人机交感假手的肌电解耦自学习动作识别及多模态信息反馈方法研究

In order to meet the demand of the growth of the EMG prosthetic hand, improve the applicability of the prosthetic hands to different individuals and enhance the capability of information interaction between the prosthetic hands and the users, the research on EMG decoupling, motion self-learning recognition and the multimodal information perception feedback is conducted in this project. Firstly, for the reason of the existence of the aliasing interference among multi-channel EMG signals, an EMG decoupling model is established and experimental calibration and multivariate analysis method are employed to suppress the aliasing interference. Then, to solve the problem of EMG motion recognition depending on the individual and measurement position, an EMG self-learning motion recognition method is proposed to remove the individual differences and the measurement position differences of the EMG signals. With the help of the EMG self-learning method, the range of the application of the EMG prosthetic hand is expanded and the adaptive time of the amputees using prosthetic hand is cut down. Finally, focusing on at the problem of lacking proprioceptive sensation, the research of multimodal information perception feedback is conducted so that the users can perceive the grasping state of the prosthetic hand and improve the control performance. The implementation of the project will promote the development of the EMG prosthetic hand to improve quality of life of the amputees.

本项目针对肌电假手功能需求的增长，以提高肌电假手对不同残疾人的适应性、增强假手和佩戴者之间的信息交互能力为目标，对肌电信号的解耦、动作自学习识别、多模态信息感知反馈等肌电假手关键技术展开研究。首先，因皮肤表面测量得到的多路肌电信号间存在混叠干扰，拟采用实验标定、多元数分析方法建立肌电信号解耦模型，消除混叠带来的信息冗余并为后续动作识别提供优质信号。然后，为解决肌电信号动作识别过程中强烈依赖于个体和测量位置的问题，研究一种肌电信号自学习动作识别方法，使假手的控制参数自主匹配残疾人的肌电信号，抑制肌电信号个体差异和假手重复穿戴造成的测量位置差异，缩短使用假手的学习适应时间。最后，针对假手缺乏本体感的问题，进行从假手到佩戴者的多模态信息感知反馈方法研究，使佩戴者及时准确地感知假手的状态，实现人机协同共融，从而提高假手的控制效果。本项目将促进肌电假手的发展，为残障人士提供更为理想的帮助与支持。

本项目针对残疾人使用肌电假手过程中存在的问题开展研究工作。首先，自主研制了多通道表面肌电信号采集系统，实现了人体微弱肌电信号的采集；采用实验标定方法开展了多通道肌电信号的解耦研究，抑制了多通道肌电信号间混叠带了的冗余信息。开展基于肌电信号和神经网络的手部动作识别研究，并采用方差分析方法得到了不同测量位置下的最优肌电特征组合和不同肌电特征下的最优测量位置组合，实现了手部动作识别的高效识别。其次，设计了一种机械传感一体化的多自由度仿生灵巧手，实现了仿生手对不同外形物体的自适应抓取；开展了基于抓取对象刚度辨识的假手控制方法研究，实现了假手对不同物体的柔顺抓取。最后，设计了一种基于微型振动器的手臂力触觉反馈装置，开展了假手握力的力触觉感知反馈研究，提高了假手的可交互性。