Movement intention decoding based on deep learning for multiuser myoelectric interfaces
Movement intention decoding based on deep learning for multiuser myoelectric interfaces
复制标题
DOI：
10.1109/iww-bci.2016.7457459
复制
发表时间：
2016-02
期刊：
影响因子：
--
通讯作者：
Ki-Hee Park;Seong-Whan Lee
中科院分区：
文献类型：
--
作者：
Ki-Hee Park;Seong-Whan Lee
近年来，实用化肌电接口的发展导致了手臂假肢等可穿戴康复机器人的出现。本文提出了一种基于人体生物信号肌电深度特征学习的运动意图解码方法。在日常生活中，用户间的变化通过调制不同用户之间的目标肌电模式来导致性能下降。因此，我们提出了一种用户自适应的译码方法，用于在用户间变化的情况下进行稳健的运动意图译码，使用卷积神经网络进行深度特征学习，由不同的用户训练。在我们的实验结果中，提出的方法比竞争方法更准确地预测手部运动意图。
Recently, the development of practical myoelectric interfaces has resulted in the emergence of wearable rehabilitation robots such as arm prosthetics. In this paper, we propose a novel method of movement intention decoding based on the deep feature learning using electromyogram of human biosignals. In daily life, the inter-user variability cause decreases in performance by modulating target EMG patterns across different users. Therefore, we propose a user-adaptive decoding method for robust movement intention decoding in the inter-user variability, employing the convolutional neural network for the deep feature learning, trained by different users. In our experimental results, the proposed method predicted hand movement intention more accurately than a competing method.