CAREER: Transparent Robot-Aided Rehabilitation (TRAIN): Robot-Aided Rehabilitation with Refined Characterization of Altered Biomechanics & Enhanced Physical Human-Robot Interaction

中风是美国长期残疾的主要原因，影响约650万美国人。随着中风死亡率的下降和老龄化人口的增加，中风后需要康复的人数预计将增加，这就迫切需要提高中风康复服务的有效性。为了应对这一挑战，这个教师早期职业发展计划（CAREER）项目引入了一个新的机器人辅助康复框架，即透明机器人辅助康复（TRAIN）。该框架在两个不同方面建立在“增强的透明度”之上：（1）在理解中风后改变的生物力学方面的透明度和（2）在物理人机交互中的透明度，即，机器人的概念是在运动任务期间人类在物理上感觉不到。该项目的重点是开发肩关节康复训练框架，并测试中风幸存者的有效性。肩关节功能障碍是中风后最常见的并发症之一，但机器人辅助肩关节康复尚未得到充分探讨。TRAIN框架在机器人肩关节运动治疗中的成功应用将直接受益于上肢的整体运动功能，包括改善的运动范围、力量和稳定性。此外，成功的应用将带来改善生活质量的次要好处，例如减少运动任务期间的疲劳和提高日常活动的独立性。该项目的研究活动和成果将无缝集成到各种教育和推广计划中，以激发和吸引不同的学生群体，激励他们从事STEM职业，并培养机器人和人体运动科学方面的下一代科学家和工程师。一个独特的“按需外展”计划将促进代表性不足的少数民族学生的外展机会。调查员的长期研究目标是通过机器人设计，控制器设计和神经肌肉系统的精确量化（系统识别）的综合创新来推进机器人辅助康复。为了实现这一目标，该项目将使用机器人设计（轻量级，并行驱动的肩部外骨骼机器人），系统识别（3D肩部阻抗的精确量化）和控制器设计（基于生物力学的主动阻抗控制器）的集成创新来产生肩部康复的变革框架。集成一种新型的轻量级，并行驱动的肩部外骨骼机器人，最大限度地减少对自然手臂动力学的影响，结合快速和强大的系统识别算法，将促进对中风导致的脑损伤如何改变3D肩部阻抗的理解。 研究计划分为四个方面。 第一个目标是开发一种5自由度的轻型、并联、驱动肩外骨骼机器人。 该机器人将由一个全驱动/电机驱动的3-DOF球面并联机器人（SPM）和一个2-DOF被动滑动接口。 SPM由三个并联的致动器组成，其连接到与用户耦合的肩部;滑动接口是放置在用户的上臂上的袖带，并且也耦合到肩部。 致动器被耦合以提供对球形工作空间的访问。 将确定机器人的最佳配置，并评估其应用精确扰动和模拟肩关节处大范围阻抗的能力。 第二次推力用于量化动态运动任务期间的3D肩部阻抗。 使用优化的机器人，一个强大的系统识别算法，以及开发量化的3D肩阻抗（刚度，阻尼和惯性）在手臂运动的方向在年轻的受试者没有神经肌肉疾病的历史。 机器人和开发的算法将被用于确定如何肩阻抗在正常的肩功能是改变中风患者慢性轻偏瘫。第三个重点是开发一种基于生物力学的主动阻抗控制器。主动阻抗控制器将通过响应于用户的运动意图而改变来自机器人的阻尼电阻来进一步增强透明度。例如，在一个实施例中，降低阻尼以减少当用户打算在一个方向上移动时不希望的阻力的体验。 然后将在Thrust 2卒中患者中评估控制器的有效性，即其通过降低肌肉活动/努力来增加透明度的能力。 第四个重点是根据TRAIN框架开发和评估患者特异性的适应性运动疗法。 基于参与早期推力的每例卒中患者的肩部阻抗改变的量化，将开发旨在将肩部阻抗改变纠正至未受损基线（根据年龄匹配的未受影响对照确定）的机器人运动疗法。 将设计一个为期6周（12次）的患者特定机器人训练计划，提供一套独特的强化和伸展练习，以根据对患者运动性能的评估调整机器人阻抗。最后，我们将在设计的患者中对运动计划进行评估，并在训练后和3个月的随访中评估其对肩关节运动功能改善的影响。该奖项反映了NSF的法定使命，并通过使用基金会的智力价值和更广泛的影响审查标准进行评估，被认为值得支持。

Stroke is a leading cause of long-term disabilities in the United States, affecting about 6.5 million Americans. With a decreasing stroke mortality and increase in the aging population, the number of people requiring rehabilitation after a stroke is projected to increase, creating a critical need to improve the effectiveness of stroke rehabilitation services. To address this challenge, this Faculty Early Career Development Program (CAREER) project introduces a new robot-aided rehabilitation framework, namely Transparent Robot-Aided Rehabilitation (TRAIN). This framework builds upon "enhanced transparency" in two distinct aspects: (1) transparency in terms of understanding altered biomechanics following stroke and (2) transparency in physical human-robot interaction, i.e., the concept of a robot being physically imperceptible to a human during motor tasks. The project focuses on developing the TRAIN framework for shoulder rehabilitation and testing the effectiveness for stroke survivors. Shoulder dysfunction is one of the most common complications following stroke, but robot-aided shoulder rehabilitation has not yet been fully explored. Successful application of the TRAIN framework to robotic shoulder exercise therapy will directly benefit the overall motor function of the upper extremity including improved range of motion, strength, and stability. In addition, successful application will lead to secondary benefits of improving the quality of life, such as reduced fatigue during motor tasks and improved independence in daily activities. Research activities and outcomes of this project will be seamlessly integrated into various education and outreach programs in order to excite and attract a diverse group of students, inspire them to pursue careers in STEM, and train next-generation scientists and engineers in robotics and human movement science. A unique "Outreach on Demand" program will promote outreach opportunities for underrepresented minority students.The investigator's long-term research goal is to advance robot-aided rehabilitation through integrated innovations in robot design, controller design, and refined quantification (system identification) of the neuromuscular system. Toward this goal, this project will produce a transformative framework for shoulder rehabilitation using integrated innovations in robot design (a lightweight, parallel-actuated shoulder exoskeleton robot), system identification (refined quantification of 3D shoulder impedance), and controller design (a biomechanics-based active impedance controller). Integration of a novel lightweight, parallel-actuated shoulder exoskeleton robot that minimally impacts natural arm dynamics, in combination with a fast and robust system identification algorithm, will advance understanding of how brain injury due to stroke alters 3D shoulder impedance. The Research Plan is organized under 4 thrusts. The FIRST THRUST is to develop a 5-DOF lightweight, parallel, actuated shoulder exoskeleton Robot. The robot will consist of a fully-actuated/motor driven 3-DOF spherical parallel manipulator (SPM) and a 2-DOF passive slip interface. The SPM consists of three parallel actuators connected to a shoulder piece coupled to the user; the slip interface is a cuff placed on the user's upper arm and is also coupled to the shoulder piece. The actuators are coupled to provide access to a spherical workspace. The optimal configuration of the robot will be determined and its ability to apply precise perturbations and simulate a wide range of impedances at the shoulder joint will be evaluated. The SECOND THRUST is to quantify 3D shoulder impedance during dynamic motor Tasks. Using the optimized robot, a robust system identification algorithm well be developed to quantify 3D shoulder impedances (stiffness, damping and inertia) in the direction of arm movements in young subjects with no history of neuromuscular disorders. The robot and algorithms developed will then be used to determine how shoulder impedance during normal shoulder functions is altered in stroke patients with chronic hemiparesis. The THIRD THRUST is to develop a biomechanics-based active impedance controller. An active impedance controller will further enhance transparency by altering the damping resistance from the robot in response to the user's intent of motion, e. g., to lower damping to reduce the experience of undesired resistance when the user intends to move in a direction. The effectiveness of the controller, i.e. its ability to increase transparency by lowering muscle activity/effort, will then be assessed in the Thrust 2 stroke patients. The FOURTH THRUST is to develop and evaluate patient-specific, adaptive exercise therapy based on the TRAIN Framework. Based on quantification of altered shoulder impedance in each stroke patient participating in earlier Thrusts, robotic exercise therapy that aims to correct altered shoulder impedance towards the unimpaired baseline (as determined from age-matched unaffected controls) will be developed. A 6-week (12 sessions) patient specific robotic training program that provides a unique set of strengthening and stretching exercises will be designed to adjust robotic impedance based on assessment of the patient's motor performance. Finally, the exercise program will be evaluated in the patients for whom they were designed and the effects on improvement of shoulder motor function will be assessed post training and at a 3-month follow-up.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.