Abstract
Cities worldwide are undergoing profound transformations driven by rapid urbanisation, climate change, digitalisation, and evolving mobility patterns. In this context, ambient intelligence (AmI) and intelligent transport systems (ITS) are increasingly central to shaping responsive, sustainable, and human-centred urban systems. Advances in sensing technologies, data analytics, machine learning, and geospatial intelligence have enabled urban environments to move beyond static, rule-based management toward dynamic, adaptive ecosystems. These developments allow cities not only to monitor conditions in real time but also to anticipate change, optimise operations, and enhance quality of life. This thematic issue comprises four articles. Our guest editors, Filipe Rodrigues and Merkebe Getachew Demissie, and the editor, Ana Alves, supervised the manuscript review process, for which we thank them for their service. This thematic issue explores how data-driven intelligence can support smarter mobility systems, improved environmental quality, and more adaptive urban services. The selected contributions illustrate the breadth of ITS applications, from walkability analysis and pollution forecasting to mobility–environment interactions and context-aware recommendation systems. Together, they demonstrate how integrating mobility data, environmental sensing, and predictive analytics can help address pressing urban challenges while advancing the core objectives of AmI and Smart Environments.
The first paper, authored by Pereira et al., ‘Analysis of walkability surrounding the future bus rapid transit lines in the city of Coimbra’, examines the pedestrian accessibility and walkability conditions around planned Bus Rapid Transit (BRT) corridors. By integrating spatial analysis with walkability indicators, the study provides evidence-based insights into how future transit investments can better support active mobility and transit-oriented development. The work highlights the importance of anticipating user needs at the planning stage, ensuring that smart transport infrastructure is embedded within environments that are safe, accessible, and conducive to walking.
The second paper, authored by Mutlu, ‘Machine learning-based approaches to forecast particulate matter in the Istanbul metropolitan area’, addresses urban environmental quality through the lens of predictive intelligence. By applying machine learning techniques to forecast particulate matter concentrations, the study demonstrates how data-driven models can enhance situational awareness and support proactive environmental management in large metropolitan areas. Such predictive capabilities are essential for smart environments that aim to adapt dynamically to changing conditions and mitigate risks to public health. AmI has transitioned from static, rule-based systems to dynamic, living ecosystems that are too complex for traditional manual programming. This is mainly due to the heterogeneity and large volume of sensing data. City sensors enable traffic and air quality to be measured, providing a basis for studying the correlation between mobility and pollution in a given city, especially in extreme situations, such as during the pandemic emergency period.
The third paper, ‘Mobility effect on city pollution: A case study’, by Juma et al., predicts the major air pollutant, nitrogen dioxide (NO2), considering as input variables human activity, weather, and other air pollutants over one year of collected data, with the scenario of the city of Lisbon during the first period of the SARS-CoV-2 virus pandemic. Additionally, a correlation was established between the calculated impacts of the decreased mobility and the reduction in air pollution. Urban computing leverages geospatial data and sensing to enhance city living. The development of location-based recommender algorithms/systems has opened the door to more effective methods for recommending sites that users might find interesting.
The fourth paper, ‘Geo-Lecture: Exploiting location-based recommendation in the context of lecture attendance tracking using a Geographical Information System (GIS)’, by Asabere et al., proposes a method for tracking and monitoring classroom attendance patterns using a GIS. This method includes context-aware recommendations for lecture room assignment and real-time attendance verification through geofencing. The authors introduce a device-agnostic platform featuring integrated GIS visualisation, which was empirically validated against educational metrics and recommender system KPIs.
The future of transportation is increasingly shared, on-demand, and multimodal, prompting urban and transportation planners to redesign urban areas to accommodate diverse transportation modes. The overarching goal is to reduce dependence on private vehicles by creating urban environments where multiple mobility options are accessible, efficient, and well-integrated. In recent years, cities have introduced on-demand mobility services to address accessibility gaps, particularly in areas where conventional public transit is limited or unavailable. At the same time, shared micromobility services have expanded rapidly across cities worldwide, helping address first- and last-mile challenges and, in some cases, substituting for personal car use for short-distance trips. In this context, cities are increasingly evaluating the impacts of these services on accessibility, both as standalone modes and in integration with public transit systems. Although the introduction of micromobility can create different forms of ‘coopetition’ with public transit, such as complementary, supplementary, or competitive, most urban planners are prioritising complementary and supplementary integration strategies to enhance overall system performance and improve access.
Looking ahead, several avenues for future research emerge. To improve accessibility and walkability, cities must critically assess how existing and planned transportation services operate and whether they meet diverse mobility needs. Accessibility should be placed at the centre of evaluation, incorporating multiple equity perspectives, including utilitarian, egalitarian, and sufficientarian frameworks. Furthermore, both horizontal and vertical equity assessments should be conducted to examine access outcomes across different population groups. By systematically evaluating how current and planned services address accessibility and walkability, cities can realign their transportation systems with broader mobility goals and objectives, ensuring that future investments promote inclusive, sustainable, and equitable urban development.
Researchers are increasingly adopting machine learning, particularly deep learning, to enhance predictions of emissions and pollutants and to complement traditional statistical and physics-based models. Conventional approaches, such as regression or dispersion models, often rely on simplifying assumptions, such as linearity or limited interactions, that struggle to capture the complex, nonlinear dynamics of real-world emissions. Emission patterns are influenced by multiple interacting factors, such as traffic activity, weather, land use, topography, and human behaviour, and their sources and impacts are not always spatially aligned. These processes exhibit strong spatiotemporal correlations, including short-term fluctuations, long-term trends, and cross-location interactions, which are challenging for traditional models to represent accurately.
Deep learning architectures such as recurrent neural networks (RNN), long short-term memory (LSTM), convolutional neural network (CNN), and graph neural networks (GNN) are well-suited to modelling these temporal and spatial dependencies, while emerging approaches like transformers with attention mechanisms can capture long-range interactions across time and space. These models also allow integration of heterogeneous data sources, handle missing or noisy data, and adapt to changing conditions such as policy interventions or extreme events. As cities become increasingly instrumented, advanced spatiotemporal deep learning offers the potential to improve prediction accuracy, model complex nonlinear interactions, and support equitable, evidence-based environmental policy.
