基于主成分分析的灵巧假肢手仿生控制器

Bio-inspired controller for a dexterous prosthetic hand based on Principal Components Analysis.

作者信息

Matrone G, Cipriani C, Secco E L, Carrozza M C, Magenes G

机构信息

Department of Computer Engineering and Systems Science, University of Pavia, via Ferrata 1, Pavia, Italy.

出版信息

Annu Int Conf IEEE Eng Med Biol Soc. 2009;2009:5022-5. doi: 10.1109/IEMBS.2009.5333826.

DOI:10.1109/IEMBS.2009.5333826

PMID:19964659

Abstract

Controlling a dexterous myoelectric prosthetic hand with many degrees of freedom (DoFs) could be a very demanding task, which requires the amputee for high concentration and ability in modulating many different muscular contraction signals. In this work a new approach to multi-DoF control is proposed, which makes use of Principal Component Analysis (PCA) to reduce the DoFs space dimensionality and allow to drive a 15 DoFs hand by means of a 2 DoFs signal. This approach has been tested and properly adapted to work onto the underactuated robotic hand named CyberHand, using mouse cursor coordinates as input signals and a principal components (PCs) matrix taken from the literature. First trials show the feasibility of performing grasps using this method. Further tests with real EMG signals are foreseen.

摘要

控制具有多个自由度（DoFs）的灵巧肌电假肢手可能是一项极具挑战性的任务，这要求截肢者具备高度的专注力以及调制多种不同肌肉收缩信号的能力。在这项工作中，提出了一种新的多自由度控制方法，该方法利用主成分分析（PCA）来降低自由度空间的维度，并允许通过一个双自由度信号驱动一个具有15个自由度的手。此方法已通过测试，并经过适当调整，以应用于名为CyberHand的欠驱动机器人手，使用鼠标光标坐标作为输入信号，并采用从文献中获取的主成分（PCs）矩阵。初步试验表明了使用该方法进行抓握的可行性。预计将使用真实肌电信号进行进一步测试。