Article
Engineering, Multidisciplinary
Jiaqi Wang, Dongmei Wu, Yongzhuo Gao, Xinrui Wang, Xiaoqi Li, Guoqiang Xu, Wei Dong
Summary: This paper proposes an integral subject-adaptive real-time Locomotion Mode Recognition (LMR) method based on GA-CNN for a lower limb exoskeleton system. It shows high accuracy, low delay, and sufficient adaption to different subjects.
JOURNAL OF BIONIC ENGINEERING
(2022)
Article
Robotics
Yuepeng Qian, Yining Wang, Chuheng Chen, Jingfeng Xiong, Yuquan Leng, Haoyong Yu, Chenglong Fu
Summary: This study develops a high-level exoskeleton control method that uses a depth sensor to detect changes in locomotion mode and an adaptive oscillator to accurately estimate the user's gait phase. Experimental results show that this method has higher accuracy and predictive capability than existing methods.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2022)
Article
Engineering, Biomedical
Inseung Kang, Dean D. Molinaro, Gayeon Choi, Jonathan Camargo, Aaron J. Young
Summary: This study introduces a deep convolutional neural network-based locomotion mode classifier for hip exoskeleton applications, which overcomes the limitations of current exoskeleton systems and achieves optimized performance. The results show that the model has a high classification accuracy and can adapt to different environments.
IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING
(2022)
Article
Engineering, Biomedical
Mohammad Shushtari, Hannah Dinovitzer, Jiacheng Weng, Arash Arami
Summary: An ultra-robust accurate gait phase estimator, trained on data from hip and knee joint angles of 14 participants, was developed to estimate gait phases during treadmill and overground walking. The estimator showed uniform spatial and temporal performance across participants and gait conditions, and demonstrated robustness to various walking conditions and interactions with an exoskeleton.
IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING
(2022)
Article
Biophysics
Jianbin Zheng, Mingpeng Peng, Liping Huang, Yifan Gao, Zefang LI, Binfeng Wang, Yu Wang
Summary: Human activity intention is crucial for the control of wearable powered lower extremity exoskeletons. This study proposes a hybrid CNN-SVM model based on multi-channel IMU signals to identify different locomotion modes of the human body, achieving high recognition rates and real-time performance.
JOURNAL OF MECHANICS IN MEDICINE AND BIOLOGY
(2022)
Article
Chemistry, Analytical
Jiale Ren, Aihui Wang, Hengyi Li, Xuebin Yue, Lin Meng
Summary: Lower limb exoskeleton robots have significant research value in improving physical motion functions. In this study, a transformer-based neural network was developed to predict hip and knee joint angles using plantar pressure. The model showed improved performance compared to other models in terms of prediction accuracy.
Article
Engineering, Biomedical
Woolim Hong, Jinwon Lee, Pilwon Hur
Summary: This study proposes a new labeling method to improve the accuracy of estimating the user's gait phase based on variable toe-off onset at different walking speeds. Three models were trained using long short-term memory, and the proposed labeling method significantly improved the estimation accuracy. The proposed method also performed well in detecting heel-strike and toe-off.
IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING
(2023)
Article
Engineering, Biomedical
Shuangyue Yu, Jianfu Yang, Tzu-Hao Huang, Junxi Zhu, Christopher J. Visco, Farah Hameed, Joel Stein, Xianlian Zhou, Hao Su
Summary: This paper presents a high-accuracy gait phase estimation and prediction algorithm based on a two-stage artificial neural network. The algorithm uses a portable controller with only two IMU sensors to estimate and predict the gait cycle in real time. It can detect and classify gait phases in unrhythmic conditions, and also predict future intra-and inter-stride gait phases.
ANNALS OF BIOMEDICAL ENGINEERING
(2023)
Article
Chemistry, Analytical
Jiwoo Choi, Sangil Choi, Taewon Kang
Summary: In this paper, a smartphone authentication system based on human gait is proposed. By learning human gait features and implementing filtering techniques, the system can accurately identify legitimate users. The study demonstrates the possibility of using human gait as a new user authentication method with high reliability.
Article
Chemistry, Multidisciplinary
Zhuo Qi, Qiuzhi Song, Yali Liu, Chaoyue Guo
Summary: This research utilizes a self-developed wearable lower limb exoskeleton system to propose a locomotion mode recognition algorithm for accurately and quickly recognizing the motion pattern of an exoskeleton robot. The algorithm combines a finite state machine with a hierarchical support vector machine to recognize five typical locomotion modes and eight locomotion mode transitions. Input signals include angle information from hip and knee joints, collected by inertial sensing units, and plantar pressure information collected by force sensitive resistors. Experiments on six subjects demonstrate high accuracy and efficiency of the algorithm.
APPLIED SCIENCES-BASEL
(2022)
Article
Engineering, Biomedical
Yue Wen, Sangjoon J. Kim, Simon Avrillon, Jackson T. Levine, Francois Hug, Jose L. Pons
Summary: This study proposes a novel approach using a deep convolutional neural network to estimate the neural drive. The deep CNN can identify the cumulative spike train through general features of motor unit action potentials, allowing for generalization across different contraction tasks. The results show that the deep CNN provides accurate estimations of the cumulative spike train with low root-mean-square error.
IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING
(2022)
Article
Automation & Control Systems
Xinyu Wu, Ye Yuan, Xikun Zhang, Can Wang, Tiantian Xu, Dacheng Tao
Summary: This article presents a graph convolutional network model for gait phase classification and verifies its real-time performance on a lower limb exoskeleton. The experimental results demonstrate that the proposed model achieves higher prediction accuracy and robustness in gait phase classification for different individuals in various environments.
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS
(2022)
Article
Computer Science, Information Systems
Ming-Hang Tan, Qin-Mu Wu, Zhi-Qin He, Lin Li, Bing Qiu, Chun-Shan Luo, Yi-Ming Xu
Summary: In this study, an oscillator-based hybrid gait phase estimation method was proposed for hip assistive exoskeletons. The method processed the raw data of hip angle with Butterworth low-pass filtering, and accurately divided each gait cycle by combining hip angle, hip angular velocity, and plantar pressure. The experiments showed that the proposed method had lower phase estimation errors and improved the gait feature estimation performance compared to other methods.
Article
Environmental Sciences
Liming Pu, Xiaoling Zhang, Zenan Zhou, Liang Li, Liming Zhou, Jun Shi, Shunjun Wei
Summary: This paper introduces a robust least squares phase unwrapping method based on deep learning, which can achieve more accurate and robust results by learning global phase features and the phase gradient between adjacent pixels compared to traditional methods.
Article
Engineering, Biomedical
Jinwon Lee, Woolim Hong, Pilwon Hur
Summary: User gait phase estimation is crucial for the seamless control of lower-limb robotic assistive devices during ambulation. This study investigated different sensor setups for machine learning approach to achieve more accurate gait phase estimation for robotic transfemoral prosthesis at different walking speeds, showing robust and accurate results. The choice between the two sensor setups can depend on researchers' preference in consideration of device setup or focus of interest.
IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING
(2021)
Article
Medicine, General & Internal
Krishan Bhakta, Jonathan Camargo, Pratik Kunapuli, Lee Childers, Aaron Young
Article
Engineering, Biomedical
Dawit Lee, Eun Chan Kwak, Bailey J. McLain, Inseung Kang, Aaron J. Young
IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING
(2020)
Article
Engineering, Biomedical
Dawit Lee, Bailey McLain, Inseung Kang, Aaron Young
Summary: This study compared the biomechanical effects of three different assistance strategies on a knee exoskeleton system and found that all strategies significantly reduced metabolic cost compared to the unpowered condition, with no significant differences between strategies. Powered extension assistance during early stance was shown to improve performance, but user control of assistance may not be significant when walking on an inclined surface with a knee exoskeleton.
IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING
(2021)
Article
Engineering, Biomedical
Yi-Tsen Pan, Inseung Kang, James Joh, Patrick Kim, Kinsey R. Herrin, Trisha M. Kesar, Gregory S. Sawicki, Aaron J. Young
Summary: Powered assistance from hip exoskeletons is an effective means to increase walking speed post-stroke, and tuning the balance of assistance between the non-paretic and paretic limbs (i.e., a bilateral strategy) may be the most effective way to maximize performance gains.
ANNALS OF BIOMEDICAL ENGINEERING
(2023)
Article
Engineering, Biomedical
Inseung Kang, Dean D. Molinaro, Gayeon Choi, Jonathan Camargo, Aaron J. Young
Summary: This study introduces a deep convolutional neural network-based locomotion mode classifier for hip exoskeleton applications, which overcomes the limitations of current exoskeleton systems and achieves optimized performance. The results show that the model has a high classification accuracy and can adapt to different environments.
IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING
(2022)
Article
Engineering, Biomedical
Inseung Kang, Pratik Kunapuli, Aaron J. Young
IEEE TRANSACTIONS ON MEDICAL ROBOTICS AND BIONICS
(2020)
Proceedings Paper
Engineering, Biomedical
Inseung Kang, Pratik Kunapuli, Hsiang Hsu, Aaron J. Young
2019 IEEE 16TH INTERNATIONAL CONFERENCE ON REHABILITATION ROBOTICS (ICORR)
(2019)
Article
Robotics
Inseung Kang, Hsiang Hsu, Aaron Young
IEEE ROBOTICS AND AUTOMATION LETTERS
(2019)