原子运动神经机理与动力外骨骼的脑机协调控制方法研究

Powered exoskeleton, as a kind of intelligent mechatronic system designed and manufactured to assist or to enhance the motor function of the human body, is now stepping from its prototype to the process of industrializationn. Achieving the coordinated control of the human-machine combined exoskeleton system is difficult but decisive to make that the powered exoskeleton a practical and promising technology, the key of which is to model the motor system of humans accurately and predict the complicated and uncertain voluntary movements in the real time. Current research shows that the motor system of humans generates complex motor behavior via generating and combining discrete and simple elementary units of movement (or submovements) whose features can be modulated depending on the motor task, which is completely different with the mainstream methods of robot controlling. Studying on the neural basis of submovements in humans may, in our opinion, be a new and promising way in designing a effective coordinated controller of a human-machine combined exoskeleton system, or even a traditional robot system. According to the former analysis on achieving human-machine control of the powered exoskeleton, and based on the study of brain computer interface(BCI) in the recent years, this project plans to research on the neural basis of the submovements(including how they are generated, switched, cohered or synchronized) in the human motor system by means of electrophysiological measurements including EMG (Electromygraphy) and EEG (Electroencephalograph); to develop classification algorithms to probe and predict the human submovemtns; to achieve effective brain-computer controller of the powered exoskeleton corresponding to the neural mechanisms of motor system of humans. These studies could probably be meaningful in the realization of future human-machine warrior, or the intelligent prosthesis for handicapped people.

动力外骨骼是一种辅助和增强人体运动机能的智能化机电装置，目前正从原型系统阶段向产业化方向迈进。而人机协调控制系统设计是制约外骨骼研究走向实用的核心和难点问题，其关键在于如何对高度复杂和不确定的人体自主运动进行准确建模和实时预测。研究表明，人类神经系统通过实时产生和调制一系列简单、离散的原子运动模式而实现复杂和连续动作，这与当前机器人控制的主流方法有着完全不同的机理。研究人类这种原子运动神经控制机理，有可能为人机外骨骼协调控制和机器人控制探索新的思路。本项目立足于脑机接口领域的工作基础，围绕实现外骨骼控制的技术目标，拟通过肌电、脑电等神经电生理实验，探索人体运动协调控制的神经机理，研究原子运动产生、转换、衔接和同步等过程的内在规律，实现通过神经电信号来检测和预测原子运动的分类算法，建立符合人类运动神经机理的高效外骨骼脑机协调控制方法，为未来"机器战士"和智能假肢等研究储备相关理论和关键技术。

本项目从实现外骨骼控制的技术目标出发，在人体原子运动神经机理的电生理实验设计、原子运动模式的在线识别算法研究、外骨骼控制方法原型实现以及外骨骼控制方法的实物仿真实验等方面取得了丰硕的研究成果。设计和实施了一系列多种类型人体原子运动神经电生理实验，深入探索了人体运动协调控制的神经机理以及原子运动产生、转换、衔接和同步等过程的内在规律。研究了基于神经活动信号的人体原子运动模式在线识别算法，实现通过神经电信号来检测和预测原子运动的分类算法，并通过计算机仿真和进一步的运动电生理实验对所设计的算法进行验证和优化。通过分析EEG、EMG中所包含的不同层次人体运动控制信息，研究基于EMG、EEG等多模态信息融合的多层次的人机一体化外骨骼协调控制策略，建立原子运动模式与外骨骼控制指令之间的映射关系，实现人机协调一致的外骨骼控制方法原型。建立外骨骼控制实物/半实物仿真实验平台，提出符合人类运动神经机理的高效外骨骼脑机协调控制方法，同时引入共享控制技术，建立了基于共享控制的机械臂脑机操控系统，充分发挥了两种智能体各自的优势，提高人机交互的效率和操作性。在项目研究期间，课题组共发表SCI期刊论文10篇，会议论文2篇，并在项目资助下申请并授权脑机接口相关国家发明专利5项。本项目的研究成果为未来“机器战士”和智能假肢等研究提供了重要的理论与技术储备。