Automatic Tracking Based on Weighted Fusion Back Propagation in UWB for IoT Devices.

The global population is progressively entering an aging phase, with population aging likely to emerge as one of the most-significant social trends of the 21st Century, impacting nearly all societal domains. Addressing the challenge of assisting vulnerable groups such as the elderly and disabled in carrying or transporting objects has become a critical issue in this field. We developed a mobile Internet of Things (IoT) device leveraging Ultra-Wideband (UWB) technology in this context. This research directly benefits vulnerable groups, including the elderly, disabled individuals, pregnant women, and children. Additionally, it provides valuable references for decision-makers, engineers, and researchers to address real-world challenges. The focus of this research is on implementing UWB technology for precise mobile IoT device localization and following, while integrating an autonomous following system, a robotic arm system, an ultrasonic obstacle-avoidance system, and an automatic leveling control system into a comprehensive experimental platform. To counteract the potential UWB signal fluctuations and high noise interference in complex environments, we propose a hybrid filtering-weighted fusion back propagation (HFWF-BP) neural network localization algorithm. This algorithm combines the characteristics of Gaussian, median, and mean filtering, utilizing a weighted fusion back propagation (WF-BP) neural network, and, ultimately, employs the Chan algorithm to achieve optimal estimation values. Through deployment and experimentation on the device, the proposed algorithm's data preprocessing effectively eliminates errors under multi-factor interference, significantly enhancing the precision and anti-interference capabilities of the localization and following processes.

Neural Networks for Hyperspectral Imaging of Historical Paintings: A Practical Review.

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.

Edge-Based Transfer Learning for Classroom Occupancy Detection in a Smart Campus Context.

Studies and systems that are aimed at the identification of the presence of people within an indoor environment and the monitoring of their activities and flows have been receiving more attention in recent years, specifically since the beginning of the COVID-19 pandemic. This paper proposes an approach for people counting that is based on the use of cameras and Raspberry Pi platforms, together with an edge-based transfer learning framework that is enriched with specific image processing strategies, with the aim of this approach being adopted in different indoor environments without the need for tailored training phases. The system was deployed on a university campus, which was chosen as the case study. The proposed system was able to work in classrooms with different characteristics. This paper reports a proposed architecture that could make the system scalable and privacy compliant and the evaluation tests that were conducted in different types of classrooms, which demonstrate the feasibility of this approach. Overall, the system was able to count the number of people in classrooms with a maximum mean absolute error of 1.23.

Designing Interfaces to Display Sensor Data: A Case Study in the Human-Building Interaction Field Targeting a University Community.

The increase of smart buildings with Building Information Modeling (BIM) and Building Management Systems (BMS) has created a large amount of data, including those coming from sensors. These data are intended for monitoring the building conditions by authorized personnel, not being available to all building occupants. In this paper, we evaluate, from a qualitative point of view, if a user interface designed for a specific community can increase occupants' context-awareness about environmental issues within a building, supporting them to make more informed decisions that best suit their needs. We designed a user interface addressed to the student community of a smart campus, adopting an Iterative Design Cycle methodology, and engaged 48 students by means of structured interviews with the aim of collecting their feedback and conducting a qualitative analysis. The results obtained show the interest of this community in having access to information about the environmental data within smart campus buildings. For example, students were more interested in data about temperature and brightness, rather than humidity. As a further result of this study, we have extrapolated a series of design recommendations to support the creation of map-based user interfaces that we found to be effective in such contexts.

On Supporting University Communities in Indoor Wayfinding: An Inclusive Design Approach.

Mobility can be defined as the ability of people to move, live and interact with the space. In this context, indoor mobility, in terms of indoor localization and wayfinding, is a relevant topic due to the challenges it presents, in comparison with outdoor mobility, where GPS is hardly exploited. Knowing how to move in an indoor environment can be crucial for people with disabilities, and in particular for blind users, but it can provide several advantages also to any person who is moving in an unfamiliar place. Following this line of thought, we employed an inclusive by design approach to implement and deploy a system that comprises an Internet of Things infrastructure and an accessible mobile application to provide wayfinding functions, targeting the University community. As a real word case study, we considered the University of Bologna, designing a system able to be deployed in buildings with different configurations and settings, considering also historical buildings. The final system has been evaluated in three different scenarios, considering three different target audiences (18 users in total): i. students with disabilities (i.e., visual and mobility impairments); ii. campus students; and iii. visitors and tourists. Results reveal that all the participants enjoyed the provided functions and the indoor localization strategy was fine enough to provide a good wayfinding experience.

An alternative approach to dimension reduction for pareto distributed data: a case study.

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.

RaveGuard: A Noise Monitoring Platform Using Low-End Microphones and Machine Learning.

Urban noise is one of the most serious and underestimated environmental problems. According to the World Health Organization, noise pollution from traffic and other human activities, negatively impact the population health and life quality. Monitoring noise usually requires the use of professional and expensive instruments, called phonometers, able to accurately measure sound pressure levels. In many cases, phonometers are human-operated; therefore, periodic fine-granularity city-wide measurements are expensive. Recent advances in the Internet of Things (IoT) offer a window of opportunities for low-cost autonomous sound pressure meters. Such devices and platforms could enable fine time-space noise measurements throughout a city. Unfortunately, low-cost sound pressure sensors are inaccurate when compared with phonometers, experiencing a high variability in the measurements. In this paper, we present RaveGuard, an unmanned noise monitoring platform that exploits artificial intelligence strategies to improve the accuracy of low-cost devices. RaveGuard was initially deployed together with a professional phonometer for over two months in downtown Bologna, Italy, with the aim of collecting a large amount of precise noise pollution samples. The resulting datasets have been instrumental in designing InspectNoise, a library that can be exploited by IoT platforms, without the need of expensive phonometers, but obtaining a similar precision. In particular, we have applied supervised learning algorithms (adequately trained with our datasets) to reduce the accuracy gap between the professional phonometer and an IoT platform equipped with low-end devices and sensors. Results show that RaveGuard, combined with the InspectNoise library, achieves a 2.24% relative error compared to professional instruments, thus enabling low-cost unmanned city-wide noise monitoring.

Thyroid autoimmunity and type 1 diabetes in children and adolescents: screening data from Juvenile Diabetes Tuscany Regional Centre.

BACKGROUND AND AIMS: The incidence of autoimmune thyroiditis in patients with type 1 diabetes mellitus (T1DM) is higher than in healthy population. The aim of this study is to investigate epidemiology and natural history of thyroid autoimmunity (AIT), thyroiditis diagnosis and need for therapy in paediatric patients with T1DM and to find the most suitable timing of AIT screening. METHODS: T1DM patients (493 pts., 268 males and 225 females) treated in the Juvenile Diabetes Tuscany Regional Centre at Meyer's Children Hospital were enrolled to determine TSH, fT4, thyroid autoantibodies levels and to undergo thyroid ultrasound. Anamnestic data about T1DM onset, AIT onset, time frame between T1DM and AIT onsets and the relationship between AIT and celiac disease (CD) were studied. RESULTS: In the screened population 11.7% of patients presented with increased thyroid autoantibodies, and 63.6% of them showed positive thyroid ultrasound. AIT was significantly more frequent in females compared to males (p < 0.01). The mean age at AIT onset was 11.17 +/- 3.29 years (range 4.99-20, 30) and AIT onset before 12 yrs. of age was found in 54.5% of cases; 18.4% patients (all females) presented CD. The mean time between T1DM and AIT onset was 2.46 +/- 3.41 years (range 0-13, 41). The subgroup with increased thyroid autoantibodies was not statistically different from the whole population with regard to TDM1 duration and mean age at onset. CONCLUSIONS: AIT is frequently associated with T1DM (11.7%) regardless of age and duration of diabetes. We suggest a yearly screening for AIT after TDM1 onset, at every age.