基于多频超声检测及电刺激感知反馈的假肢手双向人机接口研究

In the field of bidirectional interface of prosthetic hands, though electromyography and electrical stimulation based bidirectional interface achieves relatively high decoding accuracy, it suffers from signal shift, electromagnetic interference and crosstalk in electromyography signals capturing period. This project proposes a novel method for bi-directionally interfacing prosthetic hands based on multi-frequency ultrasound detection and electrical stimulation feedback, taking advantage of high spatial resolution of multi-frequency ultrasound being free from electrical disturbance. Multi-frequency ultrasound signals are employed to detect the morphological changes of forearm muscle groups, further the ultrasonic signals are extracted and classified for control of prosthetic hands and electrical stimulation coding, exciting human tactile and proprioceptive feedbacks. This project gives priority to study the mapping between multi-frequency ultrasound signals and deformation of forearm muscle groups, to investigate the relation between deformation of forearm muscle groups and finger movements and grip forces, and to develop feedback strategy of combined multi-functional electrical stimulations. This project aims to realize the enhanced bidirectional interface of prosthetic hands based on multi-frequency ultrasound detection and the perceptive feedback of electrical stimulation, design a wearable bidirectional interface system with multi-frequency ultrasound detection and multi-functional electrical stimulation, and integrate sensing technique and intelligent prosthesis with a goal of a cognitive synergy between the prosthesis and the human body.

在假肢手双向人机接口研究领域，基于肌电信号和电刺激反馈的双向人机接口虽有较高解码准确率，但存在信号特征时移、电磁信号及反馈电刺激干扰肌电信号采集等问题。本项目利用超声信号可提供较高空间分辨率的肌肉动态信息且不易受电子干扰等特性，提出一种基于多频超声检测及电刺激感知反馈的假肢手双向人机接口新方法，利用多频超声检测前臂肌肉群形态，并将获取的信息进行特征提取及分类识别，用于假肢手的动作控制及感觉反馈的电刺激编码，激发人体触觉和本体感觉反馈。本项目主要研究多频超声信号与前臂肌肉群形态变化的映射规律；研究肌肉群形态变化与手指动作姿势及手指动作力量的关系；研究融合超声信号的复合电刺激感知反馈策略。通过本项目的研究，实现基于多频超声检测和电刺激感知反馈的假肢手双向人机接口，设计出可佩戴的多频超声检测及电刺激感知反馈的双向人机接口系统，实现人机接口与假肢手的集成，为假肢手与人体的有机融合奠定基础。

项目以假肢手双向人机接口为研究对象，重点开展多频超声信号与肌肉群形态变化的映射规律、超声信号对假肢手的动作与力量协同的控制方法、基于肌肉群形态变化的电刺激编码和控制方法三大关键科学技术的研究，以解决现有肌群假肢控制接口意图感知精度低、交互自然性不足、缺乏本体感觉反馈等问题，从而为高信息传输率的生机接口和假肢手本体感觉反馈提供新的技术和思路。. 项目针对骨骼肌在体测量，设计了多频穿戴式超声传感器，解决了超声骨骼肌检测过程中探测深度和精度难以兼顾的问题并提高了超声传感器的穿戴性。研究了前臂肌群形态变化与手指动作和手指力量之间的联系；重点开展了超声信号对假肢手动作与力量的协同控制和多关节腕、手运动同步控制的研究，提出了分类回归模型级联算法和多源信息融合策略实现同步的手势与手指力解码，提出了子类离散分析和线性增强训练策略实现多自由度腕、手运动同步解码。建立了基于高斯分布的手指触觉模型，揭示了手指皮肤生物力学和触觉机制的联系，并通过电刺激波形调制优化了电触觉反馈的舒适度；提出了多项式拟合和高斯混合回归级联算法，实现了肌肉神经电生理到电刺激编码的映射；探究了自主和电刺激诱发下肌肉收缩模式的相关性，建立了超声肌肉感知和电刺激参数的联系。搭建了穿戴式超声-电刺激双向假肢控制平台，包括可穿戴式超声检测系统、肌电/超声联合采集系统、肌电/电刺激双向闭环系统和轻量化经皮电触觉系统，各系统之间可无线互联。开发了嵌入式超声假肢控制算法和三维虚拟手平台，可用于生机接口康复训练。. 课题研究成果在IEEE TIE、IEEE TBME、IEEE TNSRE、IEEE SMC-A等重要学术期刊、会议发表论文37篇，SCI收录20篇。新增发明专利申请9项目，授权软件著作权1项。