Urban Currents

Application of a GA-BPNN model with Bayesian updating mechanism for pavement quality index (PQI) prediction

The study developed a hybrid model combining a Genetic Algorithm-optimized Backpropagation Neural Network (GA-BPNN) with Bayesian updating to predict the Pavement Quality Index (PQI). The model was trained on pavement performance data collected via a multi-functional road condition assessment system across five major Beijing highways (G4, G6, G7, S15, G106) from 2008 to 2021, and dynamically updated using 2022 inspection data.

Why it matters — By integrating Bayesian inference, the model introduces a self-learning mechanism that can continuously update its predictions with new, small-sample inspection data, overcoming the static limitations and poor uncertainty quantification of traditional neural networks in infrastructure maintenance.

Caveat: The model's performance and updating mechanism were validated using data from only five specific highway corridors within a single metropolitan region.

Transportation Research Interdisciplinary Perspectives · doi

A systematic review of inclusive intelligent transportation systems in smart cities

This systematic review analyzed 2,253 studies published between 2021 and 2026 to evaluate disability-inclusive intelligent transportation systems (ITS). The analysis reveals that 66.0% of the literature focuses on general accessibility frameworks, while only 10.2% addresses cognitive impairments and 5.5% targets mobility impairments. Most proposed AI and deep learning solutions remain simulation-based and lack real-world validation, standardized evaluation metrics, and accessibility-aware datasets.

Why it matters — It exposes a critical disconnect in smart-city research where advanced AI mobility models are rarely tested in real-world scenarios or designed for specific, high-need disability groups, providing a unified framework to bridge technology, policy, and user-centered design.

Caveat: The findings are based on a literature review of academic publications rather than an empirical assessment of deployed municipal transportation systems.

Discover Cities · pdf · doi

Negotiating temporalities: History, heritage, and the politics of urban regeneration in Beijing

Based on fieldwork conducted in 2019 in Beijing's Dashilar neighborhood, this study analyzes how different actors mobilize competing historical eras to contest and negotiate urban regeneration. The research demonstrates that state actors prioritize imperial history to define heritage and authenticity, residents draw on socialist histories to defend housing entitlements, and market actors leverage post-reform temporalities to treat urban space as a commodity.

Why it matters — It provides a framework showing that history is not just a passive backdrop but an active, strategically deployed resource that shapes whether urban regeneration projects succeed or stagnate.

Caveat: The findings are based on a qualitative case study of a single neighborhood in Beijing, which may limit generalizability to cities with different historical and institutional trajectories.

City Culture and Society · pdf · doi

15-Minute City and beyond: Linking local and regional accessibility with urban vitality

This study analyzes the relationship between urban vitality and two scales of accessibility in Hong Kong, using longitudinal restaurant registration records as a proxy for vitality. Local accessibility is measured by walking time to neighborhood centers, while regional accessibility is measured by public transit travel time to the mainland border. Leveraging the closure and reopening of the Hong Kong-Mainland China border as a quasi-natural experiment, the analysis reveals that local accessibility is positively correlated with restaurant density, while improved regional accessibility yields a modest 2% increase in vitality.

Why it matters — It demonstrates that local accessibility is the primary driver of urban vitality, and that regional connectivity cannot compensate for poor local walkability, though it can amplify economic growth in areas where strong local access already exists.

Caveat: The study relies on restaurant registration counts as a sole proxy for the broader concept of urban vitality.

Cities · doi

Framing artificial intelligence as a policy instrument in urban climate action: Practitioners' perspectives from Paris

This study analyzes how artificial intelligence is integrated into urban climate action in Paris, using 10 open interviews with 12 practitioners and an analysis of 13 policy documents. It maps how AI introduces incremental shifts in daily tasks, alters professional responsibilities around risk assessment, and introduces tensions between technical utility and symbolic values like accountability.

Why it matters — It reframes AI from a neutral technical tool to a socio-technical policy instrument, revealing how its deployment can depoliticize climate action and alter decision hierarchies in municipal workflows.

Caveat: The findings are based on a small qualitative sample of interviews and documents from a single European capital, which may limit generalizability to other municipal governance contexts.

Cities · pdf · doi

Impact of nature connectedness on loneliness among university students living away from home: The mediating roles of meaningfulness, peace of mind, and resilience

This study analyzed survey responses from 257 Indian urban youth aged 19 to 26 to evaluate how nature connectedness relates to loneliness. Using partial least squares structural equation modeling, the analysis measured the mediating roles of meaningfulness, peace of mind, and resilience, finding that the negative association between nature connection and loneliness is significantly mediated by meaningfulness and peace of mind.

Why it matters — It identifies the specific psychological pathways—meaningfulness and peace of mind, rather than resilience—through which contact with nature mitigates loneliness in young urban populations, providing a clearer mechanism for designing targeted nature-based mental health interventions.

Caveat: The study relies on a relatively small convenience sample of university students and uses self-reported online survey data, which may limit the generalizability of the causal pathways.

Cities · pdf · doi

Deep learning reveals cross-platform consistency in street view imagery for urban perception mapping

The study evaluates the consistency of physical streetscape elements and human perceptual scores between Baidu Street View and Google Street View using paired images from Hong Kong. At the sampling-point scale, vegetation, buildings, and sky show strong cross-platform correlations of 0.753, 0.756, and 0.720, while perceptual dimensions like safety and depressiveness correlate between 0.591 and 0.672. Robustness tests using SegFormer, DeepLabv3+, ResNet50, and CLIP reveal that physical structural consistency remains stable across segmentation pipelines, whereas perceptual consistency is highly sensitive to the chosen model backbone.

Why it matters — It establishes that physical streetscape metrics can be reliably integrated across different street-view platforms, though researchers must account for model backbone sensitivity when mapping subjective human perceptions.

Caveat: The findings and cross-platform comparison are based on a single city, Hong Kong, which may limit generalizability to cities with different urban densities or street layouts.

Computers Environment and Urban Systems · doi

Whether and where to move? Exploring the determinants of high-tech firms’ relocation decisions in the Pearl River Delta region, China

This study analyzes the relocation decisions of high-tech firms in China's Pearl River Delta region, examining the choice of whether to move, where to relocate, and how these choices shift across a firm's life cycle. The analysis uses a framework spanning four dimensions: territorial and network embeddedness, location conditions, policy factors, and innovation capacity.

Why it matters — It demonstrates that policy factors, territorial embeddedness, and innovation capacity are the primary drivers of high-tech relocation, and reveals that destination choices are heavily constrained by a firm's previous location and origin-destination disparities.

Caveat: The abstract does not disclose the specific sample size, time period, or quantitative model used to establish these relationships.

Habitat International · doi

Driving under the influence of cannabis (DUIC): A risky proposition?

This study analyzes Canadian national survey data from 2018 to 2023 to track trends in cannabis use and driving under the influence of cannabis (DUIC) following legalization. It finds that while overall cannabis use rose, DUIC prevalence among users decreased, keeping general population DUIC rates stable until slight increases emerged in 2022 and 2023. The analysis also links these trends to public health campaigns and users' risk perceptions, noting that initial post-legalization spikes in THC-related traffic injuries and fatalities have begun returning to baseline levels.

Why it matters — It demonstrates that legalizing cannabis does not automatically trigger a sustained surge in impaired driving, suggesting that targeted public health campaigns can successfully mitigate risks even as overall consumption grows.

Caveat: The findings rely on self-reported survey data regarding drug use and driving behavior, which may be subject to underreporting.

Journal of Transport & Health · pdf · doi

Divergent approaches to the professional governance of arborists and urban foresters in Canada

This study analyzes the professional governance of Canadian arborists and urban foresters using a policy review and expert interviews with governance organizations. It reveals that urban forestry is typically managed through 'governance with government' frameworks driven by public interest, whereas arboriculture relies on voluntary trade designations managed 'by and without government' to improve safety, pay, and professional standing.

Why it matters — It documents how differing regulatory pathways create jurisdictional friction and ambiguous scopes of practice between the two fields, showing that shifting toward formal government regulation can trigger trade-offs between professional recognition and reduced labor market competition.

Caveat: The findings are based on policy reviews and interviews with governance organizations rather than a direct survey of the practitioners whose autonomy and market competition are affected.

Urban forestry & urban greening · pdf · doi

Technocratic humanism and computational urbanism: Adriano Olivetti, Constantinos A. Doxiadis, and industrial settlements in Ivrea and ‘Aspra Spitia’

This historical study compares the mid-twentieth-century industrial urbanism models of Adriano Olivetti in Ivrea, Italy, and Constantinos A. Doxiadis in Aspra Spitia, Greece. It analyzes how both figures integrated architecture, social design, and early computational infrastructures—specifically Olivetti's Elea 9003 computer and the Doxiadis Associates Computer Center in Athens—to manage and plan human-scale industrial settlements.

Why it matters — It traces the historical roots of modern computational urbanism, demonstrating that the integration of information processing, governance, and spatial planning emerged much earlier than the contemporary 'smart city' era as an attempt to balance industrialization with humanistic design.

Planning Perspectives · doi

‘We’re the real shit!’ Low-income groups’ unequal access to green mobility transitions in Norway

Through 14 storytelling workshops and 22 individual interviews with 73 low-income participants across Bergen, Stavanger, and Trondheim, this study documents how low-income groups are excluded from the benefits of Norway's low-carbon mobility transition. The analysis reveals that these residents face systemic barriers to accessing electric vehicle subsidies, new technologies, and public infrastructure, while receiving no recognition or reward for their existing low-emission travel habits.

Why it matters — It demonstrates that technology-centric sustainability policies, such as subsidizing electric vehicles, actively marginalize low-income populations and ignore low-cost, highly effective alternatives like e-bikes and affordable public transit.

Caveat: The findings are based on qualitative self-reporting from storytelling workshops and interviews, rather than quantitative travel diaries or spatial accessibility modeling.

Mobilities · pdf · doi

Constitutional governance for societies of AI agents in the built environment: a research agenda

This paper establishes a research agenda for the constitutional governance of multi-agent AI systems operating within buildings and urban infrastructure. It structures this agenda around three core challenges: designing mechanisms for building retrofits under deep uncertainty using public subsidies, verifying agent behavior in building operations using physics-informed models, and coordinating urban infrastructure to remain resilient during shocks. The framework draws on three decades of multi-agent systems research to define conceptual foundations, analyze a specific governance event, and identify twelve open research problems.

Why it matters — It shifts the paradigm of urban AI governance from treating individual AI systems as isolated tools to managing them as a complex, negotiating society of strategic agents with asymmetric information. This provides a structured pathway to prevent systemic failures and coordinate collective AI behaviors in physical urban systems before widespread, uncoordinated deployment occurs.

arXiv · pdf