CRII: CHS: Healthcare Professionals Teaching Robots Communication Strategies for Effective Intervention Delivery

在美国，医疗工作者的持续短缺。行为治疗师、护士、物理治疗师和家庭护理工作者等职业是目前和未来最需要的医疗保健专业。社交辅助机器人是一种新的战略技术，有可能通过扩大现有劳动力向有需要的个人提供医疗保健干预的能力来缓解医疗保健行业的劳工问题。然而，由于医疗保健服务消费者的需求迅速变化，通常需要适应现有的干预措施或开发新的干预措施，因此这些机器人技术仍然难以扩展，难以整合到现实世界的医疗保健干预交付中。该项目将为医疗保健专业人员构建使能技术，使其能够根据消费者的需求和临床环境生成和定制机器人介导的干预措施。开发的技术还将使医疗保健专业人员能够教机器人沟通策略，以便有效地进行干预，这样社会辅助机器人就可以提供与人类教师同等水平的干预服务。因此，这项工作将通过减轻医疗保健行业现有的劳动力问题，使医疗保健服务更容易获得，并通过与社交辅助机器人的协作，为消费者定制，从而造福社会。这项研究的成果将促进有效的人机交流、机器人学习和人机教学界面的知识。为了实现这些目标，该项目将开发一个从示范中学习的启发框架，使没有编程经验的医疗保健专业人员能够教授社交辅助机器人管理干预和非语言沟通策略，以有效地进行干预。(a)教学界面：研究人员将设计并构建一个社交沉浸式教学界面，使医疗保健专业人员能够远程操作社交辅助机器人，以演示干预交付任务和干预期间有效的非语言沟通策略。主要重点将是确定远程操作模式，以最有效和高效地从医疗保健专业人员获得高质量的演示。(b)从示范中学习干预措施：将设计算法，以便从保健专业人员的示范模型中学习干预交付任务的结构、他们在这些干预期间使用的有效非语言沟通行为及其使用的背景。然后，社交辅助机器人将利用这些学习到的模型为医疗服务消费者执行学习到的干预措施。(c)干预期间的非语言交流：该项目最终将进行人机交互研究，重点是在现实世界的干预交付场景中整合和评估机器人介导的干预措施，这些干预措施利用类似人类的非语言交流策略。这项研究的成果将促进有效的人机交流、机器人学习和人机教学界面的知识。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。

There is a persistent shortage of healthcare workers in the USA. Occupations such as behavioral therapists, nurses, physical therapists, and home care workers are among the most in demand present and future healthcare professions. Socially assistive robots are a new strategic technology that has the potential to alleviate labor concerns in the healthcare industry by amplifying the capabilities of our existing labor force in delivering healthcare interventions to individuals in need. However, these robotic technologies remain difficult to scale and challenging to integrate in real-world healthcare intervention delivery due to the rapidly changing needs of healthcare service consumers that often require adaptation of existing interventions or development of new interventions. This project will build enabling technologies for healthcare professionals to generate and customize robot-mediated interventions according to the needs of their consumers as well as their clinical settings. The developed technology will also enable healthcare professionals to teach robots communication strategies for effective intervention delivery so that socially assistive robots can provide on par intervention services as their human teachers. Hence, this work will benefit society by alleviating the existing labor concerns for the healthcare industry and make healthcare services more accessible as well as customizable to consumers via collaboration with socially assistive robots. The outcomes of this research will advance knowledge in effective human-robot communication, robot learning, and human-robot teaching interfaces.To address these goals, this project will develop a Learning from Demonstration inspired framework that will enable healthcare professionals with no prior experience in programming to teach socially assistive robots to administer intervention and nonverbal communication strategies necessary for effective intervention delivery. The following fundamental research challenges will be addressed: (a)Teaching Interface: The researchers will design and build a socially immersive teaching interface which enables healthcare professionals to teleoperate a socially assistive robot to demonstrate intervention delivery tasks and effective nonverbal communication strategies during the intervention. The primary focus will be identifying teleoperation modalities that are most effective and efficient for attaining high-quality demonstrations from healthcare professionals. (b) Learning Interventions from Demonstrations: Algorithms will be designed to learn from demonstrations by healthcare professionals’ models of the structure of an intervention delivery task, effective nonverbal communication behaviors they use during these interventions, and the context of their use. These learned models will then be utilized by a socially assistive robot to carry out the learned interventions for healthcare service consumers.(c) Nonverbal Communication during Interventions: The project will culminate in a human-robot interaction study focusing on integrating and evaluating robot-mediated interventions, which utilize human-like nonverbal communication strategies, in real-world intervention delivery scenarios.The outcomes of this research will advance knowledge in effective human-robot communication, robot learning, and human-robot teaching interfaces.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.