Predictive Maintenance (PdM) has emerged as a key enabler of Industry 4.0, taking advantage of data-driven techniques to anticipate equipment failures and enhance maintenance strategies. This PRISMA-guided systematic literature review examines 115 peer-reviewed publications from 2012 to 2025, focusing on two core PdM tasks: anomaly detection (AD) and remaining useful life (RUL) prediction. The findings reveal a methodological shift from traditional machine learning to deep learning, a growth in real-world industrial use cases, and a transition toward proprietary, domain-specific datasets. Despite its importance for early fault identification, anomaly detection remains comparatively underrepresented and lacks standardized evaluation frameworks. As a contribution, this paper presents a structured assessment of AD and RUL methods paired with practical guidance for selecting algorithms given typical data and shop-floor constraints, along with a concise overview of commonly used datasets and evaluation measures. Moreover, this review highlights current research gaps and proposes future directions to better align academic contributions with practical demands of industrial environments.

This paper presents the design and development of an alternative, cost-effective automated piece positioning system, specifically tailored for Small and Medium-sized Enterprises (SMEs), which integrates computer vision with EtherCAT-controlled servo motors. The proposed method combines a robust vision system with an AI-enhanced algorithm based on edge detection to precisely identify object contours. This enables a Programmable Logic Controller (PLC) to control the servo motor, adjusting the piece's angle with high accuracy. Experimental results demonstrate the solution's practical viability, achieving a minimal angular oscillation of less than 0.0012 degrees and a promising low image processing time of approximately 20ms, showcasing its potential for enhancing manufacturing efficiency and quality in industrial applications.

The increasing complexity of modern manufacturing systems poses significant challenges for production planning and control. While technological advancements have separately enhanced predictive maintenance and production scheduling, existing approaches still lack their simultaneous integration. As a result, inefficient resource utilization, increased downtime, and inaccurate decisions are presented. This research proposes a conceptual framework for joint adaptive and intelligent predictive maintenance and production scheduling. A literature review is conducted to identify current challenges and integration gaps, which serve as the foundation for the framework. The proposed solution is structured around three core modules: a simulation platform for performance evaluation, a meta-learning system for adaptive prognostics, and a reinforcement learning system for real-time production planning optimization. Supporting elements for ensuring ease of integration, data management and system interoperability are also introduced. The framework aims to enhance operational efficiency in dynamic manufacturing environments through artificial intelligence driven methods.

Trajectory planning for unmanned aerial vehicles (UAVs) in dynamic and partially observable environments becomes more complex when extended from two-dimensional to three-dimensional navigation. Although Deep Reinforcement Learning (DRL) methods have shown strong performance in 2D scenarios, their application to 3D spaces requires redesigned observation models, action representations, and safety mechanisms. This paper extends a 2D DRL-based trajectory planning framework to 3D environments using Proximal Policy Optimization (PPO), Deep Q-Network (DQN) and Deep Deterministic Policy Gradient (DDPG). UAV agents are trained to reach randomly placed 3D targets while avoiding static and dynamic obstacles using only local sensory information. The observation space combines a local 3D occupancy representation with a relative 3D goal vector, preserving partial observability and avoiding reliance on a global map. This article proposes that the simulation results demonstrate robust, collision-aware navigation and improved safety and trajectory efficiency in each one of the DRL algorithms implemented, each having positive and negative specifics.

Path planning in unknown and partially observable environments remains a major challenge in autonomous unmanned aerial vehicles (UAVs) navigation. This paper presents a Deep Reinforcement Learning (DRL) approach for two-dimensional navigation with dynamic obstacles and limited sensory input. The proposed strategy relies on Proximal Policy Optimization (PPO) to train agents to reach randomly placed target positions while avoiding collisions with static and moving obstacles. The agent perceives the environment using a local laser-based map and a relative vector to the target, without access to the global map. To evaluate performance, the PPO approach is compared with classical A* and DRL algorithms, such as Deep Q-Network (DQN), Deep Deterministic Policy Gradient (DDPG), under the same simulation conditions. Metrics include trajectory length, goal-reaching success rate, number of collisions, and cumulative reward. Experimental results show that PPO can generalize to novel scenarios, outperforming other DRL-based methods in terms of safety and goal efficiency.

Wireless communication systems enable flexibility, mobility, and scalability for industrial applications, but there are still concerns about their performance in critical scenarios. This paper proposes a testing tool and methodology focused on application-oriented performance evaluation to asses if communication system can cope with requirements from applications. Exemplary results of uplink transmission in a 5G campus network demonstrate that optimized configurations reduced transmission time by 48 % compared to default configurations, highlighting the potential of customized parametrization. The presented methodology offers practical testing strategies and can be scaled to evaluate future wireless technologies such as 6G and non-terrestrial networks, promoting performance-oriented optimization in wireless systems.

The increasing use of Unmanned Aerial Vehicles (UAVs) in applications such as logistics, environmental monitoring, infrastructure inspections, and precision agriculture has brought new safety and operational efficiency challenges. In this context, ensuring collision avoidance in operations with multiple UAVs in shared spaces has become a priority. This work proposes a framework based on Deep Reinforcement Learning (DRL), designed to optimize collaborative trajectory planning and collision avoidance in dynamic environments, which include static and moving obstacles. The approach combines LiDAR sensors with Detection And Avoidance (DAA) algorithms. These systems allow UAVs to identify and avoid obstacles accurately, even in dense urban scenarios and with inaccurate sensors. In addition, the framework promotes collaboration between UAVs through decentralized architectures, allowing greater efficiency in the use of airspace and reducing overlap in shared missions. Performed simulations have shown that the proposed approach provides improvement in movement safety. The results show that the system is adaptable in dynamic scenarios, making it promising for applications in complex airspaces.

The Digital Twin (DT) has emerged as a crucial component for advancing Industry 4.0, Industry 5.0, and the Metaverse, attracting substantial interest from both academia and industry. By digitalizing real-world entities, DTs enable capabilities such as monitoring, simulation, prediction, and optimization. However, accurately modeling real-world systems remains challenging due to the diversity of asset representations and the differing perspectives of stakeholders. This paper presents an integrated framework with a unified language to align these perspectives across various system levels, from individual devices to complex workflows. It introduces a methodology for semantic DT models and outlines a 4-layer architecture to clearly define system responsibilities.

The effective management of process deviations and abnormal events depends on operational actions in providing the appropriate responses to each situation. Applications with requirements focused on the provision of resources tend to be handled by a Human-Machine Interface (HMI) along with Supervisory Control and Data Acquisition (SCADA). Process reliability is achieved if the operation ensures that actions will be performed according to the required response and priority levels. This paper presents a novel architecture entitled the Heuristic-Based Recommendation System (HB-RS). The main goal is to provide resources capable of streamlining the process of actively handling abnormal situations. For recommendation purposes, the use of probabilistic networks is proposed, which are well suited to represent uncertainty elements present in engineering applications. To create the inference mechanism for these systems, a Multi-Entity Bayesian Network (MEBN) is proposed, highlighting the importance of semantic characterization via knowledge-based, probabilistic graphical model formalism, which is closely linked to the modeling paradigm in which we can predict situations. The developed capabilities have been applied to a real case study in the operation of a metro train system, and the results obtained indicate the value of the proposed method.

Hydroponic systems have been shown to be an alternative to conventional growing systems in recent years. One of its main challenges became the individual measurement of nutrients as opposed to the EC and pH measurement systems. ISE have proven to be a promising way to measure nutrients, but face the challenges of drift and interference from other ions in their measurements. To overcome this, calibration and sampling techniques in conjunction with machine learning have demonstrated promising results. The results obtained open new perspectives for future investigations into the integration of IoT and AI with sensors in hydroponic systems.

Industry 4.0 emphasizes the flexibility of production systems, requiring dynamic adaptation to varied product requirements. This paper presents a framework for managing production systems by integrating application and communication systems. Central to it is the use of Asset Administration Shell (AAS) to create Digital Twins of assets, enabling continuous adaptation and optimization of production processes through effective resource management. The paper discusses a use case involving negotiation between communication and application systems, proposing a model-based solution leveraging AAS. It details the implementation of the proposed model with AAS components, providing an analysis of the use case and an automation solution. Validation results show how the AAS can improve the performance and dependability of wireless communication systems. Industrie 4.0 betont die Flexibilit & auml;t von Produktionssystemen und erfordert eine dynamische Anpassung an unterschiedliche Produktanforderungen. Dieses Dokument stellt einen Rahmen f & uuml;r die Verwaltung von Produktionssystemen durch die Integration von Anwendungs- und Kommunikationssystemen vor. Im Mittelpunkt steht dabei die Verwendung von Asset Administration Shell similar to(AAS) zur Erstellung digitaler Zwillinge von Assets, die eine kontinuierliche Anpassung und Optimierung von Produktionsprozessen durch effektives Ressourcenmanagement erm & ouml;glichen. Das Dokument er & ouml;rtert einen Anwendungsfall, der die Verhandlung zwischen Kommunikations- und Anwendungssystemen beinhaltet und schl & auml;gt eine modellbasierte L & ouml;sung vor, die AAS nutzt. Es beschreibt detailliert die Implementierung des vorgeschlagenen Modells mit AAS-Komponenten und bietet eine Analyse des Anwendungsfalls und eine Automatisierungsl & ouml;sung. Validierungsergebnisse zeigen, wie die AAS die Leistung und Zuverl & auml;ssigkeit von drahtlosen Kommunikationssystemen verbessern kann.

The growing use of unmanned aerial vehicles (UAVs) covers several applications in the civilian and military domains. One prominent application of this technology is precision agriculture, which offers innovative solutions to complex challenges. This paper focuses on integrating UAVs in precision agriculture, emphasizing the cooperative transfer of tasks while mitigating interference in GNSS systems. The proposed methodology is based on autonomous UAV navigation, using onboard cameras to ensure precise maneuvers along crop rows. This approach reduces the exclusive reliance on GNSS signals, overcoming a common limitation in drone applications, including precision agriculture settings. The study demonstrates the system's effectiveness through simulations on the Gazebo platform. The system manages the transfer of tasks, allowing UAVs to assume, execute, and transfer tasks autonomously, considering factors such as battery level and remaining agrochemical payload during spraying. This paper contributes to the evolution of precision agriculture by mitigating GNSS interference challenges through a cooperative UAV approach.

This article presents a method for generating a dataset from CAN protocol communication using a test bench based on the IEC 62228-2019 standard. Modern automotive embedded systems feature many electronic control units for engine control, ABS, traction control, and power steering. Ensuring reliability and effectiveness, consistent testing, and data analysis are essential. This article proposes a testing approach to verify and record CAN communication under disturbances, using a redundant channel and a fault injection method with specially developed hardware. Different injected noise values were defined based on prior work. A sequence of tests applying the hardware-in-the-loop approach generates valuable preprocessed datasets. The results show that the proposed method records all CAN communication data, producing around 15 million records. This dataset enables detailed analysis of EFT fault propagation in the CAN network from various vehicle noise sources, providing a foundation for future data-driven diagnostic systems development.

This paper explores the Asset Administration Shell (AAS) for modelling and managing application requirements in industrial automation. Parameters are defined as influencing quantities for traffic modelling and performance parameters for communication requirements. These form a comprehensive digital twin of the application, aligning its needs with the communication system capabilities. A use case demonstrates managing competing logical links, prioritizing periodic traffic over non-periodic. The AAS dynamically adjusts data rates to meet stringent timing requirements, ensuring uninterrupted critical data flow. Results demonstrate the potential of AAS in efficient resource management and operational optimization for industrial applications.

Multicore systems emerged as an alternative to traditional monocore systems. Although these systems possess high performance, they have more complexity. Moreover, their performance can be degraded because shared hardware resources introduce scheduling delays. To reduce this delay or contention, there are several techniques that can be applied both when allocating tasks to cores and when scheduling tasks within each core. In this paper, we propose a scheduling algorithm that combines different known scheduling policies to obtain a temporal plan that reduces the interference. In addition, we propose an artificial neural network to predict which allocation policy should be applied to minimize the length of the intervals that compose the temporal plan and thus reduce the scheduling complexity of each interval.

This paper explores the fault susceptibility of in-vehicle communication protocols. It proposes integrating aspect-oriented concepts to model common faults in the design phases of embedded systems and faults as non-functional requirements NFR. An analytical case study was conducted considering the tests of electrical fast transients (EFT) injection presented in previous works and based on data collected during these experiments. A fault tree analysis was conducted to determine the effect of the faults in the networked control system. Results based on a single-time approach reveal adverse effects (CAN: 0.622, CAN-FD: 0.352, FlexRay: 0.730). The research contributes to mitigating the impact of EFT faults on in-vehicle communication networks, enabling more specific fault analysis. Copyright (c) 2024 The Authors.

Precision agriculture is a domain in fast expansion, demanding flexible approaches to deploy customized solutions to specific needs in short time windows. In this context, this work proposes a service-oriented architecture and microservices to support Unmanned Aerial Vehicles systems development for this sector. The proposal presents a framework and middleware designed to offer modular and reusable software components with an intermediate layer to facilitate integration, communication, control, and data management between different systems in an agricultural application setting. This paper presents the design of this approach with a glance to the roadmap to its implementation.

Network Functions Virtualization (NFVs) and Information-Centric Networking (ICN) are promising networking paradigms for the future of the Internet. In addition, microservice architecture offers an attractive alternative to monolithic architecture for developing applications. Employing ICN NFVs implemented as microservices enables the use of ICN in generic devices with the ability to scale on demand to address specific application requirements. However, working in this context is challenging due to the complexity of managing and coordinating microservices at runtime, especially for a cluster of nodes. This paper proposes Micro-Chain, an architecture for deploying, scaling, and chaining ICN microservices. The architecture consists of five modules and their interrelationships. The basic operations of the proposed architecture were validated in a scalability scenario within a node cluster, where CPU and memory thresholds were utilized as parameters for scaling decisions. The performed experiment revealed a microservices placement issue that resulted in lower throughput.

The military usage of Unmanned Aerial Vehicles (UAVs) has garnered attention, especially after their employment in the Ukrainian war. Despite the most commented lethal usage, they have many other applications from which surveillance for imagery acquisition is one of primal importance. Using a standalone UAV for this purpose is well-known, but to cope with the scale of battlefield operations, using multi- UAV systems is an asset of great value. However, these systems rely on ad hoc networks that require solutions beyond conventional ones based on the Internet Protocol (IP). This paper addresses this concern by proposing a communication support mechanism for multi-UAV military surveillance systems based on the Information-Centric Networks (ICN) paradigm. The proposed approach consists of the dynamic deployment of an ICN network based on microservices architecture, where the communication services of each UAV are deployed according to their resources. The solution is validated in a simulated battlefield scenario where a surveillance UAV provides data demanded by other nodes. The results demonstrate that the proposed solution minimizes the data delivery delays by successfully deploying the customized set of microservices to support the transmission, even when a UAV with a low battery level is replaced at runtime.

Wireless Sensor Networks (WSNs) play a crucial role in several applications ranging from precision agriculture to health wearable devices. A significant issue regarding WSN is energy consumption. To tackle this problem, clustering technique is frequently considered. In this context and according to the literature, the inclusion of a mobile sink node, such as an Unmanned Aerial Vehicle (UAV), can dramatically decrease energy consumption due to the communication, since the UAV can collect the data from the static nodes. This paper addresses this problem by modifying the A ∗ path planning algorithm for a fixed-wing UAV considering wind field constraints. The position of the cluster heads is also optimized in the problem in order to further reduce the energy consumption of static network nodes and the flying data collector dynamic node. The results demonstrate that the modified A ∗ algorithm and the deployed clustering technique efficiently find an optimal path with a gain in energy of up to 20% for the UAV and 25% for the network.