The video game industry faces significant challenges in content creation, with AAA games requiring extensive time and resources. This research addresses these challenges through the development of a Unity module for automatic 3D kitchen model generation using a Machine Learning-based Procedural Kitchen Generation (PKG) model. The module significantly reduces the time needed for game-level design, achieving designs over five times faster than traditional methods. A comparative study shows that the module produces results comparable to industry-standard tools in terms of user preference. This Unity module offers promising potential for commercial use, streamlining the design process and enhancing efficiency in game development.
This article investigates the different challenges and phases of building an ecosystem to utilise data centre waste heat. Data centres consume an increasing amount of energy, and currently, in many cases, excess heat is not utilised beyond heating the data centre property. The research examines how a circular economy can be integrated into the operations of a data centre’s waste heat business ecosystem and how a business ecosystem based on utilising waste heat from a data centre should be built. The research is based on a literature review, interviews with experts and stakeholders in the case ecosystem and a survey. Based on the research, a concept for a business ecosystem based on the utilisation of waste heat from a pre-existing data centre perspective is modelled, and the concept outlines things to be considered when building such an ecosystem. Key findings were the challenges, opportunities and prerequisites of developing a data centre waste heat ecosystem and the roadmap of constructing one.
This study investigates how sustainable supply chain management (SCM) practices and information technology (IT) drive firm performance in small and medium enterprises (SMEs), with a focus on the mediating roles of supply chain agility (SCA) and sustainable supply chain integration (SSCI). Drawing on data from 219 managers and supply chain experts in Iranian SMEs, structural equation modeling (SEM) revealed that both sustainable SCM practices and IT adoption significantly enhance firm performance directly and indirectly through SCA and SSCI. Sustainable SCM practices demonstrated a robust influence on SSCI (β = 0.35, p < 0.001), SCA (β = 0.49, p < 0.001), and firm performance (β = 0.44, p < 0.001), while IT similarly impacted SSCI (β = 0.34, p < 0.001), SCA (β = 0.39, p < 0.001), and performance (β = 0.21, p < 0.001). Critically, SCA and SSCI emerged as pivotal mediators, explaining 48% of the variance in firm performance. The findings underscore that agility and integration transform sustainable and technological investments into competitive outcomes, enabling SMEs to navigate dynamic markets. Practical insights highlight the need for firms to prioritize cross-departmental coordination, IT-driven customer insights, and adaptive decision-making frameworks.
In the modern soccer world, also referred to as football, analytics has become integral IT serves a role akin to assistant coaches, contributing significantly to the analysis of team and individual player performance during both games and training sessions. Despite the abundance of advanced technologies, there is still a need for concrete measurements in soccer analytics. A prime example is Expected Goals (xG), a widely embraced metric that goes beyond mere scorelines to offer in-depth insights into player and team dynamics, and has proven its value in recent years. The identification of key features from a dataset to predict xG is a critical aspect of machine learning research applied to soccer analytics. However, prior investigations have either overlooked certain crucial features or failed to recognize their significance. This study proposes a novel approach by incorporating features related to coaches, coaching tenure, and tactics, which have the potential to improve predictive accuracy and provide actionable insights for teams and analysts. Using a dataset of 2,917 observations covering the top five European leagues, we applied regression-based machine learning models, employing a preprocessing pipeline and k-fold cross-validation to ensure robust evaluation. The findings reveal that the xG values for the English Premier League (EPL) surpass those of the other four leagues studied, with an average xG of approximately 1.93. This indicates that, on average, teams in the EPL tend to have a higher expected goal count per match than teams in other top European leagues. This superiority can be attributed to various factors, such as the higher average quality of teams and players, tactical nuances, and the overall competitiveness inherent within the EPL compared with the other leagues under scrutiny. In our study, we introduced new features such as coaches, coaching durations (years), and tactics. Incorporating these features enhanced the model performance across MSE, EVS, and R², thereby demonstrating the efficacy of our approach. Lasso and Ridge Regression models achieved improved predictive accuracy, with EVC and R² reaching up to 96%, while the Decision Tree model showed a nearly 6% reduction in the MSE.
Inefficient municipal solid waste management represents a significant obstacle on the transition towards a circular economy in densely populated urban environments, such as the district of San Juan de Lurigancho, Lima, Peru. This study presents the implementation of advanced technologies, specifically smart containers integrated with sensors and real-time monitoring systems, to optimize waste collection. This resulted in decrements in operating costs and unrecycled waste by 20% and 27%, respectively, whilst increasing citizen engagement by 18% regarding segregation and recycling activities, highlighting positive changes in community behavior. However, the present technological infrastructure, technical training and necessary funds are all insufficient, which presents limitations on the deployment and scalability of these measures. This study thusly indicates the necessity to elaborate public policies, environmental education and engagement strategies to guarantee system stability. These findings both provide a replicable model to address these issues and offer a roadmap for the integration of smart technologies in sustainable city planning, bridging technological innovation and sustainability.
The Geological Overview Map of Bosnia and Herzegovina, at a scale of 1:200,000, is the first comprehensive and detailed geological map meticulously created for the entire country. This paper examines the mathematical and geodetic basis of the map and evaluates its positional accuracy. A method was developed to compute coordinate corrections for arbitrary points based on the differences between the Austro-Hungarian and Yugoslav triangulations, in order to determine the position of any point on the old geological map with respect to the current Bosnian-Herzegovinian state coordinate system. After resolving the issues related to map projection, mathematical and geodetic foundations of the map and coordinate systems, the georectification procedure was applied to the map scan. The results show that the TPS transformation effectively minimizes geometric distortions caused by scanning errors, deformations of the map's base material, and systematic errors from the original map elaboration. After applying the georectification, most remaining positional distortions were attributed to random errors.
Additive technologies enable the production of elements with complex geometry, which opens up new possibilities in the design of lightweight and strong structures. However, the additive manufacturing process is associated with limitations that must be taken into account at the design stage. This article presents the results of research on the design and analysis of skeletal structures intended for additive manufacturing. CAD models of the robot gripper were developed and then optimized using three different approaches: the use of a shape generator, generative design, and finite element analysis (FEM). The aim of the study was to compare the strength of a robot gripper that was lightened using three different design approaches, with a fixed mass reduction of 40%, in order to determine which method provides the greatest mechanical strength. Experimental studies included numerical analysis and real tests conducted on a testing machine. The test results showed that generative design allows for the most optimal material distribution, which translates into a more favorable strength-to-weight ratio compared to other methods of reducing the weight of the structure.
This research investigates the influence of copper plate dimensions on temperature distribution during TIG (Tungsten Inert Gas) welding of Cu-ETP (Electrolytic Tough Pitch Copper). When welding copper and lightly alloyed copper alloys with high thermal conductivities, the appropriate type of current and shielding gas must be selected to ensure sufficient heat input. This high heat input compensates for the rapid heat dissipation and cooling in the localized weld zone. The required heat input is determined by the dimensions of the copper sheets to be welded. Experimental measurements using thermocouples were performed on plates with identical thicknesses but varying dimensions, while numerical simulations via ANSYS complemented the analysis. The results highlight the significant impact of plate dimensions on heat dissipation and temperature gradients, providing valuable insights for improving welding settings.
This project focused on the development of a machine learning model to classify rice plants based on their growth stages, specifically identifying whether the plants are in the "Raw" or "Ripe" stage. The research was conducted using a dataset obtained from Roboflow, which provided annotated images of rice plants. The dataset was divided into training, testing, and validation subsets to ensure the model's robustness and generalization capability. The project involved a comprehensive data preparation process, which included consolidating images into class-based folders, handling file conflicts, and stratifying the dataset into appropriate splits for training, testing, and validation. Several convolutional neural networks (CNN) architectures were explored, including ResNet50, InceptionV3, and MobileNetV2, each leveraging transfer learning from pre-trained models on the ImageNet dataset. ResNet50 achieved an accuracy of 87.3% with a log loss of 0.33, demonstrating good performance but with some misclassifications between similar classes. InceptionV3 outperformed the other models, achieving an accuracy of 95.1% and a log loss of 0.13, indicating superior classification capability and better calibration of predicted probabilities. MobileNetV2 also performed well with an accuracy of 93.5% and a log loss of 0.22, offering a balance between accuracy and computational efficiency. The results highlight InceptionV3 as the most effective model for this task, with a strong ability to differentiate between the rice growth stages. The findings underscore the importance of model selection and data preparation in developing accurate and reliable machine-learning models for agricultural applications. The project demonstrates the potential of CNNs in improving agricultural practices through precise crop monitoring and classification.
The research focused on the effects of abrasive type, grain size, and wheel structure on surface roughness during external cylindrical longitudinal grinding of D2 steel. Experimental research was conducted using a full factorial design. The measured surface roughness values, depending on the combination of input parameters, ranged from 0.22 to 1.31 μm, corresponding to ISO roughness grades N4 to N7. Among the factors examined, grain size had the most significant impact on surface roughness, followed by the type of abrasive, while the grinding wheel structure had the least effect. The best surface roughness is obtained by utilizing corundum with the highest aluminium oxide content as an abrasive, along with the smallest grain size and a dense structure of the grinding wheel. Additionally, modelling of the grinding process was performed. The validity of the obtained regression equation for predicting surface roughness was confirmed through five additional verification experiments, which supported the accuracy of the modelling.
Maize, a staple cereal crop globally, faces significant yield challenges due to foliar diseases. This study presents an innovative real-time maize leaf disease detection system integrating a YOLOv8 deep learning model with a custom-designed Unmanned Aerial Vehicle (UAV). The system targets three prevalent maize diseases: Grey leaf spot, common rust, and northern leaf blight. Trained on 10,239 images, the YOLOv8 model, comprising 168 layers and 3,006,233 parameters, achieves 8.1 GFLOPs with an inference speed of 6.3ms per image. Performance evaluation reveals a mean Average Precision (mAP@0.5) of 0.836 and mAP@0.5:0.95 of 0.681 across all classes, with high accuracy for common rust (Precision: 0.953, Recall: 0.979) and grey leaf spot (Precision: 0.919, Recall: 0.892). The custom UAV, designed for agricultural surveying, features a thrust-to-weight ratio of 5.85, ensuring stable flight with the 140g payload of a Raspberry Pi 5 and Camera Module V2. With a total weight of 605g and a 3S 5000mAh LiPo battery, the drone achieves an estimated flight time of 5.3 minutes, balancing survey coverage with real-time disease detection capabilities. The integration of this high-performance model with an efficient UAV platform represents a significant advancement in precision agriculture, enabling early disease intervention and targeted treatment strategies, thus promoting sustainable farming practices through optimized resource allocation and potential reduction in pesticide usage.
One of the most important problems today is the formation of greenhouse gases, mainly caused by the emission of flue gases in thermal power plants: carbon dioxide (CO2), sulfur dioxide (SO2) and nitrogen oxides (NOx) ("acid rain"). CO2 emissions are inevitable and related to the fuel combustion process (carbon capture techniques). Sulfur dioxide emissions are reduced by cleaning coal and by applying wet and dry flue gas desulfurization. The injection of absorbents can be applied in the combustion chamber or in the flue gases. Nitrogen oxide emissions are reduced by proper fuel selection and advanced solutions in the construction of the combustion chamber. Burners with low NOx emissions are used.
The subject of the article is to investigate the potential to adjust the stiffness of 3D printed plastic implants to match bone stiffness by analyzing design parameters and mechanical properties, considering the spongy structure of the bone. The study consisted of theoretical and practical parts, including original research. In the theoretical part, the definition of implants, the materials used for their production, and the requirements they must meet were discussed, as well as 3D printing methods, with particular emphasis on the FFF method. The static compression test was also described. In the practical part, original research was conducted to verify the possibilities of adjusting the stiffness of the implants. Two 3D models of the L2 lumbar vertebra with different vertebral body thicknesses and internal lattice structures were developed based on literature dimensions. Thirty models with three different lattice densities and two wall thicknesses were printed and subjected to a static compression test. The results showed that an increase in lattice density increased the compression strength of the samples. Samples with thin walls exhibited lower compression strength compared to those with thick walls, regardless of the spongy lattice density. In each group of samples, the results were consistent, indicating good repeatability of the prepared samples. The findings suggest that appropriately selecting the lattice density and wall thickness can significantly improve the mechanical strength of bone implants, meeting the compression strength criteria for the spongy tissue of lumbar vertebrae.
The purpose of the proposed study is to provide the required data transmission characteristics in the Industrial Internet of Things (IIoT) system. Our efforts are primarily aimed at increasing the probability of message delivery, as well as reducing the likelihood of their duplication in wireless networks based on the Message Queuing Telemetry Transport (MQTT) protocol. The article offers a scheme and description of the automated control system, which is based on the use of original models to assess the probability of delivery and the likelihood of duplicate messages. The control parameters are the quality of service (QoS) level and the number of allowed retransmissions (Nretry). Conducting computational experiments and deploying the system in a real environment have shown that its application makes it possible to improve the requirements for the probability of message delivery by 7.2 percent and the probability of their duplication by 8.6 percent compared to existing adaptive QoS approaches. Due to the results achieved, the proposed system is recommended for use in IIoT networks, which place increased demands on the reliability of data delivery.
The combat power of military troops is highly dependent on resilient supply chain management. Recent trends in the application of offensive drone technologies elevate the risks of supply chain failures along the last mile and demand pre-emptive resilience strategies. This paper presents a concept for increased resilience in military logistics that considers requirements elicited with methods from process engineering in collaboration with experts from the Austrian military logistics school. On a basic level, the concept is defined by a) decentralization of repositories at the battalion and company level achieved by cluster-forming technologies, and b) cross-docking for short-time supply points. The presented concept depends on the successful integration of cluster-forming technologies, but if done properly, it has the potential to improve the resilience of modern military logistics.
Humans need a comfortable and healthy home to carry out activities properly. Low-income people's homes in the Highlands have poor indoor air quality due to stoves that produce smoke. People also use stoves as a heating tool to achieve thermal comfort. This study aimed to investigate thermal comfort and poor air quality in Wooden Homes in the Highlands. The study used a mixed method. Qualitative methods were used to obtain data on the thermal comfort of occupants. Quantitative methods were used to obtain data on indoor air quality. The study results showed opposing aspects between thermal comfort in people in the highlands and the fulfillment of clean and healthy air. The study also created solutions for thermal comfort and clean air quality by simulating ventilation placement.
This study analyzes the influence of photopolymer plate hardness and mounting tape on print quality in flexography, with a focus on evaluating the optical and colorimetric parameters of prints on white PLA film. The experiment was carried out in six variations of printing: two hardness of polymer plates in combination with three hardness of mounting tapes. A test image was printed using solvent cyan ink and all key print quality parameters were analyzed, including ink density, tonal values, and CIELAB values in accordance to ISO 12647-6:2012 standard. The results showed that combinations using the softer plate with harder tapes provided the optimal balance between halftone precision, solid density, and colorimetric accuracy. Combinations using the harder plate resulted in lower ink transfer but also less dot gain. Microscopic analysis revealed that combinations of harder plates with softer tapes led to lower ink film uniformity. The study provides useful guidelines for optimizing flexographic printing on PLA substrates.
According to some previous researches the English language knowledge, turned out to be important, both in professional and private life of adult population, including students, from the Republic of Croatia. This paper presents a study of the influence of knowledge and application of the English language on the choice of students’ mobility mode during their stay in foreign cities. The research is based on a survey of students studying civil engineering and architecture from two Croatian towns - Osijek and Rijeka. The survey included a total of 221 students from both towns, and based on the created databases, a prediction model of the choice of mobility modality was developed (based on the data gathered from the students from Osijek), which was validated (based on the data gathered from the students from Rijeka). The mobility modality prediction model was developed using a neural network and with 27 input parameters related to the knowledge and use of the English language, it achieved a correlation of 77% (R2 0.57, MAE 0.16), and the validation results gave a correlation of 63% (R2 0.39, MAE 0.18). The influence of certain input parameters on the choice of mobility modality of students in foreign cities was analyzed by applying statistical tools in each urban area separately. The main limitation concerns both the sample size and the unequal distribution of respondents between the two cities, which should be taken into account when interpreting the comparison.
The application of Machine Learning (ML) methods represents a significant aspect in the advancement of Industry 4.0. The creation of an appropriate data set for these applications has been identified as the most time-consuming step in the underlying end-to-end pipeline. One of the major obstacles in this process step is to bridge the gap between business understanding and data understanding. To address this challenge, we propose a novel methodology to bridge this gap based on a systematic literature review. Our methodology begins with the construction of an ontology that depicts the underlying manufacturing process along with its parameters. We then show how this ontology can be utilized to deepen the understanding of the manufacturing process. Subsequently, we demonstrate how appropriate target variables for ML-models and suitable data sources can be determined with the support of our ontology. We further elucidate our methodology through a real-world example.
Milkrun systems serve as the backbone of material supply for multi-line or parallel manufacturing production, ensuring synchronized and efficient logistics. Despite their importance, route optimization is still largely done manually, leading to inefficiencies. If too few routes are used, transportation time increases, and individual vehicles become overloaded, which is a common issue in current practices. Finding the optimal number of routes is a complex NP-hard problem, requiring advanced computational approaches. While assignment matrices are often available and useful for direct transport allocation, they are not used for optimizing milkrun loops. To address this, a heuristic-based model has been developed that optimizes the number of routes while considering travel length and vehicle capacity. The model and case study integrates genetic algorithms to explore feasible solutions efficiently, balancing the trade-off between too many and too few routes. The proposed approach improves both cost-effectiveness and service reliability in production logistics.
