神经假肢手系统的感知反馈及柔顺控制基础研究

Rehabilitation of hand functions for amputee caters to significant social benefits and demand, for which intelligent prosthetic hands seem to provide innovative solutions. Yet, even the most advanced myoelectric prosthetic hand still lacks tactile feedback and human-like compliant grip. We previously found that the amputation stumps exist a phenomenon of “evoked tactile sensation”, from which we could develop a technology of artificial sensory feedback using transcutaneous electrical nerve stimulation (TENS). Moreover, based on the principle of spinal reflex we developed a “virtual reflex” model for compliant movement control. The subsequent key issue is to obtain a control signal from the central nervous system, and produce compliant movements on prosthetic hands. The dexterous manipulation of human hand relies on continuous feedback from sensory receptors and regulation of motor outputs. Therefore when creating sensation and motion for prosthetic hands, the scientific questions of interest are how to convey specific sensory information to somatosensory cortex according to the mechanical pressure at prosthetic fingers, and how to guide cortical plasticity toward improving the controllability of amputee. This project will conduct innovative fundamental research on sensory feedback based on evoked-tactile-sensation using TENS, and virtual reflex control of compliant movement coupled with novel neural interface technology. This study will establish a two-way communication between prosthetic hand and amputee for exchanging information of sensory feedback and motor feedforward control. We seek to acquire control command signal of the central nervous system from peripheral nerve recording using the recently available carbon nanowire microelectrodes. Leveraging on the latest technology of neuromorphic model of “virtual reflex”, we aim to achieve human-like control of compliant movements in prosthetic hands. Integrating these technologies will significantly enhance the functionality and performance of neural prosthetic hand.

针对截肢者的手功能康复具有重大社会意义和需求，假肢手为他们提供了智能康复的全新手段。但目前最先进的肌电假肢手仍因缺乏触觉感知反馈和如人体般的柔和抓力。我们前期发现了截肢末端的“诱发指感”现象，据此开发了表面电刺激感知反馈技术；还根据脊髓神经反射原理建立了“虚拟反射”式柔顺运动模型。进一步的关键问题是获取中枢神经的控制信号，驱动假肢手的仿生柔顺运动。人手的灵巧操作依赖连续感觉反馈与运动控制输出。假肢手感知与控制的科学问题是如何依机械触觉在大脑皮层诱发特定的感觉效应，进而引导皮层可塑性向提高截肢者控制功能的方向进行。本项目将开展创新的表面电刺激诱发指感及神经控制和接口技术的基础研究，建立假肢手感知反馈及前馈控制的双向通道；采用新兴的碳纳米线微电极，在外周神经提取大脑的前馈控制信号；并利用最新的神经拟态芯片脊髓虚拟反射技术，实现假肢手的仿生柔顺控制。集成这些技术将大幅提高假肢手的功能和灵巧操作。

本项目针对当前商业假肢手普遍功能性低，截肢者接受率不高，使用电机刚性驱动和开环控制，缺乏感知反馈等难点，开展新一代假肢手的柔性驱动和非植入式触觉感知反馈的基础研究和关键技术开发。本项目旨在（一）探明基于诱发指感现象的电刺激触觉感知的神经机制和可行性，及其对假肢手抓物功能和辨识能力的提升；（二）利用神经肌肉反射模型，实现具有生物真实特性假肢手的实时控制，验证生物真实假肢手的类人柔顺性，评估指尖压力和手握力的可控性，抓取易碎易滑物体的精细控制能力；（三）探索人体神经系统与假肢交互中所遵循的神经规律，发现并凝练人机交互中的“神经类同性”普适原则。.项目获得成果包括：1.实现了多通道电刺激触觉反馈系统的集成；2.完成了基于脑磁图和核磁共振对诱指感现象的大脑响应研究，探明了诱指感区与大脑感觉初级皮层（primary somatosensory cortex, SI）的连续性和对应性；3.将诱指感反馈系统与商业假肢（Bebionic hand）进行了集成，并适配于两个截肢者；4.开发了硬软件去除电刺激干扰技术和间隔多通道电刺激(Interleaved stimulation)的改进方案；5.评估了触觉反馈对截肢者抓取和辨识物体的功能提升；6.在柔顺驱动控制研究上，建立了神经拟态反射模型与线驱动假肢手耦合的力学模型；7.在虚拟现实中全面评估了一对神经拟态拮抗肌控制的手指关节系统的刚度自适应可调性；8.集成了一台单侧柔顺控制的假肢手雏型机，并在5名截肢者操控下评估了手指尖力和手握力的可控能力，显示了与健侧手相近的力控能力；9.招募11名截肢者，开展了抓物功能的评测，展示了生物真实假肢系统前所未有的精细抓取能力；10.发现并凝练出了神经康复中人机交互的“神经类同性”普适原则。该原则不仅适用于假肢手设计，也适用于人体功能康复训练和辅助的神经康复系统。.本项目首次展示了发展新一代无创感知柔顺控制假肢手的可行性；首次验证了感知反馈和柔顺控制赋予假肢手抓物功能和辨识能力的跨越提升；为提升我国假肢产业水平，奠定了理论基础和技术储备。