Hasil untuk "Machine design and drawing"

Yugo Matsui, Shota Yoshida, Chihiro Nakagawa et al.

Abstract This study investigates the effectiveness of an augmented reality (AR) goggle-based information display system for autonomous driving in enhancing the safety of a stand-up personal mobility vehicle (PMV) during autonomous driving. Stand-up PMVs, due to their high center of gravity, pose an increased risk of falling—particularly when the user is distracted or engaged in tasks that reduce situational awareness. Previous research on the use of AR goggles during autonomous driving to optimize travel time utilization revealed potential safety risks associated with visual distraction and impaired balance. To address these concerns, we developed a simplified information display system that utilizes AR goggles to provide the rider with advance notice of upcoming turns, mitigating discomfort and risk from unexpected vehicle maneuvers. We conducted driving experiments with three participants, using a four-wheeled stand-up PMV equipped with an autonomous driving system. The experiment involved three routes with varying turn timings to minimize anticipatory movements. We evaluated the system's effectiveness in reducing fall risk by analyzing foot pressure data and administering a questionnaire survey. The results show that the AR goggle-based information display system for autonomous driving effectively reduces fall risk, particularly when users are engaged in tasks requiring focused attention. The system subconsciously alerts users to upcoming turns, even when their attention is diverted. Positive questionnaire feedback further highlights the system’s potential to enhance both comfort and safety during autonomous PMV operation.

Jamal Alotaibi

The latest developments in Advanced Driver Assistance Systems (ADAS) have greatly enhanced the comfort and safety of drivers. These technologies can identify driver abnormalities like fatigue, inattention, and impairment, which are essential for averting collisions. One of the important aspects of this technology is automated traffic accident detection and prediction, which may help in saving precious human lives. This study aims to explore critical features related to traffic accident detection and prevention. A public US traffic accident dataset was used for the aforementioned task, where various machine learning (ML) models were applied to predict traffic accidents. These ML models included Random Forest, AdaBoost, KNN, and SVM. The models were compared for their accuracies, where Random Forest was found to be the best-performing model, providing the most accurate and reliable classification of accident-related data. Owing to the black box nature of ML models, this best-fit ML model was executed with explainable AI (XAI) methods such as LIME and permutation importance to understand its decision-making for the given classification task. The unique aspect of this study is the introduction of explainable artificial intelligence which enables us to have human-interpretable awareness of how ML models operate. It provides information about the inner workings of the model and directs the improvement of feature engineering for traffic accident detection, which is more accurate and dependable. The analysis identified critical features, including sources, descriptions of weather conditions, time of day (weather timestamp, start time, end time), distance, crossing, and traffic signals, as significant predictors of the probability of an accident occurring. Future ADAS technology development is anticipated to be greatly impacted by the study’s conclusions. A model can be adjusted for different driving scenarios by identifying the most important features and comprehending their dynamics to make sure that ADAS systems are precise, reliable, and suitable for real-world circumstances.

Marián Gogola, Ján Ondruš

This study investigates drivers’ visual attention strategies during naturalistic urban driving using mobile eye-tracking (Pupil Labs Neon). A sample of experienced drivers participated in a realistic traffic scenario to examine fixation behaviour under varying traffic conditions. Non-parametric analyses revealed substantial variability in fixation behaviour attributable to driver identity (H(9) = 286.06, <i>p</i> = 2.35 × 10⁻⁵⁶), stimulus relevance (H(7) = 182.64, <i>p</i> = 5.40 × 10⁻³⁶), and traffic density (H(4) = 76.49, <i>p</i> = 9.64 × 10⁻¹⁶). Vehicles and pedestrians elicited significantly longer fixations than lower-salience categories, reflecting adaptive allocation of visual attention to behaviourally critical elements of the scene. Compared with the fixed-rule method, which produced inflated anomaly rates of 7.23–14.84% (mean 12.06 ± 2.71%), the DSGP algorithm yielded substantially lower and more stable rates of 1.62–3.33% (mean 2.48 ± 0.53%). The fixed-rule approach over-classified anomalies by approximately 4–6×, whereas DSGP more accurately distinguished contextually appropriate fixations from genuine attentional deviations. These findings demonstrate that fixation behaviour in driving is strongly shaped by individual traits and environmental context, and that driver-specific modelling substantially improves the reliability of attention monitoring. Therefore DSGP framework offers a robust, personalised alternative evaluated at the proof-of-concept level to fixed thresholds and represents a promising direction for enhancing driver-state assessment in future ADAS.

Ivan Janoško, Martin Krasňanský

This paper presents a detailed experimental evaluation of a newly developed diesel fuel additive, specifically formulated to enhance the energy efficiency and emission characteristics of internal combustion engine (ICE) vehicles, with particular emphasis on its applicability to aging vehicle fleets. Diesel engines are known for producing significant amounts of harmful emissions, necessitating the development of effective mitigation strategies. One such approach involves the use of fuel additives. The additive under investigation is a proprietary formulation containing 1-(N,N-bis(2-ethylhexyl)aminomethyl)-1,2,4-triazole and other compounds. To the best of our knowledge, this specific additive composition has not yet been tested or reported in the existing scientific literature. To evaluate the real-world contribution of such additives, a comprehensive set of controlled measurements was conducted in a certified chassis dynamometer laboratory, including an exhaust gas analyser and supplementary diagnostic equipment. The testing protocol comprised repeated measurement cycles under identical driving conditions, both without and with the additive. Exhaust gas concentrations of CO₂, CO, and NOx were continuously monitored. Simultaneously, fuel consumption and engine performance were tracked over a cumulative driving distance of 2000 km. The results indicate measurable improvements across all monitored domains. CO₂ emissions decreased by 4.57%, CO by 14.29%, and NOx by 3.12%. Fuel consumption was reduced by 4.79%, while engine responsiveness and power delivery showed moderate but consistent enhancements. These improvements are attributed to more complete combustion and an increased cetane number enabled by the additive’s chemical structure. The findings support the adoption of advanced additive technologies as part of transitional strategies towards low-emission transportation systems.

Gonçalves Bruno S. F., Lopes Erik Teixeira, Fernandes Leonor Taborda et al.

The article analyses studies on the impact of the COVID-19 pandemic on outpatient services in a large hospital, aiming to provide insights for resource management amidst disruptive events. The objectives include identifying challenges and proposing solutions to optimize service delivery and address spatial constraints using discrete-event simulation. Utilizing a case study approach, the research employs simulation as a key methodology to analyse outpatient service scenarios. Scenarios are generated by combining different probabilities of patient return to check-in with various team parameterizations. The researchers analysed historical data and key performance indicators from the simulation. The study focuses on a collaborative approach with the hospital team to ensure the relevance and applicability of proposed solutions. The research identifies bottlenecks induced by social distancing measures, particularly in patient reception and check-in areas. Uneven service distribution throughout the day leads to a misallocation of resources and reduction of available physical space. Telemedicine emerges as a significant response, effectively addressing both service optimization and physicians’ workload despite spatial constraints. Additionally, the study underscores the role of simulation in crisis decision-making for hospital operations management. Practical applications emanating from the study emphasize the need for healthcare institutions to adopt adaptable strategies and leverage simulation tools for effective resource management during disruptive events. Hospital administrators can draw insights to inform resource reallocation and workflow optimization, with a focus on negotiating flexible scheduling and exploring telemedicine to enhance service delivery.

M. Emre Sahin, Edoardo Altamura, Oscar Wallis et al.

We present Qiskit Machine Learning (ML), a high-level Python library that combines elements of quantum computing with traditional machine learning. The API abstracts Qiskit's primitives to facilitate interactions with classical simulators and quantum hardware. Qiskit ML started as a proof-of-concept code in 2019 and has since been developed to be a modular, intuitive tool for non-specialist users while allowing extensibility and fine-tuning controls for quantum computational scientists and developers. The library is available as a public, open-source tool and is distributed under the Apache version 2.0 license.

Christoforos N. Spartalis, Theodoros Semertzidis, Petros Daras et al.

We introduce SAFEMax, a novel method for Machine Unlearning in diffusion models. Grounded in information-theoretic principles, SAFEMax maximizes the entropy in generated images, causing the model to generate Gaussian noise when conditioned on impermissible classes by ultimately halting its denoising process. Also, our method controls the balance between forgetting and retention by selectively focusing on the early diffusion steps, where class-specific information is prominent. Our results demonstrate the effectiveness of SAFEMax and highlight its substantial efficiency gains over state-of-the-art methods.

Mohammad Javad Sobouti, Amirhossein Mohajerzadeh, Haitham Y. Adarbah et al.

Unmanned aerial vehicles (UAVs) have emerged as a promising technology to enhance the performance and functionality of mobile networks. UAVs can act as flying base stations, relays, or users to provide wireless services to ground users or devices. However, the optimal placement and trajectory design of UAVs in mobile networks is a challenging problem, as it involves multiple objectives, constraints, and uncertainties. In this paper, we provide a comprehensive survey of the state-of-the-art research on UAV placement and trajectory optimization in cellular networks. We first introduce the main objectives and challenges of UAV placement and trajectory optimization, such as maximizing coverage, throughput, energy efficiency, or reliability, while minimizing interference, delay, or cost. We also examine the primary models and assumptions employed for UAV placement and trajectory optimization, including channel models, mobility models, network architectures, and constraints. Additionally, we discuss the main methods and algorithms employed for UAV placement and trajectory optimization. These include optimization techniques, heuristic algorithms, machine learning approaches, and distributed solutions. Analytical results, numerical simulations, or experimental tests are further discussed as the main performance metrics and evaluation methods used for UAV placement and trajectory optimization. We also highlight the main applications and scenarios of UAV placement and trajectory optimization, such as cellular offloading, emergency communications, or aerial base stations. Finally, we identify some open problems and future research directions on UAV placement and trajectory optimization in cellular networks.

Świt Grzegorz, Ulewicz Małgorzata, Pała Robert et al.

This article presents a comprehensive improvement in the experimental analysis of cracking processes in smooth and sharp V-notched samples taken from gas transport pipelines, utilizing the acoustic emission (AE) method. The research aimed to establish a robust correlation between the failure mechanisms of uni-axially tensile samples and the distinct characteristics of AE signals for enhanced quality management in pipeline integrity. The study encompassed materials from two different straight pipe sections, encompassing both long-term used materials and new, unused materials. Through the application of the k-means grouping method to AE signal analysis, we achieved the identification of AE signal parameters characteristic of various stages of the material destruction process. This advancement introduces a significant improvement in monitoring and managing the operational safety of pipeline networks, offering a methodology that leverages advanced acoustic emission signal analysis. The outcomes present significant implications for the pipeline industry by proposing methods to enhance safety systems and more effectively manage the integrity and quality of gas infrastructure.

Tim Z. Xiao, Robert Bamler, Bernhard Schölkopf et al.

Motivated by the progress made by large language models (LLMs), we introduce the framework of verbalized machine learning (VML). In contrast to conventional machine learning (ML) models that are typically optimized over a continuous parameter space, VML constrains the parameter space to be human-interpretable natural language. Such a constraint leads to a new perspective of function approximation, where an LLM with a text prompt can be viewed as a function parameterized by the text prompt. Guided by this perspective, we revisit classical ML problems, such as regression and classification, and find that these problems can be solved by an LLM-parameterized learner and optimizer. The major advantages of VML include (1) easy encoding of inductive bias: prior knowledge about the problem and hypothesis class can be encoded in natural language and fed into the LLM-parameterized learner; (2) automatic model class selection: the optimizer can automatically select a model class based on data and verbalized prior knowledge, and it can update the model class during training; and (3) interpretable learner updates: the LLM-parameterized optimizer can provide explanations for why an update is performed. We empirically verify the effectiveness of VML, and hope that VML can serve as a stepping stone to stronger interpretability.

Sourojit Ghosh, Sarah Coppola

Traditional and currently-prevalent pedagogies of design perpetuate ableist and exclusionary notions of what it means to be a designer. In this paper, we trace such historically exclusionary norms of design education, and highlight modern-day instances from our own experiences as design educators in such epistemologies. Towards imagining a more inclusive and sustainable future of design education, we present three case studies from our own experience as design educators in redesigning course experiences for blind and low-vision (BLV), deaf and hard-of-hearing (DHH) students, and students with other disabilities. In documenting successful and unsuccessful practices, we imagine what a pedagogy of care in design education would look like.

Zachary Dana, Ahmed Ammar Naseer, Botros Toro et al.

Chronic kidney disease (CKD) is a significant public health challenge, often progressing to end-stage renal disease (ESRD) if not detected and managed early. Early intervention, warranted by silent disease progression, can significantly reduce associated morbidity, mortality, and financial burden. In this study, we propose a novel approach to modeling CKD progression using a combination of machine learning techniques and classical statistical models. Building on the work of Liu et al. (2023), we evaluate linear models, tree-based methods, and deep learning models to extract novel predictors for CKD progression, with feature importance assessed using Shapley values. These newly identified predictors, integrated with established clinical features from the Kidney Failure Risk Equation, are then applied within the framework of Cox proportional hazards models to predict CKD progression.

Wesley C. Fraser

The scientific study of the Solar System's minor bodies ultimately starts with a search for those bodies. This chapter presents a review of the use of machine learning techniques to find moving objects, both natural and artificial, in astronomical imagery. After a short review of the classical non-machine learning techniques that are historically used, I review the relatively nascent machine learning literature, which can broadly be summarized into three categories: streak detection, detection of moving point sources in image sequences, and detection of moving sources in shift and stack searches. In most cases, convolutional neural networks are utilized, which is the obvious choice given the imagery nature of the inputs. In this chapter I present two example networks: a Residual Network I designed which is in use in various shift and stack searches, and a convolutional neural network that was designed for prediction of source brightnesses and their uncertainties in those same shift-stacks. In discussion of the literature and example networks, I discuss various pitfalls with the use of machine learning techniques, including a discussion on the important issue of overfitting. I discuss various pitfall associated with the use of machine learning techniques, and what I consider best practices to follow in the application of machine learning to a new problem, including methods for the creation of robust training sets, validation, and training to avoid overfitting.

Ruda Zhang, Negin Alemazkoor

In system analysis and design optimization, multiple computational models are typically available to represent a given physical system. These models can be broadly classified as high-fidelity models, which provide highly accurate predictions but require significant computational resources, and low-fidelity models, which are computationally efficient but less accurate. Multi-fidelity methods integrate high- and low-fidelity models to balance computational cost and predictive accuracy. This perspective paper provides an in-depth overview of the emerging field of machine learning-based multi-fidelity methods, with a particular emphasis on uncertainty quantification and optimization. For uncertainty quantification, a particular focus is on multi-fidelity graph neural networks, compared with multi-fidelity polynomial chaos expansion. For optimization, our emphasis is on multi-fidelity Bayesian optimization, offering a unified perspective on multi-fidelity priors and proposing an application strategy when the objective function is an integral or a weighted sum. We highlight the current state of the art, identify critical gaps in the literature, and outline key research opportunities in this evolving field.

Jonathan Stray

Researchers are starting to design AI-powered systems to automatically select and summarize the reports most relevant to each analyst, which raises the issue of bias in the information presented. This article focuses on the selection of relevant reports without an explicit query, a task known as recommendation. Drawing on previous work documenting the existence of human-machine feedback loops in recommender systems, this article reviews potential biases and mitigations in the context of intelligence analysis. Such loops can arise when behavioral “engagement” signals such as clicks or user ratings are used to infer the value of displayed information. Even worse, there can be feedback loops in the collection of intelligence information because users may also be responsible for tasking collection. Avoiding misalignment feedback loops requires an alternate, ongoing, non-engagement signal of information quality. Existing evaluation scales for intelligence product quality and rigor, such as the IC Rating Scale, could provide ground-truth feedback. This sparse data can be used in two ways: for human supervision of average performance and to build models that predict human survey ratings for use at recommendation time. Both techniques are widely used today by social media platforms. Open problems include the design of an ideal human evaluation method, the cost of skilled human labor, and the sparsity of the resulting data.

Zhenge Jia, Dawei Li, Cong Liu et al.

The first ACM/IEEE TinyML Design Contest (TDC) held at the 41st International Conference on Computer-Aided Design (ICCAD) in 2022 is a challenging, multi-month, research and development competition. TDC'22 focuses on real-world medical problems that require the innovation and implementation of artificial intelligence/machine learning (AI/ML) algorithms on implantable devices. The challenge problem of TDC'22 is to develop a novel AI/ML-based real-time detection algorithm for life-threatening ventricular arrhythmia over low-power microcontrollers utilized in Implantable Cardioverter-Defibrillators (ICDs). The dataset contains more than 38,000 5-second intracardiac electrograms (IEGMs) segments over 8 different types of rhythm from 90 subjects. The dedicated hardware platform is NUCLEO-L432KC manufactured by STMicroelectronics. TDC'22, which is open to multi-person teams world-wide, attracted more than 150 teams from over 50 organizations. This paper first presents the medical problem, dataset, and evaluation procedure in detail. It further demonstrates and discusses the designs developed by the leading teams as well as representative results. This paper concludes with the direction of improvement for the future TinyML design for health monitoring applications.

Georgia Papacharalampous, Hristos Tyralis, Nikolaos Doulamis et al.

Merging satellite and gauge data with machine learning produces high-resolution precipitation datasets, but uncertainty estimates are often missing. We addressed the gap of how to optimally provide such estimates by benchmarking six algorithms, mostly novel even for the more general task of quantifying predictive uncertainty in spatial prediction settings. On 15 years of monthly data from over the contiguous United States (CONUS), we compared quantile regression (QR), quantile regression forests (QRF), generalized random forests (GRF), gradient boosting machines (GBM), light gradient boosting machine (LightGBM), and quantile regression neural networks (QRNN). Their ability to issue predictive precipitation quantiles at nine quantile levels (0.025, 0.050, 0.100, 0.250, 0.500, 0.750, 0.900, 0.950, 0.975), approximating the full probability distribution, was evaluated using quantile scoring functions and the quantile scoring rule. Predictors at a site were nearby values from two satellite precipitation retrievals, namely PERSIANN (Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks) and IMERG (Integrated Multi-satellitE Retrievals), and the site's elevation. The dependent variable was the monthly mean gauge precipitation. With respect to QR, LightGBM showed improved performance in terms of the quantile scoring rule by 11.10%, also surpassing QRF (7.96%), GRF (7.44%), GBM (4.64%) and QRNN (1.73%). Notably, LightGBM outperformed all random forest variants, the current standard in spatial prediction with machine learning. To conclude, we propose a suite of machine learning algorithms for estimating uncertainty in spatial data prediction, supported with a formal evaluation framework based on scoring functions and scoring rules.

Yuta Ishikawa, Hiroyuki Nabae, Gen Endo et al.

Abstract Artificial multiarticular musculoskeletal systems consisting of serially connected links driven by monoarticular and multiarticular muscles, which are often inspired by vertebrates, enable robots to elicit dynamic, elegant, and flexible movements. However, serial links driven by multiarticular muscles can cause unstable motion (e.g., buckling). The stability of musculoskeletal mechanisms driven by antagonistic multiarticular muscles depends on the muscle configuration, origin/insertion of muscles, spring constants of muscles, contracting force of muscles, and other factors. We analyze the stability of a multi-serial-link mechanism driven by antagonistic multiarticular muscles aiming to avoid buckling and other undesired motions. We theoretically derive the potential energy of the system and the stable condition at the target point, and validate the results through dynamic simulations and experiments. This paper presents the static stability criteria of serially linked robots, which are redundantly driven by monoarticular and multiarticular muscles, resulting in the design and control guidelines for those robots.

Corejova Tatiana, Jucha Peter, Padourova Anna et al.

Society, companies and institutions are involved in a digital transformation that can be pervaded in various industries or sectors, and this also applies to communication, sales and distribution channels. The possibilities of e-commerce have also increased and world trade has been further developed. In 2020, more than two billion people bought goods or services over the Internet. Customer satisfaction depends on the solution of the last mile process, the method of picking up shipments as well as the time and place of picking up the shipment. The most common forms of off-premises delivery are automated parcel locker or machine (APM) and pick-up and drop-off delivery (PUDO). The aim of the paper is to analyse the level of the PUDO and APM network in European countries and in the V4 countries with regard to the size of the country and the population. For this purpose, it was necessary to focus on determining the population per 1 PUDO and the number of inhabitants per 1 APM in individual European countries and subsequently in the V4 countries. The obtained data were processed and recalculated in Excel. The results showed that within European countries the best values were achieved by Finland with 526 inhabitants per 1 PUDO and Spain with 188 inhabitants per 1 APM. Regarding the V4 countries, the Czech Republic achieved the best value in the case of inhabitants on PUDO with 729 inhabitants per 1 PUDO and in the case of APM Poland with 3,184 inhabitants per 1 APM.

Yu Yamauchi, Yuichi Ambe, Hikaru Nagano et al.

Abstract Continuum robots can enter narrow spaces and are useful for search and rescue missions in disaster sites. The exploration efficiency at disaster sites improves if the robots can simultaneously acquire several pieces of information. However, a continuum robot that can simultaneously acquire information to such an extent has not yet been designed. This is because attaching multiple sensors to the robot without compromising its body flexibility is challenging. In this study, we installed multiple small sensors in a distributed manner to develop a continuum-robot system with multiple information-gathering functions. In addition, a field experiment with the robot demonstrated that the gathered multiple information has a potential to improve the searching efficiency. Concretely, we developed an active scope camera with sensory functions, which was equipped with a total of 80 distributed sensors, such as inertial measurement units, microphones, speakers, and vibration sensors. Herein, we consider space-saving, noise reduction, and the ease of maintenance for designing the robot. The developed robot can communicate with all the attached sensors even if it is bent with a minimum bending radius of 250 mm. We also developed an operation interface that integrates search-support technologies using the information gathered via sensors. We demonstrated the survivor search procedure in a simulated rubble environment of the Fukushima Robot Test Field. We confirmed that the information provided through the operation interface is useful for searching and finding survivors. The limitations of the designed system are also discussed. The development of such a continuum robot system, with a great potential for several applications, extends the application of continuum robots to disaster management and will benefit the community at large.