Hasil untuk "Machine design and drawing"

Ahmed Alqaraawi, M. Schuessler, Philipp Weiß et al.

Convolutional neural networks (CNNs) offer great machine learning performance over a range of applications, but their operation is hard to interpret, even for experts. Various explanation algorithms have been proposed to address this issue, yet limited research effort has been reported concerning their user evaluation. In this paper, we report on an online between-group user study designed to evaluate the performance of "saliency maps" - a popular explanation algorithm for image classification applications of CNNs. Our results indicate that saliency maps produced by the LRP algorithm helped participants to learn about some specific image features the system is sensitive to. However, the maps seem to provide very limited help for participants to anticipate the network's output for new images. Drawing on our findings, we highlight implications for design and further research on explainable AL In particular, we argue the HCI and AI communities should look beyond instance-level explanations.

Rodrigo Benenson, Stefan Popov, V. Ferrari

Manually annotating object segmentation masks is very time consuming. Interactive object segmentation methods offer a more efficient alternative where a human annotator and a machine segmentation model collaborate. In this paper we make several contributions to interactive segmentation: (1) we systematically explore in simulation the design space of deep interactive segmentation models and report new insights and caveats; (2) we execute a large-scale annotation campaign with real human annotators, producing masks for 2.5M instances on the OpenImages dataset. We released this data publicly, forming the largest existing dataset for instance segmentation. Moreover, by re-annotating part of the COCO dataset, we show that we can produce instance masks 3x faster than traditional polygon drawing tools while also providing better quality. (3) We present a technique for automatically estimating the quality of the produced masks which exploits indirect signals from the annotation process.

Jingxiang Feng, Peiran Zhao, Haoran Zheng et al.

Robust object detection in autonomous driving is challenged by inherent limitations of conventional frame-based cameras, such as motion blur and limited dynamic range. In contrast, event-based cameras, which operate asynchronously and capture rapid changes with high temporal resolution and expansive dynamic range, offer a promising augmentation. While the previous research on event-based object detection has predominantly focused on algorithmic enhancements via advanced preprocessing and network optimizations to improve detection accuracy, the practical engineering and integration challenges of deploying these sensors in real-world systems remain underexplored. To address this gap, our study investigates the integration of event-based cameras as a complementary sensor modality in autonomous driving. We adapted a conventional frame-based detection model (YOLOv8) for event-based inputs by training it on the GEN1 dataset, achieving a mean average precision (mAP) of 70.1%, a significant improvement over previous benchmarks. Additionally, we developed a real-time object detection pipeline optimized for event-based data, integrating it into the CARLA simulation environment and ROS for system prototyping. The model was further refined using transfer learning to better adapt to simulation conditions, and the complete pipeline was validated across diverse simulated scenarios to address practical challenges. These results underscore the feasibility of incorporating event cameras into existing perception systems, paving the way for their broader deployment in autonomous vehicle applications.

Ślusarczyk Beata, Grondys Katarzyna, Sałek Robert

The current preselection vehicle weighing control system in Poland is not functional enough in many respects to effectively identify and eliminate overloaded heavy vehicles travelling on national roads. The concept of implementing automatic weigh-in-motion (WIM) systems could bring numerous economic, social, and environmental benefits for both investors and road users. At the same time, the development of vehicle weighing control systems may pose challenges to the operations of enterprises in the transport industry. To determine how TSL (Transport-Shipping-Logistics) companies in Poland perceive the potential of implementing the WIM system, a survey was conducted among them. Statistical tests were used to analyse the results, which revealed that transport enterprises have a positive attitude toward the implementation of the WIM system and the potential benefits it offers, even if they have a tendency to take risks associated with overloading cargo.

Athanasios Mamakos, Dominik Rose, Anastasios Melas et al.

Some European Heavy Duty (HD) vehicle manufacturers have adopted Open Crankcase Ventilation (OCV) systems to improve reliability and performance. The emission compliance of HD vehicles both during certification and In-Service Conformity (ISC) testing need to also account for the crankcase ventilation. Despite that, the contribution of crankcase emissions to the overall emissions profile of modern trucks remains underexplored. This study experimentally characterizes the crankcase emissions of a Euro VI Step E HD truck equipped with an OCV system under controlled conditions on a chassis dynamometer. Emissions were measured over the World Harmonized Vehicle Cycle (WHVC) and an ISC-compliant driving cycle at two test cell temperatures. The results indicate that crankcase emissions account for up to 4% and 8% of the current regulatory limits for nitrogen oxides (NO_x) and 23 nm solid particle number (SPN₂₃), respectively. The tightening of NO_x limits under Euro 7 regulations would increase these contributions to approximately 11%. SPN₁₀ crankcase emissions were found to be on the order of 10¹¹ (11% of the Euro 7 limit). Real-time SPN₁₀ and SPN₂₃ measurements revealed that the fraction of nanosized particles increases significantly during cold start, suggesting increased oil combustion within the cylinder. These findings highlight the need to refine crankcase emissions measurement procedures within regulatory frameworks. A systematic investigation of measurement setups and ageing effects, taking into account variations in OCV system designs and piston ring wear, is essential to determine whether characterization during certification is sufficient or if ISC testing throughout the vehicle’s useful life will be required.

Yi Yu, Yufei Wang, Wenhan Yang et al.

Recent advancements in deep learning-based compression techniques have demonstrated remarkable performance surpassing traditional methods. Nevertheless, deep neural networks have been observed to be vulnerable to backdoor attacks, where an added pre-defined trigger pattern can induce the malicious behavior of the models. In this paper, we propose a novel approach to launch a backdoor attack with multiple triggers against learned image compression models. Drawing inspiration from the widely used discrete cosine transform (DCT) in existing compression codecs and standards, we propose a frequency-based trigger injection model that adds triggers in the DCT domain. In particular, we design several attack objectives that are adapted for a series of diverse scenarios, including: 1) attacking compression quality in terms of bit-rate and reconstruction quality; 2) attacking task-driven measures, such as face recognition and semantic segmentation in downstream applications. To facilitate more efficient training, we develop a dynamic loss function that dynamically balances the impact of different loss terms with fewer hyper-parameters, which also results in more effective optimization of the attack objectives with improved performance. Furthermore, we consider several advanced scenarios. We evaluate the resistance of the proposed backdoor attack to the defensive pre-processing methods and then propose a two-stage training schedule along with the design of robust frequency selection to further improve resistance. To strengthen both the cross-model and cross-domain transferability on attacking downstream CV tasks, we propose to shift the classification boundary in the attack loss during training. Extensive experiments also demonstrate that by employing our trained trigger injection models and making slight modifications to the encoder parameters of the compression model, our proposed attack can successfully inject multiple backdoors accompanied by their corresponding triggers into a single image compression model.

Cheng Qian, Ming Zhang, Yuanping Nie et al.

As the machine learning-related technology has made great progress in recent years, deep neural networks are widely used in many scenarios, including security-critical ones, which may incura great loss when DNN is compromised. Starting from introducing several commonly used bit-flip methods, this paper concentrates on bit-flips attacks aiming DNN and the corresponding defense methods. We analyze the threat models, methods design, and effect of attack and defense methods in detail, drawing some helpful conclusions about improving the robustness and resilience of DNN. In addition, we point out several drawbacks to existing works, which can hopefully be researched in the future.

Xiangpei Gu, Nicola Bianchi, Zhuoran Zhang

Variable flux reluctance machines (VFRMs) are increasingly attracting research interest due to their magnetless and robust brushless structure. Under the modulation effect of the airgap permeance, the VFRM operates with a series of field harmonics, distinguishing it from conventional AC synchronous machines. This paper deals with the analysis and preliminary design of the VFRM from the perspective of multiple working airgap field harmonics. Firstly, the spatial and temporal order of the working field harmonics are defined. The systematic winding theory, including the unified star of slots and winding factor calculation method, is established to consider all these working harmonics. Then, an average torque model is built and simplified. The key role of 1st-order rotor permeance, 1st- and 3rd-order polarized stator permeance is deduced. The relationship between key parameters and average torque is computed, providing a guideline for the preliminary design of the VFRM.

Tutak Magdalena

The article addresses a critical and timely issue: improving safety in underground coal mining. The primary objective of the paper was to develop a research methodology based on modelling studies to identify and assess the state of methane hazards during mining operations. To achieve this, structural modelling of the physical and chemical phenomena occurring in mining regions was conducted using Computational Fluid Dynamics. The core research was performed using the finite volume method on a real longwall exploitation site ventilated by a Y-system. This approach enabled the determination of methane and oxygen concentration distributions in the mining region and goafs, treated as a porous and permeable medium. Based on these findings, potential fire and/or methane explosion hazard zones were identified in the goaf. The model test results underwent a validation process, comparing them with actual measurements. The determined errors were within an acceptable range, confirming the accuracy of the developed model of the mining region and the phenomena within it. Furthermore, the model was used to predict the locations of zones at risk of fire and/or methane explosion in the goafs, particularly in areas with potentially increased gas emissions. The results clearly demonstrate the significant potential of using model studies to diagnose and forecast methane hazards in underground mining operations. Identifying these potential danger zones allows for the implementation of preventive measures to reduce the likelihood of dangerous incidents.

Ramirez-Zuñiga Eyder James, Castro-Silva Hugo Fernando, Velásquez-Pérez Torcoroma et al.

The coffee sector stands as a cornerstone of Colombia’s economy, ranking third in the nation’s export portfolio. Despite the Colombian coffee esteemed global reputation, it has yet to fully exploit its potential for diversification into differentiated products. Present agro-industrial paradigms emphasize trade and sustainable, efficient agricultural practices, underscoring the imperative for innovation across production, marketing, and distribution channels. This study aims to pinpoint the pivotal innovation factors within coffee farm production processes. To this end, a sample of 66 coffee farms was selected through simple random sampling. Drawing from the 2018 Oslo model, innovation types associated with sustainable specialty coffee certifications were delineated. Within this framework, seven fundamental factors emerged for investigation: economic, social, environmental, production, knowledge, technology, and change management. Through cluster analysis, it became evident that economic, environmental, knowledge, technological, and change management factors are indispensable for fortifying the coffee industry.

Bruce W. Jo

High-speed capsular vehicles are firstly suggested as an idea by Elon Musk of Tesla Company. Unlike conventional high-speed trains, capsular vehicles are individual vessels carrying passengers and freight with the expected maximum speed of near 1200 [km/h] in a near-vacuum tunnel. More individual vehicle speed, dispatch, and position control in the operational aspect are expected over connected trains. This numerical study and investigation evaluate and analyze inter-distance control and their characteristics for high-speed capsular vehicles and their operational aspects. Among many aspects of operation, the inter-distance of multiple vehicles is critical toward passenger/freight flow rate and infrastructural investment. In this paper, the system’s equation, equation of the motion, and various characteristics of the system are introduced, and in particular control design parameters for inter-distance control and actuation are numerically shown. As a conclusion, (1) Inter-distance between vehicles is a function of error rate and second car start time, the magnitude range is determined by second car start time, (2) Inter-distance fluctuation rate is a function of error rate and second car start time, however; it can be minimized by choosing the correct second car start time, and (3) If the second car start time is chosen an integer number of push-down cycle time at specific velocity error rate, the inter-distance fluctuation can be zero.

Yuyao Lu, Geng Yang, Yajing Shen et al.

Abstract In the past decade, the global industry and research attentions on intelligent skin-like electronics have boosted their applications in diverse fields including human healthcare, Internet of Things, human–machine interfaces, artificial intelligence and soft robotics. Among them, flexible humidity sensors play a vital role in noncontact measurements relying on the unique property of rapid response to humidity change. This work presents an overview of recent advances in flexible humidity sensors using various active functional materials for contactless monitoring. Four categories of humidity sensors are highlighted based on resistive, capacitive, impedance-type and voltage-type working mechanisms. Furthermore, typical strategies including chemical doping, structural design and Joule heating are introduced to enhance the performance of humidity sensors. Drawing on the noncontact perception capability, human/plant healthcare management, human–machine interactions as well as integrated humidity sensor-based feedback systems are presented. The burgeoning innovations in this research field will benefit human society, especially during the COVID-19 epidemic, where cross-infection should be averted and contactless sensation is highly desired.

Hąbek Patrycja, Lavios Juan J., Krupah Edward

Sustainable supplier development helps to improve mutually the supplier’s as well as the buying company sustainability performance. The producer could choose guidance, compliance or capacity building activities to develop its supplier or implement them all. This paper aims to present how the car producers practice sustainable supplier development taking into account different types of approaches and implementation tools. The authors applied content analysis to investigate approaches of six car producers from EU member states. The data was collected from the sustainability reports and complemented with the available information of the supplier sustainability requirements and the code of conduct of each car producer. The findings revealed that analysed car producers use similar approaches to develop their suppliers in the context of sustainability. All of them use mix of activities from all identified categories and collaborate within industry initiatives devoted to spread sustainability in supply chain.

Wei-Hsun Lee, Teng-Jyun Liang, Hsuan-Chih Wang

In recent years, the wireless sniffing technique (WST) has become an emerging technique for collecting real-time traffic information. The spatiotemporal variations in wireless signal collection from vehicles provide various types of traffic information, such as travel time, speed, traveling path, and vehicle turning proportion at an intersection, which can be widely used for traffic management applications. However, three problems challenge the applicability of the WST to traffic information collection: the transportation mode classification problem (TMP), lane identification problem (LIP), and multiple devices problem (MDP). In this paper, a WST-based intelligent traffic beacon (ITB) with machine learning methods, including SVM, KNN, and AP, is designed to solve these problems. Several field experiments are conducted to validate the proposed system: three sensor topologies (X-type, rectangle-type, and diamond-type topologies) with two wireless sniffing schemes (Bluetooth and Wi-Fi). Experiment results show that X-type has the best performance among all topologies. For sniffing schemes, Bluetooth outperforms Wi-Fi. With the proposed ITB solution, traffic information can be collected in a more cost-effective way.

Tramy Nguyen, Nicholas Walczak, Daniel Sumorok et al.

Communicating information about experimental design among a team of collaborators is challenging because different people tend to describe experiments in different ways and with different levels of detail. Sometimes, humans can interpret missing information by making assumptions and drawing inferences from information already provided. Doing so, however, is error-prone and typically requires a high level of interpersonal communication. In this paper, we present a tool that addresses this challenge by providing a simple interface for incremental formal codification of experiment designs. Users interact with a Google Docs word-processing interface with structured tables, backed by assisted linking to machine-readable definitions in a data repository (SynBioHub) and specification of available protocols and requests for execution in the Open Protocol Interface Language (OPIL). The result is an easy-to-use tool for generating machine-readable descriptions of experiment designs with which users in the DARPA SD2 program have collected data from 80 208 samples using a variety of protocols and instruments over the course of 181 experiment runs.

Masayoshi Tanishita, Takashi Kobayashi

The Worldwide harmonized Light duty Test Procedure saw its light first as the United Nations Economic Commission for Europe Global Technical Regulation in 2017. However, it remains unclear how much the deviation is between the actual and test fuel economy. In this study, we analyzed the deviation between the actual and test (JC08 and WLTC) fuel economy and examined how well regional characteristics such as average travel speed and temperature could explain the deviation using 182–1035 drivers and 19–52 car models data in Japan. As a result, (1) more than a 30% discrepancy was observed between the actual and JC08 mode test fuel economy, and the higher the test fuel economy, the larger the deviation; (2) regarding WLTC mode fuel economy, the deviation is 19% and constant regardless of the test fuel economy; (3) average travel speed and temperature can explain approximately 8% of the discrepancy.

Sirama Sirama, Simon Parekke

The 1.5 inch pipe rolling machine utilizes an electric motor to drive the pipe holder and a 2ton jack as a pipe press. Although the roller machine is widely used in conventional workshops by using human power, it causes the pipe rolling time to be long and the machine user gets tired quickly. This is due to lack of resources to think about developing and improving the performance of these machines. Therefore, the focus of this research is the design of a 1.5 inch pipe rolling machine using an electric motor as a drive and a 2 ton jack as a pipe press. This research aims to produce a design and machine for 1.5 inch pipe rolling and to help accelerate the pipe rolling process and reduce worker fatigue. This pipe rolling machine consists of several components, namely, the engine frame, electric motor, gearbox reducer, chain-sprocket, shaft, pipe drive wheel, hydraulic pump and so on. The working process of this machine begins by preparing a 1.5 inch x 3 meter pipe then the pipe is positioned above the drive wheel, lowers the upper pressure wheel to the desired depth with a 2ton jack power, turn on the electric motor and press the button to roll the pipe back and forth and so on. up to pipe roll size as needed. The stages in making this machine are design, detailed drawing, manufacturing process, quality control, assembly and testing, data retrieval and data analysis. The research resulted in drawings of the design and manufacture of the machine, the calculation of the rotational speed of the machine for optimal rolling obtained 20.46 rpm and an average pipe rolling time of 6.80 minutes for one workpiece. This machine can still be optimized by adding a workpiece movement limiter when rolling.

Soheil Habibian, Mehdi Dadvar, Behzad Peykari et al.

An amendment to this paper has been published and can be accessed via the original article.

Nobutsuna Endo

Abstract Few physical models of oral and laryngeal systems for human speech movement exist for computer or mechanical simulators. In particular, a robot tongue mechanism that fully reproduces the deformation motion of the human tongue is lacking. The human tongue is an aggregate of muscles that is devoid of a skeleton. It possesses only a small hyoid. A mechanism that can drive and control the deformation of a soft body, such as the human tongue, along multiple degrees of freedom has not been realized to date. To solve this problem, a wire-pulling mechanism with embedded soft tubes is proposed. Using this mechanism, a flexible tongue that can be deformed along multiple degrees of freedom without breaking the wire is achieved. A prototype planar mechanism with two degrees of freedom that is capable of contraction and bending was fabricated. A deformation model that assumes a piecewise constant curvature (PCC) was formulated. Deformation tests confirmed that the prototype is capable of contraction and bending movements that are consistent with those of the model. Variations in the error with respect to the hardness of the deformable part are discussed, and the limits of the deformation model based on the PCC assumption are presented.

Axel Diewald, Clemens Kurz, Prasanna Venkatesan Kannan et al.

Automotive radar sensors play a vital role in the current development of autonomous driving. Their ability to detect objects even under adverse conditions makes them indispensable for environment-sensing tasks in autonomous vehicles. As their functional operation must be validated in-place, a fully integrated test system is required. Radar Target Simulators (RTS) are capable of executing end-of-line, over-the-air validation tests by looping back a received and afterward modified radar signal and have been incorporated into existing Vehicle-in-the-Loop (ViL) test beds before. However, the currently available ViL test beds and the RTS systems that they consist of lack the ability to generate authentic radar echoes with respect to their complexity. The paper at hand reviews the current development stage of the research as well as commercial ViL and RTS systems. Furthermore, the concept and implementation of a new test setup for the rapid prototyping and validation of ADAS functions is presented. This represents the first-ever integrated radar validation test system to comprise multiple angle-resolved radar target channels, each capable of generating multiple radar echoes. A measurement campaign that supports this claim has been conducted.