic control, reduce congestion, and enhance urban mobility. Additionally, incorporating weather data improves prediction accuracy and highlights the significant influence of environmental factors on traffic flow, supporting the development of more holistic and adaptive models [8,11]. In line with the challenges highlighted for effective traffic management in smart cities, Liu et al. [38] proposed a low-power monitoring system for urban roads. They carefully analyzed the functional requirements and designed the system architecture around key components such as a networked sharing platform, data flows, and a cloud storage database. Additionally, they evaluated the energy consumption of wireless sensor networks, quantified topological parameters, and implemented energy-saving techniques such as data compression, along with sleep and wake modes. While these strategies show promise, further experiments are necessary to validate their effectiveness and ensure that smart traffic management solutions remain sustainable and efficient in the long term.

The second set of papers on KST2₂ focuses on future-oriented traffic prediction using advanced techniques to forecast traffic conditions, optimize flow, and enable proactive decision-making by anticipating future scenarios through statistical analysis and predictive modeling. Aligned with predictive analysis efforts aimed at understanding and forecasting traffic behavior to inform better management strategies, Ait Ouallane et al. [6] reviewed approaches to traffic congestion, focusing on analyzing, classifying, and optimizing systems for predicting and managing traffic flow. They emphasized that increasing vehicle numbers and slow infrastructure growth lead to severe congestion, especially at intersections, highlighting the need for ongoing research to improve smart infrastructure and traffic management. By analyzing various congestion reduction methods, they provided valuable engineering insights for developing intelligent traffic systems. The study also identified gaps in addressing all aspects of urban traffic management, particularly data collection, processing, and control, and outlined future research directions and unresolved challenges essential for creating more realistic and effective solutions. In this context, Mrad et al. [39] focused on forecasting short-term traffic flow on motorways using hybrid AI techniques, wavelet transforms, and neural networks to address the nonlinear and non-stationary nature of traffic data. Their emphasis on data decomposition and forecasting horizons reflects an understanding of the temporal dynamics in traffic systems, highlighting their potential for real-time traffic monitoring. The study aims to create accurate models to predict future traffic volumes and improve transportation efficiency. It uses a broad approach by testing how sensitive the models are across different time frames, data levels, and performance measures. Other studies focused on evaluating model performance, allowing dynamic adjustments, and forecasting congestion, supporting predictive analytics to anticipate traffic patterns and improve decision-making in smart cities [19,40,41]. Saleem et al. [19] developed a system that utilizes smart technologies and explainable AI to predict traffic congestion, promote transparency, and support proactive traffic management within AV networks, while ensuring data security and privacy. They achieved high accuracy (96%) and a low miss rate, outperforming previous methods and providing valuable insights into vehicle-to-vehicle communication. The emphasis on accuracy, low miss rate, and advanced technologies for transparency and security aligns with the broader goal of developing predictive and secure traffic management solutions for smart cities. In turn, Sheeba and Selvaganesan [40] developed a smart traffic control system that combines deep learning and advanced optimization techniques to effectively address traffic management challenges and assist regulators in making informed decisions. They conducted a comprehensive performance comparison considering error rate, accuracy, miss rate, and journey duration. The approach enabled precise diagnosis of road congestion within smart vehicular networks and successfully alleviated it with a low error rate. Key benefits such as easy deployment, rapid congestion forecasting, and low computational complexity highlight growth opportunities through the adoption of advanced technologies and data analytics. Future improvements can be achieved by leveraging local data processing, enabling more efficient control, quicker responses, and reduced dependence on central systems [5].

Aligned with the sub-theme considered, Robinsha and Amutha [41] proposed Velocious, an innovative IoT architecture designed to create a fast and efficient transportation system that minimizes delays and improves obstacle avoidance. This architecture integrates various IoT devices and sensors within the transportation network to enable real-time data collection, communication, and coordination. The process of data acquisition, transfer, and interpretation required a federated learning (FL) approach with context awareness, considering factors such as location, time, and situation [8]. By analyzing real-time data, the system can dynamically adjust traffic signals, routes, and other parameters to ensure smooth vehicle movement. Using sensors and advanced decision-making allows cities to better manage parking, traffic, and operations, underscoring the importance of adopting AI technologies for smart city development. Furthermore, Musa et al. [42] proposed an IoT- and ITS-based framework that emphasizes data-driven decision-making to optimize traffic flow, reduce congestion, and promote environmentally sustainable practices. Their system demonstrates how integrating AI, data hosted on remote servers, and sensor networks can significantly streamline urban traffic management as a decision-making process. In this context, Huang et al. [43] developed a framework for transportation scenarios and reviewed the evolution of smart parking and traffic management systems, supported by real-world examples from Wuxi City’s Xizhang District. They identified traffic challenges and predicted future trends in smart parking, road systems, and traffic management. However, the complexity of these large networks presents challenges such as high energy consumption, difficulties in data exchange between subsystems, and issues with system updates and optimization.

Recent advances have improved real-time traffic monitoring across various areas, including large-scale assessments [34], security [35], video analytics [36], traffic forecasting [37], and energy-efficient monitoring [38]. The integration of AI, IoT, and big data also enhances forecasting and proactive management, emphasizing the importance of combining predictive models with real-time responses [6,19,39,40,41,42,43]. Moving forward, addressing stakeholder perceptions, user acceptance, and policy considerations is crucial for developing more reliable smart traffic systems.

3.3. Autonomous Vehicles

The articles related to the Autonomous Vehicle (KT3) theme, based on joint topic research and text analysis, focus on developing advanced obstacle recognition and decision-making systems for navigation and control, utilizing cutting-edge AI-driven communication technologies. The primary goal is to enhance vehicle-infrastructure interactions and promote safer roads through intelligent, adaptive systems equipped with real-time perception and decision-making capabilities. The articles are classified into two sub-themes: Navigation and Control (KST3₁), which includes studies on obstacle recognition, image analysis, decision-making, nighttime traffic sign detection, and adaptive navigation systems [44,45,46]; and Advanced Communication Technologies for AVs (KST3₂), covering smart data exchange, sensing, secure networks, digital twins, and mobile communication [47,48,49,50,51].

Table 3. Recent research on Autonomous Vehicles (KT3) across the identified KST3_1,2.

Authors (Country 1, Year)	Study Focus	Methodology	Findings and Challenges
Navigation and Control (KST31)
Dewi et al. (Taiwan, 2025) [44]	Improving traffic sign detection at night for AVs.	Applying YOLO model and techniques to handle image contrast.	Research needed for real-time, reliable networks, environmental variability, and interoperability.
Chen et al. (Taiwan, 2023) [45]	Design road signs for mark detection during the day and at night.	Applying YOLO model and contrast enhancement algorithms.	Issue on lighting and optimization persist.
Saleh and Fathy (Egypt, 2023) [46]	Exploring integration of IoT and AI techniques in ITS and AVs	Developing a framework to combine 5G technologies with DL	Despite advances, issues on real-road testing and security persist.
Advanced Communication Technologies for AVs (KST32)
Liu et al. (China, 2023) [47]	Use AI to improve interaction of reality with digital information.	Survey on ITS with 5G networks and AI technologies.	Integration issues in transportation persist.
Chen and Lv (2022) [48]	Analyzing performance of the Digital Twins for AVs.	Combining DL and digital twin technology through simulation.	Model achieves 93% accuracy, but extensive validation is needed.
Hamza et al. (Saudi Arabia, 2022) [49]	AI-driven system for IoV for smart transportation in urban areas	Network configuration, vehicle communication and route optimization.	Security, data integrity, user privacy, managing costs persist.
Mahrez et al. (Morocco, 2022) [50]	Integration of ITS and AI approaches for smart mobility.	Survey of AI-driven ITS supporting urban planning and mobility.	Data privacy, investment, and integration issues pose hurdles.
Reebadiya et al. (India, 2021) [51]	AI-enabled sensing and tracking architecture for driving.	Integrating blockchain technology with AI and advanced communication networks	Safer system excels but needs security improvements.

¹ With multiple authors’ affiliations, either the country of the first author or the funding organization acknowledged as supporting the authors is indicated.

summarizes these studies, noting that some articles blend elements from both categories, but they are primarily classified based on their main focus, contributing to safer, more efficient, and integrated autonomous vehicle systems.

Autonomous navigation and control leverage emerging technologies to facilitate mapping, localization, and real-time decision-making [44,45,46]. Dewi et al. [44] employed the YOLO model for traffic sign detection at night using public datasets to train AVs.

They integrated a contrast enhancement technique to improve image visibility in low-light conditions while preventing noise amplification. The system achieved a mean average precision of 76%, demonstrating its potential for real-time nighttime autonomous navigation [44]. Despite these advances, issues such as real-time processing, network reliability, environmental variability, and system interoperability remain. Similarly, Chen et al. [45] developed a comprehensive Taiwanese dataset for daytime and nighttime traffic signs and employed the YOLO model with contrast enhancement algorithms to ensure robust detection at night. They suggested conducting further testing on mobile or embedded devices to assess performance in real-world scenarios. Additionally, they highlighted the need to expand the dataset by incorporating data from diverse environments beyond nighttime conditions and to compare the benefits of image enhancement with other road marking sign standards and detection methods.

Aligned with the core focus on vehicle control operations within autonomous systems, Saleh and Fathy [46] proposed a framework leveraging 5G and DL to enhance AV control. Their system detects driver drowsiness using AI techniques and enables remote vehicle control by a central command center in emergency situations, such as stopping the vehicle or switching to teleoperation mode. The framework employs network technology that centralizes management and control, allowing a single controller to manage data flow efficiently and flexibly, thereby enhancing ultra-low latency and reliability for real-time feedback signals. Simulations demonstrated significant improvements in throughput and latency, supporting safe and reliable control operations. This approach underscores the importance of advanced communication technologies and AI-driven safety systems in managing AVs within the evolving ITS landscape. The integration of Advanced Communication Technologies (KST3₂), including AI-driven Vehicle-to-Everything (V2X) systems, is essential for smart transport networks, enhancing situational awareness and performance, but concerns about reliability due to vulnerabilities in critical safety decisions persist [47]. Aligned with KST3₂, Liu B. et al. [47] explored the contribution of AI approaches to AV technology through a comprehensive survey on ITS to assess the effects of 5G technologies on road transportation and smart cities. While challenges remain in fully developing V2X, 5G-based vehicle image recognition shows promising results: 92% reduction in traffic congestion, 84% improvement in transportation, 88% increase in vehicle image monitoring, and 85% enhancement in V2X communication. However, validation across diverse environments is lacking, and issues like coverage gaps, latency, interoperability, legal, and infrastructure challenges must be resolved for safe deployment. Meanwhile, Chen and Lv [48] advanced this domain by proposing an AV Digital Twins prediction model, integrating DL with digital twin technology through simulation. Their results demonstrated a prediction accuracy of 93%, over 3% higher than existing models, along with high data transmission security. Despite these promising findings, the study highlights the need for extensive validation and application to ensure robustness and safety in real-world scenarios. Continued research and investment are essential to fully realize autonomous driving systems’ benefits [25]. In turn, Hamza et al. [49] proposed an AI-driven system for the Internet of Vehicles (IoV) to enhance smart transportation in urban areas. The system includes network setup, selecting specific vehicles to manage communication, and optimizing routes for interactions. Its energy efficiency was validated through experiments, showing better performance than existing methods. The authors highlighted the need for future research to develop trust-based frameworks within IoV networks to improve security, data integrity, and user privacy. Despite associated costs, this work supports recent advances in IoV and cloud computing that enhance urban transportation, traffic management, and resource efficiency, promoting sustainable smart city growth [8]. In this context, Mahrez et al. [50] reviewed approaches like ITS that use communication technology to enhance road safety and autonomous driving. They found that integrating advanced AI improves urban planning, crowd analysis, and traffic prediction, despite challenges related to big data. These advancements can enhance routing, reduce congestion, and support sustainable urban growth. The authors emphasized the importance of eco-friendly measures, political support, and open data sharing for sustainable urban mobility, while also highlighting cybersecurity as a critical concern with the rise of connected automated vehicles (CAVs) [8]. Aligned with this aspect, Reebadiya et al. [51] proposed a secure sensing and tracking architecture utilizing next-generation networks, blockchain technology [8], and AI to manage sensitive data among CAVs. Their system demonstrated improved security and lower delays compared to traditional models, making it suitable for safety-critical applications that require high reliability and low latency. Integrating autonomous navigation with infrastructure communication is key to developing sustainable smart cities by enhancing mobility, safety, and reducing congestion and emissions. Addressing challenges like coverage gaps, latency, and interoperability is crucial for robustness across diverse environments. Technologies such as blockchain, Digital Twins, and AI-driven routing further bolster security and efficiency, supporting scalable solutions [5,8]. Overcoming these hurdles will enable reliable, sustainable, and safe autonomous transportation, transforming urban communities into more efficient and eco-friendly environments.

3.4. Safety Enhancements

Safety Enhancements (KT4) primarily emphasize targeted interventions tailored to specific contexts. These interventions include autonomous driving systems for crash prevention and emergency response, and consideration of how user communities and environmental conditions impact safety performance and outcomes [45]. Based on thematic analysis, the articles were classified as summarized in

Table 4. Recent research on Safety Enhancements (KT4) across the identified KST4_1,2.

Authors (Country 1, Year)	Study Focus	Methodology	Findings and Challenges
Autonomous Driving for Crash Prevention (KST41)
Martínez and Insuasti (Colombia, 2025) [16]	AI application in vehicle license plate recognition.	Systematic literature review from bibliometric databases.	Variations in plate designs and environmental conditions need research.
Wang et al. (China, 2025) [52]	Proactive prediction of crashes near intersections.	DL model complemented by data sampling technique.	High accuracy; challenges persist in uneven data distribution and integrating diverse information.
Wang et al. (China, 2025) [53]	Assessing wireless communication technology for collision avoidance.	Scenarios analysis to analyze conflicts through simulation.	38% decrease in traffic conflicts; data loss and security are challenging.
Jagatheesaperumal et al. (India, 2024) [54]	Introducing a framework for safe smart cities using AI of Things.	Combining sensors and communication technologies.	Bolstered road safety; sensor reliability and data security are challenging.
Djazia et al. (Algeria, 2023) [55]	Developing a smart driver assistance system for crash prediction.	ML and Internet of vehicles.	Enhanced prediction; data privacy and cybersecurity issues persist.
Bokolo (Norway, 2023) [56]	Examining senior citizens’ mobility and safety concerns.	Systematic literature review and case studies across countries.	Inclusive initiatives still face gaps in policy implementation.
Domínguez and Sanguino (Spain, 2021) [57]	Predicting pedestrian intentions and behaviors near crosswalks.	Identifying safe routes using AI, sensor data, and optimization.	Achieved 99% accuracy, challenges in data variability and sensor limitations remained open.
Liu et al. (China, 2020) [58]	Improving urban expressway pattern recognition.	Image processing and AI algorithms	Despite prediction accuracy, further tests in the field should be made.
Emergency Response (KST42)
Pathik et al. (Saudi Arabia, 2022) [59]	Developing a crash detection and rescue system for smart cities.	Using IoT, AI, DL.	Achieved 98% accuracy. Cybersecurity and data privacy issues persist.

¹ When articles have multiple authors’ affiliations, the country of the first author or the funding organization acknowledged as supporting the authors is indicated.

, covering their focus, methods, results, and challenges across the KSTs [16,52,53,54,55,56,57,58,59]. Specifically, Autonomous Driving for Crash Prevention (KST4₁) involves predicting and mitigating traffic risks using AI, IoT, and communication techniques [16,52,53,54,55,56,57,58], while Emergency Response (KST4₂) emphasizes rapid detection and management of crashes to facilitate response [59].

Aligned with the first sub-theme, Martínez and Insuasti [16] reviewed 90 studies on AI in vehicle license plate recognition, highlighting progress in road safety through DL models and advanced image processing techniques [8,15]. While these innovations support proactive law enforcement, reduce violations, and prevent crashes in smart city projects, variations in plate designs and environmental conditions can compromise recognition accuracy, emphasizing the need for continued research to enhance image preprocessing and system robustness. In turn, Wang et al. [52] developed a comprehensive DL model to predict road crashes at intersections using AI and big data. Their three-phase process begins with extracting key features based on road network topology, establishing a structural foundation. The second phase integrates environmental, traffic, weather, and risk data using graph convolutional networks to model relationships between nearby intersections, enhancing contextual understanding [8]. Finally, the third phase analyzes sequential data to capture temporal dependencies, further improving prediction accuracy. A data sampling technique was introduced to address data imbalance and diversity. While the model achieved high accuracy and outperformed existing methods on New York traffic data, challenges remain due to uneven data distribution and the complexity of integrating various information types. While mainly focused on crash prediction, this approach also offers valuable insights for emergency response planning. However, additional research is needed to improve robustness and address current limitations. Wang et al. [53] conducted a scenario analysis through simulations and demonstrated that AI-enabled Cellular Vehicle-to-Everything (C-V2X) technology significantly enhances AV communication, reducing latency by over 99% compared to traditional systems and decreasing traffic conflicts by 38% at 60% AV adoption. While the simulations indicate improved flow and safety, challenges such as data loss, delays, and security remain. The study emphasizes the potential of wireless communication technology to enhance collision avoidance, reduce congestion, and improve traffic management. It also underscores the importance of continuous model development and stakeholder collaboration to fully realize these benefits in smart cities.

Two studies explored innovative approaches to enhancing road safety and traffic management in smart cities [54,55]. Jagatheesaperumal et al. [54] proposed a vehicle safety framework using AI of Things (AIoT), combining sensors such as eye blink, ultrasonic, and alcohol detectors with advanced communication technology to enable faster data exchange between vehicles. The eye blink sensor proactively detects hazards and alerts drivers, while ultrasonic sensors monitor nearby vehicle speeds to improve traffic flow. The system also checks alcohol levels, adjusting vehicle speed via Global Positioning System (GPS) and Global System for Mobile communications (GSM) technology if high levels are detected, and notifies authorities. This progress in road safety provides a foundation for further advancements in smart urban transportation. However, challenges such as sensor reliability and data security remain, necessitating further research. In turn, Djazia et al. [55] developed a smart driver assistance system using ML models and the Internet-of-Vehicles (IoV) for crash prediction and prevention. Their framework addresses IoV limitations such as internet disconnection and response delays by utilizing vehicular edge computing [8]. This approach leverages the processing and storage capabilities of resources closer to end users, including vehicles and roadside infrastructure, to improve system responsiveness and reliability. While this approach improved prediction accuracy, ensuring data privacy and cybersecurity remains crucial for building public trust, especially in countries developing intelligent mobility systems.

Further studies [56,57] emphasize creating safe, inclusive mobility solutions for vulnerable users. Bokolo [56] examined the mobility needs and safety concerns of senior citizens aged 65 and above, analyzing policies from Norway, Canada, the US, UK, Sweden, and Northern Ireland. The research highlights how emerging technologies, particularly AI-driven ML, can improve mobility inclusion and safety for older adults. However, challenges such as policy gaps and barriers to technology adoption remain, underscoring the need for further efforts to ensure equitable, accessible public transit, age-friendly infrastructure, assistive technologies, and supportive policy frameworks that promote mobility equity and safety for seniors, especially in aging societies. Building on Domínguez & Sanguino [57], a mobile app was created to predict pedestrian crossing intentions and enhance safety before crashes occur, using smartphone sensors and achieving an accuracy of 99%. It guided users along safer routes by integrating pedestrian area data from a cloud database. The app’s routes were safer than those from Google Maps, increasing the use of safe pedestrian areas by at least 183%. Nonetheless, ongoing challenges related to environmental and sensor limitations continued to affect performance across different settings. Similarly, Liu et al. [58] aimed to enhance pattern recognition in urban expressways to improve traffic detection accuracy for emergency management, utilizing image processing, AI, and advanced clustering algorithms [8,11]. Their approach achieved a recognition accuracy of 97% and provided valuable technical insights and a theoretical foundation for recognizing traffic patterns on urban motorways. However, to enhance robustness and broader applicability, it should be tested in more complex traffic environments, especially in areas prone to rapid road deterioration, congestion, and secondary accidents. This would help prevent erroneous traffic capacity forecasts and safety underestimations.

In the field of developing effective emergency response tools, Pathik et al. [59] proposed an IoT- and AI-driven crash detection and rescue system focused on early risk identification and prevention. The system utilized IoT sensors to collect data on position, pressure, gravitational force, and speed, which was then sent to the cloud for validation through DL algorithms. This approach achieved 98% accuracy in crash detection and activated rescue modules to promptly notify emergency services. However, challenges such as limited datasets—requiring data generation from online videos—and the risk of false detections remain. While the system improves response times, future efforts should address cybersecurity and data privacy concerns to ensure safe and widespread adoption.

Advancements in AI, IoT, and sensor technologies for community safety show great promise but face three main challenges. First, cybersecurity issues and data privacy concerns threaten system integrity and public trust, requiring stronger security protocols. Second, environmental factors affecting sensor reliability, such as lighting and weather, can lead to false alarms or missed detections, necessitating improved algorithms and sensor fusion. Third, operational issues like system disconnections, response delays, and data imbalances hinder seamless deployment and real-time performance, highlighting the need for standardized, optimized system architectures. Addressing these challenges is crucial to fully realizing the potential of ITS and enhancing urban crash prevention.

3.5. Environmental Impact

Environmental Impact (KT5) emphasizes innovative strategies to reduce emissions and enhance energy efficiency, thereby promoting sustainable transportation [9,60,61,62,63,64]. As shown in

Table 5. Recent research on Environmental Impact (KT5) across the identified KST5_1,2.

Authors (Country 1, Year)	Study Focus	Methodology	Findings and Challenges
Emissions Reduction (KST51)
Liu et al. (China, 2025) [9]	Strategies to achieve emission reduction in road transport.	Analysis of measures, practices, and energy system upgrades	Implementation of advancements, regulations, and renewable energy remains complex
Rauniyar et al. (Norway, 2023) [60]	Developing a real-time noise and emissions monitoring system.	Cloud-based data collection, AI algorithms for emission classification, and real-world testing.	Hotspots identified; data accuracy and response management pose challenges.
Energy Efficiency (KST52)
Kumar Reddy et al. (India, 2023) [61]	Improving the Quality of Service of traffic using context-aware AI to reduce real-time data transfers	A three-layered learning model with platoon control tested via simulations	Prediction improved (8–24%); managing context migration is challenging.
Al-Selwi et al. (Malaysia, 2022) [62]	Investigating the impact of weather data to improve traffic prediction.	Analyzing models with and without weather data to assess performance.	Weather data enhances prediction accuracy; integrating external factors remains complex.
Kumar et al. (India, 2020) [63]	Proposing an electric-powered public bus system for smart cities.	Discrete event-based simulation with multi-objective optimization	Reduced waiting times (0.2 to 0.7 min); issues on resource management and security persist.
Reid et al. (Mexico, 2018) [64]	Develop an IoT-based system for accurate vehicular detection.	ML and analysis of benchmark studies to assess transport issues.	Accuracy can be achieved; mitigation strategies need further research.

¹ When articles have multiple authors’ affiliations, the country of the first author or the funding organization acknowledged as supporting the authors is indicated.

, articles under KT5 are categorized into two subthemes: Emissions Reduction (KST5₁) and Energy Efficiency (KST5₂).

Emissions Reduction (KST5₁) focuses on decreasing greenhouse gases and pollutants using techniques like real-time monitoring, traffic inference, and acoustic analysis to support environmental sustainability [9,60]. In contrast, Energy Efficiency (KST5₂) involves optimizing energy consumption in transportation systems and infrastructure through advanced vehicle technologies, improved traffic management, and eco-friendly driving practices. Well-designed roads and smart systems further contribute to energy savings and better transportation performance [61,62,63,64]. Although distinct, energy efficiency measures are closely intertwined with overall environmental sustainability efforts, often complementing emissions reduction strategies for a more comprehensive approach.

Liu et al. [9] explored strategies to reduce emissions in road transportation, focusing on technological advancements, organizational management, and energy system upgrades. Their study highlights improvements in vehicle efficiency, the adoption of sustainable energy sources, and innovative technologies like 5G, alongside urban planning and priority-setting for public transit. The authors emphasized the potential of alternative fuels such as hydrogen and biogas, underscoring the importance of pollution regulations, fuel flexibility, and integrated planning to maximize environmental benefits. Nonetheless, they acknowledged ongoing implementation challenges, particularly concerning data accuracy and response management, and stressed the need for a comprehensive approach that combines technical improvements, management strategies, and a shift toward renewable energy sources to effectively mitigate transportation’s environmental footprint.

As part of a European project aimed at promoting sustainable, intelligent mobility through continuous tracking and reduction of pollutants from various transportation modes, Rauniyar et al. [60] contributed to emission reduction efforts by developing real-time monitoring systems for noise and exhaust emissions. These systems utilize cloud-based data collection and AI algorithms to classify pollution sources and were tested in European cities. Their findings indicate that real-time monitoring can effectively identify pollution hotspots and enable prompt interventions; however, challenges related to data reliability and response strategies remain.

In the field of energy efficiency, Kumar Reddy et al. [61] aimed to improve service management in smart transportation by employing context-aware AI to reduce real-time data transfers, thereby lowering energy consumption and environmental impact. Their approach involves a three-layered learning model combined with platoon control, tested through scenario analysis using cloud computing, which demonstrated an improvement of 8–24% in context prediction accuracy. This led to reductions in service times, energy use, and CO₂ emissions. Managing external data, however, remains a challenge affecting system performance. Similarly, Al-Selwi et al. [62] improved traffic prediction models by integrating weather data, which significantly increased prediction accuracy and supported more energy-efficient traffic management. Nonetheless, the integration of external factors complicates deployment, necessitating further refinement. Additionally, Kumar et al. [63] proposed an electric public bus system for Dehradun’s smart city, aiming to enhance energy efficiency through discrete event simulation and multi-objective optimization based on real-time passenger needs and road conditions. Their results show reductions in energy consumption and shorter passenger waiting times by 0.2 to 0.7 min. However, resource management and security challenges, including data confidentiality and authentication, pose significant hurdles for real-world implementation. Reid et al. [64] focused on traffic inference as a key strategy for optimizing flow and reducing emissions. Their adaptive traffic control systems, based on tailored mathematical models, achieved 99% accuracy in vehicle detection during benchmark tests. While promising, further refinement is necessary to fully realize emission mitigation benefits.

The deployment of IoT-enabled multi-agent systems, equipped with distributed sensors and machine learning algorithms, presents a promising solution for autonomous traffic management, congestion reduction, and pollution control, collectively advancing urban environments toward sustainable mobility. However, challenges persist, primarily due to the high costs of technologies like AI and IoT, which limit widespread adoption. Issues related to data accuracy and reliability, along with the need for extensive infrastructure upgrades and cross-sector collaboration, further impede progress. Continued research and development are crucial to overcoming these barriers and realizing broader, real-world impacts.

3.6. User Experience

Papers investigating the connections between Artificial Intelligence, Road Transportation Systems, and Smart Cities within the User Experience (KT6) theme mainly focus on two key subthemes [65,66,67,68,69,70,71,72,73]: Enhanced Mobility (KST6₁) and Personalized Services (KST6₂). Enhanced Mobility (KST6₁) emphasizes technological and systemic advancements that improve the ease, efficiency, and accessibility of transportation [65,66,67,68,69,70,71,72]. Innovations such as AVs, multimodal transport, and smart infrastructure contribute to smoother, more reliable, and flexible urban mobility. These systems also enhance safety, reduce congestion, and support digitalization, with the goal of creating safer and more user-friendly transportation options through the integration of social perception and traffic prediction models. Conversely, Personalized Services (KST6₂) focus on tailoring transportation experiences to individual preferences and behaviors by leveraging data and technology to provide customized route planning, information, and notifications, thus boosting user satisfaction and engagement [73]. Both sub-themes aim to deliver a more engaging, personalized, and enjoyable transportation experience, contributing to a smarter, more user-centric future for city mobility.

Table 6. Recent research on User Experience (KT6) across the identified KST6_1,2.

Authors (Country 1, Year)	Study Focus	Methodology	Findings and Challenges
Enhanced Mobility (KST61)
Yung and Kim (China, 2025) [65]	Studying how AI influences trust and acceptance of AVs.	Data analysis of 392 Chinese vehicle owners.	Trust varies with automation levels; public education and building trust are essential for AV adoption.
Issaouı and Selmi (Saudi Arabia, 2025) [66]	AI-powered face mask detection for smart cities.	Using advanced AI methods to detect faces and classify masks, and optimization.	Achieved high accuracy; issues on data and real-world deployment persist
Lv et al. (Sweden, 2023) [67]	Physical social intelligence in cyber–physical social ecosystems within the context of smart cities.	Analyzed behavior modeling, AI learning, and adaptation strategies across various sectors.	Behavior modeling advances; security, data techniques, AI integration pose obstacles for development.
Kumar et al. (India, 2022) [68]	Developing an IoT-based real-time face mask detection system for smart city public transportation.	Designed hybrid DL and ML models tested on multiple datasets	The proposed model outperformed with error rate of 1.1%, but resource limitations remain a concern.
Sepasgozar and Pierre (Canada, 2022) [69]	Traffic prediction using key road features while preserving privacy.	Combining long short-term memory with FL on vehicular network dataset in simulation.	The model predicts accurately, preserves privacy, outperforms other AI algorithms, but faces complexity.
Gollapalli et al. (Saudi Arabia, 2022) [70]	Designing an automated traffic control system to minimize user delays.	Utilized IoT sensor data processed on cloud, integrated with Neuro-Fuzzy approach	Simulation achieved 99% accuracy; issues include data integration and system complexity
Singh et al. (India, 2021) [71]	Exploring the potential of AI technologies in highway applications through digitalization.	Proposed architecture integrating smart lighting, traffic management, renewable energy, and AI.	Digitalization improves efficiency; challenges include system integration and technology adoption
Domínguez et al. (Spain, 2020) [72]	Improving vehicle detection in smart crosswalks using ML models	Trained and tested various ML models with real traffic data from Portugal and Spain.	ML models outperform fuzzy classifiers, but issues related to data variability and model robustness remain.
Personalized Services (KST62)
Ahmed et al. (Malaysia, 2020) [73]	Understanding the factors influencing smart mobility adoption in Malaysia.	Combining advanced statistics and neural networks	Understanding of technology acceptance; behavioral evaluation is key to smart mobility success

¹ When articles have multiple authors’ affiliations, the country of the first author or the funding organization acknowledged as supporting the authors is indicated.

summarizes the key insights from these studies.

Aligned with KST6₁, Yang and Kim [65] examined how levels of driver-AI collaboration influence trust and acceptance of AVs, with a particular emphasis on personalized experiences. Their study, involving 392 vehicle owners across Chinese cities, shows that trust varies with the degree of automation: manual control depends on ease of use and reliability, while increased AI involvement requires greater AI literacy. Conversely, fears about the technology can reduce acceptance. As AI takes on more driving tasks, building trust becomes crucial, highlighting the importance of public education and phased automation. These insights support the development of safer, more inclusive, and sustainable urban transportation systems through AI integration.

Within the same theme, Issaoui and Selmi [66] explored how advanced AI and DL can improve user experience and health monitoring in smart city public spaces. They developed an AI-driven face mask detection system, essential for managing safety during pandemics in crowded urban areas. Using sophisticated AI techniques to identify faces and assess mask-wearing, the system’s accuracy was enhanced through optimization algorithms. Despite its high precision, challenges remain in data management and real-world deployment. In the context of enhanced mobility in smart cities, developing systems that are more responsive, efficient, and aligned with social behaviors and needs is crucial. Building on this, Kumar et al. [68] created an IoT-based real-time face mask detection system to boost safety during COVID-19, achieving high accuracy with minimal error (1.1%) but facing resource limitations. Lv et al. [67] also highlighted that social perception is crucial for enabling systems to understand and respond to social interactions in both online and real-world environments.

The study in [67] focuses on connecting these interactions via mobile networks, leveraging behavior modeling, AI learning, and adaptation strategies across various sectors. Although progress has been achieved, challenges like security, data management, and AI integration still remain. As smart cities continue to grow, effective infrastructure management, using technologies such as computer vision and transfer learning, becomes crucial for automating health and safety measures. Their system outperformed other models in testing, demonstrating its potential despite the hurdles associated with practical implementation.

Further advancing urban mobility, Sepasgozar and Pierre [69] developed a vehicular network traffic prediction model using FL, enabling computers to learn from data without sharing it directly, thereby helping to protect privacy [8]. When combined with data from Vehicular Ad Hoc Networks—wireless vehicle communication systems—this approach can accurately forecast traffic patterns. These predictions were tested in a simulated environment, preparing the system for deployment in real-world traffic management through virtual model evaluation. Although the system predicts traffic accurately, outperforms existing AI methods, and maintains data privacy, it still faces challenges related to complexity and computational demands. Similarly, Gollapalli et al. [70] predicted road congestion using IoT sensor data processed on the cloud with a hybrid Neuro-Fuzzy approach [11]. Their simulation achieved 99% accuracy, surpassing other methods, with training accuracy reaching 99.21%, supporting automated traffic control. However, challenges related to data integration and system design remain. Despite these issues, this research provides a valuable tool for smart city environments, supporting automated traffic management and congestion reduction to enhance urban mobility.

In the digitalization of urban mobility, Singh et al. [71] proposed an integrated architecture that combines smart highway lighting, traffic management, renewable energy, and AI. This approach significantly enhances safety and operational efficiency, despite challenges related to system integration and widespread technology adoption. The work supports the United Nations’ 2030 Agenda by promoting sustainable, intelligent highway systems through advanced digital technologies, contributing to safer and more efficient transportation networks. Complementing these efforts, Domínguez et al. [72] focused on improving vehicle detection at smart crosswalks to boost road safety. They trained and tested various ML models [8,11], including random forest, time-series forecasting, multi-layer perceptron, support vector machine, and logistic regression, using real traffic data from Portugal and Spain. They also developed a DL program that learns and improves over time, outperforming traditional machine learning models. However, issues related to data variability and model robustness still persist, necessitating further refinement for deployment in dynamic traffic environments.

In a different vein, focusing primarily on personalized user experiences, Ahmed et al. [73] examined factors influencing the adoption of smart mobility in Malaysia, emphasizing IoT and personalization. Analyzing data from Klang Valley using advanced statistical methods and neural networks [11], they identified five critical factors: digital skills, IoT service quality, privacy concerns, online reviews, and social influence. Their findings underscore that understanding user acceptance is essential for successful implementation. These insights can guide policymakers in developing strategies to increase the adoption of IoT and AI in urban transportation, making smart mobility more accessible. The study also highlights challenges such as building trust, improving AI literacy, addressing fears of technology, and securing public support for the effective integration of AI and IoT in autonomous vehicles and smart transportation systems.

Overall, these studies demonstrate important progress in advancing smarter, safer, and more efficient urban transportation through AI, IoT, and digital technologies. However, challenges remain, such as building user trust across diverse groups, ensuring data security and privacy, and overcoming the complexity and resource limitations of AI systems. Future efforts should aim to develop adaptable, scalable solutions capable of reliable real-world operation to fully realize the potential of smart mobility and support sustainable, inclusive urban transportation.

3.7. Road Maintenance

The Road Maintenance theme within AI, road transportation systems, and smart cities emphasizes a holistic approach focused on improving technology, user experience, and sustainability [12,74,75,76,77,78,79].

Table 7. Recent research on Road Maintenance (KT7) across the identified KST7_1,2.

Authors (Country 1, Year)	Study Focus	Methodology	Findings and Challenges
Predictive Maintenance (KST71)
Mahmudah et al. (Indonesia, 2025) [12]	Using AI models to automatically detect road damage in real-time	Adjusting YOLO model settings for use on low-power devices.	Fast performance model; scaling and hardware issues persist.
Qin and Pournaras (UK, 2023) [74]	Developing an energy-efficient coordination model with drones.	Using decentralized multi-agent collective learning algorithms.	Achieved 46% accuracy; Energy management and reliable communication require further deepening.
Hijji et al. (Saudi Arabia, 2023) [75]	Developing an AI-driven hierarchical framework for road maintenance	Using advanced AI models to integrate images and sensors data.	Advanced pothole detection; challenges of real-time deployment.
Swarnkar et al. (South Africa, 2023) [76]	Lithium-ion battery health prediction for proactive maintenance	Comparing ML algorithms with modified support vector machine to predict health battery	Reduced error; challenged by complex battery degradation modeling
Smart Materials Integration and Performance Analysis (KST72)
Gabbar et al. (Canada, 2023) [77]	Smart monitoring for real-time road condition inspection	Using sensors, digital signal processing, and AI algorithms	Accurate real-time monitoring; data integration remains challenging.
Jagatheesaperumal et al. (India, 2023) [78]	AI acoustic and ultrasonic system for real-time road monitoring.	Hardware design in vehicle rims, data collection, ML algorithms.	Achieved accuracy (99%); real-time processing remains challenging.
Liu et al. (USA, 2023) [79]	Developing a multi-task, edge-based sensing system.	Implementing “Sensing as a Service” with cooperative sensing.	Achieved high accuracy; challenges in handling data.

¹ When articles have multiple authors’ affiliations, the country of the first author or the funding organization acknowledged as supporting the authors is indicated.

provides a summary of the key studies on this theme. In particular, Predictive Maintenance (KST7₁) is revolutionizing urban infrastructure management by using AI to analyze data from road sensors and traffic patterns, enabling accurate deterioration forecasts and timely repairs [74,75,76]. Additionally, the second sub-theme (KST7₂) involves combining smart materials with AI to evaluate their performance under various conditions [77,78,79]. These strategies support sustainability and infrastructure resilience while reducing environmental impact and downtime [77].

Within the context of KT7, Mahmudah et al. [12] advanced AI-driven methods for automatically detecting road damage in real-time, thereby reducing maintenance costs.

They optimized the YOLO model for low-power devices, achieving nearly 99% accuracy and fast image processing, making it suitable for small hardware and even faster on high-end devices. Despite limitations related to system scaling and hardware requirements, this technology generally offers a promising solution for efficient road monitoring and maintenance, supporting the development of digital twins and smarter, safer cities by enabling early damage detection and timely repairs. In turn, Qin and Pournaras [74] developed a scalable, energy-efficient coordination model for large drone swarms used in smart city applications such as traffic monitoring and disaster response. Their decentralized multi-agent learning approach increased accuracy by 46% and efficiency by 3%, addressing challenges in energy management and large-scale sensing. The model enhances drone resilience, flexibility, and scalability, enabling effective data collection over large areas with battery-limited drones. Experiments demonstrated improved performance and showed how coordinated drone mobility boosts sensing accuracy. However, further improvements in energy management, communication, and stability in dynamic environments are necessary to support smarter urban management through advanced sensing. Building on these advancements, Hijji et al. [75] developed a novel framework for road maintenance using AI, 6G, and edge computing, integrating image and sensor data to improve pothole detection accuracy [11]. Designed to support CAV driving, it features a new CNN model that combines visual and sensor inputs for precise pothole detection [8]. Although high detection accuracy and computational efficiency have been achieved, issues related to processing speed and data storage requirements remain, particularly for real-time implementation and managing large data volumes. Overall, this work highlights the potential of connected vehicles and advanced communication technologies to enhance the effectiveness of urban road maintenance. Complementing these efforts, Swarnkar et al. [76] focused on AI-driven battery health forecasting, using ML algorithms to accurately predict degradation [11]. Their modified support vector machine achieved precise predictions with low errors, aiding maintenance planning before failures occur. Despite the complexity of modeling wear processes due to multiple factors, these predictive insights are vital for extending equipment lifespan and reducing unexpected failures. Overall, these studies highlight how AI-powered predictive maintenance is transforming smart road upkeep, making solutions more efficient, accurate, and scalable for safer, smarter cities.

AI-driven approaches are increasingly used to evaluate smart materials under various traffic conditions, improving understanding of their performance. Some research also employs automated inspection tools and advanced technologies to accurately assess the condition of road materials and infrastructure, supporting more efficient maintenance and enhanced durability. In this context, Gabbar et al. [77] developed a smart system for real-time road condition monitoring by integrating sensors, digital processing, and AI to optimize maintenance activities. Achieving over 97% accuracy in detecting snow coverage and predicting safety indices across various weather conditions, the system enables better-informed maintenance strategies and resource allocation. It can also be adapted for inspecting other infrastructures such as bridges and tunnels using autonomous drones, enhancing monitoring capabilities. To ensure long-term effectiveness, it is important to evaluate costs, energy efficiency, and address issues related to data accuracy and system integration to facilitate efficient workflows. This system advances smart city development by enabling rapid hazard detection and supporting technologies like self-driving vehicles, streamlining maintenance operations with hardware, data processing software, and a web interface for real-time management.

Similarly, Jagatheesaperumal et al. [78] developed an AI-driven acoustic and ultrasonic system embedded in vehicle rims to monitor road conditions in real time, utilizing ML algorithms such as multi-layer perceptron, support vector machine, and random forest [11], achieving accuracy levels of up to 99% in monitoring road quality. This system enhances road quality assessment and supports emission reduction efforts. Although managing the large volume of real-time data remains a significant challenge, it offers a promising approach for accurate, real-time road monitoring with valuable environmental benefits. In turn, Liu et al. [79] developed a multi-task, edge-based traffic management system called the cooperative and comprehensive smart edge node, which enhances perception accuracy by combining cooperative sensing and parallel computation through “Sensing as a Service”. Despite limited edge resources, the system effectively performs vehicle counting, road surface classification, visibility estimation, and real-time communication among traffic controllers and users. Field tests confirmed its high accuracy, bridging sensing capabilities with traffic management needs. While challenges remain in optimizing computational limits and managing increasing data volumes, this approach significantly enhances real-time traffic monitoring and supports more efficient transportation systems.

Predictive maintenance and smart materials are essential for enhancing infrastructure resilience and sustainability in ITS. To unlock their full potential, efforts should focus on optimizing AI, reducing costs, and minimizing environmental impacts, thereby supporting the development of safer, more efficient, and adaptable roads to meet the increasing demands of smart cities.

3.8. Road Intersection

Designing road intersections as vital components of citizen-focused traffic systems is essential in cities, and incorporating advanced technologies can help bridge the gap between traditional intersection design and modern, intelligent road planning. This approach involves combining established methods, such as fixed layouts and signals, with innovative strategies that leverage technologies like intelligent traffic management systems, sensors, artificial intelligence, and data analysis to optimize traffic flow and adapt dynamically to real-time traffic conditions [5,53,57]. In this context, simulation tools are indispensable for urban planning and design, as they allow for testing, optimizing, and visualizing smart infrastructure concepts before implementation. They facilitate modeling adaptive systems and support data-informed decisions, helping to develop future-ready, technology-driven transportation systems [27,53,84,85,86]. However, some aspects of road network design need to evolve. Instead of fixed intersection geometries, layouts should become adaptable and sensor-integrated, capable of responding dynamically to real-time conditions. The analysis of articles related to KT8 [80,81,82,83,84] identified two subthemes: Innovative Intersection Design through Simulation (KST8₁) [80,81,82] and Smart Intersection Management (KST8₂) [83,84], as summarized in

Table 8. Studies on road intersections (KT8) across the identified KST8_1,2.

Authors (Country 1, Year)	Study Focus	Methodology	Findings and Challenges
Innovative Intersection Design through Simulation (KST81)
Anitha et al. (India, 2024) [80]	Designing road traffic solutions before implementation.	Integrating IoT sensors, AI models, and video/image computing	Accuracy, scalability, and data processing issues need improvement.
Cai et al. (China, 2024) [81]	Infrastructure and pedestrian behavior analysis.	ML and image recognition techniques for predictions.	Despite accuracy, data complexity and variability pose challenges.
Li et al. (USA, 2023) [82]	Analysis of physical infrastructure attributes (marked crosswalks).	Computer vision and DL on Street View images.	High-visibility crosswalks exist locally, but regional adoption varies.
Smart Intersection Management (KST82)
Aydin et al. (UK, 2023) [83]	Comparing traditional fixed signals vs. AI-driven smart systems.	Data-driven analysis to test camera-based smart intersections.	16% delays and 20% speed reduction but driver adaptation needs research.
Wan and Hwang (Taiwan, 2018) [84]	Managing adaptive isolated intersection signal control.	Applying learning models and microscopic simulation.	Reducing delays by 20%; traffic fluctuations remain a challenge

¹ When articles have multiple authors’ affiliations, the country of the first author or the funding organization acknowledged as supporting the authors is indicated.

.

Aligned with KST8₁, simulations in ITS enable planners to evaluate infrastructure options safely and cost-effectively, optimize intersection design, and test new strategies. However, increased sensor use introduces challenges such as system integration, data sharing, security, and adaptability, necessitating ongoing advancements to improve navigation and traffic efficiency at intersections. In this context, Anitha et al. [80] developed an AI-powered congestion-free transportation system through simulation, aiming to design innovative intersections before real-world deployment. Their proposed system, called Enhanced Transportation Technologies, integrates IoT sensors, AI models, and video/image-based computing to improve traffic control at a busy intersection. The model successfully reduced waiting times, overcoming traditional sensor limitations.

However, this research underscores the potential of innovative intersection design to optimize urban traffic, supporting safer and more efficient transportation, particularly in overcrowded cities. It exemplifies how simulation-based design efforts can contribute to operational, real-time management strategies, making it relevant for both subthemes within KT8. While these technologies improve decision-making and congestion management, challenges related to accuracy, scalability, and data processing persist.

Cai et al. [81] proposed an innovative approach to analyze pedestrian crossing actions and simulate crossing violations. They employed ML models and advanced image recognition techniques to study pedestrian behavior. Using real data from signalized intersections in Chinese cities, they identified the most accurate model for predicting crossing probabilities and speeds. While their insights are valuable for crash prevention and the design of smarter intersections, future research should expand data collection to include various vehicle types, such as buses, bicycles, and trucks, as well as diverse urban environments. This would enable a more comprehensive analysis and support smart intersection management. Additionally, data complexity and variability in real-world scenarios remain challenging.

With specific reference to crosswalk markings and visibility, which are key components of intersection design and safety, Li et al. [82] developed a longitudinal crosswalk dataset using Google Street View imagery and deep learning to analyze over 38,000 intersections near transit stations across the U.S. from 2007 to 2020. Their approach monitored changes in intersection-level marked crosswalks around more than 4000 transit-oriented development stations, providing insights into crosswalk visibility over time. The study found that high-visibility crosswalks are increasingly replacing parallel-line crosswalks, especially in high-density residential areas with more zero-vehicle households, for better crash prevention. This research highlights the potential of advanced visual technologies, such as image-based analysis, in designing and improving intersection features to support smarter city initiatives. It also acknowledges limitations and geographical differences [25]. Looking ahead, opportunities exist to integrate these systems with other urban infrastructure, such as public transportation, parking, and emergency response, to create a comprehensive approach to enhancing urban mobility and meeting the needs of modern cities.

Over the past decade, cities worldwide have implemented intelligent intersections with adaptive traffic signals to improve traffic flow using real-time data. Aligned with KST8₂, Aydin et al. [83] compared the performance of traditional fixed signals with AI-driven systems at three isolated intersections in Samsun, Turkey. The intelligent systems reduced control delays by 16%, vehicle speeds by 20%, and decreased daily CO₂ and PM₁₀ emissions. Initially, drivers were impatient and struggled to adapt, but their behavior improved over time. While camera-based AI systems can significantly enhance traffic flow at existing intersections, driver adaptation remains a challenge during implementation. In turn, Wan and Hwang [84] developed an advanced system that adapts signal timings based on traffic conditions, learning from past data to respond more effectively to changing patterns. The model was tested through simulation on an isolated intersection, evaluated using total system delay and average delay per vehicle. Preliminary results show that the trained system significantly outperforms fixed timing plans, achieving a 20% reduction in total delay across all scenarios. Future research aims to extend these applications to multiple intersections to enhance traffic management, efficiency, and urban transportation systems.

However, research in this area remains limited, and intersection design continues to pose significant challenges. Traditional geometries lack flexibility, restricting sensor placement, data collection, and AI responsiveness, which can adversely affect safety and efficiency. Traffic simulation tools are vital for supporting road and street space design, enabling engineers and planners to evaluate configurations, predict traffic flow, and identify potential issues before construction [85,86]. These tools help optimize intersection layouts, enhance safety, improve efficiency, and facilitate the integration of smart infrastructure, ultimately creating more effective and adaptable transportation systems. Nonetheless, upgrading intersections with advanced features remains costly and complex, especially in dense urban areas where space and visibility are limited. In line with this approach, crowdsensing through sentinel vehicles equipped with onboard smartphones can provide continuous, real-time data for monitoring roadway variables, supporting the development of smarter cities without the need for additional infrastructure [87]. Looking ahead, forward-thinking planning and the integration of intelligent management systems will be essential to overcome existing limitations and fully realize the potential of smart transportation systems for safer, more adaptable, and more efficient urban intersections.

3.9. Major Findings from the Literature on the Joint Topic

The authors acknowledge that the proposed methodology may have influenced the number of articles included in the review; it followed a systematic approach based on a joint research framework focused on the relevant themes. Nonetheless, the review encompasses studies from diverse countries, highlighting global efforts to develop smart road systems in urban areas. The reviewed articles demonstrate an interconnected network of emerging technologies, strategies, and challenges shaping ITS solutions for smart cities [5,11,14]. Topic selection was also guided by the reading of the selected papers and was thus influenced by the knowledge acquired through a detailed examination of the studies. However, the central focus of this development is understanding how themes (KTs) interrelate, collectively fostering safer, sustainable, and adaptive mobility. Their mutual reinforcement creates a holistic, multi-layered transportation system that underscores the importance of cross-domain collaboration.

At the core is smart infrastructure (KT1), where sensor networks, big data, and AI should form the backbone of modern transportation. Experts like Alanazi [24] highlighted the need for data-rich systems, inspired by advanced countries such as South Korea and Japan, while warning that rapid technological progress can quickly render data obsolete. Consequently, designing flexible, interoperable infrastructures that enable real-time data collection, such as DL-based pedestrian detection [25], dynamic modeling of vehicle technologies, scenario analysis and driving strategy simulations [27,28,33], is essential for advancing smart mobility. These interconnected innovations support smart traffic management (KT2), enabling real-time congestion monitoring and adaptation [19,38,39].

AI-driven rerouting [10], vision sensors [30], and vehicle recognition [31] optimize traffic [38,39], reduce delays [10,29], and cut emissions [28]. However, enhancing mobility also depends on improved behavioral real-time data utilization for more effective system performance [34,37]. With enhanced infrastructure and management systems in place, AVs (KT3) can be increasingly integrated into the smart city ecosystem. Studies employing advanced communication technologies [47] and digital twins [48] have demonstrated the sophisticated capabilities of AVs to navigate complex urban environments safely. These systems both benefit from and contribute to the broader infrastructure, creating a cyclic relationship where infrastructure supports AV navigation, and AV data, in turn, enhances infrastructure responsiveness. However, environmental variability [44], security [8,46,51], interoperability, and cost issues [44,49] highlight the need for robust, scalable solutions before advanced systems can be fully implemented or adopted on a large scale.

In turn, safety (KT4) overlaps with all areas, focusing on risk reduction through proactive crash prediction models [52], advanced safety solutions [16,53,54,55], and real-time emergency detection systems [59]. These rely on robust infrastructure and traffic management, creating a layered approach where safety enhancements reduce crashes, benefiting road infrastructure durability and environmental sustainability through the synergy of multiple and inclusive innovations [56,57]. Consequently, environmental impact mitigation, including emissions reduction and energy-efficient systems, is increasingly integrated with smart infrastructure [9,60,61,62,63,64]. Techniques such as real-time exhaust monitoring using AI classifiers can enable hotspot identification, although data accuracy and response times require further research [60].

Advanced traffic control and eco-driving strategies show promising energy savings but require broader deployment to improve reliability and address infrastructural costs [62,63,64]. While progress is being made in road safety and environmental mitigation, the success of these technologies also fundamentally depends on user experience (KT6). Studies have highlighted that acceptance, trust, and inclusivity are crucial for adopting high-tech systems. Research on driver trust [56,65] and the factors influencing smart mobility adoption [68,71] indicate that technology acceptance depends on perceived safety, transparency, and cultural factors [73].

Personalized route suggestions improve acceptance, increasing transportation appeal [73], but trust and behavioral adaptation remain obstacles to the effectiveness of road traffic systems. Similarly, smart road maintenance (KT7) promotes sustainability through AI-driven predictive maintenance [12,74], smart materials, and drone-based inspections [74,77], enabling continuous monitoring, timely interventions, and an extended infrastructure lifespan within ITS. This approach shifts management from reactive to proactive, although challenges such as scaling, energy consumption, and data integration still persist.

The integration of smart management practices with innovative road planning and design should foster a comprehensive approach to sustainable, future-ready urban mobility. In this context, road intersection design (KT8), which is vital for urban mobility, can benefit from simulation tools and AI control systems that monitor pedestrian behavior [81] and optimize signal timings [84]. Developing adaptable intersections that respond to traffic and environmental changes can improve safety and traffic flow, but infrastructure limitations, upgrading costs, and the integration of sensor data for decision-making require further study. In this context, traffic simulation tools can be essential for integrating AI technologies with road planning, enabling testing, optimization, and visualization of smart infrastructure before deployment. This supports data-driven decisions and the development of adaptive, future-ready road networks aligned with technological progress [27,28,82,84,85,86].

Overall, these themes are not discrete but form an integrated, multi-layered system where infrastructure progress enhances traffic management, supports autonomous vehicles, and promotes sustainability. Conversely, safety and environmental systems strengthen infrastructure resilience and build public trust, which are essential for long-term success. Addressing issues related to scalability, interoperability, cybersecurity, and costs requires a coordinated effort to effectively leverage emerging technologies. As these themes evolve synergistically, the goal of developing safer, greener, and smarter urban roads—transforming mobility in smart cities—becomes attainable, allowing this vision to be turned into reality.

4. Discussion

This review aims to synthesize a broad multidisciplinary understanding of the application of AI technologies in smart cities and road transportation systems. Based on a detailed examination of the selected studies from the joint search and the findings illustrated in the previous section, the review provides insights from 67 articles across multiple disciplines. It categorizes the field into specific key topics (KTs) such as Smart Infrastructure (KT1), Traffic Management (KT2), Autonomous Vehicles (KT3), Safety Enhancements (KT4), Environmental Impact (KT5), User Experience (KT6), Road Maintenance (KT7), and Road Intersection (KT8), with cross-references among them. This approach analyzes overall progress, challenges, and gaps in AI’s application in the areas of smart city infrastructure and intelligent road systems, providing a comprehensive understanding rather than focusing solely on individual findings [1,8,87,88].

While acknowledging that the review is limited to articles indexed in the Scopus database and published between 2013 and mid-2025, and that relevant studies from other databases or emerging research not yet indexed may have been excluded, it aims to provide a broad multidisciplinary synthesis. Consequently, the review might not delve deeply into technical details or context-specific challenges, potentially oversimplifying some complex issues as referred to in [8,11,88,89]. Additionally, the thematic classification into KTs and KSTs involved some subjective judgment, which could have influenced how studies were grouped and interpreted [20,23]. However, the paper selection process focused specifically on articles addressing the joint search theme, thereby providing valuable insights into the role of AI in enhancing urban innovation and transportation efficiency. The review explicitly maps insights to the predefined key questions (KQ1–KQ3) outlined in Section 1, offering structured, targeted answers. Addressing these key questions builds on existing research on advancements in emerging, cutting-edge AI-driven technologies, aligned with KTs and KSTs, and aimed at developing safer, more sustainable, and intelligent road systems within smart cities.

4.1. Answering the Key Questions

AI-driven technologies are now fundamental to modern smart transportation systems, significantly enhancing their intelligence and responsiveness. They are vital for solutions such as smart traffic management, real-time monitoring, AV navigation, and predictive maintenance (KT1-KT4). According to [90,91], advanced AI-driven approaches are used to recognize pedestrians and vehicles with high accuracy [8,25] and predict traffic states to optimize flow [10,29]. Vision-based systems employing algorithms like YOLO enable real-time road sign detection and collision avoidance, which are crucial for autonomous driving [44,45]. To create accurate predictive models, as demonstrated across various fields of knowledge [5,8,11,88], ML and DL also underpin path planning, congestion detection, and crash prediction, enabling the system to respond dynamically to complex urban scenarios [8,19,32,33].

The contribution of AI to the system’s intelligence is comprehensive, as it enables accurate decision-making based on vast amounts of data, enhances safety through early hazard detection, and supports adaptive, context-aware infrastructure management. Despite these advances, many studies emphasize that the full potential of AI has yet to be realized due to challenges such as data quality, environmental variability, and infrastructural limitations, particularly in real-world deployments [25,30]. Additionally, its application should be specifically directed toward city governance [87]. Nonetheless, AI fundamentally transforms road transportation into a more responsive, safe, and efficient domain, driving the evolution of nodes from static entities to intelligent systems capable of learning, adapting, and collaborating across urban environments. To focus specifically on the core role of AI-driven technologies in enhancing system intelligence, decision-making, and operational responsiveness in road transportation, Table 1, Table 2, Table 3 and Table 4 provide a concise overview of selected peer-reviewed studies on this subject (KT1–KT4). This illustrates the role of AI-driven technologies and the extent to which they contribute to making the road transportation system intelligent, addressing KQ1 (see Section 1).

Granularity in the review was also necessary to answer KQ2 in Section 1, enabling an evaluation of the boundaries, overlaps, and integration among relevant disciplines, as well as the benefits of their collaboration in developing effective AI-driven approaches for designing, implementing, and managing smart road systems in environmentally friendly and energy-sustainable cities [5,9,61,62,63]. Given the rapid pace of innovation in AI and smart city technologies, some findings may quickly become outdated, and the review may thus serve as a snapshot prone to obsolescence [24]. Additionally, emphasizing emerging AI-driven solutions might overlook practical implementation barriers, policy constraints, or socio-economic factors that influence real-world adoption [62].

The development of AI-driven smart road systems operates at the intersection of diverse disciplines—urban planning, civil engineering, computer science, data analytics, and policy-making. Each discipline contributes unique expertise: urban planners and road engineers strategize infrastructural layouts; engineers design sensor networks and communication systems; computer scientists develop algorithms; policymakers craft regulations for data privacy and interoperability. There are clear boundaries—such as technical limits of sensor hardware or policy constraints around data sharing—but also overlapping zones where collaboration yields substantial benefits for enhancing users’ experience (KT5–KT8); see Table 5, Table 6, Table 7 and Table 8. Integrating IoT sensors with AI models creates the basis for real-time traffic management [34,35] and predictive maintenance [12,13,74,75,77].

Similarly, the interplay between emerging AI-driven technologies, urban planning, and road design is a crucial and ongoing aspect discussed indirectly throughout the literature, requiring feedback from both technical and policy domains [87,92,93]. Some aspects that should evolve include informed infrastructure design, as technologies like intelligent intersection control, AV pathways, and sensor-based monitoring require rethinking traditional geometries [31,83,92]. Modern intersections should feature adaptable traffic signals, surveillance cameras, and sensor placements that support real-time data collection for AI systems (KT8). This involves creating flexible, scalable layouts capable of accommodating sensor arrays and dynamic signal control, moving away from static configurations [93]. AI and IoT can significantly enhance road network topology and connectivity by enabling more detailed traffic management, fostering interconnected and adaptable networks that support vehicle-to-infrastructure (V2I) and vehicle-to-vehicle (V2V) communication, leading to more efficient, responsive transportation systems [29,47,53]. Planning should prioritize high-connectivity corridors, dedicated lanes for AVs, and flexible routing options that facilitate AI-driven rerouting and congestion mitigation. Therefore, data-driven design criteria should integrate extensive sensor networks and big data analytics to shift from capacity-centric to systems emphasizing real-time responsiveness and environmental sustainability [5]. This impacts lane widths, intersection spacing, and traffic signal placement, underscoring the need for standards that support sensor-rich, networked environments.

Vulnerable user integration also needs AI-optimized safety systems, such as smart crosswalks and pedestrian detection, which necessitate redesigning crosswalks, sidewalks, and signage to improve visibility and sensor effectiveness [56,57,72]. According to [89,90], road infrastructure should ensure inclusivity for vulnerable users, including seniors and disabled individuals, aligning with evolving standards for smart, age-friendly mobility. Meanwhile, road design should incorporate materials and structural features that support sensors and autonomous vehicle interactions, such as embedded sensors in pavement, to enhance durability under increased technological loads [77]. Maintenance strategies should also be integrated from the start, utilizing predictive analytics to develop maintenance-friendly road networks that adapt to technological demands [75].

Overall, blending AI technology with thoughtful planning and design will enable smarter, safer, and more sustainable road networks, fostering a future of resilient urban mobility. In this view, simulation tools are essential in urban planning and road design, as they allow testing, optimization, and visualization of smart infrastructure concepts through scenario analysis before implementation, as the literature informs [27,53,80,84,85,86]. They enable planners to model adaptive intersections, traffic flow, and safety features under various scenarios, assessing how AI-driven systems interact with physical layouts [80]. However, the examination of microsimulation models goes beyond the objective of this paper and refers to the specialized literature on the topic [86].

Findings highlight that road network designs need to evolve from fixed geometries to adaptable, sensor-integrated layouts that respond to real-time conditions [93]. Connectivity should focus on high-capacity corridors enabling seamless vehicle-infrastructure communication. Embedding sensors, cameras, and communication nodes into roads will support continuous data collection and AI control, fostering interconnected and adaptive networks that support V2I and V2V communication [29,47,53]. The infrastructure must remain modular and scalable for future upgrades, accommodating sensor arrays and dynamic signal control [93]. Safety can be enhanced through redesigned crosswalks and sensor-equipped zones for pedestrians and vulnerable areas [56,57]. Additionally, integrating green infrastructure, dedicated lanes for electric and autonomous vehicles, and smart materials promotes environmental sustainability [25,61]. Developing AI-driven technologies requires flexible, connected, and eco-conscious planning supported by updated standards, interdisciplinary collaboration, and proactive policies [90,92] to create resilient and sustainable urban mobility systems. According to Yang et al. [90], knowledge could be transferred from the safety field of AVs to the other modes. Further margins of overlap are especially evident in areas like digital twins and autonomous vehicle deployment, where engineering, AI, and policy must align to ensure safety, reliability, and ethical compliance [46,48].

Dialogue among disciplines fosters a comprehensive understanding, enabling the creation of resilient, flexible, and contextually appropriate systems. For example, stakeholder engagement and policy dialogue are crucial for addressing privacy concerns [33] and ensuring system scalability across diverse urban environments. Over time, cross-disciplinary communication promotes the development of standards and protocols that enhance data interoperability, security, and system resilience in complex traffic networks, which are elements essential for scaling solutions and maintaining seamless operation in various scenarios [25,29,94]. This collaborative approach strengthens the robustness and adaptability of AI-driven methods, ensuring that technological innovations remain practically viable within socio-political contexts. Advances in AI, IoT, data analytics, and simulation technologies make it feasible to assess the adaptive capacity of smart road systems during the transition to sustainable urban environments, thereby addressing KQ3 (see Section 1).

Many recent studies in the selected articles, including those on predictive analytics and real-time monitoring, aim to evaluate and enhance system resilience. These models demonstrate how continuous data collection, environment-aware AI, and predictive algorithms can quantify responses to dynamic conditions [19,30,31]. They support simulation and stress-testing under various scenarios, assessing system performance amid environmental changes, traffic fluctuations, or infrastructure failures (see Table 2 and Table 4 related to KT2 and KT4). Methodologies such as FL [8,41,69] and digital twins [48] enable stakeholders to monitor and proactively adapt operations, strengthening resilience against environmental variability, traffic surges, and infrastructure wear. Incorporating weather data and environmental conditions further improves predictive accuracy, allowing dynamic system adjustments based on external factors, highlighting how continuous data integration and ML can quantify and enhance adaptive capacity of roads [62].

4.2. Insights and Reflections in AI Innovations for Road Transportation and Smart Cities

Building on the above, it is essential to discuss insights and reflect on the transformative potential of AI innovations for road transportation and smart cities, highlighting their prospects. Fully assessing and enhancing the adaptability of complex urban transportation systems remains a challenge. According to Mishra and Singh [91], this requires a holistic, multi-layered approach involving multiple stakeholders, such as urban planners, engineers, data scientists, and policymakers, collaborating to identify vulnerabilities, simulate future scenarios, and iterate system improvements. While technological tools have advanced significantly, gaps still exist in areas such as data interoperability, cybersecurity, system scalability, and environmental robustness.

Nonetheless, the technological maturity of AI-driven smart transportation solutions varies, with many components, such as real-time monitoring, predictive analytics, and vehicle communication systems, approaching higher technology readiness levels and demonstrating practical feasibility in pilot projects [5,8]. Widespread deployment still faces challenges related to high initial costs, infrastructural upgrades, maintenance requirements, and ensuring long-term cost-effectiveness, particularly in resource-limited urban environments. This underscores the need for scalable, phased implementation strategies and comprehensive economic assessments [95,96].

Furthermore, socio-political factors such as public acceptance, legal frameworks, and resource allocation play a crucial role in enabling long-term adaptive responses. Consequently, evaluating and improving the adaptive capacity of intelligent transportation systems becomes increasingly feasible through advanced modeling, predictive analytics, and real-time data integration. While AI and big data provide valuable solutions, limitations remain in the data used or collected by traffic models and city planning tools for intelligent systems, requiring further investigation and analysis. These efforts support the transition to sustainable urban environments by fostering resilient roads capable of accommodating evolving demands, environmental pressures, and technological changes. However, realizing this potential also depends on ongoing dialogue across disciplines, the establishment of robust standards, and active engagement of diverse stakeholders to develop adaptable, scalable, and environmentally resilient smart mobility systems. In this context, the scope of the research aims to be inclusive of a broad range of stakeholders involved in transportation and related fields.

5. Conclusions

This review highlights the role of AI in advancing road transportation systems and smart cities. It explores how cutting-edge AI technologies—intelligent infrastructure, real-time traffic management, and autonomous driving—can improve the quality of life for road users and urban communities. Examining recent advancements, it aims to provide insights for the future development of sustainable transportation infrastructure and urban road planning strategies.

Employing a joint search for “Artificial Intelligence,” “Road Transportation System,” and “Smart Cities,” and using semantic analysis of the texts, this review categorizes 67 studies across multiple disciplines to identify specific KTs and KSTs—Smart Infrastructure, Traffic Management, Autonomous Vehicles, Safety Enhancements, Environment Impact, User Experience, Road Maintenance, and Road Intersection—and offer a comprehensive overview of progress, challenges, and gaps, moving beyond individual study results. This approach directly addresses the key questions, demonstrating how AI models enhance transportation intelligence (KQ1). It highlights concrete AI applications and underscores the necessity of interdisciplinary collaboration (KQ2) to develop scalable and resilient smart road systems. It also discusses methods and models for assessing and improving the adaptive capacity of transportation systems (KQ3), using data, simulation, and collaborative approaches to address evolving urban conditions.

As urban areas grow more complex, advanced management and digital innovations can create safer, more efficient, and adaptable mobility networks. Smart cities emphasize connectivity and digital engagement, with AI systems transforming urban mobility through decision support, autonomous operations, and real-time responses. Simulation tools are essential for pre-implementation testing, optimization, and visualization of smart infrastructure concepts via scenario analysis in urban planning and road design. Road engineering can model adaptive intersections, traffic flow, and safety features to assess the interaction of AI-driven systems with physical layouts, while road design should incorporate materials and structural features that support sensors and autonomous vehicle interactions. However, ensuring long-term effectiveness requires evaluating costs, energy efficiency, and data accuracy, as well as addressing system integration for efficient workflows. Despite these innovations improving safety, traffic flow, and environmental sustainability, addressing infrastructural, environmental, and socio-political challenges remain crucial.

Although comprehensive and multidisciplinary, this review has limitations stemming from data source constraints and selection criteria. Primarily focused on the intersection of AI, transportation systems, and smart cities, the available literature consists largely of pilot projects, simulations, and small-scale implementations, rather than large-scale, real-world deployments. Consequently, these limitations may influence the overall perspectives presented. Many of the discussed technologies—such as AI-driven traffic management, AVs, and predictive maintenance—remain in experimental or theoretical stages, making it challenging to fully evaluate their long-term effectiveness and robustness across diverse urban environments. Additionally, reliance on the Scopus database and publications from 2013 to mid-2025 may have excluded relevant studies, recent conference presentations, or valuable insights from gray literature by focusing mainly on peer-reviewed journal articles. Furthermore, the authors’ assumptions regarding the relevance and interpretation of search terms could introduce additional biases. While aiming for broad synthesis, the review may not delve into all technical details, and the rapid pace of technological advancement could quickly render some findings outdated. Finally, emphasizing emerging technologies risks overshadowing practical implementation barriers, policy constraints, and socio-economic considerations. Acknowledging these limitations, including potential biases from the authors’ assumptions, helps contextualize the findings and highlights avenues for future research. However, the working protocol was described in detail and can be applied under various conditions to conduct similar literature review processes.

Going forward, future research should address the following recommendations for both the research community and policymakers:

• Future updates should expand and diversify data sources to incorporate larger-scale, regional, and empirical case studies that move beyond pilot projects and simulations. This will improve evaluations of the real-world effectiveness, long-term robustness, and scalability of AI-driven transportation solutions across diverse urban contexts, and facilitate the examination of societal acceptance and privacy concerns related to data collection and usage.
• Establish universally accepted standards for data interoperability, cybersecurity, system resilience, and privacy protection. Such frameworks will facilitate smoother integration, foster public trust, support the wider deployment of intelligent mobility solutions, and address concerns around data security and ethical use.
• Further investigation is necessary to understand and mitigate socio-economic and policy barriers, including deployment costs, energy consumption, digital divides, and socio-political challenges. Environmental sustainability could be emphasized through eco-friendly materials, green infrastructure, and renewable energy policies. It is also crucial to ensure accessibility for vulnerable groups to promote equitable benefits from smart mobility and foster social acceptance.
• Future studies should focus on creating effective models for cross-sector engagement, involving engineers, urban planners, policymakers, industry leaders, and communities, to better align technological innovation with societal values, ethical considerations, privacy protections, and adaptable regulatory frameworks. This will support a responsible and inclusive transition from research to impactful deployment.

While progress is underway, fully realizing smart road systems requires addressing ongoing technical, operational, and societal challenges. Applying lessons learned gradually—along with the increasing availability of advanced modeling, simulation, and communication tools—can support more scalable and sustainable solutions. Ultimately, greater interdisciplinary collaboration and stakeholder involvement will be key to fostering resilient, safe, and efficient urban mobility, helping to shape more adaptable and livable city environments in the future.