IEEE Trans Neural Syst Rehabil Eng. 2014 Mar;22(2):249-57. doi: 10.1109/TNSRE.2013.2260172.
An ideal myoelectric prosthetic hand should have the ability to continuously morph between any posture like an anatomical hand. This paper describes the design and validation of a morphing myoelectric hand controller based on principal component analysis of human grasping. The controller commands continuously morphing hand postures including functional grasps using between two and four surface electromyography (EMG) electrodes pairs. Four unique maps were developed to transform the EMG control signals in the principal component domain. A preliminary validation experiment was performed by 10 nonamputee subjects to determine the map with highest performance. The subjects used the myoelectric controller to morph a virtual hand between functional grasps in a series of randomized trials. The number of joints controlled accurately was evaluated to characterize the performance of each map. Additional metrics were studied including completion rate, time to completion, and path efficiency. The highest performing map controlled over 13 out of 15 joints accurately.
理想的肌电假肢手应该具备像解剖手一样连续变换任意姿势的能力。本文描述了一种基于人类抓握主成分分析的变形肌电手控制器的设计和验证。该控制器使用两个到四个表面肌电图 (EMG) 电极对，指挥连续变形的手姿势，包括功能抓握。开发了四个独特的图谱来转换主成分域中的 EMG 控制信号。通过 10 名非截肢受试者进行了初步验证实验，以确定性能最高的图谱。受试者使用肌电控制器在一系列随机试验中使虚拟手在功能抓握之间变形。评估所控制的关节数量以表征每个图谱的性能。研究了其他指标，包括完成率、完成时间和路径效率。性能最高的图谱能够准确控制 15 个关节中的 13 个以上。
