Hrčak - Portal of scientific journals of Croatia
Not a member yet
321498 research outputs found
Sort by
Visibility Issues of Right-Hand Drive Vehicles on the Right-Side Traffic
No full textThis paper analyses road safety issues related to the operation of right-hand drive vehicles in right-side operated traffic conditions. The problem is statistically valid, especially in regions where the states border with road traffic organised on different sides, or due to economic considerations, when it is much cheaper to purchase a vehicle adapted to the traffic of the other side. This article focuses on visibility issues of different driving positions in the case of the most dangerous overtaking manoeuvre by geometrically based situation analysis and safe distance to opposite obstacle evaluation. An additional part of the study is the functional testing of special indirect vision equipment in real traffic conditions. This is related to the technical regulations in force in the European Union Member States for vehicles adapted to the traffic in the other direction when constantly participating in traffic. An overview of these technical requirements, as well as the existing requirements not to restrict the free market of vehicles, is also covered in the paper
Exploring Cooperative Lane Change Decisions in Vehicle-to-Infrastructure – A Potential Conflict Analysis Approach
No full textVehicle-to-infrastructure (V2I) technology enables information interaction between vehicles and among vehicles and infrastructure, significantly enhancing the efficiency of lane-changing processes and stabilising traffic flow. Current research primarily focuses on single lane-changing events in fixed micro-level scenarios or studies involving small-scale vehicle fleets, neglecting the randomness of lane-changing vehicle arrivals and potential conflicts during lane-changing. This paper proposes a lane-changing decision model based on potential conflict analysis, specifically tailored to mandatory lane-changing requirements in high-density traffic conditions. The model comprises sub-models for lane-changing decision triggering, influence range calculation and lane-changing priority determination, capable of dynamically adjusting the lane-changing sequence, mitigating lane-changing conflicts, and improving driving safety and traffic efficiency. Simulation experiments indicate that, when compared to lane-changing patterns in real-world traffic scenarios, this model reduces travel time by 23.30%, delays by 21.95% and the number of stops by 23.84%, thereby providing a novel approach for lane-changing decision-making and control in V2I environments
Predictors of Affirmative Attitudes toward the Use of Artificial Intelligence in Science within a Mertonian Ethos Framework
No full textThis article examines public attitudes towards the use of artificial intelligence (AI) tools in science, starting from the concept of epistemological rights of citizens and Merton’s institutional ethos of science. Various predictors were examined to explain citizens’ positive attitudes towards the impact of AI on objectivity, quality, efficiency, judgment, and ethics of scientific practice. Regression analysis showed that significant predictors are sex, age, frequency of AI use, trust in AI tools, and perceptions of similarity between the human brain and computers. Also, the strongest single predictor of affirmative attitudes is the belief that AI can make ethically correct decisions, which indicates the presence of the phenomenon of dataism in a part of the population. Despite general concerns about the possible misuse of technology, citizens express moderate trust in the ability of scientific institutions to use AI tools reliably and ethically, implicitly confirming the credibility of science in the digital age
Enhanced Structure-from-Motion 3D Reconstruction through Deep Learning Feature Fusion and Optimization
No full textThis paper presents Mixum, a novel 3D reconstruction framework for Structure-from-Motion (SfM), which combines traditional feature extraction and matching techniques with deep learning-based optimization. The Mixum framework enhances the accuracy of feature matching and eliminates redundant feature points. Additionally, the integration with PixSfM, a deep-learning accuracy optimization algorithm, further reduces reprojection error and enhances multi-view consistency. Experiments on multiple public datasets reveal that Mixum significantly improves 3D reconstruction density and reduces reprojection error by up to 23%, demonstrating its applicability for complex scenes in applications like cultural heritage preservation, virtual reality, and autonomous navigation
A Graph Network for High-speed Railway Operation Risk Prevention and Control
No full textWith the rapid expansion of China's high-speed railway (HSR) network, safety concerns in HSR operations have garnered increasing attention. Currently, various railway departments have devised numerous feasible risk control plans. However, these plans predominantly exist in an unstructured textual format, leading to challenges in automation, standardization, content updates, and cross-departmental collaboration. To address these limitations, this study introduces a novel graph-based Risk Scenario Decision (RSD) model, which systematically structures unstructured emergency risk scenarios into an explicit graphical format. The RSD model utilizes graph theory principles and incorporates LLM for precise knowledge extraction and alignment. This approach significantly enhances automation, consistency, and efficiency in railway operational risk management. By transforming traditional risk control plans into a graph-based network, the RSD model facilitates efficient decision-making and rapid response, thereby improving the overall management and effectiveness of HSR operational risk control. Experimental validation demonstrates the high accuracy (up to 98.38%) of the RSD construction process. Ultimately, this research provides a robust, interpretable, and automated framework that substantially enhances proactive risk management in HSR operations, ensuring greater safety and operational efficiency
Advancements in Photogrammetric Modelling: Boundary Marking for Improved Accuracy
No full textPhotogrammetric modelling, a widely used technique for reconstructing 3D models from photographic data, relies on achieving high geometric accuracy to minimize post-processing efforts. This study proposes an enhancement to the photogrammetric reconstruction workflow through the introduction of the Boundary Marking Technique. By visibly marking the edges surrounding object surfaces, this method aims to improve edge definition and overall reconstruction quality. The technique was applied to both metal and plastic objects of identical geometry. 3D models were generated using close-range photogrammetry for both marked and unmarked versions, and deviation analyses were conducted against their CAD references. A total of 16 analyses comprising one 3D and three 2D per object were performed using computer-aided comparison tools. Results indicate that the Boundary Marking Technique significantly enhances geometric accuracy, particularly along edges and adjacent surfaces, ultimately improving model topology and dimensional fidelity across the entire reconstruction
Investigation of Dimensional Deviations in SLA 3D-Printed Cylindrical Features
No full textDimensional accuracy is a critical challenge across many additive manufacturing (AM) technologies, directly impacting the assembly of 3D-printed parts. This study investigates these deviations for stereolithography (SLA) on a desktop printer. To analyse the influence of part geometry and the key process parameter of layer thickness, 54 specimens featuring cylindrical protrusions and holes (5, 10, and 20 mm nominal diameters) were produced with three distinct thickness values (25, 50, and 100 µm). A Coordinate Measuring Machine (CMM) was used to measure dimensional and geometric characteristics, and Analysis of Variance (ANOVA) was performed for statistical analysis. The results revealed a systematic deviation trend: protrusions were consistently oversized (up to +195 µm), while holes were consistently undersized (down to ‒250 µm). Both diameter and layer thickness were found to be statistically significant parameters influencing these deviations. Based on these findings, a compensation strategy was developed to correct CAD models by accounting for mean deviation and process variability. The strategy was successfully verified through a practical example, demonstrating its effectiveness in achieving a controlled clearance fit (26 µm). This approach provides a practical methodology for improving the functional accuracy of SLA parts, crucial for a wide range of engineering applications where precise component assembly is required, including the production of functional prototypes and custom spare parts
An Optimized Belief Propagation List Decoding for Polar Codes with Dynamic Flipping
No full textIn the context of polar codes, belief propagation list (BPL) decoding has demonstrated a substantial enhancement in parallel decoding performance, achieving high throughput. Nevertheless, a performance gap still exists between the advanced BPL decoding and successive cancellation list (SCL) decoding methods. Moreover, existing bit-flipping strategies are inefficient in accurately identifying erroneous bit positions, leading to elevated computational complexity and limiting their practical applicability. This study introduces an optimized BPL decoding algorithm with dynamic flipping (OBPL-DF) aimed at bridging this performance gap while reducing computational demands. Initially, an efficient decoding scheme is proposed to further decrease computational complexity in practical scenarios. Subsequently, to improve the precision of error position detection, a partial cyclic redundancy check (CRC) code is employed on erroneous codewords. Finally, a dynamic flipping metric is developed within the bit-flipping strategy, allowing the selection of flipped positions to be guided by this novel metric rather than being confined to a predetermined set. Simulation results demonstrate that the OBPL-DF algorithm surpasses the performance of existing BPL flip (BPLF) decoding techniques and approaches that of enhanced SCL decoding, all while achieving significantly lower latency
Early Tithonian ammonites, microfacies, biostratigraphy, and biogeography from the Mészkemence section (Zengővárkony, Mecsek Mountains, South Hungary)
No full textA rich but poorly preserved Lower Tithonian faunal assemblage is reported from the aban-doned quarry/lime kiln at Zengővárkony (Mecsek Mountains, Hungary). Some Lower Titho-nian Tethyan ammonite zones (Hybonotum, Semiforme, and Fallauxi) are recognised. The Fallauxi Zone is recorded for the first time from the region. The pelagic fauna is dominated by ammonites and their aptychi. Brachiopods, belemnites, and very rarely bivalves are ac-cessory elements of the fauna. The mollusc fauna comprises 167 specimens that represent 18 genera and 13 species. Sublithacoceras rhodaniforme, Pseudopallasiceras toucasi, Bi-plisphinctes pseudocolubrinus, and Physodoceras cf. widerai are recorded for the first time from the Mecsek Mountains. The ammonite fauna has a typical Mediterranean character. Based on cluster analysis, the Mecsek ammonite fauna is closest to the ammonite assem-blages of the Transdanubian Range (Hungary). Based on quantitative analysis, the fauna is similar to the ammonite assemblages of the Apennines Veneto – Trento (Southern Alps, Italy), and Rogozník (Pieniny Klippen Belt). The microfacies is the Globochaete – Sacco-coma microfacies. Large-sized pygopid brachiopods, serpulid tube worms on inoceramid shells, and benthic foraminifera indicate optimal bottom conditions. Pelagic faunal elements (Saccocoma sp., globuligerinid planktonic foraminifera) are also present, and calcareous dinoflagellate cysts (Carpistomiosphaera malmica/tithonica) are also recorded
Book review of Handbook of Artificial Intelligence in Higher Education
No full textThis is a review of Handbook of Artificial Intelligence in Higher Education, edited by Stefan Popenici, Jürgen Rudolph, Fadhil Ismail and Shannon Tan (Edward Elgar Publishing)