Abstract
The rapid advancement of wireless technologies, particularly Wi-Fi, has spurred significant research into indoor human activity detection across various domains (e.g., healthcare, security, and industry). This work explores the non-invasive and cost-effective Wi-Fi paradigm and the application of deep learning for human activity recognition using Wi-Fi signals. Focusing on the challenges in machine interpretability, motivated by the increase in data availability and computational power, this paper uses explainable artificial intelligence to understand the inner workings of transformer-based deep neural networks designed to estimate human pose (i.e., human skeleton key points) from Wi-Fi channel state information. Using different strategies to assess the most relevant sub-carriers (i.e., rollout attention and masking attention) for the model predictions, we evaluate the performance of the model when it uses a given number of sub-carriers as input, selected randomly or by ascending (high-attention) or descending (low-attention) order. We concluded that the models trained with fewer (but relevant) sub-carriers are competitive with the baseline (trained with all sub-carriers) but better in terms of computational efficiency (i.e., processing more data per second).

Abstract
This study systematically characterizes Do It Yourself (DIY) and open-source wheeled robotic platforms used in higher education and academic competitions. It also analyzes Robot Operating System (ROS)-based designs with respect to real-time performance and multi-sensor integration, following Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines. A total of 20 high-quality studies were identified across five major digital libraries (Dimensions, Web of Science, SpringerLink, ScienceDirect, and IEEE Xplore), which were searched on 12 January 2026. Eligibility was restricted to peer-reviewed English-language studies published between 2005 and 2026 that explicitly implement ROS-based wheeled platforms in higher education contexts. Results were synthesized through qualitative analysis using a structured data extraction form implemented in the Parsifal systematic review platform. Methodological quality and risk of bias were assessed using a structured appraisal checklist. The results show a dominant trend toward distributed dual-processor architectures, which separate low-level real-time control from high-level processing. Most platforms target an accessible price range of 50€ to 500€ for open-source and DIY platforms. ROS has emerged as the standard middleware, enabling multi-sensor integration and supporting digital twin workflows. There is also a clear shift toward open-source hardware and Three-Dimensional (3D)-printed modular designs, which reduce production costs. However, challenges remain, including software obsolescence and the lack of maintenance plans. The findings highlight the need for interoperable reference architectures and automated deployment workflows to ensure long-term sustainability. Evidence is limited by heterogeneity, inconsistent reporting, and small sample sizes, which introduce risks of bias and imprecision. This review was formally registered with protocols.io.

Abstract
Deep reinforcement learning (DRL) is a promising solution for mobile-robot navigation, yet its performance often degrades in harsh terrain where wheel-ground interactions vary rapidly. Uneven and deformable surfaces can change the effective wheel radius and introduce persistent left-right actuation asymmetries, which distort odometry and action execution and ultimately reduce goal-reaching success. We propose a two-stage, predictor-augmented DRL framework that improves robustness to these non-idealities without relying on external supervision or privileged sensors. In Stage 1, the robot collects randomized trajectories in an ideally flat environment while sampling episode-wise left/right wheel-radius factors, and trains a lightweight self-supervised predictor from onboard signals (odometry, IMU, and wheel joint states) to estimate the mismatch factors. In Stage 2, the trained predictor is frozen and its outputs are appended to the observation of a Soft Actor Critic (SAC) navigation policy, enabling the policy to condition its decisions on estimated actuation drift during navigation in harsh terrain. We evaluate the proposed approach in ROS 2/Gazebo across multiple harsh-terrain scenarios with wheel-radius perturbations. Our method outperforms all baselines in success, slip-related metrics, and cumulative roll-pitch inclination, while producing modestly longer paths as a deliberate trade-off to mitigate wheel-radius mismatch. We further validate sim-to-real transfer on a real robot under unmodeled effects (sensor noise, calibration drift, and actuator saturation). The full source code and a demonstration video of both simulation and real-world experiments are publicly available on GitHub. © 2026 IEEE.

Abstract
E-learning has been important in higher education, enabling people to continue their education with more flexibility. Virtual laboratories play a crucial role in Computer Science distance learning degrees, by enabling students to study at their rhythm and getting practical answers to practical problems immediately. Theoretical models such as finite automata, pushdown automata, context-free grammars, Turing machines, etc., are essential for understanding the grounds of languages and computability and are also the basis for the implementation of compilers. In this paper, a new virtual laboratory is presented, UAbALL-Automata Learning Lab, developed at Universidade Aberta (UAb), the Portuguese Open University. This virtual laboratory has already been tested in the curricular unit of Languages and Computation, with good feedback from the students. A comparison to other tools was performed showing that UAbALL is more complete in terms of tools provided.

Abstract
Green hydrogen is expected to play an important role for decarbonizing hard-to-abate sectors but faces regulatory, economic, and operational barriers. In the EU, strict renewable energy usages requirements and temporal and geographical criteria constrain green hydrogen production and complicate integration with electricity markets. Support mechanisms (SMs), such as premiums and quotas, aim to boost hydrogen production, yet their impacts on coupled electricity-hydrogen systems remain underexplored. This paper extends a previous joint electricity-hydrogen Cournot equilibrium model to represent and analyze the impact of different green hydrogen production SMs. Different SMs lead to different equilibrium models that were solved using equivalent quadratic optimization problems and applied to real-size Iberian case studies. Results reveal how different SMs influence hydrogen and electricity prices, production and emissions, highlighting trade-offs among stakeholders. The findings provide guidance for designing balanced policies that stimulate green hydrogen while minimizing unintended consequences and offer flexible tools to assess regulatory and economic interactions in emerging hydrogen markets

Abstract
Robust robotic manipulation is an essential task for the progress of automation, yet clothing handling remains a major challenge for robots. Despite currently going through rapid technological advancement, the textile industry still faces many obstacles regarding textile manipulation, which often requires a lot of testing and resources to build robust systems. Simulation is often explored as an alternative to real-life testing, but the unpredictability of this type of material makes the development of reliable simulated environments with fabrics very difficult. This paper presents an advancement made in textile models in the MuJoCo simulator by extending an existing macro for realistic rectangular cloth generation to support garments of arbitrary shapes. Both a T-shirt and a square cloth were placed in real manipulation scenarios, and the setup was replicated afterwards in simulation, taking 3D scans of the final state. Several recorded metrics show similarities between the two tests, with the simulated models mimicking most of the relevant features and behaviour of real-life scenarios. The results indicate that the proposed approach shows great potential as an alternative for a reliable simulation framework for robotic garment manipulation. © 2026 IEEE.