Oscar Hernán Ramírez-Agudelo

Remote operation and infrastructure-supported driving expand vehicular perception and control through distributed communication between vehicles, infrastructure, and control centres. However, this also creates the need for a cybersecurity management framework that can satisfy the requirements of the cloud and edge environment, as well as rapidly emerging adversarial AI attacks. Since communications between participants in both scenarios can be modulated as data streams, a hybrid data stream-based cybersecurity management framework extending the GALOIS architecture that satisfies requirements is proposed. So far, no concept has been proposed to model a cybersecurity management framework based on a distributed data stream processing architecture that takes into account the interoperability of the different devices in particular. Our results show that real-time analysis of adversarial attacks is feasible but needs a redesign. Our introduced online detection operator outperformed the classical offline detection by 584.3% in latency time.

Images captured in hazy and smoky environments suffer from reduced visibility, posing a challenge when monitoring infrastructures and hindering emergency services during critical situations. The proposed work investigates the use of the deep learning models to enhance the automatic, machine-based readability of gauge in smoky environments, with accurate gauge data interpretation serving as a valuable tool for first responders. The study utilizes two deep learning architectures, FFA-Net and AECR-Net, to improve the visibility of gauge images, corrupted with light up to dense haze and smoke. Since benchmark datasets of analog gauge images are unavailable, a new synthetic dataset, containing over 14,000 images, was generated using the Unreal Engine. The models were trained with an 80\% train, 10\% validation, and 10\% test split for the haze and smoke dataset, respectively. For the synthetic haze dataset, the SSIM and PSNR metrics are about 0.98 and 43\,dB, respectively, comparing well to state-of-the art results. Additionally, more robust results are retrieved from the AECR-Net, when compared to the FFA-Net. Although the results from the synthetic smoke dataset are poorer, the trained models achieve interesting results. In general, imaging in the presence of smoke are more difficult to enhance given the inhomogeneity and high density. Secondly, FFA-Net and AECR-Net are implemented to dehaze and not to desmoke images. This work shows that use of deep learning architectures can improve the quality of analog gauge images captured in smoke and haze scenes immensely. Finally, the enhanced output images can be successfully post-processed for automatic autonomous reading of gauges

Data quality plays a central role in the performance and robustness of convolutional neural networks (CNNs) for image classification. While high-quality data is often preferred for training, real-world inputs are frequently affected by noise and other distortions. This paper investigates the effect of deliberately introducing controlled noise into the training data to improve model robustness. Using the CIFAR-10 dataset, we evaluate the impact of three common corruptions, namely Gaussian noise, Salt-and-Pepper noise, and Gaussian blur at varying intensities and training set pollution levels. Experiments using a Resnet-18 model reveal that incorporating just 10\% noisy data during training is sufficient to significantly reduce test loss and enhance accuracy under fully corrupted test conditions, with minimal impact on clean-data performance. These findings suggest that strategic exposure to noise can act as a simple yet effective regularizer, offering a practical trade-off between traditional data cleanliness and real-world resilience.

Context. Massive stars can have extreme effects on their environments from local to galactic scales. Their multiplicity can affect this influence by altering how they evolve over time by causing dynamical interactions, common-envelope evolution, mergers and more. While O star multiplicity has been studied over a broad separation range (to the point where absolute masses of these systems have been determined and investigations into multiple system formation and interactions have been performed), studies of B star multiplicity are lacking, even though they dominate the production of core-collapse supernovae and neutron stars. Aims. Using interferometry, we investigated the multiplicity of a statistically significant sample of B stars over a range of separations (∼0.5–35 au, given that the average distance to our sample is 412 pc). Methods. We analysed high angular resolution interferometric taken with the PIONIER (Precision Integrated-Optics Near-infrared Imaging ExpeRiment) instrument at the Very Large Telescope Interferometer (VLTI) for a sample of 32 B stars. Using parametric modelling of the closure phases and visibilities, we determined best-fitting models to each of the systems and investigated whether each source was best represented by a single star or a higher-order system. The detection limits were calculated for companions to determine whether they were significant. We then combined our findings from the interferometric data with results from a literature search to determine whether other companions were reported at different separation ranges. Results. Within the interferometric range 72 ± 8% of the B stars are resolved as multiple systems. The most common type of system are binary systems, followed by single stars, triple systems, and quadruple systems. The interferometric companion fraction derived for the sample is 1.88 ± 0.24. When we accounted for spectroscopic companions that have been confirmed in the literature and wide companions inferred from Gaia data in addition to the companions we found with interferometry, we obtain multiplicity and companion fractions of 0.88 ± 0.06 and 2.31 ± 0.27, respectively, for our sample. The number of triple systems increases significantly to the second-most populous system when we account for spectroscopic companions. This suggests that binarity and higher-order multiplicity are as integral to the evolution of B stars as they are for O stars.

This paper introduces GAUREAD, an end-to-end computer vision system that is able to autonomously read analogic gauges with circular shapes and linear scales in unstructured environments. Existing gauge reading software still relies on some manual entry, like the gauge location and the gauge scale, or they are able to work just with a frontal view. On the contrary, GAUREAD comprises all the necessary steps to make the measurement unconstrained from previous information, including gauge detection from scene, perspective rectification and scale reconstruction. Our algorithm achieves a speed of 800 milliseconds per reading on the NVIDIA Jetson Nano 4 GB. Experimental tests show that GAUREAD can provide a measurement with an error within 3% for perspective angles below 20° and within 9% up to 50°. The system is foreseen to be implemented on mobile robotics to automatise not only safety routines, but also critical security operations.

Offshore wind farms (OWFs) are important infrastructure which provide an alternative and clean means of energy production worldwide. The offshore wind industry has been continuously growing. Over the years, however, it has become evident that OWFs are facing a variety of safety and security challenges. If not addressed, these issues may hinder their progress. Based on these safety and security goals and on a Bayesian network model, this work presents a methodological approach for structuring and organizing expert knowledge and turning it into a probabilistic model to assess the safety and security of OWFs. This graphical probabilistic model allowed us to create a high-level representation of the safety and security state of a generic OWF. By studying the interrelations between the different functions of the model, and by proposing different scenarios, we determined the impacts that a failing function may have on other functions in this complex system. Finally, this model helped us define the performance requirements of such infrastructure, which should be beneficial for optimizing operation and maintenance.

We present results from the B-type binaries characterization (BBC) programme, a multi-epoch spectroscopic study of 88 early B-type binary candidates in the 30 Doradus region of the Large Magellanic Cloud (LMC). From radial-velocity analysis of 29 observational epochs, we confirm the binary status of 64 of our targets, comprising 50 SB1 and 14 SB2 B-type binaries. A further 20 systems (classified as SB1*) show clear signs of periodicity but with more tentative periods. Orbital solutions are presented for these 84 systems, providing the largest homogeneous sample to date of the binary properties of early B-type stars. Our derived orbital-period distribution is generally similar to those for samples of more massive (O-type) binaries in both the LMC and the Galaxy. This similarity with the properties of the more massive O-type binaries is important as early B-type stars are expected to account for the majority of core-collapse supernovae. Differences in the period distributions of the different samples start to increase above 4 d, and are also present between the earliest (B0-0.7) and later-type (B1-2.5) systems within the BBC sample, although further study is required to understand if this is an observational bias or a real physical effect. We have examined the semi-amplitude velocities and orbital periods of our sample to identify potential candidates that could hide compact companions. Comparing with probability distributions of finding black hole companions to OB-type stars from a recent theoretical study, we have found 16 binaries in the higher probability region that warrant further study.

Farr and Mandel reanalyze our data, finding initial mass function slopes for high-mass stars in 30 Doradus that agree with our results. However, their reanalysis appears to underpredict the observed number of massive stars. Their technique results in more precise slopes than in our work, strengthening our conclusion that there is an excess of massive stars (>30 solar masses) in 30 Doradus.

The Tarantula region in the Large Magellanic Cloud contains the richest population of spatially resolved massive O-type stars known so far. This unmatched sample offers an opportunity to test models describing their main-sequence evolution and mass-loss properties. Using ground-based optical spectroscopy obtained in the framework of the VLT-FLAMES Tarantula Survey (VFTS), we aim to determine stellar, photospheric and wind properties of 72 presumably single O-type giants, bright giants and supergiants and to confront them with predictions of stellar evolution and of line-driven mass-loss theories. We apply an automated method for quantitative spectroscopic analysis of O stars combining the non-LTE stellar atmosphere model {{\sc fastwind}} with the genetic fitting algorithm {{\sc pikaia}} to determine the following stellar properties: effective temperature, surface gravity, mass-loss rate, helium abundance, and projected rotational velocity. We present empirical effective temperature versus spectral subtype calibrations at LMC-metallicity for giants and supergiants. In the spectroscopic and classical Hertzsprung-Russell diagrams, our sample O stars are found to occupy the region predicted to be the core hydrogen-burning phase by Brott et al. (2011) and K\"{o}hler et al. (2015). Except for five stars, the helium abundance of our sample stars is in agreement with the initial LMC composition. The aforementioned five stars present moderate projected rotational velocities (i.e., $v_{\mathrm{e}}\,\sin\,i\,<\,200\,\mathrm{km\,s^{-1}}$) and hence do not agree with current predictions of rotational mixing in main-sequence stars. Adopting theoretical results for the wind velocity law, we find modified wind momenta for LMC stars that are $\sim$0.3 dex higher than earlier results. [Due to the limitation of characters, the abstract appearing here is slightly shorter than that in the PDF file.]