Kyutacar : Kyushu Institute of Technology Academic Repository
Not a member yet
9746 research outputs found
Sort by
Maximum Torque Per Ampere Control Implemented Hardware System for Synchronous Reluctance Motor
Get PDFNowadays, motor control systems are widely used in the industrial world. Especially, induction motors (IMs) and Permanent Magnet Synchronous motors (PMSMs) are commonly used. However they have some disadvantages, for example, IM have less efficiency and PMSM use rare earth for rotors. Synchronous Reluctance Motors (SynRMs) are rare earth-free motors with high efficiency, that are drawing attention. There is a growing need for a control system that can drive them with high efficiency and precision. Here we implement Maximum Torque Per Ampere (MTPA) Control in the FPGA hardware control system for speed control of SynRM and confirm the validity of the system.journal articl
A Support System for a Visually Impaired Person Finding Bus Route Numbers Employing MY VISION
Get PDFA bus is not a very convenient means for a visually impaired person because of the difficulty in identifying its route number, although it is an economical tool for travel. This paper proposes a method of detecting and identifying a bus route number using the MYVISION system which employs an ego camera worn by a visually impaired person. The method finds a frontal area of a bus approaching a bus stop using the video which the camera provides by using optical flow and random forest employing Haar-like features. It then extracts the upper destination panel area followed by the detection of a route number at the right-hand side of the panel. Finally, the detected route number is identified by template matching. In the experiment, various kinds of videos containing the buses of a bus company were captured at different places and different weather conditions, and the effectiveness of the proposed method was shown. The method is now planned to be applied to the busses of various bus companies.conference pape
Study on Sharing Electricity using Photovoltaic Panels and Storage Batteries in Housing Complexes
Get PDFAfter the Great East Japan Earthquake, the household sector of public welfare is promoting the introduction of distributed energy systems by diversifying energy sources and applying local energy use as one of the energy supply measures in case of a disaster. This study introduced an energy-sharing system in the housing complexes to examine whether each house with different family composition and life patterns (i.e., different energy use patterns) could use energy more efficiently. The target energy in the housing complexes was set to electricity, which was generated by photovoltaic panels and stored in storage batteries. The strategy for stable electricity supply and profit generation was as follows: (1) electricity generated by photovoltaic panels is consumed first in the housing complexes; (2) the remaining electricity is stored in a large-capacity storage battery for the operation of the cooling and heating system; (3) afterward, electricity is sold directly to nearby housing complexes at a lower price than the supply price of electricity companies and at a higher price than when sold to electricity companies. The calculation results show that the profit from the sale of surplus electricity and the reduction rate of CO2 emissions were evaluated. The annual electricity purchase was 87 MWh, which decreased by 52% due to the introduction of the electricity-sharing system. Annual electricity sales were 433 MWh. The annual profit from selling surplus electricity directly to nearby houses was 1.14 times higher than selling to electricity companies. The CO2 emission reduction rate was 56.2%.journal articl
Input-to-state stabilization of endemic models for uncertainty of transmission, inflows, and immunity waning
No full textTo mitigate the impact of infectious diseases spread, feedback decision that effectively adjusts the control amount making use of current data has been anticipated. In general, feedback control design is based on models, which are inherently subject to inaccuracy and uncertainty. Control theory seeks robustness guarantees that do not rely on the model perfection which is usually required for prediction purposes. To provide such a key, this paper deals with uncertainty of disease transmission and waning immunity as well as uncertain inflows from neighboring regions. New feedback control laws are proposed to achieve robustness in the framework of input-to-state stabilization (ISS) by governing societal activity levels, vaccination, and isolation. The control addresses endemic situations, which are more practical and mathematically much harder than disease-free situations. To go beyond Jacobian linearization and local analysis, the proposed control covers the entire space of population variables by articulating the achievable globalness mathematically. The preceding ISS-based studies cannot cope with waning immunity no matter how small the waning rate is since it gives rise to supply and dissipation in different growth orders in their formulation. This paper demonstrates how a Lyapunov function and control laws can be constructed to coordinate the orders.journal articl
Interactive System for Creating Attractive Poses Using Pose Features and Statistical Models
Get PDFCreating attractive poses is crucial in generating works of art. However, designing these poses is challenging because of a lack of clear definitions or rules for attractive poses. Thus the artist must design through experience and trial and error. In this study, we propose a system for creating new attractive poses based on a set of sample poses for a class of attractive poses. Pose features for some parts of the body, rather than the entire body, make the pose attractive. By combining such multiple sample pose features, the proposed method creates a novel attractive pose that is dissimilar to any of the sample poses. Using our system, a new attractive pose is created by interactively deforming based on the pose features of the sample poses. Our system constructs 2D latent spaces that map the sample poses for each pose feature for the user to find the pose feature to apply to the editing pose.conference pape
Sign Language Recognition using Bidirectional Reservoir Computing
Get PDFSign language recognition (SLR) facilitates communication between deaf and hearing individuals. Deep learning is widely used to develop SLR-based systems; however, it is computationally intensive and requires substantial computational resources, making it unsuitable for resource-constrained devices. To address this, we propose an efficient sign language recognition system using MediaPipe and an echo state network (ESN)-based bidirectional reservoir computing (BRC) architecture. MediaPipe extracts hand joint coordinates, which serve as inputs to the ESN-based BRC architecture. The BRC processes these features in both forward and backward directions, efficiently capturing temporal dependencies. The resulting states of BRC are concatenated to form a robust representation for classification. We evaluated our method on the Word-Level American Sign Language (WLASL) video dataset, achieving a competitive accuracy of 57.71% and a significantly lower training time of only 9 seconds, in contrast to the 55 minutes and 38 seconds required by the deep learning-based Bi-GRU approach. Consequently, the BRC-based SLR system is well-suited for edge devices.conference pape
Low-Cost Human Activity Recognition Using Enhanced Reservoir Computing Architecture
Get PDF九州工業大学博士（工学）1 Introduction| 2 Literature Review| 3 Reservoir Computing for Sign Language Recognition| 4 Reservoir Computing for Endotracheal Activity Recognition| 5 Reservoir Computing for Inner Speech Recognition using EEG Signals| 6 Conclusions, Future Works and Emerging Research DirectionsHuman activity recognition (HAR) is a vital area of research in pattern recognition and computer vision. Its practical applications in daily life, particularly those aimed at improving quality of life, have attracted widespread attention. In healthcare, for example, HAR can be used to evaluate medical procedures such as endotracheal suctioning. This study focuses on three HAR tasks: sign language recognition (SLR), endotracheal suctioning (ES) recognition, and inner speech recognition using electroencephalogram (EEG) signals. The key objective is to make HAR systems more accessible and portable. Two main challenges arise in this context: (1) achieving real-time performance and (2) enabling low-cost computation. Many prior studies utilize deep learning or deep RNN; however, these approaches are not well-suited for portable devices due to their high computational demands. Conversely, traditional machine learning methods struggle with dynamic input, especially video-based data. Therefore, this study emphasizes reservoir computing (RC), particularly echo state network (ESN), due to their suitability for temporal data. However, there are some challenge that faced in this study because the standard ESN did not perform optimally to solve the complex HAR, such as: 1. Single input modality may be insufficient. 2. There is a lack of training data. 3. Low memory capacity in the standard ESN. 4. One reservoir faces difficulty in extracting various features in a complex task. 5. A single readout struggles to classify many patterns.
In the sign language task, three datasets are used in this study: the WLASL100, WLASL300 (American sign language) dataset and the LSA64 (Argentinian sign language) dataset. SLR faces problems such as the difficulty of extracting various features, the inadequacy of one modality, the low memory capacity of the ESN, and the inadequacy of one readout to solve complex problems. Four architectures are proposed to solve these problems in SLR tasks: multiple reservoir computing to extract various features, multimodality to provide additional modalities to the reservoir model, deep ESN to increase memory capacity, and ensemble learning of deep ESN and multiple readouts. The ensemble learning method demonstrated the highest level of accuracy, achieving 76.12% for WLASL100, 58.20% for WLASL300, and 91.41% for signer independence on LSA64.
ES activity is associated with complications and risks. Recognizing ES activity ensures patient safety and improves nurses’ skills when conducting complicated procedures. Data collection presents more challenges, including missing values. Using generative AI and LLM can help solve this issue by analyzing and generating data. This study uses a combination of LLM and multiple readouts to classify ES labels. The optimal parameters for this task were 16 grids and 210 nodes. The classification accuracy was 70.5%. In the inner speech recognition by using EEG signals task, there is a problem of a lack of modality. This study used bidirectional RC to improve the performance of unidirectional RC. The highest accuracy parameter in the EEG task is 18.94%. The dataset’s low accuracy is due to the need for additional preprocessing to reduce noise.九州工業大学博士学位論文 学位記番号:生工博甲第517号 学位授与年月日:令和7年9月25日令和7年度doctoral thesi
Adaptive Dressing Assistance using a Dual Arm Robot for Individual Physique and Kyphosis
Get PDFOne of the significant challenges in the development of care robots is recognizing and adapting to individual differences in users’ physiques and symptoms, thereby providing personalized assistance. Among Activities of Daily Living (ADLs), dressing assistance poses a particularly complex problem due to the physical interaction between clothing, the human body, and the robot. This complexity is further amplified in care settings, as the primary users are elderly or disabled individuals, whose physical characteristics vary widely. This study focuses on the physical traits of elderly individuals targeted by dressing assistance robots and proposes an adaptive method that accommodates variations in body size and kyphosis. The proposed approach utilizes Dynamic Movement Primitives (DMPs) to generate dressing trajectories that conform to the user’s body, based on skeletal information estimated by a vision sensor. Experiments were conducted with eight healthy individuals of varying body types simulating kyphotic posture, one healthy elderly participant, and one elderly participant with posture-related impairments due to Parkinson’ s disease. The results demonstrate that, compared to non-adaptive methods, the proposed approach enables more adaptive and less burdensome dressing assistance.journal articl
Effect of water temperature in the deep sea on the mitigation of hydrogen embrittlement by O2 impurity
No full textThe addition of trace amounts of oxygen (O2) to hydrogen gas (H2) can mitigate hydrogen embrittlement (HE) in steels by suppressing the hydrogen uptake. In this study, the influence of temperature on the HE mitigation effect of O2 was investigated using fracture toughness tests of SCM440 low-alloy steel in H2 environments containing volume ppm-levels of O2 at 293 K and 277 K. Lowering the temperature from 293 K to 277 K enhanced the HE in pure H2. However, the addition of O2 to the H2 effectively mitigated the HE, with the mitigation effect becoming more pronounced at lower temperatures. This temperature-dependent mitigation is attributed to the different adsorption behaviors of H2 and O2 on the iron surface. O2 adsorption onto iron occurs without an activation barrier, making it essentially temperature-independent. In contrast, H2 adsorption requires overcoming an activation barrier and thus decreases with a lower temperature. Consequently, decreasing the temperature selectively suppresses H2 adsorption while leaving O2 adsorption unaffected, thus enhancing the relative effect of O2. Furthermore, the presence of O2 increases the activation barrier for H2 adsorption, which amplifies the temperature dependence of the H2 adsorption. These dual effects contribute to the enhanced HE mitigation by O2 at lower temperatures.journal articl