下肢康复外骨骼机器人关键控制理论和技术研究

The human lower limb injury caused by disease and abnormal motion is ascending. Lower extremity exoskeletal robots are valuable in theoretical research and application. But the current lower extremity exoskeletal robots for rehabilitation usually have limited functions, most of which exhibit poor performance during active rehabilitation training in daily life. In this project the robust coordinated control theory research of complex systems with biological uncertainty for lower extremity exoskeletal robots would be conducted aiming to assist the rehabilitation of the human motion system disorder. The patient motion sensing system would be optimized to achieve more accurate lower limb injury rehabilitation evaluation. The human natural and active walking, as while as the reaction on external assist under motor control of central nervous system would be studied, the human machine physical and information interaction mechanism would be researched, and the characteristics and control method of the human-machine coupling system would be explored based on the human-robot-human motor control system coupled model. By elevating the human motion sensing ability of the robot and the research of robot motion feedback method to the human nervous system the two-way human machine information interaction would be improved. By exploiting the motion prediction and compensation method based on human lower limb motion coupling and nonlinearity features the fluency and robustness of human machine interaction were expected to be elevated. Thus, accurate evaluation of the patients’ rehabilitation progress, real-time sensing of the human motion, precise identification of the patients’ intention, coordinated control of the human machine coupled system would be realized which result a better rehabilitation.

由各种疾病和不良运动引起的人体下肢损伤呈现逐年增多的趋势。下肢康复外骨骼机器人具有重要的理论研究和应用价值。然而，目前的下肢康复外骨骼机器人功能相对单一，更为重要的日常下肢主动康复训练功能还很不完善。本项目拟针对人体运动系统功能障碍，开展下肢康复外骨骼机器人面向不确定性因素的复杂系统鲁棒协调控制理论研究。优化患者状态感知系统，准确评价康复进程；研究人体运动控制神经调节下自然、主动行走和助力应激反应规律、人机物理和信息交互机制，建立人-机-人体运动控制神经耦合模型，探索系统特性和控制方法；提升机器人人体运动感知能力，研究机器人运动信息向人体感觉神经反馈的机制和方法，改善人机信息双向交互；通过人体下肢运动耦合非线性特性的运动预测补偿以及多模态信息融合方法，提高人机交互流畅性和鲁棒性。以达到患者康复状态准确评价，运动状态实时感知，运动意图精确辨识，人机系统协调控制的目的，实现良好的康复治疗效果。

下肢康复外骨骼机器人是针对运动神经系统损伤患者进行功能康复训练的重要技术装备，通过辅助患者行走，有效促进神经系统恢复，从而达到治疗目的。本项目针对康复训练患者个体差异性大，训练需求不同，人机自然交互，力位柔顺控制困难等关键技术问题，开展理论研究、算法开发与样机研制工作，取得了一系列突破性进展。具体工作如下：首先，结合下肢运动类周期与强非线性特征，研究了多传感器融合的人机交互力与患者状态感知方法，将人机交互力从外骨骼运动扭矩中分离出来，从而实现了针对特异患者的评估与智能康复策略等制定；其次，构建了人体骨骼肌肉系统-运动神经控制系统模型，并研究了外骨骼机器人的非简化三维多刚体动力学建模方法与动态参数在线辨识方法，采用高斯回归研究了不同速度下的交互参数的变化规律，实现了理想的人机自然交互；然后，针对下肢康复外骨骼机器人不确定性控制问题，开展了鲁棒协调控制方法研究，并从多智能体层面开展研究工作，解决关节控制、双腿助力协调、时滞、不确定、颤振、饱和、信号滤波等诸多理论与应用问题；最后，完成了人体生物不确定性的原理样机的集成设计与优化。本项目共发表SCI期刊论文62篇，申请专利14项。康复机器人产品作为项目成果之一正在国内多个临床中心进行临床测试，预计会在2022年年底拿到产品的医疗器械注册许可证和生产许可证。