Abstract
Yield forecasting is of immeasurable value in modern viticulture to optimize harvest scheduling and quality management. The number of inflorescences and flowers per vine is one of the main components and their assessment serves as an early predictor, which can explain up to 85-90% of yield variability. This study introduces a sophisticated framework that integrates the benchmark of different advanced deep learning and classic image processing to automate the segmentation of grapevine inflorescences and the detection of single flowers, to achieve precise, early, and non-invasive yield predictions in viticulture. The YOLOv8n model achieved superior performance in localizing inflorescences ( F1-Score (Box) = 95.9%) and detecting individual flowers (F1-Score = 91.4%), while the YOLOv5n model excelled in the segmentation task ( F1-Score (Mask) = 98.6%). The models demonstrated a strong correlation (R-2 > 90.0%) between detected and visible flowers in inflorescences. A statistical analysis confirmed the robustness of the framework, with the YOLOv8 model once again standing out, showing no significant differences in error rates across diverse grapevine morphologies and varieties, ensuring wide applicability. The results demonstrate that these models can significantly improve the accuracy of early yield predictions, offering a noninvasive, scalable solution for Precision Viticulture. The findings underscore the potential for Computer Vision technology to enhance vineyard management practices, leading to better resource allocation and improved crop quality.

Abstract
With the global population on the rise and a declining agricultural labor force, the realm of robotics research in agriculture, such as robotic manipulators, has assumed heightened significance. This article undertakes a comprehensive exploration of the latest advancements in controllers tailored for robotic manipulators. The investigation encompasses an examination of six distinct controller paradigms, complemented by the presentation of three exemplars for each category. These paradigms encompass: (i) adaptive control, (ii) sliding mode control, (iii) model predictive control, (iv) robust control, (v) fuzzy logic control and (vi) neural network control. The article further introduces and presents comparative tables for each controller category. These controllers excel in tracking trajectories and efficiently reaching reference points with rapid convergence. The key point of divergence among these controllers resides in their inherent complexity.

Abstract
The paper presents a new control concept based on the process moment instead of the process states or the process output signal. The control scheme is based on separate control of reference tracking and disturbance rejection. The tracking control is achieved by additionally feeding the input of the process model by the scaled output signal of the process model. The advantage of such feedback is that the final state of the process output can be analytically calculated and used for control instead of the actual process output value. The disturbance rejection, including model imperfections, is controlled by feeding back the filtered difference between the process output and the model output to the process input. The performance of tracking and disturbance rejection is simply controlled by two user-defined gains. Several examples have shown that the new control method provides very good and stable tracking and disturbance rejection performance.

Abstract
Predictive maintenance (PDM) is emerging as a strong transformative tool within Industry 4.0, enabling significant improvements in the sustainability and efficiency of manufacturing processes. This in-depth literature review, which follows the PRISMA 2020 framework, examines how PDM is being implemented in several areas of the manufacturing industry, focusing on how it is taking advantage of technological advances such as artificial intelligence (AI) and the Internet of Things (IoT). The presented in-depth evaluation of the technological principles, implementation methods, economic consequences, and operational improvements based on academic and industrial sources and new innovations is performed. According to the studies, integrating CDM can significantly increase machine uptime and reliability while reducing maintenance costs. In addition, the transition to PDM systems that use real-time data to predict faults and plan maintenance more accurately holds out promising prospects. However, there are still gaps in the overall methodologies for measuring the return on investment of PDM implementations, suggesting an essential research direction. © 2025

Abstract

Abstract
Most human time is spent indoors, and due to the pandemic, monitoring indoor air quality (IAQ) has become more crucial. In this study, an IoT (Internet of Things) architecture is implemented to monitor IAQ parameters, including CO2 and particulate matter (PM). An ESP32-C6-based device is developed to measure sensor data and send them, using the MQTT protocol, to a remote InfluxDBv2 database instance, where the data are stored and visualized. The Python 3.11 scripting programming language is used to automate Flux queries to the database, allowing a more in-depth data interpretation. The implemented system allows to analyze two measured scenarios during sleep: one with the door slightly open and one with the door closed. Results indicate that sleeping with the door slightly open causes CO2 levels to ascend slowly and maintain lower concentrations compared to sleeping with the door closed, where CO2 levels ascend faster and the maximum recommended values are exceeded. This demonstrates the benefits of ventilation in maintaining IAQ. The developed system can be used for sensing in different environments, such as schools or offices, so an IAQ assessment can be made. Based on the generated data, predictive models can be designed to support decisions on intelligent natural ventilation systems, achieving an optimized, efficient, and ubiquitous solution to moderate the IAQ.