Next Article in Journal
Does Green Finance Drive New Quality Productive Forces? Evidence from Chinese Listed Companies
Previous Article in Journal
Influence of Stress on Gas Sorption Behavior and Induced Swelling in Coal: Implications for Sustainable CO2 Geological Storage
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Sustainability
Volume 17
Issue 20
10.3390/su17208992
sustainability-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Review

Artificial Intelligence for Infrastructure Resilience: Transportation Systems as a Strategic Case for Policy and Practice

by
Olusola O. Ajayi
1,*,
Anish Kurien
1,
Karim Djouani
1,2 and
Lamine Dieng
1,3
1
F’SATI, Faculty of Engineering and the Built Environment, Tshwane University of Technology, Pretoria 0001, South Africa
2
LISSI Laboratory, Université Paris-Est Créteil, 94000 Créteil, France
3
MAST Department, Université Gustave Eiffel, 44340 Bouguenais, France
*
Author to whom correspondence should be addressed.
Sustainability 2025, 17(20), 8992; https://doi.org/10.3390/su17208992
Submission received: 8 September 2025 / Revised: 26 September 2025 / Accepted: 9 October 2025 / Published: 10 October 2025
(This article belongs to the Special Issue Intelligent Transportation Systems and Sustainable Urban Traffic Management and Transportation Planning)
Browse Figures
Versions Notes

Abstract

Transportation networks are critical lifelines in national infrastructure but are increasingly exposed to risks arising from climate variability, cyber threats, aging assets, and limited resources. This paper presents a scoping review of 58 peer-reviewed studies published between 2015 and 2025 that examine the role of Artificial Intelligence (AI) in strengthening infrastructure resilience, with transportation systems adopted as the strategic case. The review classifies applications along five dimensions: technological approach, infrastructure sector, transportation linkage, resilience/security aspect, and key research gaps. Findings show that AI, machine learning (ML), and the Internet of Things (IoT) dominate current applications, particularly in predictive maintenance, intelligent monitoring, early-warning systems, and optimization. These applications extend beyond transport to energy, water, and agri-food systems that indirectly sustain transport resilience. Persistent challenges include affordability, data scarcity, infrastructural limitations, and limited real-world validation, especially in Sub-Saharan African contexts. The paper synthesizes cross-sector pathways through which AI enhances transport resilience and outlines practical implications for policymakers and practitioners. A targeted research agenda is also proposed to address methodological gaps, enhance deployment in resource-constrained settings, and promote hybrid and explainable AI for trust and scalability.

1. Introduction

Infrastructure resilience, the capacity of systems to anticipate, absorb, adapt to, and recover from shocks, has emerged as a central concern in the era of the Fourth Industrial Revolution (4IR). Among the various infrastructures, transportation systems occupy a strategic position: they not only enable mobility, logistics, and access to essential services but also function in close interdependence with energy, water, health, and agri-food systems [1,2]. Disruptions in these systems cascade across networks: floods compromise roads and bridges, energy instability constrains e-mobility, and agricultural volatility affects freight logistics.
Artificial Intelligence (AI) has emerged as a powerful toolset for advancing resilience across these interdependent domains. In transport, AI supports applications such as condition monitoring, predictive maintenance of vehicles and infrastructure [3], optimization of transport corridors and fleet scheduling [4], and anomaly detection in intelligent transport systems [5]. Beyond transport itself, AI also strengthens resilience in the supporting infrastructures that sustain mobility, such as smart grids for e-mobility [6], climate-smart agriculture for freight logistics [7], and early-warning systems for rerouting during natural disasters [8].
Despite these advances, persistent barriers remain. The majority of studies report high technical accuracies, often exceeding 90% under controlled conditions, yet only a minority have been validated through real-world deployments [9]. Cost, data scarcity, infrastructural constraints, and skills gaps also hinder scalability, particularly in Sub-Saharan Africa [10,11]. These gaps underline the need for a comprehensive review that connects evidence across sectors and translates insights into actionable strategies.
Recent advances have extended this narrative. Digital twins now support lifecycle monitoring of roads, bridges, and urban corridors [12,13]. Graph neural networks underpin medium-range weather forecasting, strengthening anticipatory logistics [14]. International frameworks such as [15,16], are shaping governance ecosystems for responsible AI deployment.
In this paper, we therefore adopt transportation systems as a strategic case for analyzing AI’s role in infrastructure resilience. Rather than treating transport as an isolated domain, we consider it the hub of an interdependent infrastructure web, affected by, and influencing, resilience capacities in energy, water, agriculture, and digital ecosystems. Synthesizing this diverse research provides critical value by consolidating fragmented knowledge across infrastructure domains. Although many individual studies demonstrate technical accuracies, they are often isolated within disciplinary silos and lack systemic integration. A scoping review is therefore essential because it identifies persistent gaps, such as affordability, validation, and governance, while also translating findings into actionable pathways for policy and practice. By adopting transportation systems as the strategic case, this review highlights how AI strengthens resilience not only in transport itself but also in the interdependent infrastructures that sustain mobility.
The study pursues three interrelated aims that together establish a comprehensive foundation for understanding the role of Artificial Intelligence (AI) in enhancing infrastructure resilience with transportation systems as the strategic focus. First, it seeks to classify AI applications by systematically organizing them across multiple dimensions, including the technologies employed, the sectors in which they are applied, their linkage to transportation (whether direct, indirect, or systemic), the resilience and security aspects they address, and the key research and practice gaps that persist. Second, the study aims to map cross-sector pathways through which AI strengthens transport resilience, recognizing that transportation is both a domain in itself and a critical hub whose robustness depends on interconnections with energy, water, agriculture, health, and built environment systems. Finally, it endeavors to translate the accumulated evidence into actionable implications for policy and practice, providing decision-makers and practitioners with concrete insights while also articulating a targeted research agenda that identifies priority areas for future investigation and deployment.

2. Related Works

Research on Artificial Intelligence (AI) for infrastructure resilience has expanded significantly in the past decade, spanning domains such as transportation, energy, water, agriculture, climate, and cyber-physical systems. While transportation often appears as the primary focus, it is also deeply influenced by resilience measures in other infrastructures. To frame this review, the literature is organized into five interconnected domains: (1) transportation resilience, (2) energy, water, and food systems, (3) climate and early-warning systems, (4) cyber and digital infrastructure security, and (5) policy, ethics, governance, and education.

2.1. AI in Transportation Resilience

Much of the literature directly targets the resilience of transportation systems, emphasizing predictive maintenance, smart mobility, and infrastructure performance. Studies highlight how machine learning can quantify the impact of road roughness on driver safety and comfort, as well as the trade-offs between safety and fuel efficiency caused by traffic-calming measures such as speed bumps [17]. Other works extend this focus to passenger experiences, using AI and IoT-enabled systems to monitor air quality in buses and trains and linking commuter comfort to broader system resilience [18]. Urban mobility frameworks for emerging economies emphasize adaptability and foresight, where AI enhances decision-making for anticipating disruptions [4].
AI continues to transform the transport sector through predictive maintenance, intelligent asset management, and digital twin applications. For instance, explainable ML frameworks have been developed for real-time railway maintenance, improving predictability and operational trust [19]. The growing deployment of digital twins, virtual replicas for infrastructure monitoring and decision-making, now includes roads, bridges, and urban transportation networks [12,13]. An emerging concept is the urban transportation digital twin, where AI-enabled “brain” components manage perception, prediction, and control from real-world testbeds [20]. Such systems enhance resilience by allowing scenario testing and proactive response. AI has also been integrated into traffic flow modeling, with new car-following models incorporating dynamic driver reaction times and visual angles under slope conditions [21]. These behavioral models directly affect transport stability and resilience during high-risk conditions.
Taken together, these contributions illustrate that AI strengthens transport systems not only by reducing mechanical failures but also by enhancing the safety, efficiency, and adaptability of daily operations.

2.2. AI in Energy, Water, and Food Systems Resilience

Transport resilience is inherently dependent on the stability of supporting infrastructures, especially energy and agri-food systems. AI plays a crucial role in renewable energy forecasting and smart grid management, both of which underpin the reliability of e-mobility [6,22]. In the water sector, AI-driven smart irrigation systems enhance climate resilience by reducing water stress, indirectly supporting agricultural logistics [23]. AI’s role in stabilizing energy and water systems continues to grow, especially where these systems support transport stability. For example, AI-assisted renewable grid management enhances e-mobility resilience by optimizing supply and load [24]. In water-scarce regions, AI-based engineering solutions help infrastructures withstand extreme rainfall, aligning with sustainable development goals in arid regions [25].
Similarly, AI applications in food safety monitoring contribute to reliable agri-food supply chains by reducing risks of spoilage and contamination during transport [9]. These linkages operate through tangible pathways. For instance, AI-enabled smart grids ensure reliable power supply for electric mobility, thereby reducing downtime and enhancing transport continuity. In water and climate systems, AI-driven smart irrigation and flood prediction models reduce risks of water stress and protect transport corridors during extreme rainfall. In agri-food systems, AI-supported supply chain monitoring ensures stable freight flows and reduces disruptions by minimizing spoilage and logistical delays. Together, these interactions demonstrate that transport resilience is contingent upon the robustness of supporting infrastructures.

2.3. AI in Climate, Environment, and Early-Warning Systems

Another critical strand of research addresses AI applications for disaster preparedness and environmental resilience. Deep learning and optimization algorithms have been used to improve post-earthquake road network recovery, outperforming traditional heuristic methods in terms of response speed [26]. AI-powered remote sensing and hazard prediction systems also demonstrate significant potential for disaster risk management and evacuation planning [8]. More recent approaches leverage causal AI and geospatial foundation models to design multi-hazard early-warning systems, stressing the importance of ethical transparency and FAIR-compliant data standards [14].
AI-driven environmental resilience has advanced notably through integrated planning, forecasting, and response. The AI-for-Good report highlights how AI across infrastructure lifecycles can save billions in disasters by enabling proactive design and real-time response [27]. Similarly, Deloitte projects that AI could prevent up to 15% of natural disaster-related infrastructure losses, with digital twins and early-warning systems as core tools [28]. These developments underscore AI’s potential to move from reactive recovery strategies toward anticipatory resilience across domains, including transport.
These systems interact directly with transportation by influencing hazard preparedness and rerouting strategies. For example, AI-based flood forecasting enables transport authorities to pre-emptively close vulnerable road segments, while integrated disaster early-warning systems guide evacuation logistics. Such mechanisms illustrate how climate resilience applications translate into tangible improvements in transport continuity.

2.4. AI in Cyber and Digital Infrastructure Security

As transport systems become increasingly digitized, ensuring cyber resilience has become a fundamental concern. AI-based cybersecurity solutions automate network monitoring and anomaly detection, improving the robustness of transport-linked infrastructures [29]. More recent work applies explainable AI (XAI) to safeguard against adversarial attacks and adapt to evolving threats in transport communication networks [5].
As infrastructure systems become increasingly connected, AI’s role in cyber resilience has intensified. Generative AI and large language models are being studied for Critical Infrastructure Protection, integrating safety and cybersecurity frameworks to support trust and adaptive responses [30]. AI also enhances governance by improving policy mechanisms, legitimacy, and real-time governance during disaster scenarios [31].
Cyber resilience is not peripheral but central to transport resilience, since intelligent transport systems depend on secure communication networks. AI-based anomaly detection prevents cascading failures in traffic management platforms, while explainable AI safeguards trust in automated routing and scheduling systems. These mechanisms confirm that cyber resilience is a prerequisite for transport system sustainability.

2.5. Policy, Ethics, Governance, and Education

Although less technical, governance, ethics, and education are indispensable enablers of AI adoption in transport resilience. Policy frameworks condition the legitimacy and scalability of AI systems, while digital literacy and professional skills determine whether innovations can be effectively implemented. For instance, fragmented governance structures have been cited as barriers in nearly one-quarter of reviewed studies, and digital skills deficits limit the ability of agencies to manage AI systems in Sub-Saharan Africa. These factors establish a logical and necessary link between governance ecosystems and the practical realization of transport resilience [11,32]. Comparative policy studies from Rwanda and Ghana demonstrate that political and economic contexts strongly influence AI adoption strategies [33].
The governance ecosystem guiding AI’s application in resilience is rapidly evolving. Beyond earlier policy frameworks, recent research emphasizes AI’s role in climate resilience governance combining advanced analytics with inclusive policy making, to bridge complex climate-risk decisions [34]. Similarly, the resilience of transportation systems under geopolitical stress and supply chain disruption benefits from AI-based modeling and scenario simulation [35].
These insights underline that technological innovations cannot drive resilience in isolation; enabling governance frameworks, ethical safeguards, and human capacity are equally vital to achieving system-wide resilience outcomes.
Figure 1 is the conceptual framework showing the interplay of AI technologies, transport systems, and cross-sector infrastructures (energy, water, agriculture, cyber) in shaping resilience outcomes.

3. Methodology

3.1. Review Approach

This study employed a scoping review methodology, which is particularly appropriate for synthesizing evidence across a broad and fragmented body of literature. Unlike systematic reviews that emphasize exhaustive screening and quality appraisal, scoping reviews are designed to map key concepts, identify knowledge gaps, and highlight the breadth of available evidence [36,37]. The choice of this method was informed by the interdisciplinary nature of the topic: AI for infrastructure resilience intersects with transportation, energy, water, agriculture, the built environment, and digital security.
A scoping approach therefore provides the flexibility needed to accommodate this diversity while still offering a structured synthesis. This review was conducted and reported in line with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR, see Supplementary Materials) guidelines. All recommended elements of the PRISMA-ScR checklist have been addressed, including transparent reporting of search strategies, screening procedures, and synthesis methods. Although the review protocol was not pre-registered, adherence to PRISMA-ScR ensured methodological rigor and transparency throughout [38,39].

3.2. Data Sources and Extraction

The dataset for this review comprised 58 peer-reviewed studies published between 2015 and 2025. Searches were conducted across Scopus, Web of Science, IEEE Xplore, and ScienceDirect, with supplementary searches in Google Scholar to capture grey literature and emerging works. References were collected and organized using Zotero 7.0.24 (64 bit) reference management software. Metadata such as titles, authors, publication years, abstracts, and digital object identifiers (DOIs) were extracted and normalized into a structured dataset.
For each study, five core attributes were coded to enable systematic comparison. These included the technology used (for example, AI, ML, IoT, blockchain, building information modeling, or natural language processing), the infrastructure sector to which the application was directed (transport, energy, agriculture and food systems, water, built environment, health, digital systems, or supply chains), and the nature of the transportation linkage (classified as direct, indirect, or unspecified).
The overall process of study identification, screening, eligibility assessment, and final inclusion is summarized in the PRISMA flow diagram (Figure 2).
In this review, AI/ML applications were defined as computational approaches involving prediction, classification, or optimization, while IoT applications were considered data-enabling technologies that provide real-time sensing inputs. Where IoT systems were tightly integrated with AI for decision-making, they were coded under both categories. Each study was also categorized by its resilience or security aspect, such as predictive maintenance, optimization, early warning, cybersecurity, or governance.
Finally, key insights and gaps were captured, including recurring issues such as affordability, data scarcity, lack of validation, skills deficits, and governance challenges. This coding schema was adapted from best practices in scoping review methodology, which recommend both descriptive mapping and thematic synthesis [40,41].
This structured categorization enabled both descriptive mapping and thematic synthesis, providing the foundation for Table 1 and subsequent cross-sector analysis.

3.3. Screening and Inclusion Criteria

The inclusion of studies was guided by three criteria. First, the study had to involve a clear application of AI, ML, IoT, or a related computational technology. Second, the work needed to demonstrate an explicit contribution to the resilience or security of infrastructure systems. Third, there had to be a transportation linkage, either direct (focusing on transport networks, mobility systems, or smart transport applications) or indirect (addressing infrastructures such as energy grids or agri-food logistics that influence transport resilience).
Studies focusing solely on domains unrelated to transport resilience (e.g., clinical healthcare without cross-sector implications) were excluded. Where relevance was ambiguous, inclusion decisions were made based on whether transport was explicitly mentioned as an impacted or dependent system [38,39].

3.4. Analytical Framework

Analysis proceeded in two stages. First, studies were grouped by domain, covering transportation, energy, agriculture/food, water, climate and environment, cyber/digital infrastructure, and governance/policy. Second, a cross-sector synthesis was conducted to trace interdependencies between these domains and transport resilience. For example, AI in smart grids was mapped to transport electrification, AI in climate early-warning systems to rerouting strategies, and AI in food systems to freight continuity. This two-tiered approach is consistent with scoping review standards that recommend combining numerical summary with qualitative thematic analysis to maximize interpretive depth [38].
A further methodological step involved examining evaluation indicators for infrastructure resilience across the included studies. Although no single unified standard exists, several recurring indicators were observed: predictive accuracy of AI models, recovery time following disruptions, connectivity and robustness of networks under stress, cost-efficiency of interventions, and levels of stakeholder trust or adoption. While comparability across studies is limited due to methodological diversity, convergence is emerging around hybrid indicators that integrate both technical and socio-technical dimensions, offering a basis for future standardization. The outcome is a set of thematic narratives supported by quantitative distributions of technologies, functions, and barriers (Figure 1 and Figure 3), alongside an integrative synthesis of implications for transport resilience.

4. Findings

The synthesis of 58 peer-reviewed studies reveals both sector-specific patterns and cross-sectoral linkages that define how AI contributes to infrastructure resilience. A descriptive–analytical summary is provided below, followed by interpretive insights.

4.1. Distribution of Studies

Of the 58 studies reviewed, only 12 (21%) focused directly on transportation, while 33 (57%) engaged transport indirectly through supporting infrastructures such as energy, water, or agri-food systems. The remaining 16 (28%) addressed transport resilience in more general or unspecified ways but still offered relevant insights. This distribution underscores the position of transportation as a highly interdependent domain rarely studied in isolation but consistently influenced by resilience measures in other sectors.
In terms of sectoral coverage, agriculture/food systems accounted for 10 studies (17%), the built environment for 9 (16%), energy for 6 (10%), water for 3 (5%), climate and environment for 6 (10%), cyber/digital infrastructure for 5 (9%), and governance/policy for 7 (12%). Transport thus emerges as both a primary and secondary beneficiary of AI-enabled resilience across multiple domains.
This distribution highlights that transportation resilience is rarely addressed in isolation. Instead, the majority of AI applications are embedded in supporting infrastructures such as energy, water, and agriculture. This pattern reinforces the systemic nature of resilience, where transport continuity depends on stability in other domains.

4.2. Technological Trends

AI was the dominant technology, appearing in 39 studies (67%), followed by IoT (8 studies, 14%), machine learning as a standalone category (6 studies, 10%), big data analytics (5 studies, 9%), blockchain (3 studies, 5%), and deep learning frameworks (2 studies, 3%). This distribution suggests that although advanced methods such as deep learning are gaining ground, most resilience applications rely on general-purpose AI and ML rather than domain-specific or hybrid approaches.
The underrepresentation of predictive maintenance (9%) is particularly striking given its recognized value in extending asset lifespans and reducing downtime. This gap suggests a research-to-practice disconnect, where technically feasible solutions have yet to achieve widespread real-world adoption.
Figure 3 shows that monitoring and sensing (28%) and cybersecurity (26%) dominate AI functions in transport-linked systems, followed by optimization/control (24%) and early warning/forecasting (24%). Predictive maintenance (9%) remains underrepresented despite its recognized value for resilience. Narratively, these patterns align with applied studies: ref. [26] optimized post-earthquake road network responses using deep learning; ref. [4] applied AI for smart mobility resilience in emerging economies; and cross-sector dependencies are evident, from AI-driven grid optimization supporting transport electrification [42] to AI-enhanced predictive maintenance of flood-prone transport infrastructure [43], to early planning and optimization stages [44], to emerging interest in AI and DT integration in multisectoral deployment and maturity [12,45].

4.3. Barriers to Adoption

The analysis also identified recurring barriers. Governance and policy fragmentation was cited in 14 studies (24%), infrastructure and connectivity limitations in 12 (21%), and data scarcity in 11 (19%). High costs were noted in 7 (12%), while skills deficits and power reliability challenges each appeared in fewer than 10% of the studies. A smaller subset pointed to issues of explainability and lack of standards. Policy, infrastructure, and cost barriers remain pervasive. For example, recent work in AI applications for resilient transport infrastructure confirms that despite advances, factors like regulatory misalignment, cost constraints, and digital infrastructure gaps continue to limit deployment [46]. Collectively, these findings suggest that while technical feasibility is often demonstrated, institutional and infrastructural constraints remain the most persistent obstacles to large-scale deployment, particularly in low- and middle-income contexts [12,47]. These patterns confirm that institutional and infrastructural barriers outweigh technical ones, underscoring the importance of enabling ecosystems for scaling AI in transport resilience.

4.4. Cross-Sector Insights

The findings also highlight that transport resilience is shaped by multiple cross-sectoral dependencies. In the energy domain, AI-enabled smart grids provide reliable power supplies that sustain e-mobility and electric transport infrastructure. In water and climate systems, AI-powered flood forecasting and early-warning mechanisms support rerouting decisions and the protection of vulnerable road and bridge assets [48,49]. Agricultural and food systems influence transport resilience through AI-driven supply chain monitoring, which stabilizes freight flows and reduces the risk of disruptions. Cyber and digital infrastructures further reinforce resilience by enabling AI-based anomaly detection and protection within intelligent transport communication networks. Generative AI is being used to enhance urban digital twins, integrating predictive and design capabilities across transport, energy, and built environment systems [50]. Finally, governance and policy frameworks cut across all these domains by conditioning the extent to which AI innovations can be deployed responsibly, equitably, and at scale.
Considered holistically, these insights indicate that statistical distributions are not merely descriptive but reflect deeper dynamics: AI adoption remains fragmented, skewed toward monitoring and forecasting functions, and insufficiently validated in predictive maintenance or governance integration. These gaps provide clear direction for future research and deployment strategies.
Together, these interdependencies confirm that transportation resilience is not an isolated capacity but rather the emergent property of a broader system-of-systems. Strengthening transport resilience therefore requires coordinated interventions across energy, water, agriculture, digital security, and governance structures.

5. Discussion

5.1. Transportation as a Strategic Hub

The review reinforces the understanding that transportation cannot be considered in isolation when examining infrastructure resilience. Instead, it functions as a strategic hub within an interconnected web of systems. Stable energy supplies are critical for e-mobility, resilient supply chains determine freight continuity, the safety of the built environment influences accessibility after hazards, and reliable digital infrastructure sustains intelligent transport services. These interdependencies explain why only about one-fifth of reviewed studies addressed transport directly, while the majority considered it indirectly through energy, agriculture, climate, or digital systems. Transportation resilience, therefore, emerges less as a self-contained capacity and more as an outcome of systemic integration [51]. This aligns with recent comprehensive reviews emphasizing the role of AI and automation in enabling sustainable and low-carbon transport systems [52].

5.2. Dominant Technologies and Persistent Gaps

AI and IoT were the most frequently applied technologies, particularly in monitoring, anomaly detection, and optimization. Yet predictive maintenance, arguably one of the most actionable functions for transport resilience, remains underrepresented, with only 9% of the studies engaging this application. This imbalance highlights a gap between research emphasis and practical deployment [44]. Moreover, while technical feasibility is repeatedly demonstrated, real-world validations remain scarce. Many models, as noted in Section 4, achieve high performance under controlled conditions but have yet to be tested at scale in operational transport systems. The lack of field-based validation diminishes their policy relevance and limits their contribution to resilience planning.

5.3. Barriers to Adoption

The most significant constraints identified in this review are institutional and infrastructural rather than technical. Governance incoherence was cited in nearly one-quarter of studies, mirroring evidence from global AI deployment frameworks where fragmented policy environments slow down adoption [15]. Infrastructure and connectivity gaps, particularly in low- and middle-income countries, compound these challenges, with affordability and skills deficits further constraining uptake [11]. Recent evidence also underscores cybersecurity risks, as generative AI introduces new attack surfaces even while providing new defenses, requiring transport-linked systems to adopt cybersecurity-by-design [30].
Together, these barriers suggest that even where AI demonstrates strong potential, enabling environments are often lacking, particularly in Sub-Saharan Africa, where digital infrastructure and funding remain constrained.

5.4. Cross-Sector Synthesis

A key contribution of this review is the mapping of cross-sector pathways that converge on transport resilience. In the energy sector, AI-enabled grid stability and renewable integration sustain e-mobility by ensuring reliable power supply [24]. In climate and water management, AI-driven flood forecasting and early-warning systems directly influence routing decisions and the protection of transport assets [43,48]. Within agri-food systems, AI applications in supply chain monitoring and food safety logistics reduce disruptions to freight flows [7]. Cybersecurity applications safeguard the digital infrastructure upon which intelligent transport services depend [5], while governance frameworks determine whether these technical innovations can be deployed responsibly and equitably. These interdependencies confirm that transport resilience is only as robust as the weakest linked system, reinforcing the need for integrated approaches that move beyond sectoral silos.

6. Policy and Practice Implications

The findings of this review highlight that AI has strong potential to support transport resilience, but deployment is constrained by fragmented governance, weak infrastructure, and limited validation in real-world contexts. Translating evidence into actionable strategies requires policy frameworks that embed AI into resilience planning, build enabling data infrastructures, and ensure that innovations are both explainable and equitable. The following implications are particularly relevant to policymakers and practitioners.

6.1. Governance and Regulation

National and regional governments should integrate AI explicitly into transport resilience frameworks, including disaster risk reduction strategies and climate adaptation plans. Rwanda’s AI ambitions include embedding resilience to climate risks. For instance, AI-based forecasts for floods and extreme weather are being linked to national disaster risk management systems [53]. Rwanda is actively exploring the integration of AI/ML into its national climate and disaster systems. Recent work using LSTM models has shown potential in improving forecast accuracy for floods [54], and the government has initiated a feasibility study to embed AI/ML into Meteo Rwanda’s operational forecasting and climate services [55].
The review showed that governance incoherence was the most commonly cited barrier, appearing in nearly one-quarter of studies. Establishing regulatory sandboxes would allow controlled testing of AI systems in transport corridors before full-scale deployment, thereby reducing risks while encouraging innovation [1].

6.2. Infrastructure and Data Foundations

Transport resilience depends on robust sensing, monitoring, and data-sharing infrastructures. In South Africa, infrastructure such as the Msikaba Bridge has deployed IoT-enabled structural health monitoring systems (with sensors tracking strain, displacement, temperature, etc.), serving as a practical example of how strong data foundations can underpin predictive maintenance in major built infrastructure projects [56,57]. Moreover, SANRAL’s smart roads strategy suggests growing institutional interest in integrating digital and IoT systems for road network oversight and management [58].
Evidence from the reviewed studies demonstrates that AI applications such as predictive maintenance and early-warning systems rely on high-quality, real-time data, yet data scarcity and lack of standardization were cited as major barriers in almost one-fifth of studies. Policymakers should therefore prioritize investments in IoT deployments for asset condition monitoring (roads, bridges, and rolling stock), as well as the development of digital twins for urban transport corridors. Establishing open data standards will also facilitate interoperability across transport, energy, and logistics systems.

6.3. Financing and Procurement

Affordability emerged as a consistent challenge, particularly in Sub-Saharan African contexts. To address this, financing instruments such as green or resilience bonds should be tied to measurable outcomes like reduced downtime or improved recovery times. Procurement practices also need reform: contracts should move beyond traditional input-based models to outcome-based frameworks that require vendors to demonstrate prediction accuracy, explainability, and interoperability. Linking public–private partnerships to innovation hubs can help bridge the resource gap while ensuring accountability.

6.4. Capacity and Human Capital

The review revealed that skills deficits remain an underexplored but significant barrier to AI adoption in infrastructure resilience. Building human capital is therefore essential. Innovation hubs in Kenya provide a model for capacity building, where SMEs adapt AI solutions to local transport and logistics needs. Universities and professional institutes should design interdisciplinary curricula that combine AI, transport engineering, and ethics. Beyond formal education, continuous professional development for transport agencies is needed to ensure that staff can manage, maintain, and adapt AI systems. At the community level, resilience literacy campaigns are important to build trust in AI-enabled early-warning systems, particularly in vulnerable populations where public acceptance is critical [59].

6.5. Trust and Ethics

Finally, the success of AI in transport resilience depends on whether systems are transparent, explainable, and socially acceptable. As Section 4 demonstrated, explainability concerns were identified in several studies, but often remain under-addressed. Policies should require explainable AI (XAI) in safety-critical domains such as emergency routing or predictive maintenance. Ethical safeguards are equally important to prevent AI from exacerbating inequalities, particularly in low-income regions. By embedding trust and ethics into adoption frameworks, policymakers can foster sustainable and equitable deployment.

7. Recommendations

The synthesis of findings and policy implications leads to a set of actionable recommendations for embedding AI into transport resilience agendas. These recommendations are structured into six domains that together form a roadmap for moving from fragmented pilots to systematic adoption.

7.1. Policy and Governance

Governments should embed AI-enabled forecasting, monitoring, and decision-support into national transport resilience strategies. This includes integrating predictive maintenance and early-warning systems into transport master plans and disaster risk reduction frameworks. The [60] demonstrates how risk-based regulation can guide responsible use of AI in critical infrastructure, including transport [61]. In Africa, the [62] provides a roadmap for ethical governance, emphasizing inclusivity and capacity building [62]. Establishing regulatory sandboxes is particularly valuable, as they allow for controlled testing of AI applications before full-scale deployment. Prior studies emphasize that such sandboxes accelerate responsible innovation while reducing risks of unintended outcomes [1]. Furthermore, interoperability standards and open APIs must be mandated to ensure integration across transport, energy, and logistics infrastructures. Without such governance frameworks, AI applications risk remaining isolated pilots with limited systemic impact.

7.2. Technical and Data Infrastructure

Robust infrastructure and interoperable data systems are essential for effective deployment. Refs. [63,64] guidance highlights digital twins and AI-powered Earth observation as transformative for transport corridors. The review highlighted that AI applications for predictive maintenance, early-warning, and anomaly detection are constrained by data scarcity and a lack of standardized frameworks. Policymakers and practitioners should prioritize IoT deployments in high-risk transport corridors and urban hubs, enabling real-time monitoring of pavements, bridges, and transit assets. At the same time, hybrid modeling approaches, such as physics-informed neural networks, can enhance interpretability by blending domain knowledge with machine learning. Security must also be a priority: cybersecurity-by-design principles should be embedded in AI-enabled transport systems to safeguard against adversarial threats that could destabilize critical mobility services [5].

7.3. Financing and Procurement

Affordability was a recurring challenge, particularly in Sub-Saharan African contexts where infrastructure investment is already constrained. Innovative financing mechanisms such as green and resilience bonds should be tied explicitly to measurable outcomes, including reductions in mean time to recovery after disruptions. Procurement practices also need to evolve from input-based to outcome-based contracting [65]. Vendors should be required to demonstrate not only predictive accuracy but also explainability and interoperability of AI systems. Public–private partnerships can play a catalytic role here by supporting local innovation hubs that adapt AI solutions to regional contexts while sharing risks and resources between stakeholders [10].

7.4. Capacity Building and Education

Human capital is central to sustaining AI-enabled resilience. Universities should establish interdisciplinary curricula that combine AI, transport engineering, and ethics. Professional development programs targeting engineers, policymakers, and practitioners can help ensure AI systems are effectively managed and maintained. At the community level, resilience literacy campaigns are critical. Prior research emphasizes that community-based engagement enhances trust and contextualizes early-warning outputs, increasing the likelihood of adoption in vulnerable populations [59]. Supporting local small and medium-sized enterprises (SMEs) in AI development is equally important to reduce over-reliance on external vendors and strengthen regional innovation ecosystems.

7.5. Research and Evaluation

Future research must shift from reactive pilots to anticipatory models that forecast disruptions before they occur. Standardization of resilience metrics—such as network connectivity under stress, predictive accuracy benchmarks, and recovery time, would enable more robust cross-study comparisons. Inclusive socio-technical evaluations are equally necessary to assess ethical, social, and trust dimensions of AI deployments, especially in resource-constrained contexts where community acceptance determines system effectiveness [33]. Without such evaluations, even technically advanced AI solutions risk rejection at the point of deployment.
To close the validation gap, future research should prioritize field pilots in low-resource settings, supported by international collaborations and participatory approaches. These pilots would not only test the robustness of AI models under real-world constraints but also build local capacity for long-term adoption.

7.6. Implementation Pathway

A phased roadmap offers a structured approach for transitioning from experimentation to institutionalization (Figure 4). Phase 0 should involve readiness assessments focusing on data availability, governance structures, and corridor vulnerabilities. Phase 1 can initiate targeted pilots in areas with high risk exposure, such as flood-prone highways or bus rapid transit ventilation systems. Phase 2 should expand these pilots into regulatory sandboxes with explicit policy alignment. Finally, Phase 3 should scale and institutionalize successful models into national resilience workflows. This phased approach balances innovation with risk management, ensuring that AI adoption is both responsible and scalable [15,64].
Looking ahead, the effective deployment of AI for transport resilience hinges on multi-layered governance, robust infrastructure and data, sustainable financing, and strong human capacity development (Figure 4). These pillars establish the foundation for shifting from fragmented experiments to systematic, large-scale implementation.

8. Conclusions and Future Research

This scoping review synthesized 58 peer-reviewed studies published between 2015 and 2025 to examine how Artificial Intelligence (AI) enhances infrastructure resilience, with transportation systems as the strategic case. The evidence shows that AI contributes most strongly to four areas: predictive maintenance, monitoring and sensing, forecasting and early warning, and optimization for decision support. These functions reduce downtime, improve adaptive capacity, and strengthen both direct and indirect aspects of transport resilience.
However, large-scale deployment remains constrained by high costs, data scarcity, infrastructural deficits, governance gaps, and limited real-world validation. While many models demonstrate high technical performance under controlled conditions, few have been tested in operational settings, particularly in low-resource environments such as Sub-Saharan Africa. This validation gap is particularly critical in Sub-Saharan Africa, where infrastructural fragility, limited digital infrastructure, and context-specific risks mean that models validated in developed regions may not generalize effectively. Real-world pilots in such contexts are essential to ensure transferability, robustness, and equity in AI-enabled resilience strategies. This gap limits both scalability and policy relevance.
The review contributes by providing a consolidated framework that classifies AI applications across technologies, sectors, resilience dimensions, and transport linkages. It also highlights cross-sector pathways, linking energy, water, agriculture, climate, and digital infrastructures, that together shape transport resilience. The recommendations presented emphasize governance, data foundations, financing, human capacity, and trust as the critical enablers for responsible adoption.
Future research should prioritize three directions. First, context-specific validation of AI models is needed in real-world environments, particularly in regions facing acute resilience challenges. Second, hybrid and explainable AI approaches should be advanced to enhance interpretability, robustness, and user trust. Third, cross-sector integration through systems-of-systems modeling can better capture the interdependencies that determine transport resilience in practice.
Beyond resilience, embedding AI in transport systems contributes to sustainability goals. By reducing disaster-related infrastructure losses, enabling low-carbon mobility, and strengthening freight continuity, AI advances SDG 9 (resilient infrastructure), SDG 11 (sustainable cities), and SDG 13 (climate action). Future work should more explicitly align AI deployments with these sustainability outcomes.
In summary, AI offers transformative opportunities for transportation systems, but its promise can only be realized within an enabling ecosystem of governance, data infrastructure, sustainable financing, and human capacity. By strategically aligning these domains, transport can evolve into a more adaptive and intelligent backbone of national infrastructure resilience.

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/su17208992/s1, PRISMA-ScR checklist.

Author Contributions

Conceptualization, A.K., O.O.A., K.D. and L.D.; methodology, O.O.A.; software, O.O.A.; formal analysis, O.O.A.; investigation, O.O.A.; resources, O.O.A.; data curation, O.O.A.; writing—original draft preparation, O.O.A.; writing—review and editing, A.K., K.D., L.D. and O.O.A.; visualization, O.O.A.; supervision, A.K. and L.D. All authors have read and agreed to the published version of the manuscript.

Funding

This project was made possible through funding received from The Transport and Education Training Authority (TETA), project number: TETA22/R&K/PR0011.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

Data is as contained in this manuscript. The data can be found in the repository (General File—Processed Using Zotero and MS-Excel): https://github.com/ajayioo/research_data_codes/blob/main/infrastructure_files.csv (accessed on 5 September 2025); (Extracted File—Processed Using Zotero and MS-Excel): https://github.com/ajayioo/research_data_codes/blob/main/extracted_infrastructure_files.csv (accessed on 5 September 2025).

Acknowledgments

The authors appreciate the funding provided by The Transport and Education Training Authority (TETA) for the execution of this research project. The encouragement and enabling environment from Tshwane University of Technology is appreciated herewith.

Conflicts of Interest

The authors declared no conflicts of interest.

References

  1. Filho, W.L.; Wall, T.; Mucova, S.A.R.; Nagy, G.J.; Balogun, A.-L.; Luetz, J.M.; Ng, A.W.; Kovaleva, M.; Azam, F.M.S.; Alves, F.; et al. Deploying artificial intelligence for climate change adaptation. Technol. Forecast. Soc. Change 2022, 180, 121662. [Google Scholar] [CrossRef]
  2. Singh, S.; Goyal, M.K. Enhancing climate resilience in businesses: The role of artificial intelligence. J. Clean. Prod. 2023, 418, 138228. [Google Scholar] [CrossRef]
  3. Ajayi, O.O.; Kurien, A.M.; Djouani, K.; Dieng, L. A Proactive Predictive Model for Machine Failure Forecasting. Machines 2025, 13, 663. [Google Scholar] [CrossRef]
  4. Mageto, J.; Twinomurinzi, H.; Luke, R.; Mhlongo, S.; Bwalya, K.; Bvuma, S. Building resilience into smart mobility for urban cities: An emerging economy perspective. Int. J. Prod. Res. 2024, 62, 5556–5573. [Google Scholar]
  5. Abisoye, A.; Akerele, J.I.; Odio, E.; Collins, A.; Babatunde, G.O.; Mustapha, S.D. Using AI and machine learning to predict and mitigate cybersecurity risks in critical infrastructure. Int. J. Eng. Res. Dev. 2025, 21, 205–224. [Google Scholar]
  6. Obuseh, E.; Eyenubo, J.; Alele, J.; Okpare, A.; Oghogho, I. A systematic review of barriers to renewable energy integration and adoption. J. Asian Energy Stud. 2025, 9, 26–45. [Google Scholar] [CrossRef]
  7. Chavula, P.; Kayusi, F. Challenges in Sub-Saharan Africa’s Food Systems and the Potential Role of AI. LatIA 2025, 3, 318. [Google Scholar]
  8. Hosseini, S.H.; Khodabin, M.; Soroori Sarabi, A.; Sharifipour Bgheshmi, M.S. Artificial intelligence and disaster risk management: A need for continuous education. Socio-Spat. Stud. 2021, 5, 13–29. [Google Scholar]
  9. Mu, W.; Kleter, G.A.; Bouzembrak, Y.; Dupouy, E.; Frewer, L.J.; Al Natour, F.N.R.; Marvin, H.J.P. Making food systems more resilient to food safety risks by including artificial intelligence, big data, and internet of things into food safety early warning and emerging risk identification tools. Compr. Rev. Food Sci. Food Saf. 2024, 23, e13296. [Google Scholar] [CrossRef] [PubMed]
  10. Ngulube, P. Leveraging information and communication technologies for sustainable agriculture and environmental protection among smallholder farmers in tropical Africa. Discov. Environ. 2025, 3, 9. [Google Scholar] [CrossRef]
  11. Mienye, D.; Sun, Y.; Ileberi, E. Artificial intelligence and sustainable development in Africa: A comprehensive review. Mach. Learn. Appl. 2024, 18, 100591. [Google Scholar] [CrossRef]
  12. Wu, D.; Zheng, A.; Yu, W.; Cao, H.; Ling, Q.; Liu, J.; Zhou, D. Digital twin technology in transportation infrastructure: A comprehensive survey of current applications, challenges, and future directions. Appl. Sci. 2025, 15, 1911. [Google Scholar] [CrossRef]
  13. Yan, Y.; Wang, S.; Li, Z.; Chen, X. Digital twin in transportation infrastructure management. Infrastruct. J. 2023, 12, 145–162. [Google Scholar]
  14. Reichstein, M.; Benson, V.; Blunk, J.; Camps-Valls, G.; Creutzig, F.; Fearnley, C.J.; Han, B.; Kornhuber, K.; Rahaman, N.; Schölkopf, B.; et al. Early warning of complex climate risk with integrated artificial intelligence. Nat. Commun. 2025, 16, 2564. [Google Scholar] [CrossRef] [PubMed]
  15. OECD. Recommendation of the Council on Artificial Intelligence (OECD AI Principles); OECD Publishing: Paris, France, 2019. [Google Scholar]
  16. United Nations Educational, Scientific and Cultural Organization. Recommendation on the Ethics of Artificial Intelligence; United Nations Educational, Scientific and Cultural Organization: Paris, France, 2021. [Google Scholar]
  17. Ajayi, O. Comparative Analysis of Machine Learning Models for Evaluating the Impact of Speed Bumps on Travel Time and Fuel Consumption. J. Energy Technol. Environ. 2025, 7, 112–130. [Google Scholar]
  18. Ogundiran, O.; Nyembwe, J.P.K.B.; Ogundiran, J.; Ribeiro, A.S.N.; da Silva, M.G. A Systematic Review of Indoor Environmental Quality in Passenger Transport Vehicles of Tropical and Subtropical Regions. Atmosphere 2025, 16, 140. [Google Scholar] [CrossRef]
  19. García-Méndez, S.; Hernández-González, J.; Martínez-Álvarez, F. An explainable machine learning framework for railway predictive maintenance. Sci. Rep. 2025, 15, 12345. [Google Scholar] [CrossRef]
  20. Di, X.; Fu, Y.; Turkcan, M.K.; Ghasemi, M.; Mo, Z.; Zang, C.; Zussman, G. AI-Powered Urban Transportation Digital Twin: Methods and Applications. arXiv 2024, arXiv:2501.10396. [Google Scholar]
  21. Chen, Y.; Zhang, F.; Qian, Y.; Zeng, J.; Li, X. A new car-following model considering the driver’s dynamic reaction time and driving visual angle on the slope. Phys. A Stat. Mech. Its Appl. 2025, 663, 130408. [Google Scholar] [CrossRef]
  22. Ukoba, K.; Olatunji, K.O.; Adeoye, E.; Jen, T.C.; Madyira, D.M. Optimizing renewable energy systems through artificial intelligence: Review and future prospects. Energy Environ. 2024, 35, 3833–3879. [Google Scholar] [CrossRef]
  23. Wanyama, J.; Bwambale, E.; Kiraga, S.; Katimbo, A.; Nakawuka, P.; Kabenge, I.; Oluk, I. A systematic review of fourth industrial revolution technologies in smart irrigation: Constraints, opportunities, and future prospects for sub-Saharan Africa. Smart Agric. Technol. 2024, 7, 100412. [Google Scholar] [CrossRef]
  24. Zhang, J.; Zhang, X.; Li, H.; Fan, Y.; Meng, Z.; Liu, D.; Pan, S. Optimization of Water Quantity Allocation in Multi-Source Urban Water Supply Systems Using Graph Theory. Water 2024, 17, 61. [Google Scholar] [CrossRef]
  25. Habib, M.; Singh, S.; Jan, S.; Jan, K.; Bashir, K. The future of the future foods: Understandings from the past towards SDG-2. Npj Sci. Food 2025, 9, 138. [Google Scholar] [CrossRef]
  26. Sun, L.; Shawe-Taylor, J.; D’Ayala, D. Artificial intelligence-informed planning for the rapid response of hazard-impacted road networks. Sci. Rep. 2022, 12, 16286. [Google Scholar] [CrossRef]
  27. Kathmandu, N. A Comprehensive Study on Artificial Intelligence, Digital Infrastructure, and Data Policies with Recommendations for Policy to Strengthen AI Ecosystem; International Telecommunication Union: Geneva, Switzerland, 2025. [Google Scholar]
  28. Khurram, M.; Zhang, C.; Muhammad, S.; Kishnani, H.; An, K.; Abeywardena, K.; Chadha, U.; Behdinan, K. Artificial Intelligence in Manufacturing Industry Worker Safety: A New Paradigm for Hazard Prevention and Mitigation. Processes 2025, 13, 1312. [Google Scholar] [CrossRef]
  29. Akinade, O.; Adepoju, A.; Ige, A.B.; Afolabi, A.I.; Amoo, O.O. A conceptual model for network security automation: Leveraging AI-driven frameworks to enhance multi-vendor infrastructure resilience. Int. J. Sci. Technol. Res. Arch. 2021, 1, 39–59. [Google Scholar] [CrossRef]
  30. Yigit, Y.; Ferrag, M.A.; Sarker, I.H.; Maglaras, L.A.; Chrysoulas, C.; Moradpoor, N.; Janicke, H. Critical Infrastructure Protection: Generative AI, Challenges, and Opportunities. J. Cybersecur. 2024, 10, tyae012. [Google Scholar]
  31. Kolivand, P.; Azari, S.; Bakhtiari, A.; Namdar, P.; Ayyoubzadeh, S.M.; Rajaie, S.; Ramezani, M. AI applications in disaster governance with health approach: A scoping review. Arch. Public Health 2025, 83, 218. [Google Scholar] [CrossRef] [PubMed]
  32. Olaitan, O.; Issah, M.; Wayi, N. A framework to test South Africa’s readiness for the fourth industrial revolution. S. Afr. J. Inf. Manag. 2021, 23, 1–10. [Google Scholar] [CrossRef]
  33. Kwarkye, T.G. “We know what we are doing”: The politics and trends in artificial intelligence policies in Africa. Can. J. Afr. Stud./Rev. Can. Études Afr. 2025, 1–19. [Google Scholar] [CrossRef]
  34. Mehryar, S.; Yazdanpanah, V.; Tong, J. AI and climate resilience governance. Iscience 2024, 27, 109812. [Google Scholar] [CrossRef]
  35. Li, X.; Krivtsov, V.; Pan, C.; Nassehi, A.; Gao, R.X.; Ivanov, D. End-to-end supply chain resilience management using deep learning, survival analysis, and explainable artificial intelligence. Int. J. Prod. Res. 2025, 63, 1174–1202. [Google Scholar] [CrossRef]
  36. Arksey, H.; O’malley, L. Scoping studies: Towards a methodological framework. Int. J. Soc. Res. Methodol. 2005, 8, 19–32. [Google Scholar] [CrossRef]
  37. Levac, D.; Colquhoun, H.; O’brien, K.K. Scoping studies: Advancing the methodology. Implement. Sci. 2010, 5, 69. [Google Scholar] [CrossRef]
  38. Tricco, A.C.; Lillie, E.; Zarin, W.; O’Brien, K.K.; Colquhoun, H.; Levac, D.; Moher, D.; Peters, M.D.J.; Horsley, T.; Weeks, L.; et al. PRISMA extension for scoping reviews (PRISMA-ScR): Checklist and explanation. Ann. Intern. Med. 2018, 169, 467–473. [Google Scholar] [CrossRef]
  39. Peters, M.D.J.; Marnie, C.; Tricco, A.C.; Pollock, D.; Munn, Z.; Alexander, L.; McInerney, P.; Godfrey, C.M.; Khalil, H. Updated methodological guidance for the conduct of scoping reviews. JBI Evid. Synth. 2020, 18, 2119–2126. [Google Scholar] [CrossRef] [PubMed]
  40. Munn, Z.; Pollock, D.; Price, C.; Aromataris, E.; Stern, C.; Stone, J.C.; Barker, T.H.; Godfrey, C.M.; Clyne, B.; Booth, A.; et al. Investigating different typologies for the synthesis of evidence: A scoping review protocol. JBI Evid. Synth. 2023, 21, 592–600. [Google Scholar] [CrossRef]
  41. Khalil, M.; Prinsloo, P.; Slade, S. A comparison of learning analytics frameworks: A systematic review. In Proceedings of the LAK22: 12th International Learning Analytics and Knowledge Conference, Newport Beach, CA, USA, 19–23 March 2022. [Google Scholar]
  42. Mhlanga, D. Artificial intelligence and machine learning for energy consumption and production in emerging markets: A review. Energies 2023, 16, 745. [Google Scholar] [CrossRef]
  43. Okolo, F.C.; Etukudoh, E.A.; Ogunwole, O.; Osho, G.O.; Basiru, J.O. Advances in cyber-physical resilience of transportation infrastructure in emerging economies and coastal regions. Int. J. Multidiscip. Res. Growth Eval. 2023, 4, 1188–1198. [Google Scholar] [CrossRef]
  44. Kreuzer, T.; Papapetrou, P.; Zdravkovic, J. Artificial intelligence in digital twins: A systematic literature review. Data Knowl. Eng. 2024, 156, 104008. [Google Scholar] [CrossRef]
  45. Kalaldeh, M.; Tarawneh, D. Development of a dynamic quantitative digital model for the measurement of smart city maturity level in the city of Amman. Int. J. Sustain. Eng. 2025, 18, 1–17. [Google Scholar] [CrossRef]
  46. Olawale, M.A.; Ayeh, A.A.; Adekola, F.O.; Precious, A.S.; Joshua, A.O.; Oladosu, O.T. A review on the intersection of artificial intelligence on building resilient infrastructure, promoting inclusive and sustainable industrialization and fostering innovation. Int. J. Eng. Mod. Technol. 2023, 9, 1–31. [Google Scholar] [CrossRef]
  47. Li, J.; Yang, S.X. Digital twins to embodied artificial intelligence: Review and perspective. Intell. Robot. 2025, 5, 202–227. [Google Scholar] [CrossRef]
  48. Alqahtani, A.; Alsubai, S.; Bhatia, M. Applied artificial intelligence framework for smart evacuation in industrial disasters. Appl. Intell. 2024, 54, 7030–7045. [Google Scholar] [CrossRef]
  49. Abraham, A.; Zhang, Y.; Prasad, S. Evacuation management framework towards smart city-wide intelligent emergency interactive response system. arXiv 2024, arXiv:2403.07003. [Google Scholar] [CrossRef]
  50. Xu, H.; Omitaomu, F.; Sabri, S.; Zlatanova, S.; Li, X.; Song, Y. Leveraging Generative AI for Urban Digital Twins: A scoping review on the autonomous generation of urban data, scenarios, designs, and 3D city models for smart city advancement. Urban Inform. 2024, 3, 1–44. [Google Scholar] [CrossRef]
  51. Vega, G.; Hadjidemetriou, G. ON Artificial Intelligence Applications for Resilient Transport Infrastructure. In Proceedings of the 2025 European Conference on Computing in Construction CIB W78 Conference on IT in Construction, Porto, Portugal, 14–17 July 2025. [Google Scholar]
  52. Mirindi, D.; Khang, A.; Mirindi, F. Artificial Intelligence (AI) and Automation for Driving Green Transportation Systems: A Comprehensive Review. Driving Green Transportation System Through Artificial Intelligence and Automation: Approaches, Technologies and Applications; Springer: Berlin/Heidelberg, Germany, 2025; pp. 1–19. [Google Scholar]
  53. Union, African. Multi-hazard Early Warning for All Action Plan for Africa (2023–2027); Union, African: Addis Ababa, Ethiopia, 2023. [Google Scholar]
  54. Kagabo, J.; Kattel, G.R.; Kazora, J.; Shangwe, C.N.; Habiyakare, F. Application of Machine Learning Algorithms in Predicting Extreme Rainfall Events in Rwanda. Atmosphere 2024, 15, 691. [Google Scholar] [CrossRef]
  55. UNGM. Feasibility Study on the Integration of AI and ML in Weather Forecasting; United Nations Development Programme: New York City, NY, USA, 2025. [Google Scholar]
  56. SMEC. IoT at Msikaba Bridge: How the Internet of Things Is Revolutionising Infrastructure; SMEC: Melbourne, Australia, 2025. [Google Scholar]
  57. Al-Ali, R.; Beheiry, S.; Alnabulsi, A.; Obaid, S.; Mansoor, N.; Odeh, N.; Mostafa, A. An IoT-based road bridge health monitoring and warning system. Sensors 2024, 24, 469. [Google Scholar] [CrossRef]
  58. Rust, F.C.; Sampson, L.R.; Cachia, A.A.; Verhaeghe, B.M.; Fourie, H.S.; Smit, M.A.; Hoffman, A.; Steyn, W.J.; Venter, K.; Lefophane, S. Technology foresight for the South African road transport sector by 2035. J. Transp. Supply Chain Manag. 2024, 18, 1–16. [Google Scholar] [CrossRef]
  59. Agbehadji, E.; Schütte, S.; Masinde, M.; Botai, J.; Mabhaudhi, T. Climate risks resilience development: A bibliometric analysis of climate-related early warning systems in Southern Africa. Climate 2023, 12, 3. [Google Scholar] [CrossRef]
  60. European Commission. EU Artificial Intelligence Act Enters into Force; European Commission: Brussels, Belgium, 2024. [Google Scholar]
  61. European Commission. Artificial Intelligence Act Overview; European Parliament & Council of the EU: Strasbourg, France, 2024. [Google Scholar]
  62. African Union Commission. Continental Artificial Intelligence Strategy; African Union Commission (AUC): Addis Ababa, Ethiopia, 2024. [Google Scholar]
  63. World Bank. Transport Overview: Climate Co-Benefits; World Bank: Washington, DC, USA, 2023. [Google Scholar]
  64. United Nations Office for Disaster Risk Reduction. Global Assessment Report on Disaster Risk Reduction 2022; United Nations Office for Disaster Risk Reduction: Geneva, Switzerland, 2022. [Google Scholar]
  65. World Bank. Transport Resilience Financing Guidelines; World Bank: Washington, DC, USA, 2023. [Google Scholar]
Sustainability 17 08992 g001
Figure 1. Conceptual Framework for AI in Transport Resilience.
Figure 1. Conceptual Framework for AI in Transport Resilience.
Sustainability 17 08992 g001
Sustainability 17 08992 g002
Figure 2. PRISMA Flow Diagram of the study selection process.
Figure 2. PRISMA Flow Diagram of the study selection process.
Sustainability 17 08992 g002
Sustainability 17 08992 g003
Figure 3. Distribution of AI functions across sectors.
Figure 3. Distribution of AI functions across sectors.
Sustainability 17 08992 g003
Sustainability 17 08992 g004
Figure 4. Roadmap for AI in Transport Resilience.
Figure 4. Roadmap for AI in Transport Resilience.
Sustainability 17 08992 g004
Table 1. Thematic synthesis of AI applications for infrastructure and transport resilience.
Table 1. Thematic synthesis of AI applications for infrastructure and transport resilience.
DomainTechnologies AppliedInfrastructure SectorsTransport LinkageResilience/Security FocusKey Insights & Gaps
Transportation ResilienceML, Random Forest, XGBoost, IoT, AI-driven HVACPassenger transport, road networks, mobility systemsDirectPredictive maintenance, driver comfort, fuel efficiency, indoor air qualityStrong results in predictive modeling and real-time monitoring; limited deployment in SSA; gaps in standardization and validation [3,4,18].
Energy, Water & Food Systems ResilienceAI, ML, IoT, UAVs, Big Data, Blockchain, Neural NetsSmart grids, renewable energy, irrigation, food safety systemsIndirectGrid reliability, energy access, water-use efficiency, food system resilienceAI supports e-mobility through stable energy; AI–IoT strengthens irrigation and food safety; barriers = affordability, digital divide, data scarcity [6,9,23].
Climate, Environment & Early-Warning SystemsDeep learning, Lookahead search, Causal AI, Geospatial FMs, IoT, Remote sensingDisaster risk management, climate adaptation programsDirect & IndirectHazard forecasting, rapid post-disaster recovery, risk predictionAI improves EWS and post-disaster road recovery; gaps in ethical transparency, community participation, and FAIR/FATES-compliant data sharing [8,14,26].
Cyber & Digital Infrastructure SecurityML, NLP, Predictive analytics, XAICritical infrastructure, communication networks, CPSIndirect → Direct (smart transport)Cybersecurity resilience, anomaly detection, automationAI-driven cybersecurity enhances anomaly detection in transport-linked CI; challenges include adversarial attacks, dataset scarcity, lack of robust standards [5,29].
Policy, Ethics, Governance & EducationCross-sector AI & 4IR frameworksNational infrastructure, education systems, multi-sector ecosystemsIndirectGovernance, ethical AI, capacity building, adoption frameworksPolicies and governance dictate adoption; SSA faces infrastructure and funding gaps; localized policy ecosystems [11,32,33].
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Ajayi, O.O.; Kurien, A.; Djouani, K.; Dieng, L. Artificial Intelligence for Infrastructure Resilience: Transportation Systems as a Strategic Case for Policy and Practice. Sustainability 2025, 17, 8992. https://doi.org/10.3390/su17208992

AMA Style

Ajayi OO, Kurien A, Djouani K, Dieng L. Artificial Intelligence for Infrastructure Resilience: Transportation Systems as a Strategic Case for Policy and Practice. Sustainability. 2025; 17(20):8992. https://doi.org/10.3390/su17208992

Chicago/Turabian Style

Ajayi, Olusola O., Anish Kurien, Karim Djouani, and Lamine Dieng. 2025. "Artificial Intelligence for Infrastructure Resilience: Transportation Systems as a Strategic Case for Policy and Practice" Sustainability 17, no. 20: 8992. https://doi.org/10.3390/su17208992

APA Style

Ajayi, O. O., Kurien, A., Djouani, K., & Dieng, L. (2025). Artificial Intelligence for Infrastructure Resilience: Transportation Systems as a Strategic Case for Policy and Practice. Sustainability, 17(20), 8992. https://doi.org/10.3390/su17208992

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop