Abstract
In the last decades most efforts to catalog and characterize the built environment for multi-hazard risk assessment have focused on the exploration of census data, cadastral datasets, and local surveys. The first approach is only updated every 10 years and does not provide building locations, the second type of data is only available for restricted urban centers, and the third approach requires surveyors with an engineering background, which is cost-prohibitive for large-scale risk studies. It is thus clear that methods to characterize the built environment for large-scale risk analysis at the asset level are currently missing, which hampers the assessment of the impact of natural hazards for the purposes of risk management. Some recent efforts have demonstrated how deep learning algorithms can be trained to recognize specific architectural and structural features of buildings, which is needed for earthquake risk analysis. In this paper we describe how convolutional neural networks can be combined with data from OpenStreetMap and Google Street View to help develop exposure models for multi-hazard risk analysis. This project produced an original comprehensively annotated (15 characteristics) dataset of approximately 5000 images of buildings from the parish of Alvalade (Lisbon, Portugal). The dataset was used to train and test different deep learning networks for building exposure models. The best results were obtained with ResNet50V2, InceptionV3 and DenseNet201, all with accuracies above 82%. These results will support future developments for assessing exposure models for seismic risk analysis. The novelty of our work consists in the number of characteristics of the images in the dataset, the number of deep learning models trained and the number of classes that can be used for building exposure models.

Abstract
Low-code tools are a trend in software development for business solutions due to their agility and ease of use. There are a certain number of vendors with such solutions. Still, in most Western countries, there is a clear need for the existence of greater quantities of certified and experienced professionals to work with those tools. This means that companies with more resources can attract and maintain those professionals, whilst other smaller organizations must rely on an endless search for this scarce resource. We will present and validate a model designed to transform ChatGPT into a low-code developer, addressing the demand for a more skilled human resource solution. This innovative tool underwent rigorous validation via a focus group study, engaging a panel of highly experienced experts. Their invaluable insights and feedback on the proposed model were systematically gathered and meticulously analysed.

Abstract
Traditional use of predictive control techniques require the knowledge of the systems model to control and the use of constant cycle-time. In the case of a switched reluctance motor its model is highly nonlinear and time-varying with current magnitude and rotor position. The use of look-up tables has been one solution, but requires a complete knowledge of the motor and mismatches from the original model used in the design can happen due temperature variation or changes in operating regimes. To address these issues as well as to increase the tracking performance of current control, a model-free predictive algorithm is developed by updating the next cycle time of the next time step of the predictive control. A new parameter estimation method is proposed that identifies the parameters of the switched reluctance model with low computational burden. Based on knowledge of the parameters at real time, not only the ideal voltage vector is applied at each cycle but the ideal time that each cycle must have is also calculated. As result, the advanced current controller requires almost no knowledge of the motor in use. The performance of the proposed schemes is validated through simulation and by a prototype experimental setup. Experimental data shows a decreasing in prediction error around 78 per cent, when comparing to the predefined model controller.

Abstract

Abstract
Information and communication technologies, such as serious games, have contributed to addressing the gaps in cognitive rehabilitation for individuals with acquired brain injury (ABI), particularly in the context of the COVID-19 pandemic. Although there are effective software programs and games available for cognitive rehabilitation, they have certain limitations. Most current programs have difficulties to adapt to individual performance, a critical factor in promoting neuroplasticity. Additionally, these programs typically only offer single-player modes. However, patients experience difficulties in social interactions leading to social isolation. To overcome these limitations, we propose a novel platform called CogniChallenge. It introduces multiplayer serious games designed for cognitive and psychosocial rehabilitation, offering competitive and cooperative game modes. This platform facilitates engagement with other patients, family members, caregivers, and virtual agents that simulate human interaction. CogniChallenge consists of three games based on activities of daily life and incorporates a multi-agent game balance system. Future research endeavors will focus on evaluating the usability and gameplay experience of CogniChallenge among healthcare professionals and individuals with ABI. By proposing this innovative platform, we intend to contribute to expanding the application of serious games and their potential to solve problems and limitations in the specific field of cognitive rehabilitation. © 2023, Serious Games Society. All rights reserved.

Abstract
Selective thinning is a crucial operation to reduce forest ignitable material, to control the eucalyptus species and maximise its profitability. The selection and removal of less vigorous stems allows the remaining stems to grow healthier and without competition for water, sunlight and nutrients. This operation is traditionally performed by a human operator and is time-intensive. This work simplifies selective thinning by removing the stem selection part from the human operator's side using a computer vision algorithm. For this, two distinct datasets of eucalyptus stems (with and without foliage) were built and manually annotated, and three Deep Learning object detectors (YOLOv5, YOLOv7 and YOLOv8) were tested on real context images to perform instance segmentation. YOLOv8 was the best at this task, achieving an Average Precision of 74% and 66% on non-leafy and leafy test datasets, respectively. A computer vision algorithm for automatic stem selection was developed based on the YOLOv8 segmentation output. The algorithm managed to get a Precision above 97% and a 81% Recall. The findings of this work can have a positive impact in future developments for automatising selective thinning in forested contexts.