Curriculum Design in Robotic Exoskeleton-mediated Motor Training

这个名为Mind， Machine, and Motor Nexus （M3X）的项目计划使用一种可穿戴的上肢机器人外骨骼来解决如何最好地训练人们执行新的运动任务的开放性问题。将追求两个目标。首先，项目团队将在一系列人体实验中使用机器人外骨骼，以确定可以组合起来学习动态挑战运动技能的“源任务”。其次，团队将使用确定的源任务来探索“基于课程”的任务培训的有效性，该培训旨在通过增加挑战的子任务的进展来高效地建立技能。要学习的实验任务需要一个高度的手动协调操作一个具有挑战性的内部连续体动力学的手持工具。这项研究涉及多个学科，包括机械工程、计算机科学、神经科学、医学和教育。该项目将通过开发基础知识来支持机器人介导的运动技能训练指导系统，从而促进科学进步和国民健康。教育和推广活动吸引了来自研究中代表性不足群体的本科生的参与。该项目将追求两个目标，以推进机器人介导的运动技能训练的最新技术。第一个目标是，项目团队将使用可穿戴的上肢机器人外骨骼和沉浸式虚拟现实系统，在人们执行具有挑战性的动态任务时，分析工具操作过程中的运动运动学、动力学和表面肌电图活动。这一努力的结果将是一组确定的“源任务”，这些任务将构成运动技能训练“课程”的基石，该课程将在项目的第二个目标中开发和测试。为了实现这一目标，额外的人类受试者研究将评估各种课程（即源任务的组合、源任务顺序和训练任务过渡时间）的有效性，这些课程将使用强化学习代理开发，该代理将在训练期间对源任务进行排序。该研究探索了通过与机器人外骨骼的相互作用来塑造和指导人类运动功能的机制。这项工作的主要成果将是一个机器人介导的运动技能训练指导系统，它具有在机器人辅助手术、运动和身体康复等各种场景中改善人类运动表现的潜力。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 Mind, Machine, and Motor Nexus (M3X) project plans to use a wearable upper extremity robotic exoskeleton to address the open question of how best to train people to perform new motor tasks. Two objectives will be pursued. First, the project team will use the robotic exoskeleton in a series of human subject experiments to identify "source tasks" that can be combined to learn a dynamically challenging motor skill. Second, the team will use the identified source tasks to explore the efficacy of "curriculum-based" task training that seeks to build skills efficiently and effectively through a progression of sub-tasks of increasing challenge. The experimental task to be learned requires a high degree of manual coordination in the manipulation of a hand-held tool with challenging internal continuum dynamics. This research involves several disciplines including mechanical engineering, computer science, neuroscience, medical sciences, and education. The project will promote the progress of science and advance the national health by developing fundamental knowledge in support of a robot-mediated coaching system for motor skills training. Education and outreach activities engage the participation of undergraduate students from underrepresented groups in the research.This project will pursue two objectives to advance the state-of-the-art in robot-mediated coaching for motor skills training. For the first objective, the project team will use a wearable upper extremity robotic exoskeleton and an immersive Virtual Reality system to analyze movement kinematics, kinetics, and surface electromyographic activity during tool manipulation as people perform a challenging dynamic task. The result of this effort will be an identified set of "source tasks" that will form the building blocks of a motor skill training "curriculum" to be developed and tested in the project's second objective. For that objective, additional human subject studies will evaluate the efficacy of a variety of curricula (i.e., combinations of source tasks, source task orders, and training task transition times) to be developed using a Reinforcement Learning agent that will sequence the source tasks during training. The research explores the mechanisms by which human motor function can be shaped and directed by interactions with robotic exoskeletons. The major outcome of this work will be a robot-mediated coaching system for motor skills training, which holds potential to improve human motor performance in a variety of scenarios ranging from robot-assisted surgery, sports, and physical rehabilitation.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.