Urban environments, university campuses, and public and private buildings often present architectural barriers that prevent people with disabilities and special needs to move freely and independently. This paper presents a systematic mapping study of the scientific literature proposing devices, and software applications aimed at fostering accessible wayfinding and navigation in indoor and outdoor environments. We selected 111 out of 806 papers published in the period 2009–2020, and we analyzed them according to different dimensions: at first, we surveyed which solutions have been proposed to address the considered problem; then, we analyzed the selected papers according to five dimensions: context of use, target users, hardware/software technologies, type of data sources, and user role in system design and evaluation. Our findings highlight trends and gaps related to these dimensions. The paper finally presents a reflection on challenges and open issues that must be taken into consideration for the design of future accessible places and of related technologies and applications aimed at facilitating wayfinding and navigation.

Deep learning models are tools for data analysis suitable for approximating (non-linear) relationships among variables for the best prediction of an outcome. While these models can be used to answer many important questions, their utility is still harshly criticized, being extremely challenging to identify which data descriptors are the most adequate to represent a given specific phenomenon of interest. With a recent experience in the development of a deep learning model designed to detect failures in mechanical water meter devices, we have learnt that a sensible deterioration of the prediction accuracy can occur if one tries to train a deep learning model by adding specific device descriptors, based on categorical data. This can happen because of an excessive increase in the dimensions of the data, with a correspondent loss of statistical significance. After several unsuccessful experiments conducted with alternative methodologies that either permit to reduce the data space dimensionality or employ more traditional machine learning algorithms, we changed the training strategy, reconsidering that categorical data, in the light of a Pareto analysis. In essence, we used those categorical descriptors, not as an input on which to train our deep learning model, but as a tool to give a new shape to the dataset, based on the Pareto rule. With this data adjustment, we trained a more performative deep learning model able to detect defective water meter devices with a prediction accuracy in the range 87–90%, even in the presence of categorical descriptors.

This work presents an architecture to help designing and deploying smart mobility applications. The proposed solution builds on the experience already matured by the authors in different fields: crowdsourcing and sensing done by users to gather data related to urban barriers and facilities, computation of personalized paths for users with special needs, and integration of open data provided by bus companies to identify the actual accessibility features and estimate the real arrival time of vehicles at stops. In terms of functionality, the first “monolithic” prototype fulfilled the goal of composing the aforementioned pieces of information to support citizens with reduced mobility (users with disabilities and/or elderly people) in their urban movements. In this paper, we describe a service-oriented architecture that exploits the microservices orchestration paradigm to enable the creation of new services and to make the management of the various data sources easier and more effective. The proposed platform exposes standardized interfaces to access data, implements common services to manage metadata associated with them, such as trustworthiness and provenance, and provides an orchestration language to create complex services, naturally mapping their internal workflow to code. The manuscript demonstrates the effectiveness of the approach by means of some case studies.

The Nobel laureate Niels Bohr once said that: “Predictions are very difficult, especially if they are about the future”. Nonetheless, models that can forecast future COVID-19 outbreaks are receiving special attention by policymakers and health authorities, with the aim of putting in place control measures before the infections begin to increase. Nonetheless, two main problems emerge. First, there is no a general agreement on which kind of data should be registered for judging on the resurgence of the virus (e.g., infections, deaths, percentage of hospitalizations, reports from clinicians, signals from social media). Not only this, but all these data also suffer from common defects, linked to their reporting delays and to the uncertainties in the collection process. Second, the complex nature of COVID-19 outbreaks makes it difficult to understand if traditional epidemiological models, such as susceptible, infectious, or recovered (SIR), are more effective for a timely prediction of an outbreak than alternative computational models. Well aware of the complexity of this forecasting problem, we propose here an innovative metric for predicting COVID-19 diffusion based on the hypothesis that a relation exists between the spread of the virus and the presence in the air of particulate pollutants, such as PM_2.5, PM₁₀, and NO₂. Drawing on the recent assumption of 239 experts who claimed that this virus can be airborne, and further considering that particulate matter may favor this airborne route, we developed a machine learning (ML) model that has been instructed with: (i) all the COVID-19 infections that occurred in the Italian region of Emilia-Romagna, one of the most polluted areas in Europe, in the period of February–July 2020, (ii) the daily values of all the particulates taken in the same period and in the same region, and finally (iii) the chronology according to which restrictions were imposed by the Italian Government to human activities. Our ML model was then subjected to a classic ten-fold cross-validation procedure that returned a promising 90% accuracy value. Finally, the model was used to predict a possible resurgence of the virus in all the nine provinces of Emilia-Romagna, in the period of September–December 2020. To make those predictions, input to our ML model were the daily measurements of the aforementioned pollutants registered in the periods of September–December 2017/2018/2019, along with the hypothesis that the mild containment measures taken in Italy in the so-called Phase 3 are obeyed. At the time we write this article, we cannot have a confirmation of the precision of our predictions. Nevertheless, we are projecting a scenario based on an original hypothesis that makes our COVID-19 prediction model unique in the world. Its accuracy will be soon judged by history—and this, too, is science at the service of society.

Hyperspectral imaging (HSI) has become widely used in cultural heritage (CH). This very efficient method for artwork analysis is connected with the generation of large amounts of spectral data. The effective processing of such heavy spectral datasets remains an active research area. Along with the firmly established statistical and multivariate analysis methods, neural networks (NNs) represent a promising alternative in the field of CH. Over the last five years, the application of NNs for pigment identification and classification based on HSI datasets has drastically expanded due to the flexibility of the types of data they can process, and their superior ability to extract structures contained in the raw spectral data. This review provides an exhaustive analysis of the literature related to NNs applied for HSI data in the CH field. We outline the existing data processing workflows and propose a comprehensive comparison of the applications and limitations of the various input dataset preparation methods and NN architectures. By leveraging NN strategies in CH, the paper contributes to a wider and more systematic application of this novel data analysis method.