業績情報

• Synergy-Based Evaluation of Hand Motor Function in Object Handling Using Virtual and Mixed Realities
  Yuhei Sorimachi; Hiroki Akaida; Kyo Kutsuzawa; Dai Owaki; Mitsuhiro Hayashibe
  Sensors, 2025年03月, ［査読有り］
  研究論文（学術雑誌）
  DOI：https://doi.org/10.3390/s25072080
  DOI ID：10.3390/s25072080, ORCID：181217590
• Versatile graceful degradation framework for bio-inspired proprioception with redundant soft sensors
  Taku Sugiyama; Kyo Kutsuzawa; Dai Owaki; Elijah Almanzor; Fumiya Iida; Mitsuhiro Hayashibe
  Frontiers in Robotics and AI, 巻：11, 2025年01月, ［査読有り］
  Reliable proprioception and feedback from soft sensors are crucial for enabling soft robots to function intelligently in real-world environments. Nevertheless, soft sensors are fragile and are susceptible to various damage sources in such environments. Some researchers have utilized redundant configuration, where healthy sensors compensate instantaneously for lost ones to maintain proprioception accuracy. However, achieving consistently reliable proprioception under diverse sensor degradation remains a challenge. This paper proposes a novel framework for graceful degradation in redundant soft sensor systems, incorporating a stochastic Long Short-Term Memory (LSTM) and a Time-Delay Feedforward Neural Network (TDFNN). The LSTM estimates readings from healthy sensors to compare them with actual data. Then, statistically abnormal readings are zeroed out. The TDFNN receives the processed sensor readings to perform proprioception. Simulation experiments with a musculoskeletal leg that contains 40 nonlinear soft sensors demonstrate the effectiveness of the proposed framework. Results show that the knee angle proprioception accuracy is retained across four distinct degradation scenarios. Notably, the mean proprioception error increases by less than 1.91°(1.36%) when of the sensors are degraded. These results suggest that the proposed framework enhances the reliability of soft sensor proprioception, thereby improving the robustness of soft robots in real-world applications.
  Frontiers Media SA, 研究論文（学術雑誌）
  DOI：https://doi.org/10.3389/frobt.2024.1504651
  DOI ID：10.3389/frobt.2024.1504651, ISSN：2296-9144, eISSN：2296-9144, ORCID：178592332
• Learning-based object's stiffness and shape estimation with confidence level in multi-fingered hand grasping
  Kyo Kutsuzawa; Minami Matsumoto; Dai Owaki; Mitsuhiro Hayashibe
  Frontiers in Neurorobotics, 巻：18, 2024年11月, ［査読有り］, ［筆頭著者, 責任著者］
  Introduction
  
  When humans grasp an object, they are capable of recognizing its characteristics, such as its stiffness and shape, through the sensation of their hands. They can also determine their level of confidence in the estimated object properties. In this study, we developed a method for multi-fingered hands to estimate both physical and geometric properties, such as the stiffness and shape of an object. Their confidence levels were measured using proprioceptive signals, such as joint angles and velocity.
  
  Method
  
  We have developed a learning framework based on probabilistic inference that does not necessitate hyperparameters to maintain equilibrium between the estimation of diverse types of properties. Using this framework, we have implemented recurrent neural networks that estimate the stiffness and shape of grasped objects with their uncertainty in real time.
  
  Results
  
  We demonstrated that the trained neural networks are capable of representing the confidence level of estimation that includes the degree of uncertainty and task difficulty in the form of variance and entropy.
  
  Discussion
  
  We believe that this approach will contribute to reliable state estimation. Our approach would also be able to combine with flexible object manipulation and probabilistic inference-based decision making.
  Frontiers Media SA, 研究論文（学術雑誌）
  DOI：https://doi.org/10.3389/fnbot.2024.1466630
  DOI ID：10.3389/fnbot.2024.1466630, ISSN：1662-5218, eISSN：1662-5218, ORCID：178592244
• Synergy quality assessment of muscle modules for determining learning performance using a realistic musculoskeletal model　　　　　　　　　　　　　　　
  Akito Fukunishi; Kyo Kutsuzawa; Dai Owaki; Mitsuhiro Hayashibe
  Frontiers in Computational Neuroscience, 2024年05月, ［査読有り］
  How our central nervous system efficiently controls our complex musculoskeletal system is still debated. The muscle synergy hypothesis is proposed to simplify this complex system by assuming the existence of functional neural modules that coordinate several muscles. Modularity based on muscle synergies can facilitate motor learning without compromising task performance. However, the effectiveness of modularity in motor control remains debated. This ambiguity can, in part, stem from overlooking that the performance of modularity depends on the mechanical aspects of modules of interest, such as the torque the modules exert. To address this issue, this study introduces two criteria to evaluate the quality of module sets based on commonly used performance metrics in motor learning studies: the accuracy of torque production and learning speed. One evaluates the regularity in the direction of mechanical torque the modules exert, while the other evaluates the evenness of its magnitude. For verification of our criteria, we simulated motor learning of torque production tasks in a realistic musculoskeletal system of the upper arm using feed-forward neural networks while changing the control conditions. We found that the proposed criteria successfully explain the tendency of learning performance in various control conditions. These result suggest that regularity in the direction of and evenness in magnitude of mechanical torque of utilized modules are significant factor for determining learning performance. Although the criteria were originally conceived for an error-based learning scheme, the approach to pursue which set of modules is better for motor control can have significant implications in other studies of modularity in general.
  研究論文（学術雑誌）
  DOI：https://doi.org/10.3389/fncom.2024.1355855
  DOI ID：10.3389/fncom.2024.1355855, ISSN：1662-5188, ORCID：166927718
• Integrated Quantitative Evaluation of Spatial Cognition and Motor Function with HoloLens Mixed Reality
  Kenya Tada; Yuhei Sorimachi; Kyo Kutsuzawa; Dai Owaki; Mitsuhiro Hayashibe
  Sensors, 巻：24, 号：2, 開始ページ：528, 終了ページ：528, 2024年01月, ［査読有り］
  The steady increase in the aging population worldwide is expected to cause a shortage of doctors and therapists for older people. This demographic shift requires more efficient and automated systems for rehabilitation and physical ability evaluations. Rehabilitation using mixed reality (MR) technology has attracted much attention in recent years. MR displays virtual objects on a head-mounted see-through display that overlies the user’s field of vision and allows users to manipulate them as if they exist in reality. However, tasks in previous studies applying MR to rehabilitation have been limited to tasks in which the virtual objects are static and do not interact dynamically with the surrounding environment. Therefore, in this study, we developed an application to evaluate cognitive and motor functions with the aim of realizing a rehabilitation system that is dynamic and has interaction with the surrounding environment using MR technology. The developed application enabled effective evaluation of the user’s spatial cognitive ability, task skillfulness, motor function, and decision-making ability. The results indicate the usefulness and feasibility of MR technology to quantify motor function and spatial cognition both for static and dynamic tasks in rehabilitation.
  MDPI AG, 研究論文（学術雑誌）
  DOI：https://doi.org/10.3390/s24020528
  DOI ID：10.3390/s24020528, eISSN：1424-8220
• Muscle Control Analysis of Human Walking and Cycling in Speed and Load Variations with Time-Varying Synergy
  Takumi Matsumura; Eiji Inomata; Kyo Kutsuzawa; Dai Owaki; Mitsuhiro Hayashibe
  IEEE Sensors Journal, 開始ページ：1, 終了ページ：1, 2024年, ［査読有り］, ［責任著者］
  Institute of Electrical and Electronics Engineers (IEEE), 研究論文（学術雑誌）
  DOI：https://doi.org/10.1109/jsen.2024.3486294
  DOI ID：10.1109/jsen.2024.3486294, ISSN：1530-437X, eISSN：1558-1748, ORCID：178592315
• Shared Synergyを利用した高い汎化能力をもたらす模倣学習
  沓澤京
  日本ロボット学会誌, 巻：41, 号：8, 開始ページ：661, 終了ページ：664, 2023年10月, ［査読有り］, ［招待有り］, ［筆頭著者, 最終著者, 責任著者］
  The Robotics Society of Japan, 日本語, 研究論文（学術雑誌）
  DOI：https://doi.org/10.7210/jrsj.41.661
  DOI ID：10.7210/jrsj.41.661, ISSN：0289-1824, eISSN：1884-7145, 共同研究・競争的資金等ID：33406670
• Latent Representation-Based Learning Controller for Pneumatic and Hydraulic Dual Actuation of Pressure-Driven Soft Actuators
  Taku Sugiyama; Kyo Kutsuzawa; Dai Owaki; Mitsuhiro Hayashibe
  Soft Robotics, 2023年08月, ［査読有り］
  Mary Ann Liebert Inc, 研究論文（学術雑誌）
  DOI：https://doi.org/10.1089/soro.2022.0224
  DOI ID：10.1089/soro.2022.0224, ISSN：2169-5172, eISSN：2169-5180
• Transhumeral Arm Reaching Motion Prediction through Deep Reinforcement Learning-Based Synthetic Motion Cloning
  Muhammad Hannan Ahmed; Kyo Kutsuzawa; Mitsuhiro Hayashibe
  Biomimetics, 巻：8, 号：4, 開始ページ：367, 終了ページ：367, 2023年08月, ［査読有り］
  The lack of intuitive controllability remains a primary challenge in enabling transhumeral amputees to control a prosthesis for arm reaching with residual limb kinematics. Recent advancements in prosthetic arm control have focused on leveraging the predictive capabilities of artificial neural networks (ANNs) to automate elbow joint motion and wrist pronation–supination during target reaching tasks. However, large quantities of human motion data collected from different subjects for various activities of daily living (ADL) tasks are required to train these ANNs. For example, the reaching motion can be altered when the height of the desk is changed; however, it is cumbersome to conduct human experiments for all conditions. This paper proposes a framework for cloning motion datasets using deep reinforcement learning (DRL) to cater to training data requirements. DRL algorithms have been demonstrated to create human-like synergistic motion in humanoid agents to handle redundancy and optimize movements. In our study, we collected real motion data from six individuals performing multi-directional arm reaching tasks in the horizontal plane. We generated synthetic motion data that mimicked similar arm reaching tasks by utilizing a physics simulation and DRL-based arm manipulation. We then trained a CNN-LSTM network with different configurations of training motion data, including DRL, real, and hybrid datasets, to test the efficacy of the cloned motion data. The results of our evaluation showcase the effectiveness of the cloned motion data in training the ANN to predict natural elbow motion accurately across multiple subjects. Furthermore, motion data augmentation through combining real and cloned motion datasets has demonstrated the enhanced robustness of the ANN by supplementing and diversifying the limited training data. These findings have significant implications for creating synthetic dataset resources for various arm movements and fostering strategies for automatized prosthetic elbow motion.
  MDPI AG, 研究論文（学術雑誌）
  DOI：https://doi.org/10.3390/biomimetics8040367
  DOI ID：10.3390/biomimetics8040367, eISSN：2313-7673
• A Survey of Sim-to-Real Transfer Techniques Applied to Reinforcement Learning for Bioinspired Robots
  Wei Zhu; Xian Guo; Dai Owaki; Kyo Kutsuzawa; Mitsuhiro Hayashibe
  IEEE Transactions on Neural Networks and Learning Systems, 巻：34, 号：7, 開始ページ：3444, 終了ページ：3459, 2023年07月, ［査読有り］
  Institute of Electrical and Electronics Engineers (IEEE), 研究論文（学術雑誌）
  DOI：https://doi.org/10.1109/tnnls.2021.3112718
  DOI ID：10.1109/tnnls.2021.3112718, ISSN：2162-237X, eISSN：2162-2388
• Imitation Learning With Time-Varying Synergy for Compact Representation of Spatiotemporal Structures
  Kyo Kutsuzawa; Mitsuhiro Hayashibe
  IEEE Access, 巻：11, 開始ページ：34150, 終了ページ：34162, 2023年04月, ［査読有り］, ［筆頭著者, 責任著者］
  Institute of Electrical and Electronics Engineers (IEEE), 研究論文（学術雑誌）
  DOI：https://doi.org/10.1109/access.2023.3264213
  DOI ID：10.1109/access.2023.3264213, eISSN：2169-3536
• Multimodal bipedal locomotion generation with passive dynamics via deep reinforcement learning
  Shunsuke Koseki; Kyo Kutsuzawa; Dai Owaki; Mitsuhiro Hayashibe
  Frontiers in Neurorobotics, 巻：16, 2023年01月, ［査読有り］
  Generating multimodal locomotion in underactuated bipedal robots requires control solutions that can facilitate motion patterns for drastically different dynamical modes, which is an extremely challenging problem in locomotion-learning tasks. Also, in such multimodal locomotion, utilizing body morphology is important because it leads to energy-efficient locomotion. This study provides a framework that reproduces multimodal bipedal locomotion using passive dynamics through deep reinforcement learning (DRL). An underactuated bipedal model was developed based on a passive walker, and a controller was designed using DRL. By carefully planning the weight parameter settings of the DRL reward function during the learning process based on a curriculum learning method, the bipedal model successfully learned to walk, run, and perform gait transitions by adjusting only one command input. These results indicate that DRL can be applied to generate various gaits with the effective use of passive dynamics.
  Frontiers Media SA, 研究論文（学術雑誌）
  DOI：https://doi.org/10.3389/fnbot.2022.1054239
  DOI ID：10.3389/fnbot.2022.1054239, ISSN：1662-5218, eISSN：1662-5218, ORCID：156819161
• Motor synergy generalization framework for new targets in multi-planar and multi-directional reaching task　　　　　　　　　　　　　　　
  Kyo Kutsuzawa; Mitsuhiro Hayashibe
  Royal Society Open Science, 巻：9, 号：5, 2022年05月, ［査読有り］, ［筆頭著者, 責任著者］
  
  Humans can rapidly adapt to new situations, even though they have redundant degrees of freedom (d.f.). Previous studies in neuroscience revealed that human movements could be accounted for by low-dimensional control signals, known as
  motor synergies
  . Many studies have suggested that humans use the same repertories of motor synergies among similar tasks. However, it has not yet been confirmed whether the combinations of motor synergy repertories can be re-used for new targets in a systematic way. Here we show that the combination of motor synergies can be generalized to new targets that each repertory cannot handle. We use the multi-directional reaching task as an example. We first trained multiple policies with limited ranges of targets by reinforcement learning and extracted sets of motor synergies. Finally, we optimized the activation patterns of sets of motor synergies and demonstrated that combined motor synergy repertories were able to reach new targets that were not achieved with either original policies or single repertories of motor synergies. We believe this is the first study that has succeeded in motor synergy generalization for new targets in new planes, using a full 7-d.f. arm model, which is a realistic mechanical environment for general reaching tasks.
  
  The Royal Society, 英語, 研究論文（学術雑誌）
  DOI：https://doi.org/10.1098/rsos.211721
  DOI ID：10.1098/rsos.211721, ISSN：2054-5703, ORCID：113628813
• Individual deformability compensation of soft hydraulic actuators through iterative learning-based neural network
  Taku Sugiyama; Kyo Kutsuzawa; Dai Owaki; Mitsuhiro Hayashibe
  Bioinspiration & Biomimetics, 巻：16, 号：5, 開始ページ：056016, 終了ページ：056016, 2021年08月, ［査読有り］
  
  Robotic devices with soft actuators have been developed to realize the effective rehabilitation of patients with motor paralysis by enabling soft and safe interaction. However, the control of such robots is challenging, especially owing to the difference in the individual deformability occurring in manual fabrication of soft actuators. Furthermore, soft actuators used in wearable rehabilitation devices involve a large response delay which hinders the application of such devices for at-home rehabilitation. In this paper, a feed-forward control method for soft actuators with a large response delay, comprising a simple feed-forward neural network (FNN) and an iterative learning controller (ILC), is proposed. The proposed method facilitates the effective learning and acquisition of the inverse model (i.e. the model that can generate control input to the soft actuator from a target trajectory) of soft actuators. First, the ILC controls a soft actuator and iteratively learns the actuator deformability. Subsequently, the FNN is trained to obtain the inverse model of the soft actuator. The control results of the ILC are used as training datasets for supervised learning of the FNN to ensure that it can efficiently acquire the inverse model of the soft actuator, including the deformability and the response delay. Experiments with fiber-reinforced soft bending hydraulic actuators are conducted to evaluate the proposed method. The results show that the ILC can learn and compensate for the actuator deformability. Moreover, the iterative learning-based FNN serves to achieve a precise tracking performance on various generalized trajectories. These facts suggest that the proposed method can contribute to the development of robotic rehabilitation devices with soft actuators and the field of soft robotics.
  IOP Publishing, 研究論文（学術雑誌）
  DOI：https://doi.org/10.1088/1748-3190/ac1b6f
  DOI ID：10.1088/1748-3190/ac1b6f, ISSN：1748-3182, eISSN：1748-3190
• Reinforcement Learning for Robotic Assembly Using Non-Diagonal Stiffness Matrix　　　　　　　　　　　　　　　
  Masahide Oikawa; Tsukasa Kusakabe; Kyo Kutsuzawa; Sho Sakaino; Toshiaki Tsuji
  IEEE Robotics and Automation Letters, 巻：6, 号：2, 開始ページ：2737, 終了ページ：2744, 2021年04月, ［査読有り］
  Institute of Electrical and Electronics Engineers ({IEEE}), 研究論文（学術雑誌）
  DOI：https://doi.org/10.1109/lra.2021.3060389
  DOI ID：10.1109/lra.2021.3060389, ISSN：2377-3766, ORCID：98562557
• Spiking Neural Network Discovers Energy-Efficient Hexapod Motion in Deep Reinforcement Learning
  Katsumi Naya; Kyo Kutsuzawa; Dai Owaki; Mitsuhiro Hayashibe
  IEEE Access, 巻：9, 開始ページ：150345, 終了ページ：150354, 2021年, ［査読有り］
  Institute of Electrical and Electronics Engineers (IEEE), 研究論文（学術雑誌）
  DOI：https://doi.org/10.1109/access.2021.3126311
  DOI ID：10.1109/access.2021.3126311, ISSN：2169-3536, eISSN：2169-3536, ORCID：156819059
• Simultaneous Estimation of Contact Position and Tool Shape using an Unscented Particle Filter　　　　　　　　　　　　　　　
  Kutsuzawa Kyo; Sakaino Sho; Tsuji Toshiaki
  IEEJ Journal of Industry Applications, 巻：9, 号：5, 開始ページ：505, 終了ページ：514, 2020年, ［査読有り］, ［筆頭著者］
  

  The robots used in our daily lives come in contact with the environment not only directly, but also through grasped objects and tools. In such cases, the shape of the grasped objects could be unknown or uncertain; thus, the shape must be estimated using information about the contact. However, previous studies could not estimate the shape of the grasped objects without knowledge about the contact environment. In this study, unscented particle filters were used to estimate the contact positions, contact forces, and shape of the tools, simultaneously. In addition, we verified that the proposed method can estimate these characteristics by measuring the force and torque in the robots.

  
  一般社団法人 電気学会, 英語
  DOI：https://doi.org/10.1541/ieejjia.9.505
  DOI ID：10.1541/ieejjia.9.505, ISSN：2187-1094, CiNii Articles ID：130007895437
• Motion Planning With Success Judgement Model Based on Learning From Demonstration
  Daichi Furuta; Kyo Kutsuzawa; Sho Sakaino; Toshiaki Tsuji
  IEEE Access, 巻：8, 開始ページ：73142, 終了ページ：73150, 2020年, ［査読有り］, ［責任著者］
  Institute of Electrical and Electronics Engineers (IEEE), 研究論文（学術雑誌）
  DOI：https://doi.org/10.1109/access.2020.2987604
  DOI ID：10.1109/access.2020.2987604, ISSN：2169-3536, eISSN：2169-3536, ORCID：72634437
• Trajectory adjustment for nonprehensile manipulation using latent space of trained sequence-to-sequence model
  Kyo Kutsuzawa; Sho Sakaino; Toshiaki Tsuji
  Advanced Robotics, 巻：33, 号：21, 開始ページ：1144, 終了ページ：1154, 2019年11月, ［筆頭著者, 責任著者］
  Informa UK Limited, 英語, 研究論文（学術雑誌）
  DOI：https://doi.org/10.1080/01691864.2019.1673204
  DOI ID：10.1080/01691864.2019.1673204, ISSN：0169-1864, eISSN：1568-5535, ORCID：68949498
• Success/Failure Identification of Skill Movement by Neural Network Using Force Information.　　　　　　　　　　　　　　　
  Koyo Sato; Masahide Oikawa; Kyo Kutsuzawa; Sho Sakaino; Toshiaki Tsuji
  IECON 2019 - 45th Annual Conference of the IEEE Industrial Electronics Society(IECON), 開始ページ：3641, 終了ページ：3646, 2019年, ［査読有り］
  IEEE, 研究論文（国際会議プロシーディングス）
  DOI：https://doi.org/10.1109/IECON.2019.8927708
  DOI ID：10.1109/IECON.2019.8927708, DBLP ID：conf/iecon/SatoOKST19
• LSTM Learning of Inverse Dynamics with Contact in Various Environments　　　　　　　　　　　　　　　
  Daichi Furuta; Kyo Kutsuzawa; Sho Sakaino; Toshiaki Tsuji
  Proceedings - 2018 12th France-Japan and 10th Europe-Asia Congress on Mechatronics, Mecatronics 2018, 開始ページ：149, 終了ページ：154, 2018年10月, ［査読有り］
  A machine learning method has been introduced to solve the problem of inverse dynamics with contact in various environments. Conventional methods need multiple contact models to switch according to situations, while such methods have a difficulty in dealing with different environments. We propose a machine learning method that can handle various environments with a single learning model. We use long short-term memory as a learning model with high expression ability. From the verification, the proposed method showed higher performance than Gaussian processes. In addition, the performance of the model was improved by using training data collected under various environmental conditions.
  Institute of Electrical and Electronics Engineers Inc., 英語, 研究論文（国際会議プロシーディングス）
  DOI：https://doi.org/10.1109/MECATRONICS.2018.8495698
  DOI ID：10.1109/MECATRONICS.2018.8495698, SCOPUS ID：85057186097
• Sequence-to-Sequence Model for Trajectory Planning of Nonprehensile Manipulation Including Contact Model
  Kyo Kutsuzawa; Sho Sakaino; Toshiaki Tsuji
  IEEE Robotics and Automation Letters, 巻：3, 号：4, 開始ページ：3606, 終了ページ：3613, 2018年10月, ［査読有り］, ［筆頭著者］
  Institute of Electrical and Electronics Engineers (IEEE), 研究論文（学術雑誌）
  DOI：https://doi.org/10.1109/lra.2018.2854958
  DOI ID：10.1109/lra.2018.2854958, eISSN：2377-3766, ORCID：62121681, SCOPUS ID：85063307055
• Trajectory planning by variable length chunk of sequence-to-sequence using hierarchical decoder.　　　　　　　　　　　　　　　
  Tetsugaku Okamoto; Kyo Kutsuzawa; Sho Sakaino; Toshiaki Tsuji
  IEEE 15th International Workshop on Advanced Motion Control(AMC), 開始ページ：209, 終了ページ：214, 2018年, ［査読有り］
  IEEE, 研究論文（国際会議プロシーディングス）
  DOI：https://doi.org/10.1109/AMC.2019.8371089
  DOI ID：10.1109/AMC.2019.8371089, DBLP ID：conf/amc/OkamotoKST18
• Optimized Trajectory Generation based on Model Predictive Control for Turning Over Pancakes
  Toshiaki Tsuji; Kyo Kutsuzawa; Sho Sakaino
  IEEJ Journal of Industry Applications, 巻：7, 号：1, 開始ページ：22, 終了ページ：28, 2018年, ［査読有り］
  Institute of Electrical Engineers of Japan (IEE Japan), 研究論文（学術雑誌）
  DOI：https://doi.org/10.1541/ieejjia.7.22
  DOI ID：10.1541/ieejjia.7.22, ISSN：2187-1094, eISSN：2187-1108, ORCID：62121683, SCOPUS ID：85041957480
• 動力学的拘束を考慮したモデル予測制御のディープニューラルネットワーク学習　　　　　　　　　　　　　　　
  古田 大地; 沓澤 京; 岡本 哲学; 境野 翔; 辻 俊明
  ロボティクス・メカトロニクス講演会講演概要集, 巻：2017, 開始ページ：1P1, 終了ページ：I12, 2017年
  

  This paper proposes a trajectory planning method with deep neural network (DNN) which is trained by model predictive control (MPC) for dynamic manipulation. The novelty of this method is that trained DNN can receive target positions and environmental parameters to generate trajectories. The proposed method solves dynamic manipulation using dynamic constraint of the object with low calculation cost. This paper shows the effectiveness of the proposed method by demonstrations that a robot turns over pancakes under various parameters.

  
  一般社団法人 日本機械学会, 日本語
  DOI：https://doi.org/10.1299/jsmermd.2017.1P1-I12
  DOI ID：10.1299/jsmermd.2017.1P1-I12, CiNii Articles ID：130006220573
• Sequence-to-sequence models for trajectory deformation of dynamic manipulation.　　　　　　　　　　　　　　　
  Kyo Kutsuzawa; Sho Sakaino; Toshiaki Tsuji
  IECON 2017 - 43rd Annual Conference of the IEEE Industrial Electronics Society(IECON), 開始ページ：5227, 終了ページ：5232, 2017年, ［査読有り］, ［筆頭著者］
  IEEE, 研究論文（国際会議プロシーディングス）
  DOI：https://doi.org/10.1109/IECON.2017.8216904
  DOI ID：10.1109/IECON.2017.8216904, DBLP ID：conf/iecon/KutsuzawaST17
• Model predictive control based deep neural network for dynamic manipulation.　　　　　　　　　　　　　　　
  Daichi Furuta; Kyo Kutsuzawa; Tetsugaku Okamoto; Sho Sakaino; Toshiaki Tsuji
  IECON 2017 - 43rd Annual Conference of the IEEE Industrial Electronics Society(IECON), 開始ページ：5215, 終了ページ：5220, 2017年, ［査読有り］
  IEEE, 研究論文（国際会議プロシーディングス）
  DOI：https://doi.org/10.1109/IECON.2017.8216902
  DOI ID：10.1109/IECON.2017.8216902, DBLP ID：conf/iecon/FurutaKOST17
• Acceleration Control for Dynamic Manipulation of a Robot Turning over Objects
  Tsuji, T.; Kutsuzawa, K.; Sakaino, S.
  IEEE Robotics and Automation Letters, 巻：2, 号：4, 2017年, ［査読有り］
  研究論文（学術雑誌）
  DOI：https://doi.org/10.1109/LRA.2017.2720848
  DOI ID：10.1109/LRA.2017.2720848, ORCID：62121687, SCOPUS ID：85063304504
• A Control System for a Tool Use Robot : Drawing a Circle by Educing Functions of a Compass　　　　　　　　　　　　　　　
  Kutsuzawa Kyo; Sakaino Sho; Tsuji Toshiaki
  Journal of robotics and mechatronics, 巻：29, 号：2, 開始ページ：395, 終了ページ：405, 2017年, ［査読有り］, ［筆頭著者］
  Fuji Technology Press, 英語, 研究論文（学術雑誌）
  DOI：https://doi.org/10.20965/jrm.2017.p0395
  DOI ID：10.20965/jrm.2017.p0395, ISSN：0915-3942, CiNii Articles ID：40021162761, CiNii Books ID：AA10809998, ORCID：62121686, SCOPUS ID：85018714934
• Estimation of individual force at three contact points on an end-effector by a six-axis force/torque sensor.　　　　　　　　　　　　　　　
  Kyo Kutsuzawa; Sho Sakaino; Toshiaki Tsuji
  IECON 2016 - 42nd Annual Conference of the IEEE Industrial Electronics Society(IECON), 開始ページ：6409, 終了ページ：6414, 2016年, ［査読有り］, ［筆頭著者］
  IEEE, 研究論文（国際会議プロシーディングス）
  DOI：https://doi.org/10.1109/IECON.2016.7793932
  DOI ID：10.1109/IECON.2016.7793932, DBLP ID：conf/iecon/KutsuzawaST16
• Estimation of Individual Contact Force when Two Contact Points Exist during Robotic Tool Use　　　　　　　　　　　　　　　
  Kyo Kutsuzawa; Sho Sakaino; Toshiaki Tsuji
  The Abstracts of the international conference on advanced mechatronics : toward evolutionary fusion of IT and mechatronics : ICAM, 巻：2015.6, 開始ページ：46, 終了ページ：47, 2015年, ［査読有り］, ［筆頭著者］
  Japan Society of Mechanical Engineers, 研究論文（学術雑誌）
  DOI：https://doi.org/10.1299/jsmeicam.2015.6.46
  DOI ID：10.1299/jsmeicam.2015.6.46, ISSN：1348-8961, eISSN：2424-3116, CiNii Articles ID：110010043551, CiNii Books ID：AA1208830X

• Assembly robots with optimized control stiffness through reinforcement learning
  Masahide Oikawa; Kyo Kutsuzawa; Sho Sakaino; Toshiaki Tsuji
  2020年02月27日
  There is an increased demand for task automation in robots. Contact-rich
  tasks, wherein multiple contact transitions occur in a series of operations,
  are extensively being studied to realize high accuracy. In this study, we
  propose a methodology that uses reinforcement learning (RL) to achieve high
  performance in robots for the execution of assembly tasks that require precise
  contact with objects without causing damage. The proposed method ensures the
  online generation of stiffness matrices that help improve the performance of
  local trajectory optimization. The method has an advantage of rapid response
  owing to short sampling time of the trajectory planning. The effectiveness of
  the method was verified via experiments involving two contact-rich tasks. The
  results indicate that the proposed method can be implemented in various
  contact-rich manipulations. A demonstration video shows the performance.
  (https://youtu.be/gxSCl7Tp4-0)
  ORCID put code：72634452, arXiv ID：arXiv:2002.12207
• Motion Generation Considering Situation with Conditional Generative Adversarial Networks for Throwing Robots
  Kyo Kutsuzawa; Hitoshi Kusano; Ayaka Kume; Shoichiro Yamaguchi
  2019年10月08日
  When robots work in a cluttered environment, the constraints for motions
  change frequently and the required action can change even for the same task.
  However, planning complex motions from direct calculation has the risk of
  resulting in poor performance local optima. In addition, machine learning
  approaches often require relearning for novel situations. In this paper, we
  propose a method of searching appropriate motions by using conditional
  Generative Adversarial Networks (cGANs), which can generate motions based on
  the conditions by mimicking training datasets. By training cGANs with various
  motions for a task, its latent space is fulfilled with the valid motions for
  the task. The appropriate motions can be found efficiently by searching the
  latent space of the trained cGANs instead of the motion space, while avoiding
  poor local optima. We demonstrate that the proposed method successfully works
  for an object-throwing task to given target positions in both numerical
  simulation and real-robot experiments. The proposed method resulted in three
  times higher accuracy with 2.5 times faster calculation time than searching the
  action space directly.
  ORCID put code：68949512, arXiv ID：arXiv:1910.03253
• Optimized Trajectory Generation based on Model Predictive Control for Turning Over Pancakes (メカトロニクス制御研究会・実世界ハプティクス)　　　　　　　　　　　　　　　
  Tsuji Toshiaki; Kutsuzawa Kyo; Sakaino Sho
  電気学会研究会資料. MEC = The papers of Technical Meeting on "Mechatronics Control", IEE Japan, 巻：2016, 号：2, 開始ページ：73, 終了ページ：78, 2016年09月03日
  電気学会, 英語
  CiNii Articles ID：40020974470, CiNii Books ID：AA12608873
• 2P1-F07 コンパスで円を描く動作を実現する制御　　　　　　　　　　　　　　　
  沓澤 京; 大熊 隼; 境野 翔; 辻 俊明
  ロボティクス・メカトロニクス講演会講演概要集, 巻：2015, 開始ページ："2P1, 終了ページ：F07(1)"-"2P1-F07(4)", 2015年05月17日
  Robotic tool use is one of the approaches for actualizing versatility of robots. There needs to be a controller design method of utilizing characteristic functions of a tool. In this paper, the authors selected a task of drawing a circle with a compass, which is a concrete example of utilizing constraint of contact. The authors designed a coordinate system for using a compass based on a principle. The coordinate system is composed of variables of keeping constraint and creating a circular orbit and variables of moving along the circular orbit. In addition, a method of separating each contact force partially when two point contact exists is proposed.
  一般社団法人日本機械学会, 日本語
  CiNii Articles ID：110010055740, CiNii Books ID：AA11902933

• ロボットの運動制御における深層学習の応用　　　　　　　　　　　　　　　
  沓澤京
  トライボロジー会議2024 春 東京, 2024年05月, ［招待有り］
• シナジーを利用した模倣学習　　　　　　　　　　　　　　　
  沓澤京
  2023年電気学会産業応用部門大会, 2023年08月, ［招待有り］
• Synergy-Based Motor Learning for Improving the Spatial and Temporal Generalization Ability　　　　　　　　　　　　　　　
  Kyo Kutsuzawa
  The 9th IEEJ international workshop on Sensing, Actuation, Motion Control, and Optimization (SAMCON2023), 2023年03月, ［招待有り］
  英語, 口頭発表（招待・特別）
• ロボットの運動制御と深層学習技術の応用例　　　　　　　　　　　　　　　
  沓澤京
  日本結晶成長学会 新技術・新材料分科会 第2回研究会「新技術・新材料開発に向けたデータ駆動技術の応用」, 2023年03月, ［招待有り］
• 動作の時空間的構造を利用した模倣学習　　　　　　　　　　　　　　　
  沓澤京
  第69回 自律分散システム部会研究会「若手研究者による模倣学習・強化学習の新展開」, 2022年12月, ［招待有り］
  日本語
• Motor Synergy Generalization Framework for New Targets in Multi-planar and Multi-directional Reaching Task　　　　　　　　　　　　　　　
  Kyo Kutsuzawa
  The 33rd 2022 International Symposium on Micro-NanoMechatronics and Human Science, 2022年11月, ［招待有り］
  英語, 口頭発表（招待・特別）
• ロボットの運動制御における深層学習の応用　　　　　　　　　　　　　　　
  沓澤京
  トライボロジー技術へのAIの活用を考える研究会, 2021年08月, ［招待有り］
  日本語, 口頭発表（招待・特別）
• ニューラルネットワークを用いた動作特徴量の学習とロボットへの応用　　　　　　　　　　　　　　　
  沓澤京
  計測自動制御学会東北支部, 2020年12月, ［招待有り］
  日本語, 口頭発表（招待・特別）

• 2021年04月 - 現在
  機械知能・航空実験I テーマ3「線形制御」, 東北大学
• 2020年10月 - 現在
  機械知能・航空実験II テーマ5「筋電位計測の基礎と評価」, 東北大学

• 現在, IEEE
• 現在, 計測自動制御学会
• 現在, 日本ロボット学会
• 現在, 電気学会

• 運動シナジーに基づく深層学習ベース運動制御計算の高速化と体系化　　　　　　　　　　　　　　　
  日本学術振興会, 科学研究費助成事業, 基盤研究(B), 2024年04月 - 2027年03月
  林部 充宏; 沓澤 京, 東北大学, 研究分担者
  配分額（総額）：14820000, 配分額（直接経費）：11400000, 配分額（間接経費）：3420000
  課題番号：24K00841
• 力覚情報に基づく複雑な道具形状と接触位置との効率的な同時推定　　　　　　　　　　　　　　　
  日本学術振興会, 科学研究費助成事業 若手研究, 若手研究, 2022年04月 - 2025年03月
  沓澤 京, 東北大学, 研究代表者
  配分額（総額）：3250000, 配分額（直接経費）：2500000, 配分額（間接経費）：750000
  課題番号：22K14212
• Shared synergyを利用した高い汎化能力をもたらす模倣学習　　　　　　　　　　　　　　　
  戦略的な研究開発の推進 戦略的創造研究推進事業 ACT-X, 2020年 - 2022年
  沓澤 京, 東北大学, 研究代表者
  人間の教示動作をロボットに学習させる模倣学習は人間がおこなう作業の自動化に有用であると期待されますが,新規な環境に対しては動作を教示しなおす必要があります。本研究では人間の動作生成の神経的基盤である筋シナジーを利用した模倣学習システムの実現を目指します。人間は環境が変化しても共通の筋シナジーを再利用して動作生成するという知見を利用して,人間の汎化能力を再現するシステムの実現を目指します。
  課題番号：20344387
  論文ID：48148403
• ｓｅｑｕｅｎｃｅ－ｔｏ－ｓｅｑｕｅｎｃｅモデルを用いた臨機応変な物体操作　　　　　　　　　　　　　　　
  日本学術振興会, 科学研究費助成事業 特別研究員奨励費, 特別研究員奨励費, 2018年04月 - 2020年03月
  沓澤 京, 埼玉大学, 研究代表者
  配分額（総額）：1500000, 配分額（直接経費）：1500000
  本研究はsequence-to-sequence(seq2seq)モデルと呼ばれるニューラルネットワークを用いて，複数動作の臨機応変な組み合わせによる動作生成を目指すものであった。本年度においては，単一の動作を学習したseq2seqモデルを使って，モデルが獲得した動作の低次元表現（潜在表現）を利用した動作生成の研究をおこなった。
  まずは昨年度に引き続いて，訓練済seq2seqモデルの潜在表現を与えられた目的関数で最適化する研究に取り組んだ。seq2seqモデルは動力学的制約にあわせた軌道変形を学習するが，その後は訓練時とは異なる目的関数のもとで動作生成する必要がある。本研究では誤差逆伝播法に基づいた手法を提案し，軌道の終端位置の指定や制限速度の設定といった訓練時に与えていない目標への動作生成をモデルの再学習なしに実現した。
  続いて，人間からの教示動作でseq2seqモデルを訓練する研究に取り組んだ。具体事例として，画像情報を基に動作生成するタスクを扱った。実環境では画像の撮影条件が容易に変化するので，少数の教示データで学習することが求められる。本研究では，seq2seqモデルに教示動作を学習させた後，エンコーダ部分を新たなモデルに取り換えて再学習する手法を提案した。本手法は動作の潜在表現という撮影条件に不変な表現を利用することで，新規な状況への再適応を可能とする。実験によってカメラ位置などが大きく変化しても少数の教示動作から安定的に動作の再学習ができることを確認した。
  最終的に本研究によって，不連続性をもつ物理モデルや教示動作などの様々な方法によって動作学習が可能となり，さらに潜在表現を使って新たな目的関数やデータに対して動作の再適応・再学習が可能となる。これを既存技術と組み合わせることで，様々な動作を状況に応じて生成することが可能になると期待できる。
  課題番号：18J14272