Centro de Estudios e Investigaciones Técnicas de Gipuzkoa

Abstract Foreign object detection (FOD) is a crucial concern in various industries, including aerospace, automotive, and manufacturing, where undetected foreign objects can cause severe damage and safety risks. Traditional FOD detection systems rely on visual inspection, radar, X-ray imaging, or acoustic sensing, but face challenges in complex and dynamic environments. In this work, we propose a novel methodology that utilizes neural radiance fields (NeRF) as a comparative baseline for foreign object detection in industrial settings. Our approach consists of two main phases: First, a photorealistic 3D reconstruction of the environment is generated using NeRF when no foreign objects are present. Then, in the detection phase, new images of the environment are captured, and their viewpoints are matched with the NeRF-rendered scene to identify discrepancies using a deep learning-based comparative model. The system utilizes a modified SNUNet-CD architecture, enhancing attention mechanisms for improved accuracy in detecting semantic changes. Experimental validation in a controlled environment demonstrates high detection accuracy, successfully identifying foreign objects of varying sizes, materials, and placements.

Social games offer a rich and challenging environment to develop and evaluate complex robot behaviors. Social interactions with robots benefit from personalization, which requires person identification—a process that raises privacy concerns. Soft biometrics, however, offer a privacy-preserving alternative by enabling short-term identification without relying on private or sensitive features. In this work, we leverage a zero-shot Visual Question Answering (VQA) person identification system and adapt it to make Pepper play Guess Who? The identification module performance is firstly studied offline in the AveRobot dataset. Afterwards, it is integrated in a ROS based architecture that manages the game flow and enables natural verbal interaction. Experiments conducted with 29 subjects demonstrate that the system achieves identification performance nearly equivalent to that of hard biometric-based systems while also offering engaging and entertaining gameplay. Notably, user feedback reveals a high level of acceptance, highlighting that social gaming with robots is a promising avenue for developing and testing complex behaviors.

Transformers depend strongly on insulating fluids for efficient and reliable operation. Current trends are changing towards natural esters as environmentally friendly alternatives. Understanding how these oils degrade is essential, as changes in oil quality can directly affect the performance and reliability of transformers. This work analyses the effect of degradation due to oxidative aging and the influence of contaminants (copper and cellulose) on natural ester density, viscosity, absorbance and dissolved decay products, as well as dielectric properties (real part of relative permittivity, dissipation factor and conductivity). Four aging periods (0, 164, 332 and 500 hours) and four contaminant conditions (without contaminants, with copper, with cellulose kraft paper and with copper and cellulose kraft paper combination) are considered. The results show that the combination of copper and cellulose exhibits a synergistic effect, causing the most pronounced degradation of all measured properties, causing the gelation of the 500-hour sample.

ABSTRACT One of the primary challenges in ensuring road safety is effectively alerting drivers to adverse weather conditions, such as fog, which can severely impair visibility and increase the risk of accidents. Timely and accurate fog detection is crucial for providing drivers with the necessary warnings to adapt their driving behaviour and enhance safety. This paper presents a system designed to detect foggy scenarios and classify visibility levels, thereby enabling timely alerts for drivers to minimise the risks associated with reduced visibility. To achieve this, we have developed two new image datasets of road fog scenarios – Foggy‐Ceit 2023 and an extension to the Foggy CityScapes – DBF dataset – featuring both real and synthetic fog. Additionally, we compare various algorithms developed using classical vision techniques and deep learning methods (vision transformers [ViT] and EfficientNet). Finally, eXplainable artificial intelligence techniques are utilised to provide visual explanations and evaluate the performance of these models.

Stroke is commonly accompanied by motor disorders secondary to the initial injury, and at the mitochondrial level, there is an increase in oxidative stress, causing tissue damage and relating to alteration in motor function. Functional electrical stimulation (FES) therapies can be useful in stroke patients aimed at restoring motor function and improving life quality. Several studies have recently proposed that electrical stimulation modulates mitochondrial biogenesis and physiology; however, the effect of FES on individual complexes comprising the mitochondrial electron transport chain (mETC) remains unknown. To study the effect of FES therapy on the activity of the mETC, invitro analyses were conducted using UV-spectrophotometry of the activity of the complexes (I, II, and IV), glycerol-3-phosphate dehydrogenase (G3PDH), and monoamine oxidase (MAO) in peripheral blood mononuclear cells (PBMCs) obtained from stroke patients. Patients' activity of complex II and MAO was also determined in platelets. Analyses were measured before and after 12 rehabilitation sessions based on FES therapy combined with physical exercises. The results showed that the activity of the mETC, G3PDH, and MAO was significantly reduced in pre-intervention stroke patients compared to healthy volunteers, both in PBMCs and platelets, except for complex II in platelets and MAO in PBMCs. Interestingly, FES intervention increased the mETC, G3PDH, and MAO activity in both cell types of stroke patients. This preliminary study indicates that intervention can restore motor function and mitochondrial activity, which are damaged in stroke patients. However, more research is needed to determine if changes in mitochondrial activity can be used as biomarkers of the pathology and/or the efficacy of intervention.

The objective of this work is to analyze the effect of Nd concentration and additions of Nb and Ga on the hydrogen decrepitation of gas-atomized powders after grain growth annealing at 1150 ºC for 5 h. Ga and Nb additions reduced the grain size upon atomization. Thermodynamic calculations have demonstrated that, in general, the amount of liquid at 1150 ºC increases with Nd concentration and with the addition of Nb or Nb-Ga. Grain growth during annealing is accelerated when the Nd content is raised or with the addition of Ga, which can be explained by the larger amount of liquid in the alloy. On the other hand, Nb forms precipitates that delay grain growth, especially if the amount of liquid is low. Comparing the grain size distribution of the annealed samples with the particle size distribution of hydrogen decrepitated powders allowed identifying the dominant fracture mechanism. Some transgranular cracking occurred in all compositions, increasing the fraction of fine irregular particles. When the Nd concentration is raised, the particle size is reduced. In general, the dominant mechanism of fracture is intergranular crack propagation. However, the transgranular fracture was the dominant mechanism in Ga-containing alloys. Particle shape was not sensitive to the compositional changes evaluated in this work. • Amount of liquid increases with Nd concentration and addition of Nb or Ga. • Nb delays grain growth, whereas Ga and a higher Nd content accelerate it. • Span of grain size distribution increases with Nb addition and low liquid amount. • Ternary and Nb-containing alloys show dominant intergranular fracture. • Ga-containing alloys exhibit transgranular fracture.

Innovative wood inspection technology is crucial in various industries, especially for determining wood quality by counting rings in each stave, a key factor in wine barrel production. (1) Background: Traditionally, human inspectors visually evaluate staves, compensating for natural variations and characteristics like dirt and saw-induced aberrations. These variations pose significant challenges for automatic inspection systems. Several techniques using classical image processing and deep learning have been developed to detect tree-ring boundaries, but they often struggle with woods exhibiting heterogeneity and texture irregularities. (2) Methods: This study proposes a hybrid approach combining classical computer vision techniques for preprocessing with deep learning algorithms for classification, designed for continuous automated processing. To enhance performance and accuracy, we employ a data augmentation strategy using cropping techniques to address intra-class variability in individual staves. (3) Results: Our approach significantly improves accuracy and reliability in classifying wood with irregular textures and heterogeneity. The use of explainable AI and model calibration offers a deeper understanding of the model’s decision-making process, ensuring robustness and transparency, and setting confidence thresholds for outputs. (4) Conclusions: The proposed system enhances the performance of automatic wood inspection technologies, providing a robust solution for industries requiring precise wood quality assessment, particularly in wine barrel production.

Industrial Control Systems (ICS) are critical to operating various Critical infrastructures (CIs). However, ICS communication channels connecting sensors, actuators, and local and supervisory controllers are vulnerable to network attacks compromising the system’s availability and integrity. This study proposes and compares Standalone and Hybrid Software Defined Networking (SDN) solutions to mitigate (Detect and Respond) network channel attacks in ICS environments. The methodology utilised applies a testbed designed in GNS3 following the IEC 62264 Industrial Automation Pyramid. It incorporates ICS components such as PLCs and SCADA and a Simulink-based digital twin system for a wastewater treatment plant. This research establishes a proof of concept involving detection and response to network channel attacks evaluated through packet thresholds, packet analysis, and cryptographic hashing techniques in SDN. The Mitre attack framework is implemented to provide insight into the system’s vulnerabilities through adversary emulation. The research findings reveal that both SDN solutions effectively enhance ICS network security; the Standalone SDN solution is more suitable for time-sensitive networks, while the Hybrid SDN solution better serves non-time-sensitive industrial environments. While the Standalone SDN solution offers a 75% efficiency improvement, its’ status as a nascent technology introduces unresolved vulnerabilities and limited testing favouring the Hybrid SDN solution, which provides robust security and reliability due to the integration with the Snort IDS. Thus, selecting the appropriate solution requires carefully considering the trade-offs between enhanced performance and established security. In conclusion, this study underscores the potential of SDN solutions in strengthening ICS security and suggests areas for future research.