CAREER: Modeling Human Gait to Optimize Exoskeleton Control and Understand How the Goal Changes across Walking Tasks

机器人腿外骨骼是一种可穿戴设备，可以作为放大器来增强、加强或恢复人类的表现，它可以通过帮助用户更正常地行走来改善行走困难的人的生活质量。因为人类在行走时经常改变速度，所以了解这些速度变化是如何发生的，以及外骨骼在稳定和变速行走时提供适当的帮助是很重要的。因此，这个CAREER项目的目的是确定人类的目标，即人类对最佳步行方式的潜意识想法，对于健康和中风后的个体来说，匀速行走、变速行走和外骨骼辅助行走是否相同。这将通过开发一个基于物理的人类行走预测计算机模型来实现，该模型包括推导一个描述人类行走潜意识目标的数学函数。研究旨在为人类步态控制提供基本见解，并提供设计最佳外骨骼控制器的工具。综合教育计划将利用外骨骼的吸引力，制作专业视频，突出工程学如何帮助改善行走障碍，从而使工程学人性化。一个以外骨骼为基础的项目将被开发出来，以提高本科工程师对系统建模的能力，这是许多学生都难以做到的。首席研究员的首要研究目标是研究人们是如何行走的，并利用这些知识来改善行走困难患者的康复技术。为了实现这一目标，这个CAREER项目将确定人类对于健康和中风后个体的匀速行走、变速行走和外骨骼辅助行走的目标是否相同，这将为人类步态控制提供基本见解，并为设计最佳外骨骼控制器提供工具。新的基于物理的预测模型（矢状六连杆模型，旋转的髋关节、膝关节和踝关节连接大腿、小腿和脚）将被开发出来，可以解释人与设备交互的高度非线性和非直觉性，并创建正确解释这种交互的控制器。开发的模型将更好地预测人类的步态，包括速度之间的转换，并量化人们如何改变步行速度。模型开发需要确定一个目标函数，以数学方式描述人类行走的潜意识目标。最初，目标将基于人类行走时能量消耗最小化的假设。如果最小努力与选择的步态没有很好地关联，将考虑将跌倒风险纳入函数。研究的参与者包括健康的年轻人（30人）、健康的老年人（12人）、行动缓慢的老年人（12人）和中风后的老年人（12人），他们行走困难，但能够在没有辅助设备的情况下行走并能遵循指示。本研究计划分为四个部分：1)首次量化年轻人、老年人和中风后成年人的速度转换的时空（步长、持续时间和速度）和运动学特性，并以一种新颖的方式结合机器人控制的几种方法为模型生成类似人类的速度转换；2)确定相同的目标函数是否可以预测健康的人类关节运动学稳定，变速和外骨骼辅助行走，为人类对外骨骼辅助的反应提供新的见解；3)确定高龄和中风如何改变用于预测行走的目标函数，为年龄和中风如何影响行走优先级提供新的见解；4)创建一种在模拟中设计外骨骼控制器的方法，该方法考虑了非线性人与设备的相互作用，并在没有额外调整的情况下产生所需的人类步态。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

Robotic leg exoskeletons, wearable devices that act as amplifiers to enhance, reinforce or restore human performance, could improve the quality of life for individuals who have trouble walking by helping the user walk more normally. Because humans frequently change speeds while walking, it is important to understand how these speed changes occur and for the exoskeleton to provide appropriate assistance during both steady and variable speed walking. Thus, the objective of this CAREER project is to determine if the human goal, i.e., the human’s subconscious thoughts about the best way to walk, is the same for constant speed walking, changing speeds, and exoskeleton assisted walking for both healthy and post-stroke individuals. This will be accomplished by developing a physics-based predictive computer model of human walking that includes deriving a mathematical function that describes the subconscious goal of human walking. Studies are designed to provide fundamental insight into human gait control and to provide a tool to design optimal exoskeleton controllers. The integrated educational plan will use the appeal of exoskeletons to humanize engineering by producing professional videos highlighting how engineering can help improve impaired walking. An exoskeleton-based project will be developed to improve undergraduate engineers' ability to model a system, something with which many students struggle. The principal investigator's overarching research goal is to study how people walk and to use that knowledge to improve rehabilitation techniques for individuals who have trouble walking. Towards this goal, this CAREER project will determine if the human goal is the same for constant speed walking, changing speeds, and exoskeleton assisted walking for both healthy and post-stroke individuals, which will provide fundamental insight into human gait control and a tool to design optimal exoskeleton controllers. Novel physics-based (sagittal six-link model with revolute hip, knee, and ankle joints connecting the thighs, shanks, and feet), predictive models will be developed that can account for the highly nonlinear and non-intuitive nature of human-device interaction and create controllers that correctly account for this interaction. The models developed will better predict human gait, including transition between speeds, and quantify how people change walking speed. Model development requires determination of an objective function that mathematically describes the subconscious goal of human walking. Initially, the objective will be based on the assumption that humans minimize energetic effort while walking. If minimum effort does not correlate well with chosen gaits, incorporating fall risk into the function will be considered. Participants in the studies include healthy young adults (30), healthy elderly adults (12), slow elderly adults (12), and post-stroke elderly adults (12) who walk with difficulty but are able to walk without an assistive device and can follow directions. The Research Plan is organized under four tasks: 1) Quantify the spatial-temporal (step length, duration and speed) and kinematic properties of speed transitions for young, elderly, and post-stroke adults for the first time, and generate human-like speed transitions for the model by combining several methods from robotic control in a novel manner; 2) Determine if the same objective function can predict healthy human joint kinematics for steady, variable speed, and exoskeleton-assisted walking, providing novel insights into how humans react to exoskeleton assistance; 3) Determine how advanced age and stroke alter the objective function used to predict walking, providing novel insights into how age and stroke affect walking priorities and 4) Create a method to design exoskeleton controllers in simulation that accounts for the nonlinear human-device interactions and produces the desired human gait without additional tuning.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.