Smart Intersections and Connected Autonomous Vehicles for Sustainable Smart Cities: A Brief Review

Abstract

As the importance of safety, efficiency, and sustainability in urban transportation becomes more apparent, intelligent transportation systems are changing and growing. Smart intersections play a crucial role in different parts of this context. Technologies such as Vehicle-to-Everything (V2X) communication, artificial intelligence, multi-sensor data fusion, and more are incorporated into these intersections to improve capacity and safety and reduce damage to the environment. This literature review aims to merge various recent works on advancing smart intersection technologies, their thematic application, methodological approach, and regional implementations. Highlighting adaptive traffic signal control, real-time data processing, and connected autonomous vehicle (CAV) integrations sheds light on the way the effectiveness of transportation in cities can be improved. At the same time, this study tackles questions of cybersecurity and standardization. This review provides insights for researchers, policymakers, and practitioners who aim to improve transportation systems’ sustainability, fairness, and operability.

1. Introduction

Unprecedented urbanization and advancements in transportation technologies necessitate safer, faster, and more sustainable mobility solutions. Despite recent efforts to reduce road accidents, they are still a principal problem for enhancing road transportation systems. There are estimated to be around 1.19 million road traffic deaths worldwide yearly, incurring significant financial and non-financial costs to societies and governments. For instance, traffic death costs range from one to three per cent of the Gross Domestic Product (GDP) [1]. Even in developed countries, the number of fatalities and injuries resulting from these accidents remains significant. Italy, as an advanced and industrialized country, reported 3039 deaths and 224,634 injuries due to traffic crashes in 2023 [2]. As a result, improving the safety of road networks is one of the main objectives of decision-makers. Among the diverse components of road infrastructures, intersections perform a pivotal role in urban traffic management by facilitating the safe and efficient movement of vehicles, pedestrians, and cyclists. This is despite the high rate of accidents that occur at intersections, especially in urban areas. According to the AAA Foundation for Traffic Safety report, between 2010 and 2019, about 39 per cent of urban road traffic deaths in the United States were related to intersections [3]. Meanwhile, with the rapid progression of technological advancements and the emergence of novel innovations, a promising opportunity has been created to design smarter intersections to improve safety and enhance performance.

By leveraging the technologies required for smartness, smart roads can create a transformative change in increasing sustainability, safety, innovation, and inclusiveness. These smart infrastructures will be developed to redefine interactions among vehicles, users, and road environments by integrating advanced technologies, particularly autonomous vehicles (AVs) and connected and automated vehicles (CAVs). They will leverage cooperative intelligent transportation systems (C-ITSs) to facilitate seamless vehicle-to-vehicle and vehicle-to-infrastructure interactions. In addition to using various sensors capable of detecting performance, safety, and environmental factors, these roads will also utilize complementary technologies such as dedicated AV/CAV lanes, IoT-based surveillance systems, dynamic traffic management measures, and infrastructure capable of harvesting energy.

Transportation Management, Data Analytics, and Vehicle-to-Everything (V2X) communication in real-time traffic management in smart intersections, cornerstones of intelligent transportation systems (ITSs), converge in core ITS components that lead the way towards sustainable urban mobility. These intersections combine digital technology sensors and real-time traffic monitoring systems [4,5,6], stepping beyond conventional intersections (e.g., signalized intersections). Intelligent junctions have gained attention in the scientific community (see Figure 1). The rapid development of connected vehicle technologies and emerging AI-driven traffic management for urban sustainability policies place advanced intersections as hotspots of next-generation urban mobility systems [7,8]. Most of the literature on smart intersections has highlighted isolated technological practices (e.g., adaptive signal control and connected autonomous vehicle (CAV) coordination challenges).

This review addresses the following key research questions:

• What are the most effective approaches to evaluating the performance of smart intersections?
• What are the challenges and opportunities for integrating smart intersections into smart city ecosystems?
• How should future research directions be prioritized to maximize smart intersection technologies toward sustainable, inclusive, and resilient urban mobility systems?

This article gives an overview of the state of the art of these intersections, with the following main objectives:

• Highlights the key technologies underpinning smart intersection development that facilitate traffic optimization based on AI, real-time Data Analytics, and V2X communication systems;
• Reveals approaches for analysis, including simulation models, optimization methods, and real-world deployment case studies, to evaluate smart intersections’ effectiveness;
• Investigate the smart intersection ecosystem within smart cities, including its applicability in all transport modes, sustainability, and digital governance solutions;
• Analyze the top challenges for these cyber-physical intersections, including but certainly not limited to cybersecurity threats, generic standard hurdles, and scaling advanced intersection technologies in internet-limited regions;
• Analyze the future research directions for smart intersections, with a focus on AI-based adaptive traffic control, digital twins, and next-generation mobility systems.

As a comprehensive and more holistic analysis, this review could be a useful summary of the state of the art for researchers and decision-makers in solving smart intersection deployment challenges across various urban contexts.

2. Methodology

This study follows the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines to ensure a systematic and comprehensive review [10]. This helps to make the review process more reproducible and transparent.

A structured search was conducted across the following databases to identify relevant scientific articles in transportation engineering, computer science, and urban planning: Scopus, Web of Science, IEEE Xplore, ScienceDirect, and Google Scholar. We developed a Boolean search query to improve the retrieval of relevant articles while minimizing irrelevant results as much as possible. The final search query that was used in all databases was as follows:

• (“Smart intersections” OR “intelligent intersections” OR “AI-based traffic control” OR “adaptive intersection control”);
• AND (“connected vehicles” OR “autonomous vehicles” OR “V2X communication” OR “CAVs”).

Also, references in the top-cited review papers were manually searched to identify additional studies that were not found in the initial search due to manual searches.

The quality and relevance of the selected studies were ensured using the following inclusion and exclusion criteria:

○

Journal articles (peer-reviewed), book chapters, and conference proceedings from the period 2015–2024;

○

Smart intersection, intelligent traffic control, and smart road implementation reports;

○

Articles with quantitative or technical results;

○

Position papers, non-peer-reviewed and without empirical evidence;

○

Only studies published two years prior (2015 or previously) unless they are foundational works related to concepts’ building blocks, etc.;

○

Duplicated studies across multiple databases.

Following these criteria, 350 articles were first screened according to their titles and abstracts in total, and then, we screened 213 high-quality studies for a final in-depth analysis after screening included/exclusion criteria and removing duplicates. Two independent reviewers assessed each study and discussed discrepancies to guarantee inter-reviewer reliability. This helps guarantee the robustness of the literature screening and the reliability of the results.

Though this study is systematic in nature, it has some limitations. Most importantly, the studies were largely from the academic literature, and important industry reports or government policy documents were possibly missed. However, the proposed methodology guarantees a systematic and transparent process of review that is the basis for analytically evaluating trends, problems, and tendencies in smart intersection development.

3. Integrated Technologies for Smart Roads and Intersections

Smart roads form the backbone of intelligent transportation systems, enabling connected, efficient, and sustainable urban mobility. These high-tech nodes can be equipped with IoT sensors, Vehicular Communication systems (V2X), and AI-enabled navigation and control networks so that there can be real-time decision-making and dynamically controlled traffic streams on road networks. One of the building blocks of these roads is how they can allow connected and autonomous vehicles (CAVs) to use V2X communication, which includes infrastructure communication (V2I), vehicle-to-vehicle communication (V2V), and vehicle-to-pedestrian communication (V2P) (Figure 2). These coupled systems improve situational awareness and streamline traffic flow, making roads safer by using high-precision technologies, including LiDAR/millimeter-wave radar/edge computing for real-time analytics [4]. V2X technology enables predictive traffic control, risk detection in advance, and synchronized traffic light management through exchanging information without significant delays. They also contribute to fulfilling the sustainability goal of lowering traffic congestion and carbon emissions. IoT sensors are embedded into the road, continuously surveilling traffic density, vehicle speeds, and traffic conditions, providing input data to AI-driven traffic management systems that respond to signal timing and lane allocation using real-time data [5,6]. Research has shown that these responsive control methodologies substantially reduce fuel usage and greenhouse gas emissions, further proving the environmental role of dynamically connected transportation networks [11,12,13,14,15,16,17,18,19].

One of the critical functions of innovative roads is real-time infrastructure monitoring, which enhances the durability, performance, and robustness of transportation networks. Roads with connected sensors to measure the road pavement condition [5] can detect surface deterioration [20,21,22], structural fatigue [23], and environmental wear marks [24,25]; therefore, road operators can take preventive maintenance strategies and be aware of repair costs. Using these data, AI-powered predictive maintenance models predict infrastructure failures before they happen to aid in the long-term sustainability of roadways. High-tech roads also amalgamate renewable energy systems such as solar-powered roads, kinetic energy harvesting, etc., which encourage urban sustainability as they reduce the reliance on traditional energy sources [26,27]. Adaptive road infrastructure and IoT-enabled intersections work together in an intelligent mobility ecosystem.

According to the PIARC [28], smart roads can be classified based on their capabilities to host CAVs, which can be determined with the interaction of two indicators: the Level of Service for Automated Driving (LOSAD) and the Infrastructure Support for Automated Driving (ISAD). The LOSAD indicates the physical readiness of the road segment to support automation, whereas the ISAD represents the connectivity support. The following types of roads are therefore considered [28]:

-

Human way: the road infrastructure is not able to host CAVs.

-

Assisted way: the road infrastructure is adequate to perform autonomously, but this condition may stop due to several causes.

-

Automated way: the infrastructure exhibits reasonably good connectivity and physical infrastructure capabilities, so disengagements or takeover requests would be lower compared to the Assisted way and Human way.

-

Fully Automated way: the road presents a continuous Operational Road Section (ORS), ensuring Operational Design Domain (ODD) compatibility with most level 3–4 vehicles.

-

Autonomous way: Similarly to the Fully Automated way, the connectivity infrastructure supports cooperative driving, so the infrastructure can receive and transmit tailored instructions to all vehicles, micromanaging traffic performance. This road segment type is exclusive to level 4–5 CAVs.

Other international guidelines suggest different criteria to classify smart roads [29], also in terms of the use of ITSs [30].

Nowadays, smart roads, smart intersections, sensors, and CAVs are very topical issues in the scientific community, as can be immediately noticed by examining Figure 3. This figure represents in graphic form the keywords and co-occurrences obtained with bibliographic analysis in Scopus and allows for the identification of the main areas of interest and the detection of topics/subtopics that occur most frequently in the scientific literature. Figure 3 was obtained from the Scopus database using the VOSviewer software. In this figure, the size of the nodes is proportional to the number of times a term has been used in the scientific literature. The thickness of the links between the nodes is proportional to the strength of the connection.

4. Keys Technologies for Smart Intersections

Smart intersections are important junctions in the traffic network of a city that are controlled and monitored with the help of modern technologies to increase safety, traffic capacity [31,32], and environmental friendliness.

They integrate next-generation vehicle communication, artificial intelligence, IoT sensors, edge computing, and real-time traffic modeling to maximize efficiency and enable next-generation transportation systems. The main technologies that support these intersections are as follows:

-

V2X communication, enabling connectivity for traffic optimization. The technology known as V2X establishes real-time communication between vehicles, infrastructure, and pedestrians to improve traffic flow and safety. Other communication systems include vehicle-to-infrastructure (V2I), which improves capacity and reduces congestion through dynamic signal control [33]; vehicle-to-vehicle (V2V), which may prevent accidents and optimize lane usage [34]; and vehicle-to-pedestrian (V2P), which can be used, for instance, for real-time pedestrian signals [35].

-

Traffic management and control with the help of artificial intelligence. This topic includes AI and machine learning algorithms that leverage camera, sensor, and vehicle-connected data to enhance traffic efficiency; adaptive signal control: real-time operation based on traffic congestion levels [36]; accident prediction and emergency response: AI detects potential dangers and acts before the danger occurs [37,38]; and enhanced signal control efficiency, i.e., the application of machine learning in congestion modeling [39]. Recent advances in image processing techniques also enable the precise reconstruction and measurement of the transverse profiles of worn-out tracks, significantly contributing to predictive maintenance and the early detection of railway and tramway track deterioration in an urban context [11,40].

-

IoT and multi-sensor data fusion for traffic sensing. IoT-based systems improve intersection control by integrating data from multiple sources: real-time traffic monitoring, peak-hour traffic flow, and speed [40]. Environmental monitoring: sensors that detect fog, rain, and air pollution can be used to modify traffic controls [9]. Predictive maintenance allows for the immediate identification of potential failures in infrastructure for timely replacement [41].

-

Advanced sensing technologies: improving situational awareness. These technologies include LiDAR (Light Detection and Ranging), which gives a very clear view of traffic flow and movement [42]; thermal and radar imaging, through which vehicles and pedestrians can be detected during night and fog conditions [43,44]; and multi-sensor fusion: LiDAR, cameras, and radar for better safety [45,46].

-

Edge computing: real-time data processing for low-latency decisions. Decentralized traffic management reduces latency in traffic light learning processes [47]. Cooperative vehicle maneuvering enhances the coordination between conventional and automated vehicles and real-time congestion management.

-

Cybersecurity in smart intersections. End-to-end encryption ensures secure Vehicle-to-Everything connectivity [48]. Intrusion detection systems (IDSs) prevent security threats [49] and ensure data integrity with blockchain features to prevent data tampering [50,51]. Tamper-proof identity management for vehicles and infrastructure units is one of the most important outcomes of blockchain-based authentication systems enabled by distributed ledgers [52]. For resource-constrained automotive environments, lightweight blockchain models such as Hyperledger Fabric and IOTA’s Tangle are suitable [53]. Secure and auditable data sharing requires enforcing access rules and using time-stamped transactions—both of which can be effectively implemented through smart contracts [54]. However, challenges such as latency, consensus overhead, and energy consumption raise concerns about the real-time applicability of blockchain, particularly in high-throughput V2X (Vehicle-to-Everything) scenarios.

-

Integration of connected and autonomous vehicles (CAVs). This is the control of CAVs’ movements with traffic control devices [55] and lane-level navigation assistance that improves autonomous vehicle navigation [56,57]. Automated hazard detection increases safety for all road users [58,59,60,61].

-

Dynamic traffic management: Adaptive lane reallocation: temporary lane configurations are leveraged by this approach to improve efficiency during peak hours [62,63]. Emergency vehicle prioritization: this provides a smooth passage for ambulances and fire trucks [64,65].

The safety features of innovative intersections may include the following:

-

Signal phasing is modified to avoid accidents [66,67].

-

Pedestrian and cyclist detection with thermal cameras and LiDAR for safe pedestrian and cyclist movement [68,69,70]; incident detection systems, real-time monitoring, and notifications to emergency services in real time [71,72,73]; and advanced concrete safety barriers capable of withstanding multiple heavy vehicle impacts also enhance intersection safety significantly [74].

Smart intersections will have the capacity to provide environmental benefits through traffic optimization that enhances fuel efficiency and reduces pollution [75,76]. Pollutant emissions from road traffic can be checked by image processing techniques [11] and air quality monitoring systems.

Also, smart intersections may comprise the integration of EV charging stations [77,78].

Advanced intersections will be able to enhance transportation equity by providing equal opportunities, for instance, by extending pedestrian crossing times for elderly and disabled pedestrians [79,80] and optimizing bus and tram signals for minimal delay [81,82,83,84,85]. These innovative intersections can help the mobility integration of various transport modes [5,86] and can support ride sharing and bike sharing for green transport routes in city centers [87,88,89].

Table 1. Key technologies and functionalities in smart intersections.

Category	Key Technologies	Functionality	Key Reference(s)
Connectivity	V2X communication (V2I, V2V, V2P)	Enables real-time connectivity for improved traffic flow and safety	[33,34,35]
Artificial intelligence	AI-based traffic control (adaptive signal control, accident prediction)	Optimizes signal control, reduces congestion, and enhances traffic efficiency	[11,36,37,38,39,90,91,92,93,94]
Sensing and IoT	IoT and multi-sensor fusion (traffic and environmental monitoring, predictive maintenance)	Enhances real-time monitoring and adapts traffic control to environmental conditions	[9,40,41,42,43,44,45,46]
Computational technologies	Edge computing and real-time congestion management	Processes data at the edge for low-latency, real-time traffic decisions	[47]
Security	Cybersecurity (encryption, IDS, blockchain)	Secures data transmission, prevents cyber threats, and ensures data integrity	[48,49,50,51,52,53,54]
Autonomous vehicle integration	Coordinated CAV interactions and lane-level navigation assistance	Enhances traffic coordination for autonomous and connected vehicles	[55,56,57,58,59,60,61]
Traffic optimization	Dynamic lane allocation and emergency vehicle prioritization	Improves traffic efficiency through adaptive lane management	[62,63,64,65]
Safety	Collision prediction and incident detection systems	Enhances road safety by predicting and preventing accidents	[66,67,68,69,70,71,72,73,74]
Environmental impact	Traffic signal optimization and emission control	Reduces emissions, optimizes fuel use, and integrates EV charging	[11,75,76,77,78]
Equity and accessibility	Extended pedestrian crossings and public transit prioritization	Improves accessibility for vulnerable users and prioritizes public transit	[79,80,81,82,83,84,85]
Multi-modal integration	Public transit synchronization and ride sharing support	Supports integrated mobility solutions for seamless transportation	[5,86,87,88,89]

summarizes the key technologies and functionalities in smart intersections.

Deep neural networks (DNNs), reinforcement learning (RL), and graph neural networks (GNNs) have been widely used as AI methods in connected autonomous vehicles (CAVs) and V2X applications. Although DNNs provide high accuracy in perception tasks, they require large datasets and computational resources, which limits their real-time application [91,92]. Despite the need for basic training and fine-tuning, RL approaches, especially deep RL, are well suited for dynamic traffic control scenarios [93]. GNNs can be used to model complex interactions between vehicles and infrastructure, which can capture the network dynamics in V2X communications. Still, their disadvantage is their complexity, which can increase computational demands [94]. A better understanding of these tradeoffs can help select appropriate AI techniques for specific smart intersection applications.

As depicted in Figure 4, smart intersections for urban environments are designed for efficiency and technological progress. The core of the structure is the Smart Intersection Hub which uses adaptive methods for traffic control dynamic management and sensor-based monitoring to adjust signals continuously. The setup integrates AI and edge computing to support decision-making through scenario evaluation and data-based forecasts for traffic flow. Data collected from vehicle-mounted sensors and environmental monitoring devices are vital in enhancing AI-powered traffic control methods. The system includes adaptive traffic light systems, pedestrian safety protocols, and dynamic lane adjustments to optimize traffic flow, promoting sustainability and operational efficiency. Furthermore, it seamlessly integrates into the smart city infrastructure, working cooperatively with traffic management hubs, cloud services, public transport networks, and emergency response units. This ensures Vehicle-to-Everything (V2X) communication, air quality monitoring, and real-time data exchange across the city. By fostering a data-driven traffic infrastructure, this framework aligns with the overarching goal of establishing eco-conscious urban transportation networks.

5. Methodological Approaches Used for Smart Road and Intersection Design and Analysis

Various methodological approaches, including quantitative, qualitative, and hybrid techniques, feature in the planning, implementation, and measuring of non-traditional intersections. These methodologies enable researchers and practitioners to examine the traffic volume, enhance intersections’ performance, and analyze the effect that advanced transportation technologies have on the user. The key methods are as follows:

-

Quantitative methods: normally, these methods are based on mathematical modeling, simulation, and AI-driven optimization:

(A)

Optimization models: Smart intersections utilize the optimization techniques of linear programming, mixed-integer programming, and dynamic programming for intelligent real-time traffic control. These models minimize delays, increase throughput, and optimize traffic signal phasing for varying demand [69,95,96,97,98,99,100].

(B)

Traffic simulation models: microscopic, mesoscopic, and macroscopic simulation to evaluate various configurations—traffic patterns under different intersection conditions and traffic regulation systems, including the adaptive traffic signal [101,102,103,104].

(C)

Traffic prediction through machine learning predictive models (neural networks, regression analysis, and deep learning algorithms) for high-level and real-time traffic data analysis [105,106,107,108,109,110,111].

-

Qualitative methods: These interpret human factors and implications for policy-making. Qualitative approaches are essential to understanding user behavior, stakeholder engagement, and policy frameworks. They are as follows:

(A)

User behavior studies: studies show that intelligent intersections enhance and facilitate driver/pedestrian/cyclist interactions to help design and enforce AI-based traffic control systems [112,113,114].

(B)

Interviews with stakeholders: collaboration between policymakers, transportation agencies, and community groups helps developers deploy smart intersection modes in the right way for the common good as well as city-specific/municipal demands [115].

(C)

Policy and regulatory analysis: This explains the impact of existing transportation policies and presents regulatory architectures that are necessary to support CAV adoption at intersections [4]. Qualitative methodologies are important for ensuring that technological advances align with user needs, societal expectations, and regulatory requirements.

-

Hybrid approaches: These combine quantitative and qualitative information. These methods are as follows:

(A)

Traffic gen simulations: by integrating actual driver behavior at an intersection, pedestrian demand patterns, and human reaction times in intersection simulators, we can improve the precision and expand the variety of intersection performance measurements [116,117,118,119].

(B)

Impact of smart intersection technologies: researchers assess the economic feasibility of different smart intersection technologies by combining quantitative performance metrics (e.g., travel time savings or increased safety) with qualitative assessments (e.g., public perception, sustainability benefits) [120,121,122,123,124,125,126].

-

Evaluation quantifiers: Key Performance Indicators (KPIs) are used for planning smart intersections. The most important metrics are as follows:

(A)

Traffic efficiency metrics: average delay, number of stops, total travel time, and queue length [127].

(B)

Safety indicators: accidents, conflict points, and near-miss incidents [128,129,130,131,132,133,134].

(C)

Environmental impact indicators: vehicle emission reductions, fuel cost savings, and energy-efficient gains made possible, for instance, by optimal adaptive control systems [135,136,137,138].

(D)

Satisfaction survey with users: Survey studies can be conducted on the perception, usability, and accessibility of advanced intersections with different types of road users [139,140]. This evaluation metric allows us to ensure that smart solutions are in line with more significant city sustainability, safety, and mobility goals.

-

New approaches: Research directions for smart intersections. The emerging methodologies and new approaches include the following:

(A)

Adaptive traffic control reinforcement learning: AI-powered reinforcement learning systems learn to tailor signal timings for each cycle controlling the intersection, where the basic goal is to maximize intersection performance over time [141,142,143,144].

(B)

Big Data and IoT integration: the integration of Big Data analysis with IoT sensor networks provides real-time analysis of large-scale traffic datasets, improving predictive modeling and real-time location-adaptive network operation [145,146,147,148].

(C)

Digital twin technology: digital twins allow cities to rapidly iterate across a wide range of real-time virtual representations of intersections, enabling planners to evaluate multiple traffic condition scenarios, test new traffic operations strategies, and assess system efficiencies without actual implementation [149,150,151].

In smart intersections, it is essential to sample traffic data, which can be collected via various techniques and technologies, and the “fixed-spot measurement” (i.e., inductive loops, pneumatic tubes, detectors video cameras, etc.) and “probe vehicle data” systems (i.e., floating car data, FCD) are both widely used. An unusual technique employed is the “moving car observer method” (MOM), also called “moving car observer” (MCO), established by Wardrop and Charlesworth. Both in fixed-spot measurement methods and in the MOM, the vehicle detection, tracking, and counting processes can employ numerous algorithms based on deep learning and, in particular, the YOLO algorithm [152,153]. The YOLO variants (i.e., YOLOv1, YOLOv2, YOLOv3, YOLOv4, YOLOv5, YOLOX, YOLOR) have complex network structures and many network parameters. For instance, YOLOv4-tiny can be applied to achieve faster vehicle detection [154].

6. Case Studies of Smart Roads and Intersections

There is a high degree of heterogeneity globally in deploying smart roads and smart intersections, reflecting differing policy priorities, tech capabilities, and mobility challenges in urban areas. Some cities have achieved measurable improvements in traffic congestion, safety, and sustainability, yet many infrastructural, economic, and societal factors shape their success. Some case studies are briefly considered here: Amsterdam, Toronto, London, Korea, and Barcelona are critically appraised based on their strategies for smart transportation systems; in addition, C-Road Platforms are considered.

Amsterdam has implemented and tested the so-called Shockwave jam services on the A58, the Netherlands’ first cooperative vehicle–roadside system for the motorway between the cities of Eindhoven and Tilburg [155]. Shockwave Traffic Jams A58 was the first in Europe to implement the prescribed data security measures for WiFi-P and immediately raised them to a higher level [156,157,158]. Toronto has implemented AI-driven traffic optimization systems [159,160,161].

SCOOT has been the lifeline for real-time traffic optimization in London for quite some time; the new kid on the block FUSION comes with machine learning and digital twin modeling. Even with FUSION, with its 15% improvement in travel time reliability, the deployment is a testament to AI-powered urban traffic control problem-solving when presented on such a large scale. Public perception is still off, with a certain level of mistrust in the transparency of AI decision-making. Because adaptive traffic models need constant recalibrations, the long-term financial sustainability of FUSION is possible in cities with less constrained budgets [162,163].

Pyeongtaek City (South Korea) is implementing edge AI-powered intelligent intersections in South Korea, which is a pioneering effort at AI data centralization for real-time decision-making and traffic signal coordination. Unlike cloud-based models, the edge AI conducts in-device (local) data processing and lowers latency for faster response times. The model resulted in quantifiable benefits such as a 25% reduction in congestion and an increase in pedestrian safety by an average of 30% [164].

Barcelona has hyper-optimized pedestrian areas and eliminated vehicle traffic in most city center streets with the “Superblock model” project. This model tends to favor central urban districts at the expense of neighboring road congestion. Instead, congestion is relocated by routing traffic away from the intended destination places, even in the case of AVs [165].

The case studies show that although we believe smart road and intersection solutions are game changers for traffic efficiency and safety, many things work together for successful deployment. The Amsterdam C-ITS model illustrates the difficulty in building connected vehicle systems in mixed-traffic environments. The AI signals in Toronto show what happens when the efficiency vs. equity tradeoffs come to vulnerable road users. London’s FUSION model presents the conundrum between the optimization of algorithms and transparency; South Korea’s edge AI approach introduces security and scalability concerns. Finally, Barcelona’s Superblock initiative illustrates the socio-political aspects of moving towards Western-oriented urban mobility.

In Europe, with the C-Roads Platform, various road operators join together to harmonize cooperative intelligent transport system (C-ITS) deployment activities in European countries. The main objective is the deployment of interoperable cross-border C-ITS services for road users, especially the so-called “Day 1—C-ITS service” and “Day 1.5—C-ITS service”.

The core C-Roads Platform members (Figure 1a) are Austria, Belgium/Flanders, Belgium/Wallonia, the Czech Republic, Denmark, Finland, France, Germany, Hungary, Italy, the Netherlands, Norway, Portugal, Slovenia, Spain, Sweden, and the United Kingdom.

The core members are involved with their own C-ITS pilot deployments, either in place or in preparation. In addition, many associated members closely follow the C-Roads Platform and the pilot deployments of C-ITS services [166].

In light of some intentional experiences, it is believed that, in the future, cities will have to improve their adaptive governance modalities, along with transparent AI decision-making and accessible mobility planning, so smart intersections are part of equal—and resilient—urban transport systems.

Key Performance Indicators (KPIs) such as the average vehicle delay, queue length, and total travel time can be the most important indicators in evaluating the performance of smart intersections. As mentioned earlier, one of the successful examples of smart intersection implementations is London’s SCOOT system, which has significantly reduced intersection delays through adaptive signal optimization, while Toronto’s AI-based intersections have effectively reduced queue lengths during peak periods. In Amsterdam, cooperative vehicle infrastructure systems have significantly improved overall travel efficiency. From a safety perspective, indicators such as the reduction in crashes and near misses have also been considered, which were positively demonstrated in an operational study in Pyeongtaek (South Korea) using edge AI and real-time safety monitoring. In the Barcelona Superblock model, prioritizing pedestrian zones reduced vehicle–pedestrian conflicts. On the other hand, environmental impact indicators, which are essential sustainability factors, such as reduced vehicle emissions and fuel economy, have been improved through intersection optimization strategies in Toronto and Amsterdam. In addition, according to public surveys in cities such as Toronto and Amsterdam, people’s perceptions of safety, congestion management, and the commuting experience have improved. The consistent use of these KPIs will provide planners and policymakers with the tools they need to deploy smart intersections in different urban environments.

7. Integration of Smart Roads and Smart Intersections into Smart City Ecosystems

The widespread implementation of advanced intersections in smart cities may have huge mobility benefits for several reasons. They allow for the efficient traffic flow of vehicles, pedestrians, and cyclists and work to improve traffic network operation in cities. Additionally, they coordinate with Adaptive Traffic Signal Systems (ATSSs) to change signal timing in real time to respond to traffic load fluctuations, pedestrian demand, and the need for emergency vehicles [77,167]. Advanced intersections also support various transportation modes, including cars, public transit, and micro-mobility solutions (e-scooters and bikes) coordinated with pedestrian crossings, creating sustainable urban mobility [168,169,170]. Furthermore, they play an important role in the prevention of bottlenecks and reduction in congestion [171,172].

In addition, next-generation roads are critical for improving public transit systems and Mobility-as-a-Service (MaaS) integration in urban spaces. They can plan to detect the approach of buses, trams, and High-Occupancy Vehicles (HOVs) and modulate vehicular traffic signals to reduce delays and increase transit reliability [173,174]. They also exchange real-time data with MaaS systems to optimize the routing of ride sharing [175,176]. By generating lower transportation expenses and fuel consumption, optimized traffic rules decrease commute times for everybody, from commuters to businesses [177]. Next-generation roads provide equal opportunities for transportation resources, sensitive pedestrian safety, and crossings for individuals in wheelchairs, as well as a fair transit service [178,179,180]. They contribute to increasing the quality of urban life by resulting in less traffic-related pollution, safer roads, and environmentally cleaner and safer cities [181,182,183]. Furthermore, they use data exchange systems based on 5G and 6G networks to reduce the response time for decisions and interactions between road infrastructures and autonomous vehicles [6,184,185,186], apply all benefits offered by digital twin models [187,188], and utilize vertically scalable hardware and software architecture, offering sustainable deployment across different urban environments with minimal disruptions [189,190].

Despite their potential, several challenges have to be overcome. Among these is cross-platform interoperability, as smart intersections must share data with existing transportation infrastructure, traffic control centers, and city-wide mobility platforms easily for efficient system operation [191,192]. Cybersecurity and privacy also pose significant concerns. Because they are all interconnected, smart intersections have to fend off cyberattacks and data breaches as well as prevent unauthorized access by developing secure encryption schemes, authentication protocols, and intrusion detection systems [193,194,195,196]. Lastly, effective stakeholder collaboration is essential. For successful urban integration, collaboration between government agencies, technology providers, private companies, and local communities is needed to ensure intelligent intersections align with urban planning needs and public interest [197,198]. Table 2 summarizes the key technologies, benefits, and challenges of smart roads in Smart Urban Mobility Systems.

Implementing smart intersections requires a multidisciplinary approach and will not be limited to technological capabilities. Smart intersections with specific spatial and functional characteristics require appropriate urban planning to align themselves with the main direction of urban planning. For example, compact urban designs with multi-modal integration zones are more amenable to real-time coordination systems and pedestrian prioritization. At the same time, car-centric suburban environments may face obstacles to efficient deployment due to scattered traffic flows [199,200].

Considering the behavioral perspective, such as the impact of user acceptance and trust, can also play an effective role in the success of smart intersections. Drivers are more likely to resist system opacity or unexpected delays in response to AI-controlled signals [201]. Behavioral studies, such as pedestrian and cyclist behaviors, can improve user-centered design and signal timing strategies by improving detection and prioritization algorithms [202,203].

From an economic perspective, the construction of smart intersections should be examined from several perspectives. First, a lifecycle cost analysis is necessary, taking into account the initial investment, maintenance costs, and long-term benefits such as congestion reduction, emission savings, and crash reduction [204]. Then, financial models such as public–private partnerships (PPPs), congestion pricing, and government subsidies should be explored and implemented [205]. Finally, equity considerations should ensure that vulnerable communities benefit equally from the development of this smart infrastructure while considering the rights and entitlements of all stakeholders [206,207].

Table 2. Key technologies, benefits, and challenges in Smart Urban Mobility Systems.

Category	Technology/Concept	Functionality	Benefits	Key References
Urban mobility systems	Smart intersections	Coordinates traffic flow, integrates multi-modal transportation, and reduces congestion	Ensures smooth traffic flow, accommodates diverse mobility needs, and minimizes delays	[77,167,168,169,170,171,172]
Public transport integration	Transit signal prioritization	Adjusts signal timings to prioritize buses, trams, and public transport vehicles	Reduces delays, enhances service reliability, and supports efficient transit systems	[168,169,170,173,174]
Mobility-as-a-Service (MaaS)	Real-time data sharing	Enables optimized routing for shared mobility services (e.g., ride sharing, bike sharing, on-demand shuttles)	Improves urban transport efficiency and enhances mobility accessibility	[175,176]
Economic impacts	Cost reduction	Reduces travel times, congestion, and fuel consumption	Lowers transportation costs for commuters and businesses	[177,204,205]
Social impacts	Accessibility and equity	Ensures fair access to transportation resources for vulnerable populations	Improves mobility for the elderly, disabled, and low-income groups	[178,179,180,206,207]
Environmental impacts	Pollution reduction	Reduces vehicle emissions and enhances air quality monitoring	Contributes to cleaner air, safer streets, and sustainable urban environments	[181,182,183]
Future integration	5G/6G connectivity	Enables ultra-low-latency and high-speed communication between vehicles and infrastructure	Enhances real-time traffic management and CAV coordination	[6,184,185,186]
	Digital twins	Uses real-time virtual models of intersections for simulation and predictive traffic management	Improves decision-making, congestion control, and smart city planning	[187,188,189,190]
Challenges	Cybersecurity measures	Protects traffic management systems from cyber threats and data breaches	Ensures public trust and system reliability	[193,194,195,196]
	Stakeholder collaboration	Requires cooperation among governments, tech providers, and communities	Ensures successful implementation and sustainable smart city integration	[197,198]

Table 2. Key technologies, benefits, and challenges in Smart Urban Mobility Systems.

Category	Technology/Concept	Functionality	Benefits	Key References
Urban mobility systems	Smart intersections	Coordinates traffic flow, integrates multi-modal transportation, and reduces congestion	Ensures smooth traffic flow, accommodates diverse mobility needs, and minimizes delays	[77,167,168,169,170,171,172]
Public transport integration	Transit signal prioritization	Adjusts signal timings to prioritize buses, trams, and public transport vehicles	Reduces delays, enhances service reliability, and supports efficient transit systems	[168,169,170,173,174]
Mobility-as-a-Service (MaaS)	Real-time data sharing	Enables optimized routing for shared mobility services (e.g., ride sharing, bike sharing, on-demand shuttles)	Improves urban transport efficiency and enhances mobility accessibility	[175,176]
Economic impacts	Cost reduction	Reduces travel times, congestion, and fuel consumption	Lowers transportation costs for commuters and businesses	[177,204,205]
Social impacts	Accessibility and equity	Ensures fair access to transportation resources for vulnerable populations	Improves mobility for the elderly, disabled, and low-income groups	[178,179,180,206,207]
Environmental impacts	Pollution reduction	Reduces vehicle emissions and enhances air quality monitoring	Contributes to cleaner air, safer streets, and sustainable urban environments	[181,182,183]
Future integration	5G/6G connectivity	Enables ultra-low-latency and high-speed communication between vehicles and infrastructure	Enhances real-time traffic management and CAV coordination	[6,184,185,186]
	Digital twins	Uses real-time virtual models of intersections for simulation and predictive traffic management	Improves decision-making, congestion control, and smart city planning	[187,188,189,190]
Challenges	Cybersecurity measures	Protects traffic management systems from cyber threats and data breaches	Ensures public trust and system reliability	[193,194,195,196]
	Stakeholder collaboration	Requires cooperation among governments, tech providers, and communities	Ensures successful implementation and sustainable smart city integration	[197,198]

8. Smart Pavements in Smart Intersections

This is how people usually imagine Next-Generation Intersections: optimized traffic flow, connected vehicles, and smart signal control. But the intelligence of the intersection will only be as good as the infrastructure that supports it. Therefore, the pavement infrastructure’s resilience, adaptability, and sustainability are important for the future of connected and autonomous mobility. Without an advanced pavement monitoring framework, even the most sophisticated adaptive intersections can be inefficient, dangerous, and more costly than needed over their lifetime. One of the emerging research areas in smart intersections is the role of pavement intelligence, which can further enhance connectivity and safety. This includes the following considerations:

-

The unspoken challenge: pavements as a dynamic digital asset. The current practice in pavement monitoring is mostly reactive, that is, through scheduled inspections and post-failure treatments [208]. However, the frequent loads, thermal variations, and CAVs’ operational requirements in traffic control and management at smart intersections require a change from the conventional static road infrastructure to dynamic and self-healing pavement systems. The issue is not just the detection of surface defects; it is the integration of real-time pavement intelligence into the digital environment of the new generation of these intersections.

-

High-resolution multi-sensor pavement surveillance. Current Pavement Condition Assessments (PCAs) are not very frequent or detailed. Therefore, the new generation of automated intersections must incorporate embedded multi-modal sensor grids (fiber optic, piezoelectric, acoustic emission, and thermal sensors) to monitor the condition of the surface, loading and stresses, and microstructural fatigue. This aligns with the advancements in sensor-grid infrastructure discussed by [209], which highlights multi-modal sensing for pavement health monitoring. Real-time data processing will be crucially important for edge computing-based intelligent decision-making to generate the required maintenance plans [210].

-

AI-powered predictive degradation models. The nonlinear and multivariable nature of pavement deterioration makes it necessary to develop AI-enabled digital twins that can predict the degradation trends as a function of vehicle traffic, the climate, and vehicle automation [12]. It is thus possible to use machine learning models trained with historical and real-time sensor data to predict the time and place of maintenance needs to avoid failures and minimize maintenance costs.

-

The pavement–CAV symbiosis. CAVs depend on good localization, strong vehicle-to-infrastructure communication, and a good surface [211]. Nevertheless, pavement flaws, cracks, and variations in load and dynamic deflections affect LiDAR, radar, and camera-based perception systems [212]. Therefore, it is crucial for smart intersections to incorporate “CAV-sensitive pavement intelligence”, i.e., the real-time modification of signal phasing, lane closure, and vehicle paths based on the pavement condition.

-

Embedded energy harvesting and smart self-healing materials. The next level of intelligent intersections is self-sustaining pavement systems. The next generation of intersections will incorporate piezoelectric materials that can convert the kinetic energy of vehicular movement into electricity to power the embedded IoT networks and data transmission [213,214,215,216]. At the same time, the development of nanoscale self-healing asphalt [217,218] and bio-inspired concrete [219] will allow pavements to heal themselves from small cracks and avoid major failures [220].

-

Cyber-resilient and blockchain-based pavement data. As pavement monitoring is shifting from periodic assessment to real-time analysis, properly managing data credibility, proof, and security is necessary. Using blockchain technology in pavement monitoring can generate immutable and shared records of the infrastructure’s health [221,222], which can help multiple stakeholders avoid fraud and make better decisions about the need for maintenance.

-

Smart intersections and sentient infrastructure. The next generation of AI-driven intersections must move beyond the traditional break/fix model of pavement management and move towards proactive, autonomous infrastructure. This transformation can only be realized by changing urban mobility principles and making roadways not only physical paths for vehicles but also an intelligent infrastructure of the transportation network that can sense, decide, and heal itself [223,224].

Thus, integrating AI, edge computing, self-healing materials, and cybersecurity data networks will make these intersections not only the traffic nodes but also the intelligent control points of the urban environment that will guarantee the integration of classical and automated transport.

Figure 5 presents a layout for intersection systems that combine sensors embedded in the pavement (such as acoustic sensors or fiber optic sensors), IoT processing capabilities, and edge AI technology. These components gather real-time data on traffic patterns, vehicle behaviors, and weather conditions, which are crucial for adjusting traffic signals on time and enabling communication between vehicles and the infrastructure (known as V2I communication). The system facilitates rapid decision-making with minimal delays by deploying sensor networks and digital replicas of environments (digital twins) while addressing common urban challenges, such as traffic congestion and safety concerns. Additionally, temperature and humidity sensors enable predictive maintenance algorithms that effectively prolong infrastructure lifespan and meet sustainability targets. By blending infrastructure with AI-driven analysis, this system illustrates how intelligent intersections can reduce carbon emissions, promote eco-friendly transportation options, and serve as a foundation for resilient smart cities. This collaboration is essential for pioneering transportation research and provides practical insights into data-centric strategies for urban mobility.

9. Challenges and Future Directions

While the smart intersection has the potential power to drive the transformation of the urban context, several issues must be solved before its widespread integration into urban transportation systems. In particular, the following challenges can be considered:

-

Smart component standardization gaps: incompatible data formats and protocols of various smart city components could prevent efficient data exchange and interoperability [225,226].

-

Scalability restrictions: due to more and more traffic data available, cloud and mobile data storage architectures are facing bigger scalability problems which need next-level AI-driven data compression/prioritization tactics [227,228].

-

Poor interagency coordination: smart intersection projects run into problems as a result of differences in current data-sharing policies among transport agencies, private mobility providers, and urban planners [229], and the development of standardized data protocols, interoperable APIs, and better cross-agency collaboration is needed to resolve these issues.

-

Data breaches and hacking possibility: hacking could lead to unauthorized real-time traffic data and AI-based control systems to command traffic signals, allowing system shut down or vehicle following [230].

-

Privacy for pedestrians and drivers: camera-based monitoring, along with the AI tracking of pedestrian and biker trajectories, raises concerns over surveillance, data misuse, and enforcement compliance [13].

-

Robust security protocols: effective security measures (e.g., end-to-end encryption, blockchain-based authentication, intrusion detection) are required for data protection and operational security [231,232].

-

Construction and maintenance costs: Transitioning to smart intersections gives rise to substantial investment in digital infrastructure, sensor networks, and control systems driven by AI. These include the deployment of LiDAR, edge computing units, V2X modulators, and AI-based adaptive traffic management systems that introduce financial barriers for most municipalities [233,234]. On the other hand, these intersections require periodic firmware and software updates to stay compatible with future technologies ranging from 5G/6G networks to autonomous vehicles [235,236]. Finally, funding shortfalls and policy impediments make it challenging for governments to secure long-term funding for smart infrastructure projects that work well with public–private partnership (PPP) requirements and new financing models [237].

-

Driver and pedestrian acceptability: in many countries, the current users of traffic management systems may resist AI-controlled signal optimization and intersection tuning in real time [238,239].

-

Autonomous vehicle interaction concerns: while smart intersections are conceived to be interoperable in the world of connected and autonomous vehicles (CAVs), safety, reliability, and ethical decision-making in mixed-traffic conditions remain a major issue [240,241].

-

Public engagement: municipalities must implement general awareness campaigns (pilot projects) and community engagement activities to inform drivers, cyclists, and pedestrians of the advantages of intelligent intersections and the importance of security measures [242,243].

With cities moving towards intelligent mobility ecosystems and their implementation paradigms, future research must look beyond the near-term emerging technological (AI, IoT) and policy/regulatory challenges. Some of the research directions are as follows:

-

Improving the efficiency of adaptive traffic signal control by using machine learning, deep reinforcement learning (DRL), and swarm intelligence: Future work will focus on the following research priorities: Self-tuning AI systems adapt on the fly to changing traffic situations, which means no configuration by humans, and they are continuously enabled [244,245]. Multi-agent reinforcement learning (MARL) models with multiple affordances from smart intersections cooperate to perform traffic optimization collectively over an urban network [246,247]. Neural network-based deep learning methods forecast congestion to enable proactive intersection control. Digital twins will enable critical real-time traffic monitoring, simulation, and predictive traffic modeling [248], leading to increased resilience in an intersection model [249,250]. Cloud digital twin platforms enable real-time synchronization between physical intersections and virtual simulation models for live decision-making [251,252]. Fifth-generation-based V2X communication can improve vehicle-to-vehicle communication, pedestrian safety systems, or traffic-responsive control in real time [253,254]. Edge AI enables real-time, decentralized decision-making at intersections, reducing cloud dependence and latency [255,256]. Sixth-generation-enabled mobility ecosystems can be used for fast data transfer at ultra-high speeds, and AI can be used for next-level traffic control at intersections for ground transport [257,258,259,260,261,262,263]. Standardized AI governance models can be developed for process-level-wise applications in traffic management [264,265].

-

Cross-jurisdiction interoperability agreements: ensure common data protocols for intelligent intersections between states and transportation sectors [266].

-

Autonomous vehicle interaction regulations: developing policies for smart intersections to address CAVs, liability, quantum risk assessments, and procedures for emergencies [267].

-

Commutable and self-regulating smart intersections: these intersections that use vehicle and queue detectors, smart cameras, LED road markers, and Variable Message Signs (VMSs) allow vehicles to be channeled appropriately into specific lanes, depending on the traffic demand level and distribution of traffic flows, as is the case with the novel COM-Roundabout (Figure 5) [268].

Figure 5. Example of commutable and self-regulating smart intersection: the COM-Roundabout and its functional principles [268].

Figure 5. Example of commutable and self-regulating smart intersection: the COM-Roundabout and its functional principles [268].

In this and in other smart intersection types, a traffic control area could be implemented where CAVs can start to communicate with the physical infrastructure. Within this management area, CAVs could provide kinematic information (e.g., instantaneous position, speed, and acceleration) to the smart intersection manager and gather optimized trajectories to maximize traffic performance indicators, such as the LOS [269,270]. The smart intersection manager should properly channel the CAVs between the lanes, coordinate manoeuvres, and control the traffic lights and adapt them to the traffic conditions in real time [269].

In brief, research priorities and future work will be directed towards AI-assisted optimization, security cybersecurity solutions, digital twin integration for intelligent intersections, and policy-kind frameworks to scale sophisticated junctions easily and allow them to remain scalable.

10. Conclusions

Smart intersections are emerging as pivotal components of urban transport, integrating state-of-the-art technologies to create next-generation road transportation systems that seamlessly address modern mobility challenges. These intersections harness the potential of V2X communication, AI, IoT networks, and some smart sensing for better vehicle-to-infrastructure functions that support smoother traffic management, promote safety, and regard environmental circumstances. Integrating adaptive AI into traffic signal control significantly reduces urban congestion and enhances pedestrian safety. V2X communication effectively reduces collision risks. At the same time, multi-sensor fusion techniques utilizing LiDAR, radar, and thermal imaging enable the precise and reliable tracking of pedestrians and vehicles. The significance of their role in the basic building blocks of intelligent transport systems highlights the need for wide-reaching use and advancements in terms of technology.

This comprehensive review of smart intersection designs highlights converging trends towards more innovative designs for adaptive traffic light control based on AI, predictive analytics regarding congestion management, and distributed non-sensor fusion for situational awareness.

Operationally, smart intersections effectively prioritize public transportation modes, improving reliability and reducing delays. IoT-driven predictive maintenance, although highly beneficial in minimizing operational downtime and infrastructure lifecycle costs, demands substantial initial capital investments and sustained funding strategies, necessitating innovative public–private partnerships.

Case studies from different countries display how automated intersections can be tailored to several urban contexts. These intersections have great potential. However, many challenges remain, and the most important of them is the requirement for standard communication protocols, advanced security measures, and scalable deployment in resource-deprived areas. Also, merging human factors—driver behavior and pedestrian movement patterns together or user perception— into their design is a field that must continue to be researched.

However, given the pros and cons specified in this article, some recommendations are essential to help the future use of smart roads, in terms of the following:

-

Research towards an interdisciplinary approach to integrating transportation engineering, urban planning, and user behavior science for user-centric intelligent intersections;

-

Developing international strategies to enable a better interoperable and integrated implementation of these intersections among different urban domains and transportation systems;

-

Increasing data security and privacy: more significant levels of encryption protocols for cyber-physical intersections are critical, coupled with blockchain-based control and well-designed cybersecurity strategies that secure lanes against future threats while ensuring user privacy;

-

Advocating for public–private partnerships: industry stakeholders should work alongside municipal governments and research institutions to advance innovation through public–private partnerships.

When deploying smart intersections and V2X applications, important tradeoffs, such as improved efficiency, advanced AI-based optimization, and cybersecurity measures, need to be considered. Improved efficiency through optimized traffic flows can sometimes unintentionally be at the expense of pedestrians, cyclists, or public transport users, leading to a lack of fairness considerations. Similarly, advanced AI-based optimization can lead to public distrust by reducing transparency in decision-making processes, which can reduce the demand for improved system performance. Strong cybersecurity measures are also difficult to use in V2X systems, even though secure communication is necessary. This is because they add extra work for the computers, making them less scalable and responsive in real time. Decision-makers must balance these factors to ensure that technological advances align with broader societal goals, user acceptance, and practical feasibility.

The successful implementation of smart intersections hinges upon robust cybersecurity frameworks, transparent policy regulations, standardized interoperability protocols, and proactive stakeholder collaborations, ensuring both security and broad public acceptance. In conclusion, the mass adoption of smart roads will only be driven by interdisciplinary collaboration with industries and government agencies. Using today’s winning strategy to solve yesterday’s problems and identify the most significant windows of opportunity ahead, smart intersections will be a highly relevant part of the coming live transportation network by making urban mobility safer, smarter, and more inclusive.