Abstract: Designing and constructing a highway alongside a river implies that there will be numerous interactions between them. Those points of interactions represent the locations with challenging design problems and require extensive hydrological and hydraulic assessment. Subject analyses are conducted using cutting-edge hydrological, and 1D and 2D hydraulic models that are widely acknowledged in many studies, some of which include highway design and addressing the problem of road and river intersections. This paper describes the methodology used to generate a hydrological scenario of specific probability which is to be analyzed in the design phase to minimize the effects of highway construction on the water regime by implementing measures of watercourse development and riverbank protection.

Abstract: Harvesting delicate and valuable fruits, vegetables and herbs requires a large amount of manual labor, which is less and less popular nowadays due to relatively low salaries and many health problems faced by the workforce. Traditional robotic harvesting methods, on the other hand, often struggle with the complexities of different plant structures, which results in a huge difference in the quality of harvested products, as well as in the price of hand-harvested products compared to machine-harvested ones. To address these challenges, an advanced robotic system implies a solution that integrates deep learning-based plant recognition, an adequate embedded system, and a specially designed end-effector. Deep learning, particularly Convolutional Neural Networks (CNNs), has proven effective in image processing tasks, making it suitable for accurate plant identification. The design of the end-effector is very important part and incorporates soft robotics and adaptive grippers to handle delicate produce gently, reducing damage and mimicking human touch. Finally, selecting the right hardware platform is crucial for balancing low power consumption, high accuracy, and reliability. The NVIDIA Jetson series, offers powerful GPU capabilities to accelerate deep learning inference, with each model providing different levels of performance and energy efficiency. By integrating advanced technologies, this research aims to contribute to the evolution of agrorobotics, offering a more adaptive and efficient harvesting approach, ultimately improving agricultural processes and sustainability.

Abstract: Artificial intelligence (AI) is one of the main forces behind Industry 4.0, a widely recognized paradigm that combines state-of-the-art discoveries and technology. AI enables intelligent industrial automation systems to carry out complicated automation tasks independently. Machine learning and deep learning approaches also help industries to anticipate their maintenance requirements and minimize downtime. The research and development of techniques, instruments, and algorithms that produce information and choices for AI-based systems that are understandable to humans is known as explainable artificial intelligence (XAI). In this paper an explainable artificial intelligence approach for industrial automation is considered, based on Adaptive Neuro-Fuzzy Inference System (ANFIS), which is a well-known machine learning method representing hybrid intelligent system that combines the adaptive capabilities of neural networks with the fuzzy logic reasoning. The considered XAI approach has been applied to industrial automation in the food industry, namely to the deep learning-based system for grading of apples and defective apples detection on fruit sorting lines.