研究者詳細 - 田中　悠一朗

論文 - 田中　悠一朗

分割表示55 件中 1 - 20 件目／全件表示 >>

Prediction of Timing and Amount of Houseplants Watering by an Echo State Network on Jetson 査読有り国際誌

Yoshimura W., Arimura K., Kobayashi R., Mizutani A., Fujino T., Tanaka Y., Sudo T., Ishizuka N., Ito K., Tamukoh H.

Proceedings of International Conference on Artificial Life and Robotics 529 - 534 2025年01月

記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）

Cultivating of houseplants in biophilic designed spaces requires appropriate timing and amount of watering. However, determining them is challenging, as fluctuations in ambient temperature can influence these factors. We develop a system capable of predicting ambient temperature changes and determining the appropriate timing and amount of watering. The system acquires ambient data using sensors connected to a Jetson Nano and processes the data using a neural network for the prediction and determination. We adopt an echo state network, a lightweight neural network, enabling a power-efficient system capable of running on edge devices. Additionally, we implement a function to notify the user of the timing and amount of watering via a chat service whenever the soil moisture content drops below a predefined threshold.

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85219505871&origin=inward

Lightweight Hand-Waving Action Recognition Using Reservoir Computing in a Cafeteria Environment 査読有り

Isomoto K., Fumoto S., Kobayashi R., Tanaka Y., Tamukoh H.

2025 IEEE/SICE International Symposium on System Integration, SII 2025 1034 - 1040 2025年01月

記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）

Owing to the global labor shortage and increasing need for operational efficiency, the adoption of service robots is advancing rapidly. These robots must recognize human action to understand human intention and respond appropriately based on that understanding. The action recognition systems embedded in robots need to be lightweight to operate efficiently with limited computational resources. Reservoir computing (RC) is one of the solutions for lightweight action recognition systems. Yamaguchi et al. proposed an RC-based hand-waving recognition system; however, the system cannot process multiple persons simultaneously and works only when one person is in the image. Therefore, this study proposes a lightweight hand-waving recognition system that integrates OpenPose, StrongSORT, and RC to work in complex environments with multiple individuals. Experimental results demonstrated the effectiveness of the proposed system in processing multiple people simultaneously in a crowded environment and accurately recognizing hand-waving actions with 90.75% accuracy. We also confirmed that the proposed system can process data at 24-26 FPS. We demonstrated that the proposed system can perform real-time processing. In addition, the robot with the proposed system recognized hand-waving actions in the "Restaurant"task of RoboCup@Home 2024 and obtained the second-place score.

DOI： 10.1109/SII59315.2025.10871098

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=86000251022&origin=inward

Classification of Human Activity by Spiking Neural Networks using Event-based Vision Sensors 査読有り

Rakheja A., Saini R., Toyoda R., Tanaka Y., Tamukoh H.

Proceedings of International Conference on Artificial Life and Robotics 484 - 486 2025年01月

記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）

This paper presents a human action classification system using a spiking neural network (SNN) and an event-based vision sensor (EVS). The EVS captures asynchronous data with high temporal resolution, wide dynamic range, and motion blur resistance. SNNs, inspired by biological neurons, process this data event-driven, ensuring energy efficiency, low latency, and scalability. Using a custom EVS dataset of 600 videos across four action types and the optical flow for feature extraction, the system achieved 93% accuracy, offering an efficient solution for action recognition.

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85219530000&origin=inward

Classification of Human Activity by Event-based Vision Sensors using Echo State Networks 査読有り国際誌

Saini R., Rakheja A., Toyoda R., Tanaka Y., Tamukoh H.

Proceedings of International Conference on Artificial Life and Robotics 476 - 478 2025年01月

記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）

We propose a system for human activity recognition using an event-based vision sensor (EVS) with echo state networks (ESNs). Conventional cameras are susceptible to motion blur and require computationally intensive methods, whereas EVS provides no motion blur and low latency. Our research aims to enable accurate recognition of human activities by using energy-efficient methods. Therefore, we adopt ESNs, which require low computational costs, for the classifier. Additionally, we use feature extraction algorithms such as optical flow and histogram of gradients to improve accuracy. We used an EVS activity recognition dataset created by us containing six human activities and 600 videos. The results showed that our hybrid approach outperformed several techniques. We achieved 89% accuracy when trained with ridge regression.

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85219518069&origin=inward

Application of AI Robot Technology for Biophilic Design 査読有り国際誌

Manabe K., Miyazono R., Ito K., Sudo T., Ishizuka N., Mizutani A., Tamukoh H., Tanaka Y., Ueda E., Anamizu Y., Furuya L., Tamai H., Nakano S., Tanaka H.

Proceedings of International Conference on Artificial Life and Robotics 513 - 518 2025年01月

記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）

To enhance human health and well-being, Biophilic design has been increasingly recognized in recent years. This design is characterized by the integration of natural elements such as plants, nature light, and water into spaces. However, challenges are posed by the maintenance of live plants, as their decline can be caused by insufficient environmental conditions. The aim of this study is to propose a robotic system of autonomously relocate houseplants to environments optimized based on sensor data, including light, temperature, and humidity. Through the integration of AI robotics with ecological design principles, it is aimed to enhance sustainability and redefine the relationship between humans, nature, and technology, fostering a harmonious interaction among "robots, nature, and humans.".

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85219571283&origin=inward

An Exhibition Environment with 2D Markers for Guide Robot 査読有り国際誌

Mizutani A., Hattori Y., Tanaka Y., Ishizuka N., Tanaka H., Tamukoh H.

Proceedings of International Conference on Artificial Life and Robotics 524 - 528 2025年01月

記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）

The guide robot in an exhibition environment is expected to entertain visitors and reduce the maintenance cost of updating the robot database according to the changes in exhibition contents. The exhibition comprises printed panels for humans and a 2D marker printed with ultraviolet ink for robots. 2D markers are attached to the bottom of the exhibition furniture to make the markers invisible to visitors. The exhibition point may change depending on the contents of the exhibition. By searching for 2D markers autonomously in an exhibition space, the robot can update its internal database. This enables us to keep both the printed panel and the robot database without changing the robot database by users. The experiment is conducted in an exhibition environment, and the success rate in finding the exhibition point is evaluated.

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85219566021&origin=inward

Simple open-set recognition combining metric learning and anomaly detection 査読有り国際誌

Kanaoka D., Tanaka Y., Tamukoh H.

Advanced Robotics 2025年01月

記述言語：英語掲載種別：研究論文（学術雑誌）

In real world scenarios, there is always a risk that an object recognition system will encounter unknown objects. Open-set recognition, which identifies such objects as ‘unknown,’ is a crucial research area. Most existing open-set recognition methods have complex structures. Considering the application of closed-set recognition models already used in the real world are utilized for open-set recognition, a simple structure is preferable for implementation. In this study, we propose a novel open-set recognition method that focuses on the feature space of a recognition model by combining metric learning and anomaly detection (AD). By applying metric learning, the feature vectors of the same class are clustered and those of different classes are separated, increasing the open space for the detection of unknown objects. The AD model trains on feature vectors from the intermediate layers of a recognition model, allowing unknown detection without altering the existing model structure. We evaluated various combinations of metric learning and AD techniques, identifying an effective pairing. Our results show that the proposed method performs comparably to state-of-the-art approaches on simple datasets. The simplicity of the proposed method makes it easy to adapt existing closed-set recognition models for open-set recognition in real-world applications.

DOI： 10.1080/01691864.2025.2486394

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=105002243676&origin=inward

Object recognition and grasping point detection using carbon nanotube - polydimethylsiloxane nanocomposite sensor 査読有り国際誌

Tokuno Shoshi, Kimizuka Kouki, Tanaka Yuichiro, Usami Yuki, Tanaka Hirofumi, Tamukoh Hakaru

Nonlinear Theory and Its Applications, IEICE ( 一般社団法人電子情報通信学会 ) 15 ( 4 ) 883 - 898 2024年10月

記述言語：英語掲載種別：研究論文（学術雑誌）

<p>This paper presents a system for processing tactile information using a carbon nanotube (CNT)-polydimethylsiloxane (PDMS) nanocomposite sensor designed for robotic applications. This study introduces an approach for recognizing objects and detecting optimal grasping points using tactile data from a sensor-equipped robotic hand. The sensor is expected to be more efficient than a computerized implementation using sensor dynamics directly in the computation. The experiments demonstrated the system's ability to classify nine types of objects with an accuracy of 83.56% and to discern two grasping points on four object types, achieving a 71.7% success rate.</p>

DOI： 10.1587/nolta.15.883

その他リンク： https://www.jstage.jst.go.jp/article/nolta/15/4/15_883/_pdf

ナノマテリアルランダムネットワークを利用した物理リザバー演算素子とそのセンシング応用招待有り査読有り

田中啓文, アズハリサマン, バナジーディープ, 君塚紘喜, 田中悠一朗, 宇佐美雄生

表面と真空 ( 公益社団法人日本表面真空学会 ) 67 ( 9 ) 428 - 432 2024年09月

記述言語：日本語掲載種別：研究論文（学術雑誌）

<p>The rapid advancement of software-based deep learning has led to a surge in AI applications, yet hardware limitations in silicon CMOS technology hinder performance. Consequently, interest is growing in hardware technologies and new materials for artificial neural networks (ANNs) and neuromorphic systems. Exploiting nanomaterials nonlinearity caused by spontaneous physical phenomena holds promise of reducing power consumption in AI hardware. Reservoir computing devices, derived from recurrent neural networks, play a crucial role, with material reservoir devices showcasing “material intelligence.” Tailored nanomaterials for reservoir devices hold potential for revolutionizing AI, especially in robotics. As research progresses, focusing on device functionalization and applications, recent findings underline the significance of integrating nanomaterials into AI hardware for enhanced computational capabilities and energy efficiency.</p>

DOI： 10.1380/vss.67.428

その他リンク： https://www.jstage.jst.go.jp/article/vss/67/9/67_20181334/_pdf

A Hippocampus-Inspired Environment-Specific Knowledge Acquisition System Utilizing Common Knowledge with Contextual Information 査読有り国際誌

Mizutani A., Tanaka Y., Tamukoh H., Nomura O., Tateno K., Morie T.

Proceedings of the International Joint Conference on Neural Networks 2024年07月

記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）

Home service robots acquire environment-specific knowledge through experiences (episodes) in a home to perform tasks autonomously. We propose a hippocampus-inspired memory system comprising an environment-specific knowledge module that handles episodic memory and a common knowledge module. The robot works in a home and needs to learn the locations of objects in the home with minimal user help to deliver or store objects. We employ a large language model (LLM) that runs on an edge device as the common knowledge module to protect the privacy of users. The environment-specific knowledge module acquires episodes and generates contextual information. The LLM receives contextual information generated by the environment-specific knowledge module, with which it infers an appropriate possible location. We verified that the LLM could infer the possible object locations and that the proposed system could minimize the required user effort.

DOI： 10.1109/IJCNN60899.2024.10651337

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85205013660&origin=inward

Reservoir-based 1D convolution: low-training-cost AI 査読有り国際誌

TANAKA Yuichiro, TAMUKOH Hakaru

IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences ( 一般社団法人電子情報通信学会 ) advpub ( 0 ) 941 - 944 2024年06月

担当区分：筆頭著者,　責任著者記述言語：英語掲載種別：研究論文（学術雑誌）

<p>In this study, we introduce a reservoir-based one-dimensional (1D) convolutional neural network that processes time-series data at a low computational cost, and investigate its performance and training time. Experimental results show that the proposed network consumes lower training computational costs and that it outperforms the conventional reservoir computing in a sound-classification task.</p>

DOI： 10.1587/transfun.2023eal2050

その他リンク： https://www.jstage.jst.go.jp/article/transfun/advpub/0/advpub_2023EAL2050/_pdf

Recognizing Nursing Activities in Endotracheal Suction: Utilizing Multiple Readouts Reservoir Computing and Large Language Models 査読有り国際誌

Arie Rachmad Syulistyo, Yuichiro Tanaka, Hakaru Tamukoh

6th International Conference on Activity and Behavior Computing 2024年05月

Enhancing Memory Capacity of Reservoir Computing with Delayed Input and Efficient Hardware Implementation with Shift Registers 査読有り国際誌

Hirayae S., Yoshioka K., Yokota A., Kawashima I., Tanaka Y., Katori Y., Nomura O., Morie T., Tamukoh H.

Proceedings - IEEE International Symposium on Circuits and Systems 2024年05月

記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）

To use reservoir computing (RC) for practical tasks, both a high memory capacity and nonlinearity are required; however, some RC models have the problem of a low memory capacity. We propose a delay mechanism for increasing the memory capacity in RC as well as a simple and small-scale digital circuit for implementing the delay mechanism. The proposed delay mechanism is integrated into the input layer of the RC model and is expected to be implemented in several RC models, such as material reservoirs and chaotic Boltzmann machine (CBM)-RC. We conducted experiments using a CBM-RC with a delay mechanism (CBM-RC-DL) and evaluated the performance improvement achieved by introducing a delay mechanism. We used CBM-RC as the base model because it is an appropriate model for the hardware implementation of large networks but has a low memory capacity. The experimental results for CBM-RC-DL indicated that the delay mechanism significantly increased the memory capacity of CBM-RC with the addition of a small-scale circuit. Furthermore, the entire synthesized CBM-RC-DL was sufficiently small-scale to be implemented in a field-programmable gate array for edge computing, and it outperformed conventional methods in nonlinear autoregressive moving average 10 (NARMA10) - a benchmark task for time-series data processing. The proposed delay mechanism can facilitate the use of many RC models because of its simple structure.

DOI： 10.1109/ISCAS58744.2024.10557841

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85198500940&origin=inward

A Low Computational Cost Hand Waving Action Recognition System with Echo State Network for Home Service Robots 査読有り国際誌

Yamaguchi H., Mizutani A., Syulistyo A.R., Tanaka Y., Tamukoh H.

Proceedings of International Conference on Artificial Life and Robotics 421 - 424 2024年02月

記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）

This study proposes a low computational cost hand-waving action recognition system for non-verbal communication in home service robots. The system is based on an echo state network, which requires lower computational costs than that of deep neural networks (DNNs), and processes time-series data of skeletal coordinates of humans to recognize hand-waving actions. Additionally, this study proposes and compares two types of Preprocessing ing methods of the skeletal coordinates to ensure the robustness of the human positions on the frame: one method extracts shoulder and arm angles, which are invariable regardless of the humans’ positions and the other normalizes the skeletal coordinates. The experimental result shows that the proposed system has competitive accuracy and is robust to varying human positions.

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85190300153&origin=inward

Development of A SayCan-based Task Planning System Capable of Handling Abstract Nouns 査読有り国際誌

Yamao K., Kanaoka D., Isomoto K., Mizutani A., Tanaka Y., Tamukoh H.

Proceedings of International Conference on Artificial Life and Robotics 430 - 434 2024年02月

記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）

The task planning system is required to accomplish various requests from a human in real-world environments. SayCan, one of the task planning systems, has high accuracy. However, its accuracy decreases for requests that include abstract nouns of the ambiguous word/phrase. We propose a novel task planning system based on SayCan that introduces a function for checking concrete names of abstract nouns and a rule-based skill extraction, enhancing accuracy. The proposed system facilitates the interpretation of requests and enables appropriate task planning. The effectiveness of the proposed system was demonstrated at RoboCup@Home, where it achieved high performance.

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85190283813&origin=inward

Object recognition using CNT-PDMS nanocomposite sensor on a robot hand 査読有り

Shoshi Tokuno, Kouki Kimizuka, Yuichiro Tanaka, Hirofumi Tanaka, Hakaru Tamukoh

The 22nd International Symposium on Eco-materials Processing and Design 2024年01月

Thermally Stable Ag<inf>2</inf>Se Nanowire Network as an Effective In-Materio Physical Reservoir Computing Device 査読有り国際誌

Kotooka T., Lilak S., Stieg A.Z., Gimzewski J.K., Sugiyama N., Tanaka Y., Kawabata T., Karacali A., Tamukoh H., Usami Y., Tanaka H.

Advanced Electronic Materials 10 ( 12 ) 2024年01月

記述言語：英語掲載種別：研究論文（学術雑誌）

The artificial intelligence (AI) paradigm shifts from software to implementing general-purpose or application-specific hardware systems with lower power requirements. This study explored a material physical reservoir consisting of a material random network, called in-materio physical reservoir computing (RC) to achieve efficient hardware systems. The device, made up of a random, highly interconnected network of nonlinear Ag2Se nanojunctions as reservoir nodes, demonstrated the requisite characteristics of an in-materio physical reservoir, including but not limited to nonlinear switching, memory, and higher harmonic generation. The power consumption of the in-materio physical reservoir is 0.07 nW per nanojunctions, confirming its highly efficient information processing system. As a hardware reservoir, the devices successfully performed waveform generation tasks. Finally, a voice classification by an in-materio physical reservoir is achieved over 80%, comparable to an RC software simulation. In-materio physical RC with rich nonlinear dynamics has huge potential for next-generation hardware-based AI.

DOI： 10.1002/aelm.202400443

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85210391008&origin=inward

Self-Organizing Multiple Readouts for Reservoir Computing 査読有り国際誌

Tanaka Y., Tamukoh H.

IEEE Access 11 138839 - 138849 2023年12月

担当区分：筆頭著者,　責任著者記述言語：英語掲載種別：研究論文（学術雑誌）

With advancements in deep learning (DL), artificial intelligence (AI) technology has become an indispensable tool. However, the application of DL incurs significant computational costs, making it less viable for edge AI scenarios. Consequently, the demand for cost-effective AI solutions, other than DL-based approaches, is increasing. Reservoir computing (RC) has attracted interest owing to its ability to provide low-cost training alternatives, holding great promise for edge AI applications. However, the training capability of RC is constrained by its reliance on a single linear layer, while weight connections in the remaining layers remain static during training. Moreover, accomplishing continuous learning tasks is difficult owing to the catastrophic forgetting in the linear layer. Therefore, we propose the integration of self-organizing multiple readouts to enhance RC's training capability. Our method distributes training data across multiple readouts, which prevents catastrophic forgetting of readouts and empowers each readout to adeptly assimilate new data, thereby elevating the overall training performance. The self-organizing function, which assigns similar data to the same readout, optimizes the memory utilization of these multiple readouts. Experimental results show that an RC equipped with the proposed multiple readouts successfully solved a continuous learning task by mitigating catastrophic forgetting because of the data distribution to the multiple readouts. Additionally, the RC achieved higher accuracy in a sound recognition task compared with the existing RC paradigm because of ensemble learning in the multiple readouts. Multiple readouts are effective in enhancing the training capability of RC and can contribute to the realization of RC applications.

DOI： 10.1109/ACCESS.2023.3340311

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85179802812&origin=inward

LUTNet-RC: Look-Up Tables Networks for Reservoir Computing on an FPGA 査読有り国際誌

Yoshioka K., Tanaka Y., Tamukoh H.

Proceedings - International Conference on Field-Programmable Technology, ICFPT 170 - 178 2023年12月

記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）

We propose look-up tables networks-based reservoir computing (LUTNet-RC). This work is the first trial of applying LUTNets to RC. LUTNet-RC consists of a LUT-based reservoir layer and a non-LUT-based output layer. LUTNets have disadvantages such as limited sparse connectivity and weights cannot be changed after implementation. However, when applied to a reservoir layer of RC (LUT-based reservoir layer), these disadvantages are eliminated, because this layer works with sparse connectivity and the weights are fixed, so only the advantage of small circuit resources is obtained. For the LUT-based reservoir layer, we propose and model a multi-bit weight reservoir, modifying the conventional binarized reservoir to improve calculation accuracy. In the case of LUTNets, the proposed multi-bit weight reservoir can be implemented without the increase in utilized circuit resources because LUTNets focus only on the input-output relationship on neurons. Additionally, we propose a speed-up method in the output layer with time division calculation, which compares the current network state with previous states and then calculates only status-changed neurons. As a result, we implement a LUTNet-RC with 1500 reservoir neurons on a field-programmable gate array (KR260) running at 100MHz. The utilized circuit resources are dominated by LUTs, which use approximately 26% of the total amount of LUTs. The LUTNet-RC can infer more than 106 data per second. We also verify the LUTNet-RC performance using nonlinear auto-regressive moving average 10 (NARMA10) and the performance is comparable to conventional works. We conclude that the LUTNet-RC is one of the highest-performance RC on an FPGA.

DOI： 10.1109/ICFPT59805.2023.00024

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85187557709&origin=inward

Traffic Flow Optimization using a Chaotic Boltzmann Machine Annealer on an FPGA 査読有り国際誌

Yoshioka K., Tanaka Y., Tamukoh H.

Proceedings - International Conference on Field-Programmable Technology, ICFPT 266 - 267 2023年12月

記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）

We demonstrate a chaotic Boltzmann machine annealer, which is a field-programmable gate array (FPGA)-based high-performance annealing machine, and a control system powered by two open-source software packages. The annealer is employed in solving a large-scale real-world optimization problem, the traffic flow optimization. This involves distributing 500 car routes between the Haneda Airport area to the Pacifico Yokohama area, the venue of the International Conference on Field Programmable Technology 2023, while minimizing the total driving distance and preventing traffic jams. We obtain solutions that are comparable in accuracy to solutions of simulated annealing running on a graphics processing unit (GPU-SA) and a central processing unit (CPU-SA). The annealing machine on an FPGA is approximately 571 and 97600 times as fast as the GPUSA and the CPU-SA, respectively. We will demonstrate solving traffic flow optimization in the towns familiar to the visitors.

DOI： 10.1109/ICFPT59805.2023.00038

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85187557051&origin=inward

　前のページ- 次のページ　

1
2
3