Collaborative Research: HEBB: Human-Robot Enabled System to Induce Brain Behavior Adaptations

合作研究：HEBB：诱导大脑行为适应的人机驱动系统

• 批准号: 1935500
• 负责人: Laurel Riek
• 金额: $ 45万
• 依托单位: University of California-San Diego
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2019
• 资助国家: 美国
• 起止时间: 2019-09-01 至 2023-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1935500&HistoricalAwards=false
• 关键词: Collaborative; Research; HEBB; Human; Robot

The overall research objective of this collaborative project is to create an embodied, intelligent robotic system that can induce meaningful long-term change in human motor function by providing personalized, adaptive feedback and noninvasive neural stimulation designed to induce desirable neuromotor plasticity. The motor behavior targeted for enhancement is plantarflexor power during the push-off phase of gait; stroke survivors often produce diminished plantarflexor power and rely instead on an inappropriate hip flexion "pull-off" compensation, thereby limiting the quality of their gait and quality of life. Personalized learning methods will be employed to model and optimize behavioral responses to changes in performance feedback provided by an intelligent mobile robotic coach, which will guide gait training. The project will lay the foundation to determine whether training based solely on principles of motor learning suffice to induce meaningful increases in plantarflexor power that are retained over time, or whether simultaneous targeted changes in brain excitability are required. This project advances the NSF mission to promote the progress of science and advance the national health by developing an adaptive motor learning algorithm embedded within an interactive mobile robot to induce meaningful long-term changes in human motor function through human-robot interaction. Broader impacts of the project include efforts to enhance research reproducibility and rigor, and to broaden participation in STEM for women, minorities, and persons with disabilities. The overall objective of this research is to create an embodied, intelligent system that provides personalized, adaptive feedback to induce neuromotor plasticity, mediate motor adaptation, and promote meaningful, lasting increases in plantarflexor power, which is diminished during walking in many stroke survivors. Three sets of human subject experiments are researched. The first will identify critical parameters of performance feedback that facilitate the desired behavioral change. The second will use a novel learning paradigm to model and optimize behavioral responses to changes in performance feedback provided by an intelligent robotic coach. The third will use single-pulse transcranial magnetic stimulation (TMS) and paired associative stimulation (PAS) to harness neuroplastic effects in humans such that desired behavioral changes induced by optimized feedback training are made persistent through Hebbian learning mechanisms. The envisioned system will involve bi-directional learning between the human and machine intelligences to determine how to control important, but subject-specific, variables critical for maintaining and promoting motor function across the life and health span. Understanding these bi-directional relationships within the context of neurorehabilitation may provide insights that can further advance human-robot teaming in a range of application domains, including healthcare, manufacturing, and personal transportation.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.

这个合作项目的总体研究目标是创建一个具体的，智能的机器人系统，可以通过提供个性化的，自适应的反馈和非侵入性的神经刺激，旨在诱导理想的神经运动可塑性，从而诱导人类运动功能的有意义的长期变化。针对增强的运动行为是在步态的蹬离阶段期间的跖屈肌功率;中风幸存者通常产生减少的跖屈肌功率，而是依赖于不适当的髋关节屈曲“蹬离”补偿，从而限制了他们的步态质量和生活质量。个性化的学习方法将被用来模拟和优化行为反应的变化提供的性能反馈的智能移动的机器人教练，这将指导步态训练。该项目将奠定基础，以确定仅基于运动学习原则的训练是否足以诱导随着时间的推移而保留的跖屈肌力量的有意义的增加，或者是否需要同时对大脑兴奋性进行有针对性的改变。该项目通过开发嵌入交互式移动的机器人中的自适应运动学习算法，通过人机交互诱导人类运动功能的有意义的长期变化，推进NSF的使命，以促进科学进步和促进国民健康。该项目更广泛的影响包括努力提高研究的可重复性和严谨性，并扩大妇女、少数民族和残疾人对STEM的参与。本研究的总体目标是创建一个具体的智能系统，提供个性化的自适应反馈，以诱导神经运动可塑性，介导运动适应，并促进跖屈肌力量的有意义的，持久的增加，这是在许多中风幸存者步行过程中减少。研究了三组人体实验。第一个将确定促进所需行为改变的绩效反馈的关键参数。第二个将使用一种新的学习范式来模拟和优化智能机器人教练提供的性能反馈变化的行为反应。第三项研究将使用单脉冲经颅磁刺激（TMS）和配对联想刺激（PAS）来利用人类的神经可塑性效应，从而通过赫布学习机制使优化反馈训练引起的预期行为变化持续下去。设想的系统将涉及人类和机器智能之间的双向学习，以确定如何控制重要但特定于主题的变量，这些变量对于维持和促进整个生命和健康期的运动功能至关重要。理解神经康复背景下的这些双向关系可能会提供进一步推进人类-机器人合作的见解，这些合作可以在一系列应用领域中实现，包括医疗保健、制造业和个人交通。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。

项目成果

Multitask Bandit Learning Through Heterogeneous Feedback Aggregation

通过异构反馈聚合进行多任务强盗学习

• 发表时间: 2021
• 期刊: Proceedings of the 24th International Conference on Artificial Intelligence and Statistics (AISTATS
• 作者: Wang, Z.;Chaudhuri, K.
• 通讯作者: Chaudhuri, K.

Stochastic Multi-Player Bandit Learning from Player-Dependent Feedback

随机多人强盗从玩家相关的反馈中学习

• 发表时间: 2020
• 期刊: ICML Workshop on Real World Experiment Design and Active Learning
• 作者: Wang, Z.;Singh, M.K.;Zhang, C.;Riek, L.D.;Chaudhuri, K.
• 通讯作者: Chaudhuri, K.

Methods for Robot Behavior Adaptation for Cognitive Neurorehabilitation

• DOI: 10.1146/annurev-control-042920-093225
• 发表时间: 2021-09
• 期刊: Annu. Rev. Control. Robotics Auton. Syst.
• 作者: A. Kubota;L. Riek

Facial Expression Modeling and Synthesis for Patient Simulator Systems: Past, Present, and Future

• DOI: 10.1145/3483598
• 发表时间: 2022-03
• 期刊: ACM Transactions on Computing for Healthcare (HEALTH)
• 作者: Maryam Pourebadi;L. Riek

A Robot-based Gait Training System for Post-Stroke Rehabilitation

基于机器人的中风后康复步态训练系统

• DOI: 10.1145/3434074.3447212
• 发表时间: 2021
• 期刊: HRI '21 Companion: Companion of the 2021 ACM/IEEE International Conference on Human-Robot Interaction
• 作者: Banh, Sharon;Zheng, Emily;Kubota, Alyssa;Riek, Laurel D.
• 通讯作者: Riek, Laurel D.