灵巧假肢的多源生机操控及本体感反馈研究

Towards the urgent needs for hand rehabilitation through our country, this project conducts research on the perceptional information acquisition, adaptive multi-source bio-medical signal processing and control, and the proprioceptive feedback of the dexterous hand prosthesis (DHP), to fill-up the deep gap between high hand dexterousness and low control intuitiveness. Following the normal perceptional pathway of our hand interacting with the outer environment, this project starts from solving the sensor integration issue on the multi-sensory (position, force, and tactile) prosthetic hand; then aiming to establish a more nature, suitable amputee-DHP interaction strategy, it thoroughly probes the DHP’s sensor configuration, motion decoding and sensory feedback. The research can be detailed as: 1) multi-sensor integration and fusion; 2) motion decoding on multiple peripheral nerve signals; 3) adaptive bio-machine interface; 4)high-suitability sensory feedback based on transcutaneous electrical nerve stimulation (TENS); and 5) “human-in-the-loop” interaction strategy and motor system rehabilitation. The research gain of this project can help speed up the commercialization pace of our nation’s high-performance prosthetic hand product, as well as expend the range of the mechatronic research and promote novel principles on artificial perception, human-machine interaction (HMI) and bio-mechatronic integration.

面向国家残疾人群康复的迫切需要，本项目聚焦研究灵巧型假肢的感知信息获取方法、自适应多源生物电控制策略以及本体感觉反馈机制，以期填补当前多自由度假肢本体灵活性与控制感官性之间的鸿沟。该项目遵循人手在与外部环境交互时的感知信息流向，以建立高感官性、高适宜性生机交互策略为总体目标，深入探讨感知型灵巧假肢的控制系统设计、运动意图解码和感知反馈机制。具体研究内容包括：1)多传感器信息集成及融合方法；2)人体多源外周神经信号联合解码；3）自适应生机接口；4）高适宜性经皮电刺激感知反馈；以及5)“人在闭环”生机交互策略及“本体感”康复实验。该项目的实施不仅有望推动我国智能假肢产品化进程，而且有助于拓展机械电子学科的内涵，并促进人工感知、人机交互以及“生-机-电”集成新原理的发现。

面向国家残疾人群康复的迫切需要，本项目聚焦研究灵巧型假肢的多模态外周信息融合方法、自适应生机接口和本体感康复反馈机制，缓解了当前多自由度假肢本体灵活性与控制感官性之间的矛盾。该项目遵循人手在与外部环境交互时的感知信息流向，以建立高感官性、高适宜性生机交互策略为总体目标，深入探讨了感知型灵巧假肢的控制系统设计、多自由度运动意图解码和电刺激感知反馈机制。具体上，本项目提出了适用于多源外周神经信号的人体运动信息解码算法，解决了多模信息特征提取、决策融合等技术难题；研究了生物信号时变特性对假肢控制性能的影响，构建了具有自适应调节功能的运动解码模型和非监督式机器学习算；探讨了感知反馈适宜性和多通道经皮电刺激参数之间的关联关系，揭示了不同电刺激时空模式对受损运动神经系统的康复机制。共计发表学术论文28篇（其中SCI检索16篇，EI检索12篇），授权发明专利3项，出版专著1部。