Higher Institute on Territorial Systems for Innovation
PORTO@iris (Publications Open Repository TOrino - Politecnico di Torino)Not a member yet
146173 research outputs found
Sort by
A multidimensional framework for modelling and assessing cognitive assistance systems for industry 5.0
No full textIndustry 5.0 calls for human‐centric, resilient, and sustainability‐driven production ecosystems where people and intelligent technologies collaborate to create value. Cognitive Assistance Systems (CAS), intended to support operators by guiding their actions and decisions, are central to this vision. Yet, their design and evaluation remain fragmented. This paper introduces a multidimensional framework that maps CAS onto five analysis dimensions: Cognitive Functions, Autonomy, Adaptivity, Interaction Modality, and Portability. Each dimension is operationalised through explicit rating scales, enabling profiling, benchmarking, and design trade‐off analysis across heterogeneous solutions. An experiment in an assembly process illustrates the framework’s practical application and provides a test of its ability to discriminate among different CAS configurations
TIDE: Task-Driven DNN Training and Splitting for Efficient Inference at the Mobile Edge
Full text linkThe growing demands of DNN-based inference at the mobile edge is driving the need for increasingly efficient execution. Such applications often require fast and high-quality outputs, which are hard to realize due to the limited computa- tional and communication capabilities at the edge. This paper tackles these issues focusing on a DNN for the execution of tasks that are homogeneous in nature but heterogeneous in their domains. The key idea is to start with a parent DNN of interconnected computational elements (atoms), and strategically form a collection of task-specific DNNs suitable for distributed deployment. Such task-specific DNNs may include common as well as uniquely used atoms of the parent DNN. Ultimately, the aim is that they be smaller in size – thus a better match for edge resources – and achieve low-cost inference. We solve the problem of determining the best collection of task-specific DNNs through an algorithmic framework named TIDE. Experimental results show that TIDE decreases inference cost and time by 90% and 80% (resp.) relatively to centralized approaches, and by over 60% and 70% (resp.) when compared to the best benchmark
The Urban Heat Island Under Climate Change: Analysis of Representative Urban Blocks in Northwestern Italy
Full text linkUrban populations are exposed to elevated local temperatures compared to surrounding rural areas due to the urban heat island (UHI) effect, which increases health risks and energy demand. The literature highlights that accurately quantifying UHIs at broader territorial scales remains challenging because of limited microscale climate data availability and, at the same time, the difficulty of increasing the spatial coverage of the outcomes. Within the PRIN2022-PNRR CRiStAll (Climate Resilient Strategies by Archetype-based Urban Energy Modeling) project, this work addresses these limitations by coupling Urban Building Energy Modeling with archetype-based representation of urban form and high-resolution climatic data. Urban archetypes are defined as representative microscale configurations derived from combinations of urban canyon geometries and building typologies, accounting for different climatic zones, use categories, and construction periods. The proposed methodology was applied to the city of Turin (Italy), where representative urban blocks were identified and modeled to evaluate key urban context metrics under short-, medium-, and long-term climate scenarios. The UHI effect was assessed using Urban Weather Generator, while energy simulations were performed with CitySim. The urban archetype approach enables both fine spatial resolution and extensive spatial coverage, supporting urban-scale mapping
Supporting, Not Solving: Human-Centered AI Systems in Education
Full text linkEducational technologies have evolved significantly over recent decades, with AI representing the latest frontier in this progression. Current applications range from adaptive learning platforms that personalize content delivery to automated systems that provide immediate feedback. The challenge lies in developing AI educational technologies that enhance human capabilities while respecting the autonomy and agency of learners and educators. My research aims to design, implement, and evaluate human-centered AI systems in educational contexts. By prioritizing human needs and values throughout the development process, my work seeks to advance the understanding of how AI can enhance educational practices while preserving the learner’s independence and the educator’s role
A variational analysis of nematic axisymmetric films: the covariant derivative case
No full textNematic surfaces are thin fluid structures, ideally two-dimensional, endowed with an in-plane
nematic order. In 2012, two variational models have been introduced by Giomi [5] and by
Napoli and Vergori [11, 12]. Both penalize the area of the surface and the gradient of the
director: in [5] the covariant derivative of the director is considered, while [11] deals with
the surface gradient. In this paper, a complete variational analysis of the model proposed by
Giomi is performed for revolution surfaces spanning two coaxial rings
La rigenerazione del patrimonio di edilizia scolastica: abilitare la trasformazione a partire dagli spazi
No full textItaly’s school building stock constitutes a heterogeneous and stratified spatial infrastructure that must be rethought systemically in response to evolving educational, demographic, energy, and health-related needs. This paper proposes a methodological approach aimed at strengthening the capacity of local communities and institutions to define priorities and strategies for school regeneration through shared spatial knowledge. Within a political and financial context that enables large-scale public investment, the study outlines the theoretical framework and development of a digital support tool—a web-based application—designed to assist decision-makers and school communities in identifying spatial resources and transformation potential. The approach is grounded in the concept of spatial agency and employs a typological classification based on architectural form, spatial organization, and contextual relationships. By treating school space as a shared documentary base, the proposed method supports multiscalar governance processes and fosters more informed, inclusive, and adaptable strategies for regenerating educational infrastructure
Anomaly detection and localization with state-of-the-art deep learning models to support quality inspection in car manufacturing
No full textMaintaining high quality in automotive manufacturing is essential, as even small defects can lead to safety issues, costly recalls, and increased operational costs. Manual inspection is often unreliable in fast-paced production, limited by human error and poor scalability. Advanced imaging and deep learning-based Anomaly Detection and Localization (ADL) offer effective alternatives, but their use in industry is challenged by factors like complex geometries, inconsistent lighting, and environmental noise.
This work presents an ADL framework for inspecting sealant application in car underbodies that combines a video acquisition system with four state-of-the-art deep learning models. To overcome the lack of annotated data, a synthetic defect generation module is introduced, creating realistic anomalies that improve model evaluation while reducing annotation effort.
The framework was tested on both synthetic and real-world data, achieving high localization performance (AUROC up to 99.7%, F1-score of 43.4%) with inference times ranging from 0.08 to 3.33 seconds depending on model complexity. These results highlight the trade-offs between speed and accuracy, and confirm the potential of ADL models for real-time quality control in industrial automotive settings
On Cost-Effectiveness of Language Models for Time Series Anomaly Detection
Full text linkDetecting anomalies in time series data is crucial across several domains, including healthcare, finance, and automotive. Large Language Models (LLMs) have recently shown promising results by leveraging robust model pretraining. However, fine-tuning LLMs with several billion parameters requires a large number of training samples and significant training costs. Conversely, LLMs under a zero-shot learning setting require lower overall computational costs, but can fall short in handling complex anomalies. In this paper, we explore the use of lightweight language models for Time Series Anomaly Detection, either zero-shot or via fine-tuning them. Specifically, we leverage lightweight models that were originally designed for time series forecasting, benchmarking them for anomaly detection against both open-source and proprietary LLMs across different datasets. Our experiments demonstrate that lightweight models (70 Billions)
Physics-informed machine learning for the structural health monitoring and early warning of a long highway viaduct with displacement transducers
Full text linkIn Bridge Structural Health Monitoring (SHM), damage detection is often hindered by environmental and operational variability. Confounding influences such as traffic, wind, and especially temperature can significantly affect measurements, making it difficult to distinguish true damage-related anomalies. This challenge is critical in static monitoring of long steel viaducts, where thermal effects dominate displacements, making small damage-induced perturbations difficult to detect. To address this, the study introduces a Physics-Informed Machine Learning (PIML) model that establishes a reliable baseline for the ‘normal conditions’ of the infrastructure. This baseline isolates anomalies attributable to structural damage while accounting for temperature effects. The proposed grey-box approach combines data-driven modelling with physical knowledge of thermal behaviour, enhancing both accuracy and interpretability. A real-world application is presented on a long-span highway viaduct, where longitudinal displacements are monitored using temperature and displacement sensors. By using only temperature and time as inputs, the model captures nonlinear daily and seasonal thermal cycles without additional instrumentation. To assess reliability, an Early Warning System (EWS) is developed based on displacement anomaly thresholds and Kernel Density Estimation (KDE). The PIML model is evaluated against black-box (purely data-driven) and white-box (purely physics-based) alternatives. Damage scenarios are simulated by introducing anomalies into experimental data to test each model’s capability to detect abnormal behaviour while filtering out environmental effects. Results show that the grey-box PIML consistently outperforms black- and white-box models in accuracy, robustness, and anomaly discrimination. These findings demonstrate the potential of PIML to advance SHM practices and enable reliable automated EWSs for bridge monitoring
2D multiphysics model for proton conductor ceramic technology investigating the effects of temperature, composition of reactants and electrolyte material
Full text linkThe main goal of this work is the development of a model for a protonic ceramic electrolysis cell (PCEC), a
technology emerging as a promising alternative to traditional solid oxide cells (SOCs). A two-dimensional numerical model has been developed for the simulation of a planar cell (5 × 5 cm2
). The model is set up as the
combination of a thermal-fluid-dynamic model and an electro-chemical one, also comprehending the transport of
three defects (protons, electron holes and oxygen vacancies) through the membrane, typical for barium-based
zirconate. Firstly, a state-of-the-art BaZr0.8Y0.2O3− δ (BZY) electrolyte material and cell structure (electrode
supported Ni-BZY/BZY/SFM-BZY) with available experimental performance data have been selected from
literature. Furthermore, the parameters required to model a novel material BaCe0.65Zr0.2Y0.15O3− δ (BCZY), still
not available in literature, have been introduced. The purpose of this study is to quantify the effect of different
operating conditions and modeling assumptions on the hydrogen production performance for the state of the art
BZY electrolyte, also establishing the effect of the electronic leakage on the transport number and faradaic efficiency. Finally, this material is compared to the newly introduced BCZY to assess their respective advantages
and disadvantages