HCC: Medium: Collaborative Research: Neural Control of Powered Artificial Legs

动力下肢（LL）假肢机电一体化方面的最新突破有望为越来越多的下肢截肢者提供广泛的功能恢复（例如，坐在椅子上站起来，爬楼梯，甚至跑步）。pi认为，为了实现这一潜力，有必要为人造腿提供神经控制。现有的上肢神经控制方法不适用于此目的，因为上肢神经控制机制与下肢神经控制机制存在显著差异。特别是，大多数涉及下肢的活动都需要非自愿（脊髓）和自愿（脊髓上）神经控制，具有高动态性，需要多关节协调和控制不稳定运动，这些特点使得LL假体的神经控制设计规范比UL假体要求高得多。在这个项目中，pi将通过开发一种创新的动力人造腿神经控制系统来解决这一挑战，该系统可以识别和利用多尺度用户意图（例如，一般运动命令，如预期任务与详细运动命令，如预期关节运动）来调节多个LL假肢关节的内在（自主）控制，以实现运动和非运动任务的表现。目标是支持人类运动神经控制的逆向工程，同时创建创新的神经机器接口（NMI）技术，使用户能够以自然，自适应和灵活的方式控制LL假体的动态。受目前已知的人类运动控制系统的神经组织和功能的启发，pi的方法是设计一种基于无创头皮脑电图（EEG）和表面肌电图（EMG）结合的新型NMI。假设低水平外围神经和高水平中枢神经控制源的融合可以实现多尺度用户意图识别，并且比单独使用脑电图或肌电图实现更高的准确性和更快的响应时间。动力LL假体的分层控制方案，其中由NMI识别的多尺度用户意图调节内在（自主）控制，将支持动态，多关节协调运动中直观有效的假体使用，同时显着减少运动中假体用户的精神负担，因为循环运动是自主实现的（这是需要的，因为我们很少考虑行走时膝盖和脚踝的控制）。pi还将探索脑电图和肌电图信号之间的相关性，这可能会为神经适应和皮层控制在步态开始和产生过程中的时间过程提供见解，包括大脑如何开始行走和调节运动输出，以预测关键事件，如着陆时的脚位置或上下踏步，重量接受和推入摇摆阶段。最后，pi将使用转化研究来验证他们在经股截肢患者（一个高且具有挑战性的截肢水平）中的新方法。更广泛的影响：pi的长期目标是开发真正的仿生假肢，让它们感觉和工作起来就像真腿一样。他们在这个项目中的方法代表了下肢可穿戴假肢控制的范式转变。因此，项目成果将直接影响人机交互和脑机接口研究社区。这些发现也将与神经科学和康复界相关，因为它们将有助于阐明适应性脊髓和皮层对人类运动的贡献，同时提供创新和功能性神经假肢解决方案，以改善下肢截肢者的生活。

Recent breakthroughs in the mechatronics of powered lower limb (LL) prostheses hold the promise of enabling restoration for the large and growing population of lower limb amputees of a broad spectrum of functionality (e.g., standing up when seated in a chair, climbing stairs, and even running). The PIs argue that to realize this potential it is essential to provide neural control of artificial legs. The application of existing upper limb (UL) neural control approaches is inappropriate to this end, because the UL and LL neural control mechanisms are significantly different. In particular, most activities involving the lower limbs recruit both involuntary (spinal cord) and voluntary (supra-spinal) neural control, present high dynamics, and require multi-joint coordination and control of unstable locomotion, characteristics which combine to make the design specifications for neural control of LL prostheses much more demanding than those for UL devices. In this project the PIs will address this challenge by developing an innovative neural control system for powered artificial legs that can recognize and exploit multi-scale user intent (e.g., general motor commands such as intended task vs. detailed motor commands such as intended joint motion) to modulate intrinsic (autonomous) control of multiple LL prosthetic joints for locomotor and nonlocomotor task performance. The goals are to support reverse-engineering of the neural control of human locomotion while creating innovative neural-machine interfacing (NMI) technology that enables users to control the dynamics of LL prostheses in a natural, adaptive and flexible way. Inspired by what is currently known about the neurological organization and function of the human motor control system, the PIs' approach is to design a novel NMI based on a combination of noninvasive scalp electroencephalography (EEG) and surface electromyography (EMG). The hypothesis is that fusion of low-level peripheral and high-level central neural control sources can achieve multi-scale user intent recognition with higher accuracy and more rapid response time than can be realized with either EEG or EMG alone. A hierarchical control scheme for powered LL prostheses, in which multi-scale user intent identified by the NMI modulates intrinsic (autonomous) control, will support intuitive and efficient prosthesis use in dynamic, multi-joint coordinated movements while significantly reducing the mental burden of the prosthesis user in locomotion because the cyclic motion is achieved autonomously (this is desired because we rarely think about knee and ankle control when walking). The PIs will also explore correlation across EEG and EMG signals, which may provide insight into neural adaption and the time course of cortical control during the initiation and generation of gait, including how the brain initiates walking and regulates motor output in anticipation of key events such as foot placement at landing or during stepping up and down, weight acceptance, and push-off into swing phase. Finally, the PIs will use translational research to validate their novel approach in patients with trans-femoral amputations (a high and challenging amputation level).Broader Impacts: The PIs' long-term objective is to develop true bionic prostheses that feel and work just like real legs. Their approach in this project represents a paradigm shift in the control of lower limb wearable prosthetics. As such, project outcomes will directly impact both the Human-Robot Interaction and Brain-Machine Interface research communities. The findings will also be relevant to the neuroscience and rehabilitation communities, in that they will help elucidate the adaptive spinal cord and cortical contributions to human locomotion, while providing innovative and functional neuro-prosthetics solutions to improve the lives of lower limb amputees.