Abstract
All types of sports are potential application scenarios for automatic and real-time visual object and event detection. In rink hockey, the popular roller skate variant of team hockey, it is of great interest to automatically track player movements, positions, and sticks, and also to make other judgments, such as being able to locate the ball. In this work, we present a real-time pipeline consisting of an object detection model specifically designed for rink hockey games, followed by a knowledge-based event detection module. Even in the presence of occlusions and fast movements, our deep learning object detection model effectively identifies and tracks important visual elements in real time, such as: ball, players, sticks, referees, crowd, goalkeeper, and goal. Using a curated dataset consisting of a collection of rink hockey videos containing 2525 annotated frames, we trained and evaluated the algorithm's performance and compared it to state-of-the-art object detection techniques. Our object detection model, based on YOLOv7, presents a global accuracy of 80% and, according to our results, good performance in terms of accuracy and speed, making it a good choice for rink hockey applications. In our initial tests, the event detection module successfully detected an important event type in rink hockey games, namely, the occurrence of penalties.

Abstract
This paper introduces a new variable structure controller designed for depth control of an autonomous underwater sensor platform equipped with a variable buoyancy module. To that end, the prototype linear model is presented, and a finite element-based method is used to estimate one of its parameters, the hull deformation due to pressure. To manage potential internal disturbances like hull deformation or external disturbances like weight changes, a disturbance observer is developed. An analysis of the observer steady-state estimation error in relation to input disturbances and system parameter uncertainties is developed. The locations of the observer poles according to its parameters are also identified. The variable structure controller is developed, keeping energy savings in mind. The proposed controller engages when system dynamics are unfavorable, causing the vehicle to deviate from the desired reference, and disengages when dynamics are favorable, guiding the vehicle toward the target reference. A detailed analysis determines the necessary switching control actions to ensure the system reaches the desired reference. Finally, simulations are run to compare the proposed controller's performance with that of PID-based controllers recently developed in the literature, assessing dynamic response and energy consumption under various operating conditions. Both the VBM- and propeller-actuated vehicles were evaluated. The results demonstrate that the proposed controller achieves an average energy consumption reduction of 22% compared to the next most efficient PID-based controller for the VBM-actuated vehicle, though with some impact on control performance.

Abstract

Abstract
This research contributes to the field of Human-Robot Collaboration (HRC) within dynamic and unstructured environments by extending the previously proposed Fuzzy State-Long Short-Term Memory (FS-LSTM) architecture to handle the uncertainty and irregularity inherent in real-world sensor data. Recognising the challenges posed by low-cost sensors, which are highly susceptible to environmental conditions and often fail to provide regular periodic readings, this paper introduces additional pre-processing blocks. These include two indirect Kalman filters and an additional LSTM network, which together enhance the input variables for the fuzzification process. The enhanced FS-LSTM approach is evaluated using real-world data, demonstrating its effectiveness in extracting meaningful information and accurately recognising human activities. This work underscores the potential of robotics in addressing global challenges, particularly in labour-intensive and hazardous tasks. By improving the integration of humans and robots in unstructured environments, this research contributes to the broader exploration of robotics in new societal applications, fostering connections and collaborations across diverse fields.

Abstract
By using the GRAVITY instrument with the near-infrared (NIR) Very Large Telescope Interferometer (VLTI), the structure of the broad (emission-)line region (BLR) in active galactic nuclei (AGNs) can be spatially resolved, allowing the central black hole (BH) mass to be determined. This work reports new NIR VLTI/GRAVITY interferometric spectra for four type 1 AGNs (Mrk 509, PDS 456, Mrk 1239, and IC 4329A) with resolved broad-line emission. Dynamical modelling of interferometric data constrains the BLR radius and central BH mass measurements for our targets and reveals outflow-dominated BLRs for Mrk 509 and PDS 456. We present an updated radius-luminosity (R-L) relation independent of that derived with reverberation mapping (RM) measurements using all the GRAVITY-observed AGNs. We find our R-L relation to be largely consistent with that derived from RM measurements except at high luminosity, where BLR radii seem to be smaller than predicted. This is consistent with RM-based claims that high Eddington ratio AGNs show consistently smaller BLR sizes. The BH masses of our targets are also consistent with the standard MBH-sigma* relation. Model-independent photocentre fitting shows spatial offsets between the hot dust continuum and the BLR photocentres (ranging from similar to 17 mu as to 140 mu as) that are generally perpendicular to the alignment of the red- and blueshifted BLR photocentres. These offsets are found to be related to the AGN luminosity and could be caused by asymmetric K-band emission of the hot dust, shifting the dust photocentre. We discuss various possible scenarios that can explain this phenomenon.

Abstract
Electric vehicles (EVs) are key to a sustainable future, but extending battery life is essential to reduce costs and environmental impact. Thus, this paper presents the development of an Adaptive Nonlinear Predictive Model (ANLPM), integrated with a Third Order Generalized Integrator (TOGI) flux observer, which enhances induced torque estimation and stator reactance in Permanent Magnet Synchronous Motor (PMSM) systems. The model employs a Sequential Quadratic Programming (SQP) algorithm, ensuring numerical stability and efficiency within the Model Predictive Control (MPC) framework to handle nonlinear constraints effectively. Moreover, simulation results demonstrate that the ANLPM significantly outperforms classical Adaptive Linear Predictive Models (ALPM), Seven-Dimensional LPM (SDLPM), and Proportional-Integral (PI) control strategies. It achieves marked reductions in battery discharge current and energy consumption rates. Therefore, simulation comparisons, across different scenarios, show that ANLPM reduces battery discharge current by 3% over ALPM and 44.7% over PI, while cutting energy consumption by 12.2% and 28.2%, and decreasing parallel battery cells by 14.2% and 28%, respectively. Under high temperatures, ANLPM cuts battery consumption by 45.3% and reduces cells by 43.7% compared to SDLPM, highlighting its efficiency in managing energy and extending battery life in EVs.