基于生物信息感知的双侧镜像康复机器人控制方法研究

It is reported that about 2/3 of Chinese apoplexies are disenable in their upper limb activity. How to help these rehabilitants becomes a fundamental and urgent work for the researchers. The existing rehabilitation robot systems for upper limb mainly work on the CPM principle, but only consider training a single side instead of the other healthy one. Moreover, patients have seldom chances to participate in the rehabilitation. Most of human activities depend on the cooperation of their hands. During these activities, all the effects show a synchronous trend, such as the synchronization of frequency and phase between limbs. This demonstrates that the mirror symmetrical movement is the most typical movement patterns.Through the study of physiological information perception, the modeling of the robot, the robot flexibility control, and three issues of safety, comfort and effectiveness that rehabilitation robots faced will partly solved. This project proposed the sEMG signal modeling approach based on muscle physiology and EMG characteristics, sEMG based pain analysis methods, and parameter estimation based human-robot integration initiative modeling method. Base on the above theoretical approach, biological perception based patient's active control unit technology and pain feedback-based rehabilitation robot safety control unit will be established. And further build the bilateral mirror exercise therapy based rehabilitation robot system, which should enhance the effect of rehabilitation of patients. This project will provide the theoretical approach and technical support for the national strategic goals,' Everyone has the right to rehabilitation services'.

我国2/3的脑卒中患者有一侧上肢功能障碍，如何有效促进患者上肢功能恢复,是康复领域的重要课题。现有上肢康复机器人基本是基于CPM原理，进行单侧训练，未考虑健侧运动的影响，患者主动参与度低。人们日常生活中的绝大多数活动是由双手协调来完成的,肢体同时活动时,各效应器之间会有同步化趋势,如肢体频率和相位的同步，说明镜向性对称运动是人体最典型的运动模式。本项目针对双侧镜像康复过程中的生理信息感知、机器人建模、机器人关节柔顺性控制等方面的问题开展研究，提出基于肌肉生理结构与肌电特征融合的肌电信号建模方法，基于肌电和心率变化的疼痛感分析方法，以及基于参数估计的人机一体化主动建模方法。以上述理论方法为基础构成基于生物感知的患者主动控制单元技术和基于疼痛感反馈的康复机器人安全控制单元技术，进一步构建基于双侧镜像运动疗法的康复机器人系统。为实现"人人享有康复服务"的国家战略目标提供理论方法基础和技术支撑。

脑卒中等易引起瘫痪的疾病极大地影响了患者的日常生活。约有2/3的脑卒中患者会有一侧上肢的功能障碍。现有上肢康复机器人基本是基于 CPM 原理，进行单侧训练，未考虑健侧运动的影响。本项目利用人体特有的生物信息分析人体在康复过程中的特定状态，并基于双侧镜像康复原理促进患者的运动功能康复。1）利用人体的表面肌电信号，建立了肌电信号与人体关节运动量的对应关系。针对表面肌电信号噪声大、电极易脱落的特点，研究了肌电信号在传感器故障下的容错分类方法，提高肌电应用的鲁棒性与实用性。提出了运动单元动作电位分解方法，并用于关节运动的识别，减少了所需传感器数量。2）提出了基于表面肌电信号分析的康复疗效评估方法，并应用于面瘫康复效果评估，取得了积极的效果。3）搭建了基于功能电刺激的运动功能康复实验平台，实现了刺激电流的波形、强度、脉宽、占空比、频率、通道数等参数的在线实时调整。根据肘关节动力学模型，采用迭代学习控制策略进行上肢运动功能康复实验，实现了基于功能电刺激的上肢肘关节康复运动。4）建立了人体生理信号与疼痛感觉之间的关系。通过采集多源信息，同时对被试者进行定量的疼痛诱导，采用模式识别算法建立生理信号与疼痛感之间的联系。基于识别出的人体疼痛等级，进行康复训练模式的调整，实现康复疗程的在线优化。上述成果为进一步建立更为完善的康复系统提供了坚实的理论与技术基础。