NRI: INT: Wearable Robots for the Community: Personalized Assistance using Human-in-the-loop Optimization

该项目将推进可定制的软机器人外骨骼的进展，为日常活动提供无缝协助，减轻损伤或增强正常功能。柔软的机器人外骨骼服是一种新型的功能性服装，它在给穿着者的关节和肌肉同时提供机械辅助。数百万患有神经系统行走障碍的美国人，比如近700万中风后生活的人，可以从外伤服在行走过程中提供的不显眼的帮助中受益。同样，长距离携带重物的健康人（例如，急救人员或士兵）也可以受益于使用外穿衣技术来部分减轻他们的负担。将机械辅助与佩戴者的自然节奏同步是实现这些设备潜力的关键。在这个项目中创建的软机器人外骨骼将使用穿戴式传感器监测穿戴者的行走模式，应用机器学习方法来个性化机器人辅助模式，并随着穿戴者步态的变化不断更新辅助模式。相比之下，目前的方法依赖于专家临床医生和技术人员手动调整辅助模式。这个项目将为促进国家的健康和繁荣贡献新的知识。多学科研究团队包括机器人专家、运动科学家和临床医生，他们将在实验室和临床环境中与中风后患者密切合作。最近的研究表明，通过将外骨骼控制作为一个优化问题，以直接测量代谢成本为目标，可以降低步行的代谢成本。然而，记录这样的测量需要笨重的设备，会干扰正常的呼吸，这意味着这种方法目前无法在日常环境中工作。此外，降低步行的代谢成本并不是外服对中风后患者的唯一目标，中风后患者的步态特点是缓慢运动和补偿，如髋关节徒步、绕行和跳跃。在临床人群中成功的人在循环优化将需要解决所有这些因素，而不仅仅是代谢成本。初步的基础科学探索将找到健康人群中代谢成本的有效代理，并开发卒中后人群的多目标函数。在这两种情况下，目标函数必须是不显眼的测量，准确的，并响应变化的外衣控制。第二个实施阶段将使用人在环优化来自动调整便携式系统的控制参数，以帮助健康个体和中风后患者。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。

This project will advance the progress of customizable soft robotic exosuits for seamless assistance in everyday activities, to mitigate impairment or to augment normal functionality. Soft robotic exosuits are a new class of functional clothing that apply mechanical assistance to wearers' joints in parallel with their muscles. Millions of Americans with neurologically-based walking impairments, such as the nearly seven million people living post-stroke, could benefit from the unobtrusive assistance that exosuits can provide during walking. Similarly, healthy individuals who carry heavy loads over long distances (e.g., first responders or soldiers) could benefit from using exosuit technology to partially alleviate their burden. Synchronizing mechanical assistance to the wearer's natural rhythm is essential to realizing the potential of these devices. The soft robotic exosuits created in this project will monitor the wearer's walking pattern using body-worn sensors, apply machine learning methods to personalize the robotic assistance pattern, and continuously update that assistance pattern as the wearer's gait changes. In contrast, current methods rely on expert clinicians and technicians to manually tune assistance patterns. This project will contribute new knowledge to advance the national health and prosperity. The multidisciplinary research team includes roboticists, movement scientists, and clinicians, who will work closely with persons poststroke in laboratory and clinical environments.Recent work developing human-in-the loop optimization strategies for exoskeletons suggests that it is possible to lower the metabolic cost of walking by treating the exosuit control as an optimization problem, with direct measurements of metabolic cost serving as the objective. However, recording such measurements require bulky devices that interfere with normal breathing, meaning the method cannot currently work in everyday environments. Additionally, lowering the metabolic cost of walking is not the only goal of exosuits in persons poststroke, whose gait is characterized by slow movement and compensations such as hip hiking, circumduction, and vaulting. Successful human-in-the-loop optimization in clinical populations will need to address all these factors, not just metabolic cost. An initial basic-science exploration will find effective proxies for metabolic cost in healthy populations and develop a multi-objective function in people poststroke. In both cases, the objective function must be unobtrusive to measure, accurate, and responsive to changes in exosuit control. A second implementation phase will use human-in-the-loop optimization to automatically adapt control parameters of portable systems assisting healthy individuals and people poststroke.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.