【作者】 黄波；

【导师】 熊蔡华；

【作者基本信息】 华中科技大学 ， 机械电子工程， 2023， 博士

【摘要】 人体下肢卓越的运动能力为人类的健康生活与工作提供了重要保障,探索人体运动的奥秘,用工程科学的方法修复与增强人体下肢运动能力,已成为当前穿戴式下肢外骨骼机器人领域的研究热点。本论文围绕“如何建立下肢外骨骼设计方法,保障人机运动自然匹配,支撑下肢外骨骼的有效运动辅助”这一挑战性难题,聚焦人体下肢自然运动的规律解析和人体下肢自然运动的机械复现两个共性基础问题,开展了系统深入的研究。论文的主要创新性研究如下:针对下肢自然运动生成机理不清晰的问题,研究并揭示了下肢自然运动的协同运动规律。运用多目运动捕捉系统,以行走、奔跑、跳跃、转弯、坐下-站起等常见下肢运动作为基本运动模式,以水平路面、横坡、纵坡、障碍物、楼梯等路况作为环境约束,建立了涵盖36种下肢运动模式的关节运动数据集。通过对每个运动模式的运动数据进行特征提取与特征分析发现,下肢关节间存在着协同运动关系,每个运动模式中单侧下肢的6个关节运动都可以分解为6个基本的协同运动模块(称为运动基元)。进一步研究发现,少量运动基元(1到3个)的线性组合就可以有效重构每个运动模式中的下肢关节运动(重构精度>90%)。针对不同运动模式间下肢运动基元是否相似的问题,提出了运动模式间运动基元相似性的综合评价指标,定量分析了不同运动模式两两之间的相似性与差异性。在相似性评价的基础上,运用聚类分析方法,研究了下肢运动模式的自然分组结构,发现了采用相似运动基元的运动模式及其构成的五个运动模式组。通过组间对比分析,发现了不同运动模式组间协同运动特征的相似性与差异性。研究结果表明人体可以通过在不同运动模式中采用相似的协同运动策略以实现运动控制的简化。上述运动模式组内与组间协同运动特征的相似性表明,所有下肢运动模式中可能蕴含着某些共性协同运动特征。针对这一问题,突破上述在单一或部分运动模式中进行局部观察的局限性,提出了全局视角下的肢体运动特征解析思路,建立了所有运动模式的共性协同运动特征的提取方法,发现三个共性运动基元的线性组合可以重构人体下肢在行走、奔跑、跳跃和坐下-站起中的关节运动(重构精度>90%)。相关研究发现为设计多功能通用型的下肢外骨骼提供了理论依据。基于上述发现的下肢协同运动规律,建立了拟人运动下肢外骨骼机器人的设计方法。根据运动解析结果,结合关节运动相关性的聚类分析,进一步建立了下肢关节分组耦合的运动基元提取方法。在此基础上,研究了下肢运动基元的分解与合成方法,提出了运动基元的机械复现原理,建立了具备拟人运动生成能力的下肢外骨骼机构设计方法。基于运动基元的等效运动变换,建立了运动复现机构的简化设计方法,实现了外骨骼的紧凑与轻量化设计。外骨骼样机的测试实验验证了提出的外骨骼设计方法的有效性。更多还原

【Abstract】 The excellent lower limb motor ability of humans provides an important guarantee for their healthy life and work.Exploring the mystery of human locomotion and then repairing or enhancing the lower limb motor ability with engineering science methods have become the research hotspot in the field of wearable lower limb exoskeleton robotics.Focusing on the challenging issue of how to design a lower limb exoskeleton with the motor ability to match the human natural locomotion and the capacity to assist human mobility,this dissertation carries out a systematic and in-depth study on the exploration of the movement mechanism of the lower limb and the mechanical reproduction of the lower limb natural movements.The main innovative researches of the dissertation are as follows:To uncover the unclear generation mechanism of the lower limb natural movements,we investigate and reveal the kinematic synergy underlying the lower limb natural movements.By using a motion capture system,a joint motion dataset consisting of thirtysix lower limb movement patterns is built,which takes into account five common lower limb movement patterns(i.e.,walking,running,hopping,turning,and sitting-downstanding-up)and five environment constraints(i.e.,level ground,cross slopes,longitudinal slopes,obstacles,and stairs).By the feature extraction and feature analysis of the motion data of each pattern,it can be found that there exist synergistic movement relationships among the lower limb joints,and the six joint motions of one lower limb can be decomposed into six basic motor modules(called motor primitives)in each pattern.Furthermore,it can be found that the linear combination of a small number of motor primitives(one to three)can effectively reconstruct the lower limb joint motions in each pattern(reconstruction accuracy > 90%).To address the problem of whether the motor primitives are similar among the movement patterns,a comprehensive similarity index of the motor primitives is proposed to measure the similarities and differences among the movement patterns.Based on the similarity measurements,the natural grouping structure of the movement patterns is explored using the cluster analysis method.The result reveals the movement patterns that adopt similar motor primitives and five different movement pattern groups.By comparison,the similarities and differences in the synergistic movement features are further found among the different movement pattern groups.Altogether,the findings suggest that humans can simplify the motor control by adopting similar synergistic movement strategies in various movement patterns.As suggested by the similarities of the synergistic movement features within and among the above movement pattern groups,there may be some common synergistic movement features underlying all the movement patterns.To address this problem and break through the limitation of local observation in single or partial movement patterns,a new approach for analyzing the lower limb movements from a holistic viewpoint is proposed.A method for extracting the common synergistic movement features of all the movement patterns is presented,and it can be found that three common motor primitives can reconstruct the lower limb joint motions during walking,running,hopping,and sittingdown-standing-up(reconstruction accuracy > 90%).The related findings provide a theoretical basis for the design of a multifunctional and versatile lower limb exoskeleton.Based on the above findings about the kinematic synergy of the lower limb,a design method for a lower limb exoskeleton with anthropomorphic movements is presented.According to the movement analysis results and combining with the cluster analysis of joint motion correlations,an extraction method of the motor primitives with the grouped coupling of the lower limb joints is proposed.Based on this,the method of the decomposition and synthesis of the motor primitives is studied.The mechanical reproduction principle of the motor primitives is proposed,and the mechanism design method of the exoskeleton with the ability to generate anthropomorphic movements is then built.Based on the equivalent movement transformation of the motor primitives,a design method of the movement reproduction mechanism that ensures the compactness and lightweight of the exoskeleton is proposed.The tests on an exoskeleton prototype verify the effectiveness of the proposed exoskeleton design method.更多还原