康复增力型穿戴式下肢外骨骼系统设计及其抗扰协同控制研究

In recent years, the exoskeleton robots, because of its widely applications in the help of aged people, disabled man, and the improvement of combat capability of the army single man, has become a hot research topic in international academic and engineering world. This project plans to deal with their deficiencies of its relative unitary function structure, complex control algorithms and poor general adaptations, focuses on the design and disturbance rejection collaborative control research of wearable exoskeleton system with integration of functions of health rehabilitation and human performance enhancements. Firstly, the project plans to begin with the study of health and disabled human gait, movement and action mechanism of typical force, then the design of exoskeleton system, and the establishment of the exoskeleton systems modeling. Secondly, with considerations of the expensive force and torque sensors and the difficult extraction and applications of sEMG signals, we aim to develop nonlinear observers for human motion intention based on the joint position feedbacks to reduce the system costs and improve the reliability of system. Thirdly, for highly interactive uncertain nonlinear exoskeleton system, we combine with the referred nonlinear state observer, introduce ultra-local lower order model, time delay estimation techniques etc, to build a relative new disturbance rejection collaborative control which is independent to the parameters of the controlled system, anti-external disturbances and robust to system parameters variations, and employs only the input and output datas of controlled system. Finally, with dSPACE based technology, we verify and optimize further the developed theoretical results. The successful implementation of this project will further enrich and improve the service robot exoskeleton theory, promote our country development of people-centered intelligent information processing and control technology.

近年来，外骨骼机器人因能广泛应用于助老、助残及提高单兵作战能力等，已成为国际学术与工程界的研究热点。本项目拟针对其功能结构相对单一、控制算法复杂且普适应较差等不足，重点研究集康复与増力为一体的穿戴式下肢外骨骼系统设计及其抗扰协同控制。首先，从研究人体下肢的运动机理与典型受力出发，进行符合功能要求的系统设计与建模。其次，针对力及力矩传感器昂贵，肌电信号提取使用困难，研究基于位置反馈反映人体运动意图的非线性状态观测器，以降低成本与提高可靠性。然后，针对高度交互不确定的非线性外骨骼系统，引入极局部低阶模型降阶、时延估计等，构造与被控系统参数无关、抗外界干扰与系统参数摄动仅需系统输入输出数据的新型协同控制策略。最后，基于dSPACE实时测控技术验证并优化理论成果。本项目的成功实施，将进一步丰富与完善外骨骼机器人理论，推动我国基于生物特征为基础“以人为中心”的智能信息处理和控制技术的跨越式发展。

本项目在我国老龄化加剧、由疾病灾难所造成残障人士逐年增加，新形势下军队提高单兵作战能力需要的背景下，针对现有下肢外骨骼系统本体结构功能单一、控制算法复杂、针对不同人群普适性较差等不足，重点研发集康复运动与增力型为一体的下肢外骨骼智能系统及其无模型抗扰协同控制技术，成功解决了下肢外骨骼系统存在的部分关键技术难题。首先，基于人体下肢运动机理进行了拟人化结构设计，完成了12自由度康复增力型下肢外骨骼的本体结构设计及虚拟样机实现，并建立了相应的运动学和动力学模型。其次，选取与定位外骨骼的传感、驱动及能源部件后，对下肢运动意图进行了基于表面肌电信号的离线与在线识别研究，实现了基于频率估计和基于知识库与特征匹配的下肢膝关节运动意图预测。同时建立了基于位置反馈的人机交互力/力矩信息等反映人体运动意图的非线性状态观测器，消除人体下肢颤动及环境干扰对人体运动意图识别的影响。接着，针对下肢外骨骼轨迹跟踪与辅助策略研究，系统提出了基于极局部低阶模型和时延估计技术的无模型抗扰动自适应协同控制方法。针对不同康复阶段患者的个性化的康复需求，提出了基于轨迹跟踪误差、肌肉力矩、运动速度和人机交互力等多指标人体运动能力评估方法，设计自适应按需辅助康复策略。最后，借助新一代dSPACE/Matlab/Simulink技术，构建了12自由度康复增力型下肢外骨骼智能系统。本项目从理论推导、虚拟样机联合仿真、实物样机测试等多方面验证了所设计算法的有效性和优越性。本项目所提出的外骨骼系统设计方法、人体运动意图识别方法、无模型抗扰动自适应协同控制、按需辅助控制等相关技术可进一步推广到其他类型外骨骼，更好地满足康复运动或增力的控制需求。此外，本项目在国内外核心期刊与会议上共发表学术论文48篇，其中SCI高水平期刊论文38篇，期刊杂志及国际会议EI论文10篇，申请国家发明专利9项。依托本项目共培养博士研究生6名，硕士研究生11名（其中已经毕业的博士3名，毕业的硕士11名）。