Skip to main content

SERVICIOS INTELIGENTES COORDINADOS PARA AREAS INTELIGENTES ADAPTATIVAS

PID2021-123673OB-C31

•
Funding agency name Agencia Estatal de Investigación
Funding agency acronym AEI
Program Programa Estatal para Impulsar la Investigación Científico-Técnica y su Transferencia
Subprogram Subprograma Estatal de Generación de Conocimiento
Call Proyectos de I+D+i (Generación de Conocimiento y Retos Investigación)
Call year 2021
State Plan 2021-2023 Plan Estatal de Investigación Científica y Técnica y de Innovación
Beneficiary centre UNIVERSITAT POLITÈCNICA DE VALÈNCIA

Resultados relacionados

Resultados totales (Incluyendo duplicados): 30
Encontrada(s) 1 página(s)

RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Extending the Framework for Developing Intelligent Virtual Environments (FIVE) with Artifacts for Modeling Internet of Things Devices and a New Decentralized Federated Learning Based on Consensus for Dynamic Networks

  • Rebollo Pedruelo, Miguel|||0000-0002-5115-8751
  • Hérnandez López, Luís|||0000-0001-9011-0542
  • Enguix-Andrés, Francisco|||0009-0002-8144-318X
  • Carrascosa Casamayor, Carlos|||0000-0003-3649-6530
  • Rincon, Jaime Andres
[EN] One of the main lines of research in distributed learning in recent years is the one related to Federated Learning (FL). In this work, a decentralized Federated Learning algorithm based on consensus (CoL) is applied to Wireless Ad-hoc Networks (WANETs), where the agents communicate with other agents to share their learning model as they are available to the wireless connection range. When deploying a set of agents, it is essential to study whether all the WANET agents will be reachable before the deployment. The paper proposes to explore it by generating a simulation close to the real world using a framework (FIVE) that allows the easy development and modification of simulations based on Unity and SPADE agents. A fruit orchard with autonomous tractors is presented as a case study. The paper also presents how and why the concept of artifact has been included in the above-mentioned framework as a way to highlight the importance of some devices used in the environment that have to be located in specific places to ensure the full connection of the system. This inclusion is the first step to allow Digital Twins to be modeled with this framework, now allowing a Digital Shadow of those devices., This work has been developed thanks to the funding of projects: Grant PID2021-123673OBC31 funded by MCIN/AEI/ 10.13039/ 501100011033 and by ERDF A way of making Europe , PROMETEO CIPROM/ 2021/077, TED2021-131295B-C32 and Ayudas del Vicerrectorado de Investigacion de la UPV (PAID-PD-22).
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

A crossover operator for objective functions defined over graph neighborhoods with interdependent and related variables

  • Jordán, Jaume|||0000-0003-0400-9136
  • Palanca Cámara, Javier|||0000-0002-6209-9603
  • Sanchez-Anguix, Víctor|||0000-0003-4851-0037
  • Julian, Vicente|||0000-0002-2743-6037
[EN] This article presents a new crossover operator for problems with an underlying graph structure where edges point to prospective interdependence relationships between decision variables and neighborhoods shape the definition of the global objective function via a sum of different expressions, one for each neighborhood. The main goal of this work is to propose a crossover operator that is broadly applicable, adaptable, and effective across a wide range of problem settings characterized by objective functions that are expressed in terms of graph neighbourhoods with interdependent and related variables. Extensive experimentation has been conducted to compare and evaluate the proposed crossover operator with both classic and specialized crossover operators. More specifically, the crossover operators have been tested under a variety of graph types, which model how variables are involved in interdependencies, different types of expressions in which interdependent variables are combined, and different numbers of decision variables. The results suggest that the new crossover operator is statistically better or at least as good as the best-performing crossover in 75% of the families of problems tested., This work is partially funded by the Spanish Ministry of Science and Innovation under the project PID2021-123673OBC31 and PID2021-124975OB-I00, project TED2021-131295B-C32 from the State Research Agency, and DIGITAL2022 CLOUDAI02/ S8760000 from the European Commission. Jaume Jordán is supported by the IJC2020-045683-I grant funded by MCIN/AEI/10.13039/5011 00011033 and by European Union NextGenerationEU/PRTR . Additional support is provided by project INNVA1/2024/91 funded by Agencia Valenciana de la Innovación.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

AI for Sustainable Recycling: Efficient Model Optimization for Waste Classification Systems

  • Chacón-Albero, Oriol de la Pau
  • Campos-Mocholí, Mario|||0009-0009-4721-2207
  • Julian, Vicente|||0000-0002-2743-6037
  • Botti V.|||0000-0002-6507-2756
  • Marco-Detchart, Cédric
  • Rincon, Jaime Andres
[EN] The increasing volume of global waste presents a critical environmental and societal challenge, demanding innovative solutions to support sustainable practices such as recycling. Advances in Computer Vision (CV) have enabled automated waste recognition systems that guide users in correctly sorting their waste, with state-of-the-art architectures achieving high accuracy. More recently, attention has shifted toward lightweight and efficient models suitable for mobile and edge deployment. These systems process data from integrated camera sensors in Internet of Things (IoT) devices, operating in real time to classify waste at the point of disposal, whether embedded in smart bins, mobile applications, or assistive tools for household use. In this work, we extend our previous research by improving both dataset diversity and model efficiency. We introduce an expanded dataset that includes an organic waste class and more heterogeneous images, and evaluate a range of quantized CNN models to reduce inference time and resource usage. Additionally, we explore ensemble strategies using aggregation functions to boost classification performance, and validate selected models on real embedded hardware and under simulated lighting variations. Our results support the development of robust, real-time recycling assistants for resource-constrained devices. We also propose architectural deployment scenarios for smart containers, and cloud-assisted solutions. By improving waste sorting accuracy, these systems can help reduce landfill use, support citizen engagement through real-time feedback, increase material recovery, support data-informed environmental decision making, and ease operational challenges for recycling facilities caused by misclassified materials. Ultimately, this contributes to circular economy objectives and advances the broader field of environmental intelligence., This work was partially supported with grant PID2021-123673OB-C31 funded by MCIN/AEI/10.13039/501100011033 and by "ERDF A way of making Europe" and grant CIPROM/2021/077 from the Conselleria de Innovacio, Universitats, Ciencias i Societat Digital Generalitat Valenciana-PROMETEO.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Optimization of Rural Demand-Responsive Transportation through Transfer Point Allocation

  • Martí, Pasqual|||0000-0001-6406-0641
  • Jordán, Jaume|||0000-0003-0400-9136
  • Julian, Vicente|||0000-0002-2743-6037
  • de la Prieta, Fernando
[EN] Rural mobility has a lack of innovative proposals in contrast with its urban counterpart. This research aims to bring solutions that ease the implementation of reliable and flexible rural transportation. Demand-responsive transportation is chosen to develop a public transportation service providing interurban trips among several rural settlements. Given the characteristics of rural displacement demand, a novel approach is introduced to optimize the service's economic costs: the dynamic transfer point allocation. The problem is fully formulated and an architecture is introduced describing the workflow of the whole system. Data from an interurban bus transportation service are used to build a case study of a rural area of Valencia, Spain, and develop several examples illustrating the benefits of the proposed approach. The results reveal that the dynamic creation of transfer points can simplify the transportation fleet's itineraries and boost the amount of served travel requests. Finally, a discussion of the benefits and dangers of flexible features in rural transportation is developed, underscoring the need to achieve a balance between dynamic operation and service quality., This work was partially supported with grant PID2021-123673OB-C31 funded by MCIN/AEI/10.13039/501100011033 and by ERDF A way of making Europe . Pasqual Martí is supported by grant ACIF/2021/259 funded by the Conselleria de Innovación, Universidades, Ciencia y Sociedad Digital de la Generalitat Valenciana . Jaume Jordán is supported by grant IJC2020-045683-I funded by MCIN/AEI/10.13039/501100011033 and by European Union NextGenerationEU/PRTR
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Bus Ridership Prediction and Scenario Analysis through ML and Multi-Agent Simulations

  • Martí, Pasqual|||0000-0001-6406-0641
  • Julian, Vicente|||0000-0002-2743-6037
  • Jordán, Jaume|||0000-0003-0400-9136
  • Ibáñez-Peña, Alejandro
  • Novais, Paulo
[EN] This paper introduces an innovative approach to predicting bus ridership andanalysing transportation scenarios through a fusion of machine learning (ML) techniques and multi-agent simulations. Utilising a comprehensive dataset from an urban bus system, we employ ML models to accurately forecast passenger flows, factoring in diverse variables such as weather conditions. The novelty of our method lies in the application of these predictions to generate detailed simulation scenarios, which are meticulously executed to evaluate the efficacy of public transportation services. Our research uniquely demonstrates the synergy between ML predictions and agent-based simulations, offering a robust tool for optimising urban mobility. The results reveal critical insights into resource allocation, service efficiency, and potential improvements in public transport systems. This study significantly advances the field by providing a practical framework for transportation providers to optimise services and address longterm challenges in urban mobility., This work is partially supported by grants PID2021-123673 and PDC2022-133161-C32 funded by MCIN/AEI/ 10.13039/501100011033 and by «ERDF A way of making Europe». Pasqual Martí is supported by grant ACIF/2021/259 funded by the «Conselleria de Innovación, Universidades, Ciencia y Sociedad Digital de la Generalitat Valenciana». Jaume Jordán is supported by grant IJC2020-045683-I funded by MCIN/AEI/ 10.13039/501100011033 and by «European Union NextGenerationEU/PRTR».
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Cognitive Agents in Urban Mobility: Integrating LLM Reasoning into Multi-Agent Simulations

  • Calderón-Orellana, Christián Vladimir|||0009-0004-9199-6911
  • Martí, Pasqual|||0000-0001-6406-0641
  • Jordán, Jaume|||0000-0003-0400-9136
  • Palanca Cámara, Javier|||0000-0002-6209-9603
  • Julian, Vicente|||0000-0002-2743-6037
[EN] Urban mobility systems face escalating challenges associated with sustainability, equity, and resilience, further compounded by environmental pressures. Traditional agent-based models (ABMs) often fail to capture cognitively rich, adaptive behaviors, limiting their ability to simulate realistic user responses to disruptions. In this work, we propose a cognitive agent architecture based on Large Language Models (LLMs), featuring multi-horizon memory-driven planning, reflection, and adaptation. Integrated into the SimFleet agent-based simulator with realistic sociodemographic profiles, the agents dynamically generate, adjust, and reflect upon travel plans across a 20-day simulation involving over 320 individuals. Experimental results reveal emergent adaptation patterns under both stable and disrupted transport conditions, and an ablation study under severe service disruption quantifies the contributions of short-term and long-term memory modules to memory-driven reasoning, demonstrating the potential of LLM-driven agents to enhance the realism, flexibility, and interpretability of urban mobility simulations., This work was partially supported by the Spanish Government through grant PID2021-123673OB-C31 (MCIN/AEI/10.13039/501100011033) and the European Regional Development Fund (ERDF, "A way of making Europe"). Additional support was provided by grant TSI-100930-2023-9 from the "Secretaria de Estado de Digitalizacion e Inteligencia Artificial, Catedra ENIA-UPV", and by the PAID-01-22 programme for the recruitment of doctors within the Spanish Science and Technology System at the Universitat Politecnica de Valencia.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

A New Methodological Framework for Project Design to Analyse and Prevent Students from Dropping Out of Higher Education

  • Flores, Vaneza
  • Heras, Stella|||0000-0001-6212-9377
  • Julian, Vicente|||0000-0002-2743-6037
[EN] The problem of university dropout is a recurring issue in universities that affects students, especially in the first year of studies. The situation is aggravated by the COVID-19 pandemic, which has imposed a virtual education, generating a greater amount of data in addition to historical information, and thus, a greater demand for strategies to design projects based on Educational Data Mining (EDM). To deal with this situation, we present a framework for designing EDM projects based on the construction of a problem tree. The result is the proposal of a framework that merges the six phases of the CRISP-DM methodology with the first stage of the Logical Framework Methodology (LFM) to increase university retention. To illustrate this framework, we have considered the design of a project based on data mining to prevent students from dropping out of a Peruvian university., This work is partially supported by grant PID2021-123673OB-C31 funded by MCIN/AEI/10.13039/501100011033 and by "ERDF A way of making Europe". The research was developed thanks to the support of the National University of Moquegua, which provided the information for the creation of the dataset.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Robust Multi-Sensor Consensus Plant Disease Detection Using the Choquet Integral

  • Marco-Detchart, Cedric
  • Rincón-Arango, Jaime Andrés
  • Carrascosa Casamayor, Carlos|||0000-0003-3649-6530
  • Julian, Vicente|||0000-0002-2743-6037
[EN] Over the last few years, several works have appeared that employ Artificial Intelligence (AI) techniques to improve sustainable development in the agricultural sector. Specifically, these intelligent techniques provide mechanisms and procedures to facilitate decision-making in the agri-food industry. One of the application areas has been the automatic detection of plant diseases. These techniques, mainly based on deep learning models, allow analysing and classifying plants to determine possible diseases facilitating early detection and thus preventing the propagation of the disease. In this way, this paper proposes an Edge-AI device that incorporates the necessary hardware and software components for automatically detecting plant diseases from a set of images of the plant leaf. In this way, the main goal of this work is to design an autonomous device that allows the detection of possible diseases. that can detect potential diseases in plants. This will be achieved by capturing multiple images of the leaves and implementing data fusion techniques to enhance the classification process and improve its robustness. Several tests have been carried out to determine that the use of this device significantly increases the robustness of the classification responses to possible plant diseases., This work was partially supported by grant number PID2021-123673OB-C31 funded by MCIN/AEI/ 10.13039/501100011033 and by ERDF A way of making Europe and Consellería d Innovació, Universitats, Ciencia i Societat Digital from Comunitat Valenciana (APOSTD/2021/227) through the European Social Fund (Investing In Your Future).
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

An evolutionary metaheuristic for forming teams in the classroom with constraints

  • Candel, Gonzalo
  • Sanchez-Anguix, Víctor|||0000-0003-4851-0037
  • Alberola Oltra, Juan Miguel|||0000-0002-5486-5638
  • Julian, Vicente|||0000-0002-2743-6037
  • Botti V.|||0000-0002-6507-2756
[EN] Team formation is essential for developing teamwork-related skills in educational settings. The problem of team formation in the classroom consists of partitioning a classroom into non-overlapping teams of students, including every single student. Several algorithms have been proposed to automate the formation of teams, each employing different criteria for guiding the team formation process. Traditionally, metaheuristics have been a common approach due to the combinatorial complexity of the problem. This paper introduces a novel and general evolutionary algorithm for team formation in the classroom guided by mutation, the general concept of synergy between team members, and local search. Our algorithm allows for flexible team size constraints and the inclusion of compulsory and forbidden student combinations, which are not considered in existing methods but are important for capturing human relationships in the classroom. In addition, our algorithm is independent of the specific objective function employed to evaluate the teams formed. We present experiments comparing our proposal with other state-of-the-art algorithms, demonstrating robust performance across different objective functions employed in the team formation literature, superior scalability as the problem size increases, and remarkable performance in settings with or without the aforementioned constraints., This work was partially supported by MINECO/FEDER RTI2018-095390-B-C31 project of the Spanish government, project TED2021-131295B-C32 from the State Research Agency, and DIGITAL2022 CLOUDAI02/S8760000 from the European Commission, . Some of the authors are partially supported by the Spanish Ministry of Science and Innovation under the project PID2021-123673OB-C31 COSASS and project OPRES-Realistic Optimization in Problems in Public Health (No. PID2021-124975OB-I00), partially financed with FEDER funds. Additional support is provided by project INNVA1/2024/91 funded by Agencia Valenciana de la Innovación.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Application of Machine Vision Techniques in Low-Cost Devices to Improve Efficiency in Precision Farming

  • Jaramillo-Hernández, Juan Felipe
  • Marco-Detchart, Cédric
  • Rincón, Jaime Andrés
  • Julian, Vicente|||0000-0002-2743-6037
[EN] In the context of recent technological advancements driven by distributed work and open-source resources, computer vision stands out as an innovative force, transforming how machines interact with and comprehend the visual world around us. This work conceives, designs, implements, and operates a computer vision and artificial intelligence method for object detection with integrated depth estimation. With applications ranging from autonomous fruit-harvesting systems to phenotyping tasks, the proposed Depth Object Detector (DOD) is trained and evaluated using the Microsoft Common Objects in Context dataset and the MinneApple dataset for object and fruit detection, respectively. The DOD is benchmarked against current state-of-the-art models. The results demonstrate the proposed method's efficiency for operation on embedded systems, with a favorable balance between accuracy and speed, making it well suited for real-time applications on edge devices in the context of the Internet of things., This work was partially supported with grant PID2021-123673OB-C31, TED2021-131295BC32 funded by MCIN/AEI/10.13039/501100011033 and by ERDF A way of making Europe , PROMETEO grant CIPROM/2021/077 from the Conselleria de Innovación, Universidades, Ciencia y Sociedad Digital Generalitat Valenciana and Early Research Project grant PAID-06-23 by the Vice Rectorate Office for Research from Universitat Politècnica de València (UPV).
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Asynchronous consensus for multi-agent systems and its application to Federated Learning

  • Carrascosa Casamayor, Carlos|||0000-0003-3649-6530
  • Picó-Pascual, Aarón|||0000-0002-5612-8033
  • Rebollo Pedruelo, Miguel|||0000-0002-5115-8751
  • Matagne, Miro-Manuel
  • Rincón-Arango, Jaime Andrés
[EN] Federated Learning (FL) improves the performance of the training phase of machine learning procedures by distributing the model training to a set of clients and recombining the final models in a server. All clients share the same model, each with a subset of the complete dataset, addressing size issues or privacy concerns. However, having a central server generates a bottleneck and weakens the failure tolerance in truly distributed environments. This work follows the line of applying consensus for FL as a no-centralized approach. Moreover, the paper presents a fully distributed consensus in MAS (multi-agent system) modeling and a new asynchronous consensus in MAS (multi-agent system). The paper also includes some descriptions and tests for implementing such learning algorithms in an actual agent platform, along with simulation results obtained in a case study about electrical production in Australian wind farms., This work has been developed thanks to the funding of projects: Grant PID2021-123673OB-C31 funded by MCIN/AEI/ 10.13039/ 501100011033 and by ERDF A way of making Europe PROMETEO CIPROM/2021/077 TED2021-131295B-C32 Grant of the Vicerrectorado de Investigacion from UPV (PAID-PD22)
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Interurban charging station network: An evolutionary approach

  • Jordán, Jaume|||0000-0003-0400-9136
  • Martí, Pasqual|||0000-0001-6406-0641
  • Palanca Cámara, Javier|||0000-0002-6209-9603
  • Julian, Vicente|||0000-0002-2743-6037
  • Botti V.|||0000-0002-6507-2756
[EN] In recent years, there has been a strong desire to meet the challenge of electrification of vehicles in order to achieve the decarbonization objective. However, as sales of electric vehicles have increased, there is a significant lack of infrastructure to support the charging of this type of vehicle. The infrastructural deficiencies are even more evident in the interurban environment, where the autonomy in kilometers of the battery is a critical issue. To minimize the substantial economic costs involved in installing sufficient charging points to ensure any interurban journey, it is necessary to establish mechanisms that evaluate appropriate locations to deploy the necessary stations. Accordingly, this paper proposes using an evolutionary approach to calculate the most suitable locations in an interurban environment for electric charging stations. For this purpose, different input information is taken into account in the allocation process. The proposed algorithm has been tested using real data from the USA. The results assess the current infrastructure and show the advantages of the locations proposed by the algorithm., This work is partially supported by grant PID2021-123673OB- C31 funded by MCIN/AEI/10.13039/501100011033 and by "ERDF A way of making Europe". Jaume Jordán is supported by grant IJC2020-045683-I funded by MCIN/AEI/10.13039/501100011033 and by "European Union NextGenerationEU/PRTR". Pasqual Marti is supported by grant ACIF/2021/259 funded by the "Conselleria de Innovacion, Universidades, Ciencia y Sociedad Digital de la Generalitat Valenciana".
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Multi-class weed detection based on deep learning using unmanned aerial vehicle images in potato fields

  • Sandoval-Pillajo, Ana Lucia
  • Perugachi-Chavez, Luis
  • Pusda-Chulde, Marco
  • Garcia-Santillan, Ivan
  • Giret Boggino, Adriana Susana|||0000-0002-2311-0785
[EN] Early weed detection is a critical task in precision agriculture, as weeds compete with crops for resources and significantly reduce yield and quality. This study proposes a multi-class weed detection approach based on deep learning using unmanned aerial vehicle (UAV) images acquired in real potato fields under uncontrolled environmental conditions. A custom dataset containing five classes (potato crop and four weed categories) was created, annotated, and made publicly available to support reproducibility and future comparisons. Two modern object detection architectures, based on convolutional neural networks and Transformer models, were evaluated through multiple training variants to analyze the impact of class balancing, image resolution, and hyperparameter optimization on detection performance. The models were assessed using standard object detection metrics and complemented with statistical validation to verify the reliability of the predictions in real-field conditions. Experimental results show that both architectures achieve robust and comparable performance, demonstrating their suitability for near real-time weed detection and precision agriculture applications. The findings confirm the effectiveness of combining advanced deep learning models with UAV imagery for accurate multi-class weed identification in complex agricultural environments., This research was mostly funded by the Universidad Tecnica del Norte, Ecuador, under Grant Number: InvestigaUTN-2024-1353.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Evaluation of deep learning techniques for plant disease detection

  • Marco-Detchart, Cédric
  • Rincón-Arango, Jaime Andrés
  • Carrascosa Casamayor, Carlos|||0000-0003-3649-6530
  • Julian, Vicente|||0000-0002-2743-6037
[EN] In recent years, several proposals have been based on Artificial Intelligence techniques for automatically detecting the presence of pests and diseases in crops from images usually taken with a camera. By training with pictures of affected crops and healthy crops, artificial intelligence techniques learn to distinguish one from the other. Furthermore, in the long term, it is intended that the tools developed from such approaches will allow the automation and increased frequency of plant analysis, thus increasing the possibility of determining and predicting crop health and potential biotic risks. However, the great diversity of proposed solutions leads us to the need to study them, present possible situations for their improvement, such as image preprocessing, and analyse the robustness of the proposals examined against more realistic pictures than those existing in the datasets typically used. Taking all this into account, this paper embarks on a comprehensive exploration of various AI techniques leveraging leaf images for the autonomous detection of plant diseases. By fostering a deeper understanding of the strengths and limitations of these methodologies, this research contributes to the vanguard of agricultural disease detection, propelling innovation, and fostering the maturation of AI-driven solutions in this critical domain., The authors gratefully acknowledge the financial support of Conselleria d'Innovacio,Universitats,Cienciai Societat Digital from Comunitat Valenciana (APOSTD/2021/227),the European Social Fund (Investing In Your Future), grant PID2021-123673OB-C31 funded by MCIN/AEI/ 10.13039/501100011033 and by "ERDF A way of making Europe" and grant from the Research Services of Universitat Politecnica de Valencia (PAID-PD-22).
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Semantic and modular orchestration of AI-driven digital twins for industrial interoperability and optimization

  • Juárez-Juárez, María Gabriela|||0000-0002-0982-9021
  • Giret Boggino, Adriana Susana|||0000-0002-2311-0785
  • Botti V.|||0000-0002-6507-2756
[EN] Digital Twins (DTs) are foundational in smart manufacturing, supporting data-driven monitoring and optimization. Yet, many implementations remain monolithic, limiting interoperability and reusability. This paper introduces a semantic and modular architecture for orchestrating AI-driven DTs, designed to enable scalable integration and standardized coordination across industrial systems. The system employs a semantic API aligned with NGSI-LD, to expose industrial entities such as processes, anomalies, assets, and contextual KPIs (e.g., energy usage, CO2 emissions, tool wear, product quality). AI techniques ranging from threshold adjustment to symbolic learning are encapsulated as modular agents, each performing targeted optimization tasks. These agents operate over the semantic API, which ensures consistent, interpretable interactions across modules. A Manager and a Recommender agent are defined to coordinate execution; while not yet deployed at runtime, their logic is implemented through semantic interfaces that support traceable, modular activation. The system is validated using synthetic data simulating machining, assembly, and inspection tasks. Results show measurable improvements in sustainability-related KPIs following each module¿s activation. More importantly, the semantic orchestration layer enables modularity, interoperability, and AI reuse. This work contributes a standards-compliant foundation for next-generation DTs, supporting integration with ecosystems such as FIWARE, Catena-X, and IDS, and aligned with the principles of Industry 4.0 and 5.0., This work was partially supported with grant PID2021-1236730BC31 funded by MCIN/AEI/10.13039/501100011033 and by ERDF A way making Europe .
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Towards Agrirobot Digital Twins: Agri-RO5 A Multi-Agent Architecture for Dynamic Fleet Simulation

  • Gutiérrez Cejudo, Jorge
  • Lujak, Marin
  • Fernandez, Alberto
  • Enguix-Andrés, Francisco|||0009-0002-8144-318X
  • Carrascosa Casamayor, Carlos|||0000-0003-3649-6530
  • Hérnandez López, Luís|||0000-0001-9011-0542
[EN] In this paper, we propose a multi-agent-based architecture for a Unity3D simulation of dynamic agrirobot-fleet-coordination methods. The architecture is based on a Robot Operating System (ROS) and Agrobots-SIM package that extends the existing package Patrolling SIM made for multi-robot patrolling. The Agrobots-SIM package accommodates dynamic multi-robot task allocation and vehicle routing considering limited robot battery autonomy. Moreover, it accommodates the dynamic assignment of implements to robots for the execution of heterogeneous tasks. The system coordinates task assignment and vehicle routing in real time and responds to unforeseen contingencies during simulation considering dynamic updates of the data related to the environment, tasks, implements, and robots. Except for the ROS and Agrobots-SIM package, other crucial components of the architecture include SPADE3 middleware for developing and executing multi-agent decision making and the FIVE framework that allows us to seamlessly define the environment and incorporate the Agrobots-SIM algorithms to be validated into SPADE agents inhabiting such an environment. We compare the proposed simulation architecture with the conventional approach to 3D multi-robot simulation in Gazebo. The functioning of the simulation architecture is demonstrated in several use-case experiments. Even though resource consumption and community support are still an open challenge in Unity3D, the proposed Agri-RO5 architecture gives better results in terms of simulation realism and scalability., This work was partially supported by grants PID2021-123673OB-C32, PID2021-123673OBC33, TED2021-131295B-C31 and TED2021-131295B-C33 funded by MCIN/AEI/10.13039/501100011033 and by ERDF A way of making Europe and the European Union NextGenerationEU/PRTR , respectively, as well as the Agrobots Project funded by the Community of Madrid, Spain, and the AGROBOTIX Project funded by the University Rey Juan Carlos.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Dynamic roles in AgentSpeak

  • Taverner-Aparicio, Joaquín José|||0000-0002-5163-5335
  • Carrascosa Casamayor, Carlos|||0000-0003-3649-6530
  • Duric, Bogdan Okresa
[EN] Roles are a fundamental construct in human organizations to structure behavior, distribute responsibilities and facilitate effective team cooperation. Incorporating this notion in multi-agent systems, especially in deliberative architectures such as the Belief-Desire-Intentions (BDI) model, allows for organizing more clearly the interactions between agents, promoting functional specialization and improving coordination in complex tasks. The use of dynamic roles is presented as a key mechanism to provide agents with greater organizational flexibility and adaptability. Although some approaches have addressed this idea, they tend to rely on ad hoc mechanisms, external to the agent's deliberative cycle, without offering an integrated and reusable solution in different domains. This paper presents a dynamic role model for AgentSpeak. Roles are formalized as a set of beliefs and plans within the agent's knowledge, allowing them to be directly manipulated by standard deliberative mechanisms. The BDI reasoning cycle is further extended so that plan selection is contextually dependent on the agent's active roles. A set of communicative acts is also defined that allows agents to assign and transfer roles, along with their associated beliefs and plans, during execution. Finally, a case study is presented in a multi-agent capture-the-flag scenario, where agents demonstrate their ability to strategically reorganize and adapt roles at runtime., This work is partially supported by Generalitat Valenciana CIPROM/2021/077, Grant PID2021-123673OB-C31 funded by gobierno de Espana MCIN/AEI/10.13039/501100011033 and by "ERDF A way of making Europe", MOBODL-2023-08-5618 funded by the European Union and the Croatian Science Foundation and "Funding for open access charge: CRUE-Universitat Politecnica de Valencia".
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Integrating Gamification with Demand-Responsive Transportation: An Approach to User-Driven Optimisation

  • Martí, Pasqual|||0000-0001-6406-0641
  • Jordán, Jaume|||0000-0003-0400-9136
  • Julian, Vicente|||0000-0002-2743-6037
  • Okre a-Duric, Bogdan
  • Schatten, Markus
[EN] The successful implementation of demand-responsive transportation services is often hindered due to difficulties associated with economic viability. With the aim of reducing service operational costs and environmental impact, this paper proposes an architecture that enables user-driven optimisation through the modification of their travel preferences. Specifically, the paper presents two modules which, integrated with a demand-responsive system, enable vehicle route optimisation, thus saving on costs derived from travelled kilometres and vehicle maintenance. On the one hand, the request modification module computes feasible modifications of the origin or destination stops in a customer travel request, changing them to neighbouring stops such that the transportation routes are simplified. On the other hand, a gamification module is implemented to motivate the customer¿s acceptance of their request modification. Such a motivation is achieved by offering a personalised reward coupled with each modification, computed taking into account user preferences. Upon the acceptance of a modification, the user receives the reward, which may represent social or monetary benefits within the gamified demand-responsive service. The current research formalises both modules and tests their optimisation potential through the development of a case study in transportation scenarios with variable fleet sizes and demand intensities. The results indicate that the implementation of request modifications can save up to 5 % of a fleet¿s operational costs if the right conditions are met. Moreover, when the flexibility of the request modification module is increased, savings may increase, however, at the cost of the customer¿s preferred travel parameters. Thus, a balance between optimisation potential and user satisfaction must be sought., This work is partially supported by grant PID2021-123673 and PDC2022-133161-C32 funded by MCIN/AEI/https://doi.org/10.13039/501100011033 and by «ERDF A way of making Europe». Bogdan Okre a Ðuric is supported by the project MOBODL-2023-08-5618 funded by the European Union and the Croatian Science Foundation. Markus Schatten and Bogdan Okre a Ðuric are supported by the Croatian Science Foundation under the project number IP-2019-04-5824.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Exploring Federated Learning Tendencies Using a Semantic Keyword Clustering Approach

  • Enguix-Andrés, Francisco|||0009-0002-8144-318X
  • Carrascosa Casamayor, Carlos|||0000-0003-3649-6530
  • Rincón-Arango, Jaime Andrés
[EN] This paper presents a novel approach to analyzing trends in federated learning (FL) using automatic semantic keyword clustering. The authors collected a dataset of FL research papers from the Scopus database and extracted keywords to form a collection representing the FL research landscape. They employed natural language processing (NLP) techniques, specifically a pre-trained transformer model, to convert keywords into vector embeddings. Agglomerative clustering was then used to identify major thematic trends and sub-areas within FL. The study provides a granular view of the thematic landscape and captures the broader dynamics of research activity in FL. The key focus areas are divided into theoretical areas and practical applications of FL. The authors make their FL paper dataset and keyword clustering results publicly available. This data-driven approach moves beyond manual literature reviews and offers a comprehensive overview of the current evolution of FL., This research was funded by MCIN/AEI/10.13039/501100011033 and ERDF A way of making Europe grant number PID2021-123673OB-C31 and funded by VAE-VADEN UPV grant number TED2021-131295B-C32 and funded by GUARDIA grant number PROMETEO CIPROM/2021/077 and funded by Ayudas del Vicerrectorado de Investigacion de la UPV grant number PAID-PD-22.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

A flexible approach for Demand-Responsive Public Transport in rural areas

  • Martí, Pasqual|||0000-0001-6406-0641
  • Jordán, Jaume|||0000-0003-0400-9136
  • Julian, Vicente|||0000-0002-2743-6037
[EN] Rural mobility research has been left aside in favor of urban transportation. Rural areas' low demand, the distance among settlements, and an older population on average make conventional public transportation inefficient and costly. This paper assesses the contribution that on -demand mobility has the potential to make to rural areas. First, demand -responsive transportation is described, and the related literature is reviewed to gather existing system configurations. Next, we describe and implement a proposal and test it on a simulation basis. The results show a clear potential of the demand -responsive mobility paradigm to serve rural demand at an acceptable quality of service. Finally, the results are discussed, and the issues of adoption rate and input data scarcity are addressed., This work is partially supported by grant PID2021-123673OB-C31 funded by MCIN/AEI/10.13039/501100011033 and by "ERDF A way of making Europe". Pasqual Marti is supported by grant ACIF/2021/259 funded by the "Conselleria de Innovacion, Universidades, Ciencia y Sociedad Digital de la Generalitat Valenciana". Jaume Jordan is supported by grant IJC2020- 045683-I funded by MCIN/AEI/10.13039/501100011033 and by "European Union NextGenerationEU/PRTR".
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Systematic Review of Aggregation Functions Applied to Image Edge Detection

  • Amorim, Miqueias
  • Dimuro, Graçaliz
  • Borges, Eduardo
  • Dalmazo, Bruno L.
  • Marco-Detchart, Cédric
  • Lucca, Giancarlo
  • Bustince, Humberto
[EN] Edge detection is a crucial process in numerous stages of computer vision. This field of study has recently gained momentum due to its importance in various applications. The uncertainty, among other characteristics of images, makes it difficult to accurately determine the edge of objects. Furthermore, even the definition of an edge is vague as an edge can be considered as the maximum boundary between two regions with different properties. Given the advancement of research in image discontinuity detection, especially using aggregation and pre-aggregation functions, and the lack of systematic literature reviews on this topic, this paper aims to gather and synthesize the current state of the art of this topic. To achieve this, this paper presents a systematic review of the literature, which selected 24 papers filtered from 428 articles found in computer databases in the last seven years. It was possible to synthesize important related information, which was grouped into three approaches: (i) based on both multiple descriptor extraction and data aggregation, (ii) based on both the aggregation of distance functions and fuzzy C-means, and (iii) based on fuzzy theory, namely type-2 fuzzy and neutrosophic sets. As a conclusion, this review provides interesting gaps that can be explored in future work., This work was partially supported with grant PID2021-123673OB-C31 funded by MCIN/AEI/ 10.13039/501100011033 and by ERDF A way of making Europe , Consellería d Innovació, Universitats, Ciencia i Societat Digital from Comunitat Valenciana (APOSTD/2021/227) through the European Social Fund (Investing In Your Future), grant from the Reseach Services of Universitat Politècnica de València (PAID-PD-22), FAPERGS/ Brazil (Proc. 19/2551-0001279-9,19/2551-0001660) and CNPq/Brazil (301618/2019-4, 305805/2021-5) Programa de Apoio à Fixação de Jovens Doutores no Brasil (23/2551-0000126-8).
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Best-response planning for urban fleet coordination

  • Martí, Pasqual|||0000-0001-6406-0641
  • Jordán, Jaume|||0000-0003-0400-9136
  • Julian, Vicente|||0000-0002-2743-6037
[EN] The modeling of fleet vehicles as self-interested agents brings a realistic perspective to open fleet transportation research. This feature allows us to model the fleet operation from a non-cooperative point of view. In this work, we study parcel delivery in a city with limited resources (roads and charging stations). We designed and implemented a system composed of a multi-agent planner and a game-theoretic coordination algorithm: a Best-Response Fleet Planner. The system allows for the self-organization of the transportation system by coordinating a fleet of self-interested electric vehicles. The system's operation is optimized together with resource usage while preserving the agents' private interests, allowing each agent to plan its actions. The results show that our system has higher scalability than similar approaches, allowing it to function for a considerable number of agents in settings that feature congestion and conflicts. Additionally, overall solution quality is improved compared to other coordination systems, reducing congestion and avoiding unnecessary waiting times., This work is partially supported by Grant PID2021-123673OB-C31 funded by MCIN/AEI/10.13039/501100011033 and by "ERDF A way of making Europe." Pasqual Marti is supported by Grant ACIF/2021/259 funded by the "Conselleria de Innovacion, Universidades, Ciencia y Sociedad Digital de la Generalitat Valenciana". Jaume Jordan is supported by Grant IJC2020-045683-I funded by MCIN/AEI/10.13039/501100011033 and by "European Union NextGenerationEU/PRTR".
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Federated Learning for Collaborative Robotics: A ROS 2-Based Approach

  • Gutierrez, Gerardo M.
  • Rincon, Jaime A.
  • Julian, Vicente|||0000-0002-2743-6037
[EN] This paper presents a federated learning framework for multi-agent robotic systems, leveraging the ROS 2 framework to enable decentralized collaboration in both simulated and real-world environments. Traditional centralized machine learning approaches face challenges such as data privacy concerns, communication overhead, and limited scalability. To address these issues, we propose a federated reinforcement learning architecture where multiple robotic agents train local models and share their knowledge while preserving data privacy. The framework integrates deep reinforcement learning techniques, utilizing Unity for high-fidelity simulation. Experimental evaluations compare our federated approach against classical centralized learning, demonstrating that our proposal improves model generalization, stabilizes reward distribution, and reduces training variance. Additionally, results indicate that increasing the number of robots enhances task efficiency, reducing the number of steps required for successful navigation while maintaining consistent performance. This study highlights the potential of federated learning in robotics, offering a scalable and privacy-preserving approach to distributed multi-agent learning., This work was partially supported by the Spanish government with grant PID2021-123673OB-C31 funded by MCIN/AEI/10.13039/501100011033 and by "ERDF A way of making Europe".
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

A review of internet censorship: Modern measurement and circumvention techniques

  • Grubl, Thomas
  • Stiller, Burkhard
  • Enguix-Andrés, Francisco|||0009-0002-8144-318X
[EN] As Internet censorship continues to be deployed across a number of nation-states, understanding its scope and underlying mechanisms is more important than ever. Consequently, research on censorship measurement and circumvention has attracted growing academic interest, particularly in recent years. This article provides an overview of the current state of the art in the field of Internet censorship measurement and circumvention research. First, a brief overview of the fundamentals is provided, followed by an in-depth analysis of 146 con temporary Internet censorship measurement and circumvention studies, predominantly those published within the last ten years, by applying a semi-systematic literature review methodology. Subsequently, the review briefly summarizes the ethical considerations in the field, it visualizes the geographical focus of censorship measurement studies, and it provides an overview of Internet protocols used to measure censorship. In addition, it presents a tax onomy of censorship circumvention tools, analyzes their key characteristics, and examines the prevalence of the underlying network protocols used in circumvention tools. The findings suggest that, while there are numerous solutions for circumventing censorship, many are niche or theoretical, and their practicality remains unknown. Although there is an observable trend toward large-scale longitudinal censorship measurement studies, the real-world effectiveness of (academic) censorship circumvention methods is rarely evaluated. Since both censorship measurement and circumvention research go hand-in-hand, there is an increasing number of measurement studies which directly translate their findings into practical circumvention strategies., This work was partially supported by (a) the University of Zurich UZH, Switzerland, (b) the Horizon Europe Framework Program's project Certify, Grant Agreement No. 101069471, funded by the Swiss State Secretariat for Education, Research, and Innovation SERI, under Contract No. 22.00165, and (c) the grant PID2021-123673OB-C31, PRE2022-101563, funded by MICIU/AEI/10.13039/501100011033 and by "ERDF A way of making Europe".We would like to thank Juergen Bernard for his valuable feedback on the visualizations. We are also grateful to Philipp Winter for maintaining a curated bibliography of Internet censorship research-CensorBib, which is available at https://censorbib.nymity.ch/.Special thanks to Gabriel Gegenhuber for sharing a template for the world map visualization, and to Eike Petersen for providing the template used for the boxplot.We further acknowledge Roya Ensafi and her team for their significant contributions to the field, particularly through the development of Censored Planet and for their commitment to open-sourcing its data. Similarly, we want to acknowledge the Open Observatory of Network Interference (OONI) for their continued efforts to monitor and document Internet censorship worldwide, and for also making their data openly available to the research community. We are grateful to all researchers and organizations working to advance our understanding of Internet censorship and digital freedom.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Personalized Cognitive Support via Social Robots

  • Rincón-Arango, Jaime Andrés
  • Marco-Detchart, Cédric
  • Julian, Vicente|||0000-0002-2743-6037
[EN] This paper explores the use of personalized cognitive support through social robots to assist the elderly in maintaining cognitive health and emotional well-being. As aging populations grow, the demand for innovative solutions to address issues like loneliness, cognitive decline, and physical limitations increases. The studied social robots utilize machine learning and advanced sensor technology to deliver real-time adaptive interactions, including cognitive exercises, daily task assistance, and emotional support. Through responsive and personalized features, the robot enhances user autonomy and improves quality of life by monitoring physical and emotional states and adapting to the needs of each user. This study also examines the challenges of implementing assistive robots in home and healthcare settings, offering insights into the evolving role of AI-powered social robots in eldercare., This work was partially supported by grant PID2021-123673OB-C31, TED2021-131295B-C32 funded by MCIN/AEI/10.13039/501100011033 and by "ERDF A way of making Europe", and PROMETEO grant CIPROM/2021/077 from the Conselleria de Innovacion, Universidades, Ciencia y Sociedad Digital-Generalitat Valenciana.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Developing Digital Twins with SPADE for Autonomous Traffic Control

  • Raya-López, Aarón
  • Palanca Cámara, Javier|||0000-0002-6209-9603
  • Julian, Vicente|||0000-0002-2743-6037
  • Botti V.|||0000-0002-6507-2756
  • Soler Sanz, Manel
[EN] In this paper, we introduce SPADE, a framework engineered for building Digital Twins through Multi-Agent Systems. The architecture is inherently scalable and distributed, aligning perfectly with the demands of modern Digital Twin environments. We implement the Agents and Artifacts meta-model via the SPADE Artifacts extension, which serves as a structured interface connecting autonomous agents with their physical system counterparts. To demonstrate the framework's efficacy, we detail a case study involving urban traffic management in Valencia, Spain. In this implementation, we model 386 street segments as individual agents responsible for managing traffic flows and coordinating redistribution efforts. The research delineates a MAS-based communication strategy spanning the entire network and introduces a consensus algorithm specifically designed to manage traffic rerouting when a street is closed. Finally, we present results from a series of experimental trials and evaluate the system's broader potential. By synthesizing diverse data sources and providing an interactive dashboard for visualizing network conditions, this work demonstrates how SPADE can serve as a robust foundation for Digital Twin development, illustrating its potential for real-world urban applications through a conceptual implementation grounded in open sensor data., This work is partially supported by project TED2021-131295B-C32, funded by MCIN/AEI/ 10.13039/501100011033 and the European UnionNextGenerationEU/PRTR; by grant PID2021-123673-C31 funded by MCIN/AEI/10.13039/501100011033 and by "ERDF A way of making Europe" and by grant ACIF/2021/259 funded by the "Conselleria de Innovacion, Universidades, Ciencia y Sociedad Digital de la Generalitat Valenciana".
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Flexible Agent Architecture: Mixing Reactive and Deliberative Behaviors in SPADE

  • Palanca Cámara, Javier|||0000-0002-6209-9603
  • Carrascosa Casamayor, Carlos|||0000-0003-3649-6530
  • Julian, Vicente|||0000-0002-2743-6037
  • Terrasa Barrena, Andrés Martín|||0000-0003-1991-1691
  • Rincón-Arango, Jaime Andrés
[EN] Over the years, multi-agent systems (MAS) technologies have shown their usefulness in creating distributed applications focused on autonomous intelligent processes. For this purpose, many frameworks for supporting multi-agent systems have been developed, normally oriented towards a particular type of agent architecture (e.g., reactive or deliberative agents). It is common, for example, for a multi-agent platform supporting the BDI (Belief, Desire, Intention) model to provide this agent model exclusively. In most of the existing agent platforms, it is possible to develop either behavior-based agents or deliberative agents based on the BDI cycle, but not both. In this sense, there is a clear lack of flexibility when agents need to perform part of their decision-making process according to the BDI paradigm and, in parallel, require some other behaviors that do not need such a deliberation process. In this context, this paper proposes the introduction of an agent architecture called Flexible Agent Architecture (FAA) that supports the development of multi-agent systems, where each agent can define its actions in terms of different computational models (BDI, procedural, neural networks, etc.) as behaviors, and combine these behaviors as necessary in order to achieve its goals. The FAA architecture has been integrated into a real agent platform, SPADE, thus extending its original capabilities in order to develop applications featuring reactive, deliberative, and hybrid agents. The integration has also adapted the existing facilities of SPADE to all types of behaviors inside agents, for example, the coordination of agents by using a presence notification mechanism, which is a unique feature of SPADE. The resulting SPADE middleware has been used to implement a case study in a simulated robotics scenario, also shown in the paper., This work was partially supported by the Spanish Government with grant number PID2021-123673OB-C31 through the European Social Fund (Investing In Your Future).
Academica-e. Repositorio Institucional de la Universidad Pública de Navarra
Publication Artículo científico (JournalArticle)

Personalized cognitive support via social robots

  • Rincón Arango, Jaime Andrés
  • Marco Detchart, Cedric
  • Julian, Vicente
This paper explores the use of personalized cognitive support through social robots to assist the elderly in maintaining cognitive health and emotional well-being. As aging populations grow, the demand for innovative solutions to address issues like loneliness, cognitive decline, and physical limitations increases. The studied social robots utilize machine learning and advanced sensor technology to deliver real-time adaptive interactions, including cognitive exercises, daily task assistance, and emotional support. Through responsive and personalized features, the robot enhances user autonomy and improves quality of life by monitoring physical and emotional states and adapting to the needs of each user. This study also examines the challenges of implementing assistive robots in home and healthcare settings, offering insights into the evolving role of AI-powered social robots in eldercare., This work was partially supported by grant PID2021-123673OB-C31, TED2021-131295B-C32 funded by MCIN/AEI/10.13039/501100011033 and by “ERDF A way of making Europe”, and PROMETEO grant CIPROM/2021/077 from the Conselleria de Innovación, Universidades, Ciencia y Sociedad Digital—Generalitat Valenciana.
Academica-e. Repositorio Institucional de la Universidad Pública de Navarra
Publication Artículo científico (JournalArticle)

Systematic review of aggregation functions applied to image edge detection

  • Amorim, Miqueias
  • Pereira Dimuro, Graçaliz
  • Borges, Eduardo N.
  • Dalmazo, Bruno L.
  • Marco Detchart, Cedric
  • Lucca, Giancarlo
  • Bustince Sola, Humberto
Edge detection is a crucial process in numerous stages of computer vision. This field of study has recently gained momentum due to its importance in various applications. The uncertainty, among other characteristics of images, makes it difficult to accurately determine the edge of objects. Furthermore, even the definition of an edge is vague as an edge can be considered as the maximum boundary between two regions with different properties. Given the advancement of research in image discontinuity detection, especially using aggregation and pre-aggregation functions, and the lack of systematic literature reviews on this topic, this paper aims to gather and synthesize the current state of the art of this topic. To achieve this, this paper presents a systematic review of the literature, which selected 24 papers filtered from 428 articles found in computer databases in the last seven years. It was possible to synthesize important related information, which was grouped into three approaches: (i) based on both multiple descriptor extraction and data aggregation, (ii) based on both the aggregation of distance functions and fuzzy C-means, and (iii) based on fuzzy theory, namely type-2 fuzzy and neutrosophic sets. As a conclusion, this review provides interesting gaps that can be explored in future work., This research was partially funded by: the Spanish Ministry of Science (TIN2016-77356-P, PID2019-108392GB-I00 AEI/10.13039/501100011033) ; MCIN/AEI/ 10.13039/501100011033 (grant PID2021-123673OB-C31) ; "ERDF A way of making Europe", Consellería d’Innovació, Universitats, Ciencia i Societat Digital from Comunitat Valenciana (APOSTD/2021/227) through the European Social Fund (Investing In Your Future) ; Reseach Services of Universitat Politècnica de València (PAID-PD-22) ; FAPERGS/Brazil (Proc. 19/2551-0001279-9, 19/2551-0001660,23/2551-0000126-8) ; CNPq/Brazil (Proc. 301618/2019-4,305805/2021-5).
Academica-e. Repositorio Institucional de la Universidad Pública de Navarra
Publication Artículo científico (JournalArticle)

AI for sustainable recycling: efficient model optimization for waste classification systems

  • Chacón-Albero, Oriol
  • Campos-Mocholí, Mario
  • Marco Detchart, Cedric
  • Julian, Vicente
  • Rincón Arango, Jaime Andrés
  • Botti, Vicent
The increasing volume of global waste presents a critical environmental and societal challenge, demanding innovative solutions to support sustainable practices such as recycling. Advances in Computer Vision (CV) have enabled automated waste recognition systems that guide users in correctly sorting their waste, with state-of-the-art architectures achieving high accuracy. More recently, attention has shifted toward lightweight and efficient models suitable for mobile and edge deployment. These systems process data from integrated camera sensors in Internet of Things (IoT) devices, operating in real time to classify waste at the point of disposal, whether embedded in smart bins, mobile applications, or assistive tools for household use. In this work, we extend our previous research by improving both dataset diversity and model efficiency. We introduce an expanded dataset that includes an organic waste class and more heterogeneous images, and evaluate a range of quantized CNN models to reduce inference time and resource usage. Additionally, we explore ensemble strategies using aggregation functions to boost classification performance, and validate selected models on real embedded hardware and under simulated lighting variations. Our results support the development of robust, real-time recycling assistants for resource-constrained devices. We also propose architectural deployment scenarios for smart containers, and cloud-assisted solutions. By improving waste sorting accuracy, these systems can help reduce landfill use, support citizen engagement through real-time feedback, increase material recovery, support data-informed environmental decision making, and ease operational challenges for recycling facilities caused by misclassified materials. Ultimately, this contributes to circular economy objectives and advances the broader field of environmental intelligence., This work was partially supported with grant PID2021-123673OB-C31 funded by MCIN/AEI/10.13039/501100011033 and by 'ERDF A way of making Europe' and grant CIPROM/2021/077 from the Conselleria de Innovació, Universitats, Ciencias i Societat Digital Generalitat Valenciana-PROMETEO.